3x-ui/internal/web/service/xray.go

2032 lines
67 KiB
Go
Raw Normal View History

2023-02-09 11:18:06 -08:00
package service
import (
"encoding/json"
"errors"
"fmt"
"path"
"path/filepath"
"runtime"
"slices"
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
"strings"
2023-02-09 11:18:06 -08:00
"sync"
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
"github.com/mhsanaei/3x-ui/v3/internal/amneziawg"
"github.com/mhsanaei/3x-ui/v3/internal/amneziawgnet"
refactor: focused service files, leaf subpackages, and an internal/ layout (#5167) * refactor(service): split client.go into focused files client.go had grown to 4455 lines mixing ~10 responsibilities. Split it verbatim into cohesive same-package files (no behavior change): client.go foundation: ClientService, ClientWithAttachments, ClientCreatePayload, ErrClientNotInInbound, sqlInChunk client_locks.go inbound mutation locks, delete tombstones, compactOrphans client_lookup.go read-only lookups (GetByID, List, EffectiveFlow, ...) client_link.go inbound association sync (SyncInbound, DetachInbound, ...) client_crud.go single-client CRUD + validation + protocol defaults client_inbound_apply.go low-level inbound-settings mutators + by-email setters client_bulk.go bulk attach/detach/adjust/delete/create + DelDepleted client_traffic.go traffic-reset paths client_groups.go client group management client_paging.go paged listing, filtering, sorting, summary Every declaration moved unchanged (verified: identical func/type/const/var signature set before vs after). Imports redistributed per file via goimports. go build ./..., go vet, and go test ./web/service/... all pass. * refactor(service): split inbound.go into focused files inbound.go was 4100 lines. Split it verbatim into cohesive same-package files (no behavior change): inbound.go core inbound CRUD + InboundService (keeps pkg doc) inbound_protocol.go protocol / stream capability helpers inbound_node.go node/runtime/remote coordination + online tracking inbound_traffic.go traffic accounting, reset, client stats inbound_client_ips.go per-client IP tracking inbound_clients.go client lookups within inbounds + copy-clients inbound_disable.go auto-disable invalid inbounds/clients inbound_migration.go DB migrations inbound_sublink.go subscription link providers inbound_util.go generic slice/string helpers Identical func/type/const/var signature set before vs after; package doc comment preserved on inbound.go. Imports redistributed via goimports. Build, vet, and go test ./web/service/... all pass. * refactor(service): split tgbot.go into focused files tgbot.go was 3738 lines dominated by a 1246-line answerCallback. Split it verbatim into cohesive same-package files (no behavior change): tgbot.go lifecycle, bot setup, caches, small utils tgbot_router.go incoming update / command / callback dispatch tgbot_send.go outbound messaging primitives tgbot_client.go client views, actions, subscription links tgbot_inbound.go inbound listing / pickers tgbot_report.go server usage, exhausted, online, backups, notifications Identical func/type/const/var signature set before vs after. Imports redistributed via goimports. Build, vet, and go test ./web/service/... pass. * refactor(client): dedupe single-field by-email setters ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, and ResetClientTrafficLimitByEmail shared an identical ~50-line body that resolves the inbound by email, confirms the client exists, rewrites a single-client settings payload, and delegates to UpdateInboundClient. Extract that into applyClientFieldByEmail(inboundSvc, email, mutate) and reduce each setter to a 3-line wrapper. Behavior is unchanged: same checks and error strings, same single-client payload contract, same totalGB guard. SetClientTelegramUserID (resolves by traffic id, different error text) and ToggleClientEnableByEmail/SetClientEnableByEmail (different return shape and a pre-read of the old state) intentionally keep their own bodies. * refactor(service): extract panel/ subpackage Move the panel-administration leaf services out of the flat service package into web/service/panel/ (package panel): user.go UserService (auth / 2FA / LDAP) panel.go PanelService (restart / self-update) + version helpers panel_other.go non-unix RestartPanel panel_unix.go unix RestartPanel api_token.go ApiTokenService websocket.go WebSocketService panel_test.go version/shellQuote unit tests These are leaves: they depend on core (SettingService, Release) but no core file references them, so the extraction creates no import cycle. Core references are now qualified (service.SettingService, service.Release); callers in main.go, web/web.go, and web/controller/* updated to panel.*. Build, vet, and go test ./web/... pass. * refactor(service): extract integration/ subpackage Move the external-provider integration leaves into web/service/integration/ (package integration): warp.go WarpService (Cloudflare WARP) nord.go NordService (NordVPN) custom_geo.go CustomGeoService (custom geo asset management) *_test.go custom_geo / panel-proxy tests These depend on core (SettingService, ServerService, XraySettingService) but no core file references them. xray_setting.go stays in core because it calls the unexported SettingService.saveSetting. The shared isBlockedIP SSRF helper (used by core url_safety.go and by custom_geo) now has a small copy in each package rather than being exported. Core references qualified; callers in web/web.go, web/job/*, and web/controller/* updated to integration.*. Build, vet, and go test ./web/... pass. * refactor(service): extract tgbot/ subpackage Move the Telegram bot (6 files + test) into web/service/tgbot/ (package tgbot). It is a leaf: it embeds five core services (Inbound/Client/Setting/ Server/Xray) and the core never references it, so no import cycle. To support the package boundary without changing behavior: - core exposes XrayProcess() *xray.Process so tgbot keeps calling the exact same running-process methods it used via the package-level `p`; - three core methods tgbot calls are exported: ClientService.checkIs- EnabledByEmail -> CheckIsEnabledByEmail, InboundService.getAllEmails -> GetAllEmails (callers updated in-package); - tgbot's embedded-field types and the few core type refs (Status, ClientCreatePayload, SanitizePublicHTTPURL) are now service-qualified. Callers in main.go, web/web.go, web/job/*, and web/controller/* updated to tgbot.*. Build, vet, and go test ./web/... pass. * refactor(service): extract outbound/ subpackage OutboundService (outbound.go) imports only neutral packages (config, database, model, xray) and its production code is referenced by no core or sibling service file — only by web/controller/xray_setting.go and web/job/xray_traffic_job.go. Move it to web/service/outbound/ (package outbound); no core qualification needed inside. Callers updated to outbound.*. The one coupling was a tiny pure test helper, outboundsContainTag, used by both outbound.go and the core outbound_subscription_test.go; it now has a small copy in that test file rather than being shared across the boundary. Build, vet, and go test ./web/... pass. * refactor(util): move wireguard into its own subpackage util/wireguard.go was the lone file of the root `util` package (24 lines, one exported func GenerateWireguardKeypair), while every other util concern lives in a focused subpackage (util/common, util/crypto, util/netsafe, ...). Move it to util/wireguard/ (package wireguard) for consistency; its only importer, web/service/integration/warp.go, is updated. The root `util` package no longer exists. * refactor(sub): drop redundant sub prefix from filenames Inside package sub the subXxx.go prefix just repeats the package name (like client_*.go did inside service). Rename for consistency; content and type names are unchanged: subController.go -> controller.go subService.go -> service.go subClashService.go -> clash_service.go subJsonService.go -> json_service.go (+ matching _test.go files) * refactor(controller): rename xui.go -> spa.go XUIController serves the panel's single-page-app shell; spa.go names that role plainly (the other controller files are domain-named). File rename only — the type stays XUIController. api_docs_test.go keys route base paths by filename, so its "xui.go" case is updated to "spa.go". * refactor: move backend packages under internal/ Adopt the idiomatic Go application layout: the backend packages now live under internal/ (a boundary the toolchain enforces), signalling private implementation instead of a library-style flat root. No runtime behavior changes — only import paths and a few build/config paths move. Moved: config, database, logger, mtproto, sub, util, web, xray -> internal/. main.go stays at the repo root and tools/openapigen stays under tools/ (both still import internal/* because the internal rule keys off the module root). The module path github.com/mhsanaei/3x-ui/v3 is unchanged; 149 .go files had their import prefix rewritten to .../internal/<pkg>. Couplings the Go compiler can't see, updated to the new layout: - frontend i18n imports of web/translation (react.ts, setup.components.ts) - vite outDir + eslint/tsconfig ignore globs -> internal/web/dist - Dockerfile COPY paths for web/dist and web/translation - locale.go os.DirFS("web") disk fallback -> "internal/web" - .gitignore and ci.yml go:embed stub for internal/web/dist - api_docs_test.go repo-root relative walk (one level deeper) - tools/openapigen filesystem package paths; ApiTokenView repointed to the web/service/panel subpackage and codegen regenerated (clears a stale type the ci.yml codegen check was failing on) Verified: go build/vet/test (all packages), and frontend typecheck, lint, vitest (478 tests), and production build into internal/web/dist. * fix(config): keep test runs from writing logs into the source tree GetLogFolder() returns a CWD-relative "./log" on Windows. Under `go test` the working directory is each package's own folder, so InitLogger (called by tests in web/job, web/service, xray, web/websocket) created stray log/ directories scattered through the source tree (e.g. internal/web/job/log/). Redirect to a shared temp folder when testing.Testing() reports a test run. Production behavior is unchanged: Windows still uses ./log next to the binary and Linux /var/log/x-ui. The log files were always gitignored (*.log) and never committed; this just stops the noise at the source. * docs: move subscription-template guide out of root into docs/ sub_templates/ was a top-level folder holding only a README and no actual templates (3x-ui ships none by design), referenced nowhere and unlinked from any doc — it read like an empty placeholder cluttering the repo root. Move the guide to docs/custom-subscription-templates.md (a proper docs home), reword its intro to read as documentation rather than a folder note, link it from the Features list in README.md, and drop the empty sub_templates/ folder. * fix: update stale web/ path references after the internal/ move The internal/ migration rewrote Go import paths but left some references to the old top-level layout in docs, comments, and a few runtime disk paths. Functional (dev-mode only): the disk-serving fallbacks that read the Vite build from disk when running from source still pointed at web/dist/, which moved to internal/web/dist/ — so `os.DirFS`/`os.Stat`/`os.ReadFile` in internal/web/web.go and internal/sub/{sub,controller}.go are corrected. Production was unaffected (it serves the embedded FS; verified by the Docker build), but `go run` with a live frontend build silently fell back to embed. Docs/comments: frontend/README.md, CONTRIBUTING.md, the claude-issue-bot and release workflows, the openapigen -root help text, and assorted Go comments now reference internal/web, internal/database, internal/sub, internal/xray, etc. Package-name mentions (the "web" package), root paths (main.go, frontend/, install scripts, /etc/x-ui), routes (/panel/api/xray), and the historical "web/assets no longer exists" note were intentionally left as-is. * refactor(web): remove the legacy /xui -> /panel redirect middleware RedirectMiddleware existed only for backward compatibility with the old `/xui` URL scheme (301-redirecting /xui and /xui/API to /panel and /panel/api). That cutover was long ago, so drop the middleware, its registration in initRouter, and the now-inaccurate "URL redirection" mention in the middleware package doc. Old /xui URLs now 404 like any other unknown path. HTTPS auto-redirect and auth redirects are unrelated and stay. * build: fix .dockerignore for internal/ layout and exclude runtime dir - web/dist -> internal/web/dist: the embedded frontend moved under internal/, so the stale exclude no longer matched and the locally-built dist could be sent to the build context (the frontend stage rebuilds it fresh anyway). - exclude x-ui/: the local runtime directory (SQLite db, geo .dat files, xray binaries, certs — ~150MB) was being shipped into the build context for no reason. Verified the pattern excludes only the directory and still keeps x-ui.sh, which the Dockerfile copies to /usr/bin/x-ui.
2026-06-10 06:19:22 -07:00
"github.com/mhsanaei/3x-ui/v3/internal/config"
"github.com/mhsanaei/3x-ui/v3/internal/database/model"
"github.com/mhsanaei/3x-ui/v3/internal/logger"
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
"github.com/mhsanaei/3x-ui/v3/internal/tuic"
refactor: focused service files, leaf subpackages, and an internal/ layout (#5167) * refactor(service): split client.go into focused files client.go had grown to 4455 lines mixing ~10 responsibilities. Split it verbatim into cohesive same-package files (no behavior change): client.go foundation: ClientService, ClientWithAttachments, ClientCreatePayload, ErrClientNotInInbound, sqlInChunk client_locks.go inbound mutation locks, delete tombstones, compactOrphans client_lookup.go read-only lookups (GetByID, List, EffectiveFlow, ...) client_link.go inbound association sync (SyncInbound, DetachInbound, ...) client_crud.go single-client CRUD + validation + protocol defaults client_inbound_apply.go low-level inbound-settings mutators + by-email setters client_bulk.go bulk attach/detach/adjust/delete/create + DelDepleted client_traffic.go traffic-reset paths client_groups.go client group management client_paging.go paged listing, filtering, sorting, summary Every declaration moved unchanged (verified: identical func/type/const/var signature set before vs after). Imports redistributed per file via goimports. go build ./..., go vet, and go test ./web/service/... all pass. * refactor(service): split inbound.go into focused files inbound.go was 4100 lines. Split it verbatim into cohesive same-package files (no behavior change): inbound.go core inbound CRUD + InboundService (keeps pkg doc) inbound_protocol.go protocol / stream capability helpers inbound_node.go node/runtime/remote coordination + online tracking inbound_traffic.go traffic accounting, reset, client stats inbound_client_ips.go per-client IP tracking inbound_clients.go client lookups within inbounds + copy-clients inbound_disable.go auto-disable invalid inbounds/clients inbound_migration.go DB migrations inbound_sublink.go subscription link providers inbound_util.go generic slice/string helpers Identical func/type/const/var signature set before vs after; package doc comment preserved on inbound.go. Imports redistributed via goimports. Build, vet, and go test ./web/service/... all pass. * refactor(service): split tgbot.go into focused files tgbot.go was 3738 lines dominated by a 1246-line answerCallback. Split it verbatim into cohesive same-package files (no behavior change): tgbot.go lifecycle, bot setup, caches, small utils tgbot_router.go incoming update / command / callback dispatch tgbot_send.go outbound messaging primitives tgbot_client.go client views, actions, subscription links tgbot_inbound.go inbound listing / pickers tgbot_report.go server usage, exhausted, online, backups, notifications Identical func/type/const/var signature set before vs after. Imports redistributed via goimports. Build, vet, and go test ./web/service/... pass. * refactor(client): dedupe single-field by-email setters ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, and ResetClientTrafficLimitByEmail shared an identical ~50-line body that resolves the inbound by email, confirms the client exists, rewrites a single-client settings payload, and delegates to UpdateInboundClient. Extract that into applyClientFieldByEmail(inboundSvc, email, mutate) and reduce each setter to a 3-line wrapper. Behavior is unchanged: same checks and error strings, same single-client payload contract, same totalGB guard. SetClientTelegramUserID (resolves by traffic id, different error text) and ToggleClientEnableByEmail/SetClientEnableByEmail (different return shape and a pre-read of the old state) intentionally keep their own bodies. * refactor(service): extract panel/ subpackage Move the panel-administration leaf services out of the flat service package into web/service/panel/ (package panel): user.go UserService (auth / 2FA / LDAP) panel.go PanelService (restart / self-update) + version helpers panel_other.go non-unix RestartPanel panel_unix.go unix RestartPanel api_token.go ApiTokenService websocket.go WebSocketService panel_test.go version/shellQuote unit tests These are leaves: they depend on core (SettingService, Release) but no core file references them, so the extraction creates no import cycle. Core references are now qualified (service.SettingService, service.Release); callers in main.go, web/web.go, and web/controller/* updated to panel.*. Build, vet, and go test ./web/... pass. * refactor(service): extract integration/ subpackage Move the external-provider integration leaves into web/service/integration/ (package integration): warp.go WarpService (Cloudflare WARP) nord.go NordService (NordVPN) custom_geo.go CustomGeoService (custom geo asset management) *_test.go custom_geo / panel-proxy tests These depend on core (SettingService, ServerService, XraySettingService) but no core file references them. xray_setting.go stays in core because it calls the unexported SettingService.saveSetting. The shared isBlockedIP SSRF helper (used by core url_safety.go and by custom_geo) now has a small copy in each package rather than being exported. Core references qualified; callers in web/web.go, web/job/*, and web/controller/* updated to integration.*. Build, vet, and go test ./web/... pass. * refactor(service): extract tgbot/ subpackage Move the Telegram bot (6 files + test) into web/service/tgbot/ (package tgbot). It is a leaf: it embeds five core services (Inbound/Client/Setting/ Server/Xray) and the core never references it, so no import cycle. To support the package boundary without changing behavior: - core exposes XrayProcess() *xray.Process so tgbot keeps calling the exact same running-process methods it used via the package-level `p`; - three core methods tgbot calls are exported: ClientService.checkIs- EnabledByEmail -> CheckIsEnabledByEmail, InboundService.getAllEmails -> GetAllEmails (callers updated in-package); - tgbot's embedded-field types and the few core type refs (Status, ClientCreatePayload, SanitizePublicHTTPURL) are now service-qualified. Callers in main.go, web/web.go, web/job/*, and web/controller/* updated to tgbot.*. Build, vet, and go test ./web/... pass. * refactor(service): extract outbound/ subpackage OutboundService (outbound.go) imports only neutral packages (config, database, model, xray) and its production code is referenced by no core or sibling service file — only by web/controller/xray_setting.go and web/job/xray_traffic_job.go. Move it to web/service/outbound/ (package outbound); no core qualification needed inside. Callers updated to outbound.*. The one coupling was a tiny pure test helper, outboundsContainTag, used by both outbound.go and the core outbound_subscription_test.go; it now has a small copy in that test file rather than being shared across the boundary. Build, vet, and go test ./web/... pass. * refactor(util): move wireguard into its own subpackage util/wireguard.go was the lone file of the root `util` package (24 lines, one exported func GenerateWireguardKeypair), while every other util concern lives in a focused subpackage (util/common, util/crypto, util/netsafe, ...). Move it to util/wireguard/ (package wireguard) for consistency; its only importer, web/service/integration/warp.go, is updated. The root `util` package no longer exists. * refactor(sub): drop redundant sub prefix from filenames Inside package sub the subXxx.go prefix just repeats the package name (like client_*.go did inside service). Rename for consistency; content and type names are unchanged: subController.go -> controller.go subService.go -> service.go subClashService.go -> clash_service.go subJsonService.go -> json_service.go (+ matching _test.go files) * refactor(controller): rename xui.go -> spa.go XUIController serves the panel's single-page-app shell; spa.go names that role plainly (the other controller files are domain-named). File rename only — the type stays XUIController. api_docs_test.go keys route base paths by filename, so its "xui.go" case is updated to "spa.go". * refactor: move backend packages under internal/ Adopt the idiomatic Go application layout: the backend packages now live under internal/ (a boundary the toolchain enforces), signalling private implementation instead of a library-style flat root. No runtime behavior changes — only import paths and a few build/config paths move. Moved: config, database, logger, mtproto, sub, util, web, xray -> internal/. main.go stays at the repo root and tools/openapigen stays under tools/ (both still import internal/* because the internal rule keys off the module root). The module path github.com/mhsanaei/3x-ui/v3 is unchanged; 149 .go files had their import prefix rewritten to .../internal/<pkg>. Couplings the Go compiler can't see, updated to the new layout: - frontend i18n imports of web/translation (react.ts, setup.components.ts) - vite outDir + eslint/tsconfig ignore globs -> internal/web/dist - Dockerfile COPY paths for web/dist and web/translation - locale.go os.DirFS("web") disk fallback -> "internal/web" - .gitignore and ci.yml go:embed stub for internal/web/dist - api_docs_test.go repo-root relative walk (one level deeper) - tools/openapigen filesystem package paths; ApiTokenView repointed to the web/service/panel subpackage and codegen regenerated (clears a stale type the ci.yml codegen check was failing on) Verified: go build/vet/test (all packages), and frontend typecheck, lint, vitest (478 tests), and production build into internal/web/dist. * fix(config): keep test runs from writing logs into the source tree GetLogFolder() returns a CWD-relative "./log" on Windows. Under `go test` the working directory is each package's own folder, so InitLogger (called by tests in web/job, web/service, xray, web/websocket) created stray log/ directories scattered through the source tree (e.g. internal/web/job/log/). Redirect to a shared temp folder when testing.Testing() reports a test run. Production behavior is unchanged: Windows still uses ./log next to the binary and Linux /var/log/x-ui. The log files were always gitignored (*.log) and never committed; this just stops the noise at the source. * docs: move subscription-template guide out of root into docs/ sub_templates/ was a top-level folder holding only a README and no actual templates (3x-ui ships none by design), referenced nowhere and unlinked from any doc — it read like an empty placeholder cluttering the repo root. Move the guide to docs/custom-subscription-templates.md (a proper docs home), reword its intro to read as documentation rather than a folder note, link it from the Features list in README.md, and drop the empty sub_templates/ folder. * fix: update stale web/ path references after the internal/ move The internal/ migration rewrote Go import paths but left some references to the old top-level layout in docs, comments, and a few runtime disk paths. Functional (dev-mode only): the disk-serving fallbacks that read the Vite build from disk when running from source still pointed at web/dist/, which moved to internal/web/dist/ — so `os.DirFS`/`os.Stat`/`os.ReadFile` in internal/web/web.go and internal/sub/{sub,controller}.go are corrected. Production was unaffected (it serves the embedded FS; verified by the Docker build), but `go run` with a live frontend build silently fell back to embed. Docs/comments: frontend/README.md, CONTRIBUTING.md, the claude-issue-bot and release workflows, the openapigen -root help text, and assorted Go comments now reference internal/web, internal/database, internal/sub, internal/xray, etc. Package-name mentions (the "web" package), root paths (main.go, frontend/, install scripts, /etc/x-ui), routes (/panel/api/xray), and the historical "web/assets no longer exists" note were intentionally left as-is. * refactor(web): remove the legacy /xui -> /panel redirect middleware RedirectMiddleware existed only for backward compatibility with the old `/xui` URL scheme (301-redirecting /xui and /xui/API to /panel and /panel/api). That cutover was long ago, so drop the middleware, its registration in initRouter, and the now-inaccurate "URL redirection" mention in the middleware package doc. Old /xui URLs now 404 like any other unknown path. HTTPS auto-redirect and auth redirects are unrelated and stay. * build: fix .dockerignore for internal/ layout and exclude runtime dir - web/dist -> internal/web/dist: the embedded frontend moved under internal/, so the stale exclude no longer matched and the locally-built dist could be sent to the build context (the frontend stage rebuilds it fresh anyway). - exclude x-ui/: the local runtime directory (SQLite db, geo .dat files, xray binaries, certs — ~150MB) was being shipped into the build context for no reason. Verified the pattern excludes only the directory and still keeps x-ui.sh, which the Dockerfile copies to /usr/bin/x-ui.
2026-06-10 06:19:22 -07:00
"github.com/mhsanaei/3x-ui/v3/internal/util/json_util"
feat(xray): update xray-core to v26.9.30 and adapt panel Bump xtls/xray-core to b26a91de4f (v26.9.30) and the three binary pins (DockerInit.sh, release.yml Linux + Windows) in lockstep. No deleted symbols; the sing and sing-shadowsocks indirect deps drop out with the SS2022 rewrite. XDNS finalmask (#6718) replaced its string lists with objects: domains are {name, types, edns0, lenLimit, labelLimit} and client resolvers {type, settings.addr}. The loader no longer parses the old lists, so a single stored xdns mask keeps the whole core from starting. The new leaf package internal/util/maskcompat converts them: "name[:type]" becomes a domain and "name[:type]+udp://addr" a domain plus a udp resolver. A bare name maps to TXT, the type legacy clients queried by default, and each converted domain keeps the 1232-byte EDNS0 the old code always used (without it the server caps answers at 512). It runs from: - the XdnsFinalmaskObjectsFix seeder, over inbound streams, hosts, the xray template, the global sub-JSON mask and cached subscription outbounds; - inbound save (normalizeStreamSettings) and GetXrayConfig, for rows that never went through the seeder; - both link importers, since fm= from an older panel carries the lists. The finalmask form edits the object shape (every key needs a registered field, or the finalmask watch drops it on save) and lifts legacy masks on open. The udp-mask golden fixture moves to the object shape, which TestGoldenStreamFixturesBuildInXray now builds through the core. The wire format changed as well, so pre-upgrade clients need the new core. WireGuard outbound (#6771) dropped settings.domainStrategy and the remoteDNS "local" mode. The endpoint lookup now follows sockopt.domainStrategy and in-tunnel targets the outbound's targetStrategy. The old key is silently ignored, which undid the IPv4-first endpoint lookup the WARP outbound depends on (#5205), and "local" now panics the core at startup because remoteDNS goes through netip.MustParseAddr. The WireguardDomainStrategyFix seeder moves a stored family preference to both keys (a value already set wins) and turns "local" into targetStrategy; the outbound form lifts legacy rows the same way and drops its select, the WARP modal writes the new placement, and the inbound form loses a field the server never read. ValidateOutboundConfig now refuses a non-IP remoteDNS entry, which conf.Build() lets through, on template save and for outbound subscriptions. Noise finalmask items accept type "exp" (#6862), a tag expression. The form offers it for noise items only: header-custom items go through the core's PraseByteSlice, which refuses it. TUN gained autoSystemDnsToGateway (Linux) and autoSystemWfpBlockLeak (Windows). Both pass through the settings schema so a value set in JSON survives the next form save. MASQUE (inbound, outbound, transport) and the XDRIVE transport are new protocols the panel does not offer yet; their new loader refusals only cover configs the panel never generates. The FakeDNS IPv6 pool default, the SS2022 rewrite (same gRPC account; emails are now deduped case-insensitively, as the panel already does), the restored udphop interval default and the rest change no panel-facing config.
2026-09-30 08:27:10 -07:00
"github.com/mhsanaei/3x-ui/v3/internal/util/maskcompat"
refactor: focused service files, leaf subpackages, and an internal/ layout (#5167) * refactor(service): split client.go into focused files client.go had grown to 4455 lines mixing ~10 responsibilities. Split it verbatim into cohesive same-package files (no behavior change): client.go foundation: ClientService, ClientWithAttachments, ClientCreatePayload, ErrClientNotInInbound, sqlInChunk client_locks.go inbound mutation locks, delete tombstones, compactOrphans client_lookup.go read-only lookups (GetByID, List, EffectiveFlow, ...) client_link.go inbound association sync (SyncInbound, DetachInbound, ...) client_crud.go single-client CRUD + validation + protocol defaults client_inbound_apply.go low-level inbound-settings mutators + by-email setters client_bulk.go bulk attach/detach/adjust/delete/create + DelDepleted client_traffic.go traffic-reset paths client_groups.go client group management client_paging.go paged listing, filtering, sorting, summary Every declaration moved unchanged (verified: identical func/type/const/var signature set before vs after). Imports redistributed per file via goimports. go build ./..., go vet, and go test ./web/service/... all pass. * refactor(service): split inbound.go into focused files inbound.go was 4100 lines. Split it verbatim into cohesive same-package files (no behavior change): inbound.go core inbound CRUD + InboundService (keeps pkg doc) inbound_protocol.go protocol / stream capability helpers inbound_node.go node/runtime/remote coordination + online tracking inbound_traffic.go traffic accounting, reset, client stats inbound_client_ips.go per-client IP tracking inbound_clients.go client lookups within inbounds + copy-clients inbound_disable.go auto-disable invalid inbounds/clients inbound_migration.go DB migrations inbound_sublink.go subscription link providers inbound_util.go generic slice/string helpers Identical func/type/const/var signature set before vs after; package doc comment preserved on inbound.go. Imports redistributed via goimports. Build, vet, and go test ./web/service/... all pass. * refactor(service): split tgbot.go into focused files tgbot.go was 3738 lines dominated by a 1246-line answerCallback. Split it verbatim into cohesive same-package files (no behavior change): tgbot.go lifecycle, bot setup, caches, small utils tgbot_router.go incoming update / command / callback dispatch tgbot_send.go outbound messaging primitives tgbot_client.go client views, actions, subscription links tgbot_inbound.go inbound listing / pickers tgbot_report.go server usage, exhausted, online, backups, notifications Identical func/type/const/var signature set before vs after. Imports redistributed via goimports. Build, vet, and go test ./web/service/... pass. * refactor(client): dedupe single-field by-email setters ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, and ResetClientTrafficLimitByEmail shared an identical ~50-line body that resolves the inbound by email, confirms the client exists, rewrites a single-client settings payload, and delegates to UpdateInboundClient. Extract that into applyClientFieldByEmail(inboundSvc, email, mutate) and reduce each setter to a 3-line wrapper. Behavior is unchanged: same checks and error strings, same single-client payload contract, same totalGB guard. SetClientTelegramUserID (resolves by traffic id, different error text) and ToggleClientEnableByEmail/SetClientEnableByEmail (different return shape and a pre-read of the old state) intentionally keep their own bodies. * refactor(service): extract panel/ subpackage Move the panel-administration leaf services out of the flat service package into web/service/panel/ (package panel): user.go UserService (auth / 2FA / LDAP) panel.go PanelService (restart / self-update) + version helpers panel_other.go non-unix RestartPanel panel_unix.go unix RestartPanel api_token.go ApiTokenService websocket.go WebSocketService panel_test.go version/shellQuote unit tests These are leaves: they depend on core (SettingService, Release) but no core file references them, so the extraction creates no import cycle. Core references are now qualified (service.SettingService, service.Release); callers in main.go, web/web.go, and web/controller/* updated to panel.*. Build, vet, and go test ./web/... pass. * refactor(service): extract integration/ subpackage Move the external-provider integration leaves into web/service/integration/ (package integration): warp.go WarpService (Cloudflare WARP) nord.go NordService (NordVPN) custom_geo.go CustomGeoService (custom geo asset management) *_test.go custom_geo / panel-proxy tests These depend on core (SettingService, ServerService, XraySettingService) but no core file references them. xray_setting.go stays in core because it calls the unexported SettingService.saveSetting. The shared isBlockedIP SSRF helper (used by core url_safety.go and by custom_geo) now has a small copy in each package rather than being exported. Core references qualified; callers in web/web.go, web/job/*, and web/controller/* updated to integration.*. Build, vet, and go test ./web/... pass. * refactor(service): extract tgbot/ subpackage Move the Telegram bot (6 files + test) into web/service/tgbot/ (package tgbot). It is a leaf: it embeds five core services (Inbound/Client/Setting/ Server/Xray) and the core never references it, so no import cycle. To support the package boundary without changing behavior: - core exposes XrayProcess() *xray.Process so tgbot keeps calling the exact same running-process methods it used via the package-level `p`; - three core methods tgbot calls are exported: ClientService.checkIs- EnabledByEmail -> CheckIsEnabledByEmail, InboundService.getAllEmails -> GetAllEmails (callers updated in-package); - tgbot's embedded-field types and the few core type refs (Status, ClientCreatePayload, SanitizePublicHTTPURL) are now service-qualified. Callers in main.go, web/web.go, web/job/*, and web/controller/* updated to tgbot.*. Build, vet, and go test ./web/... pass. * refactor(service): extract outbound/ subpackage OutboundService (outbound.go) imports only neutral packages (config, database, model, xray) and its production code is referenced by no core or sibling service file — only by web/controller/xray_setting.go and web/job/xray_traffic_job.go. Move it to web/service/outbound/ (package outbound); no core qualification needed inside. Callers updated to outbound.*. The one coupling was a tiny pure test helper, outboundsContainTag, used by both outbound.go and the core outbound_subscription_test.go; it now has a small copy in that test file rather than being shared across the boundary. Build, vet, and go test ./web/... pass. * refactor(util): move wireguard into its own subpackage util/wireguard.go was the lone file of the root `util` package (24 lines, one exported func GenerateWireguardKeypair), while every other util concern lives in a focused subpackage (util/common, util/crypto, util/netsafe, ...). Move it to util/wireguard/ (package wireguard) for consistency; its only importer, web/service/integration/warp.go, is updated. The root `util` package no longer exists. * refactor(sub): drop redundant sub prefix from filenames Inside package sub the subXxx.go prefix just repeats the package name (like client_*.go did inside service). Rename for consistency; content and type names are unchanged: subController.go -> controller.go subService.go -> service.go subClashService.go -> clash_service.go subJsonService.go -> json_service.go (+ matching _test.go files) * refactor(controller): rename xui.go -> spa.go XUIController serves the panel's single-page-app shell; spa.go names that role plainly (the other controller files are domain-named). File rename only — the type stays XUIController. api_docs_test.go keys route base paths by filename, so its "xui.go" case is updated to "spa.go". * refactor: move backend packages under internal/ Adopt the idiomatic Go application layout: the backend packages now live under internal/ (a boundary the toolchain enforces), signalling private implementation instead of a library-style flat root. No runtime behavior changes — only import paths and a few build/config paths move. Moved: config, database, logger, mtproto, sub, util, web, xray -> internal/. main.go stays at the repo root and tools/openapigen stays under tools/ (both still import internal/* because the internal rule keys off the module root). The module path github.com/mhsanaei/3x-ui/v3 is unchanged; 149 .go files had their import prefix rewritten to .../internal/<pkg>. Couplings the Go compiler can't see, updated to the new layout: - frontend i18n imports of web/translation (react.ts, setup.components.ts) - vite outDir + eslint/tsconfig ignore globs -> internal/web/dist - Dockerfile COPY paths for web/dist and web/translation - locale.go os.DirFS("web") disk fallback -> "internal/web" - .gitignore and ci.yml go:embed stub for internal/web/dist - api_docs_test.go repo-root relative walk (one level deeper) - tools/openapigen filesystem package paths; ApiTokenView repointed to the web/service/panel subpackage and codegen regenerated (clears a stale type the ci.yml codegen check was failing on) Verified: go build/vet/test (all packages), and frontend typecheck, lint, vitest (478 tests), and production build into internal/web/dist. * fix(config): keep test runs from writing logs into the source tree GetLogFolder() returns a CWD-relative "./log" on Windows. Under `go test` the working directory is each package's own folder, so InitLogger (called by tests in web/job, web/service, xray, web/websocket) created stray log/ directories scattered through the source tree (e.g. internal/web/job/log/). Redirect to a shared temp folder when testing.Testing() reports a test run. Production behavior is unchanged: Windows still uses ./log next to the binary and Linux /var/log/x-ui. The log files were always gitignored (*.log) and never committed; this just stops the noise at the source. * docs: move subscription-template guide out of root into docs/ sub_templates/ was a top-level folder holding only a README and no actual templates (3x-ui ships none by design), referenced nowhere and unlinked from any doc — it read like an empty placeholder cluttering the repo root. Move the guide to docs/custom-subscription-templates.md (a proper docs home), reword its intro to read as documentation rather than a folder note, link it from the Features list in README.md, and drop the empty sub_templates/ folder. * fix: update stale web/ path references after the internal/ move The internal/ migration rewrote Go import paths but left some references to the old top-level layout in docs, comments, and a few runtime disk paths. Functional (dev-mode only): the disk-serving fallbacks that read the Vite build from disk when running from source still pointed at web/dist/, which moved to internal/web/dist/ — so `os.DirFS`/`os.Stat`/`os.ReadFile` in internal/web/web.go and internal/sub/{sub,controller}.go are corrected. Production was unaffected (it serves the embedded FS; verified by the Docker build), but `go run` with a live frontend build silently fell back to embed. Docs/comments: frontend/README.md, CONTRIBUTING.md, the claude-issue-bot and release workflows, the openapigen -root help text, and assorted Go comments now reference internal/web, internal/database, internal/sub, internal/xray, etc. Package-name mentions (the "web" package), root paths (main.go, frontend/, install scripts, /etc/x-ui), routes (/panel/api/xray), and the historical "web/assets no longer exists" note were intentionally left as-is. * refactor(web): remove the legacy /xui -> /panel redirect middleware RedirectMiddleware existed only for backward compatibility with the old `/xui` URL scheme (301-redirecting /xui and /xui/API to /panel and /panel/api). That cutover was long ago, so drop the middleware, its registration in initRouter, and the now-inaccurate "URL redirection" mention in the middleware package doc. Old /xui URLs now 404 like any other unknown path. HTTPS auto-redirect and auth redirects are unrelated and stay. * build: fix .dockerignore for internal/ layout and exclude runtime dir - web/dist -> internal/web/dist: the embedded frontend moved under internal/, so the stale exclude no longer matched and the locally-built dist could be sent to the build context (the frontend stage rebuilds it fresh anyway). - exclude x-ui/: the local runtime directory (SQLite db, geo .dat files, xray binaries, certs — ~150MB) was being shipped into the build context for no reason. Verified the pattern excludes only the directory and still keeps x-ui.sh, which the Dockerfile copies to /usr/bin/x-ui.
2026-06-10 06:19:22 -07:00
"github.com/mhsanaei/3x-ui/v3/internal/xray"
2023-02-18 04:37:32 -08:00
2023-02-09 11:18:06 -08:00
"go.uber.org/atomic"
)
var (
lock sync.Mutex
isNeedXrayReconcile atomic.Bool
isNeedXrayRestart atomic.Bool // Indicates that restart was requested for Xray
isManuallyStopped atomic.Bool // Indicates that Xray was stopped manually from the panel
xrayState xrayLifecycle
)
2023-02-09 11:18:06 -08:00
type xrayLifecycle struct {
mu sync.RWMutex
process *xray.Process
result string
fix(xray): refuse a config the running core cannot bind (#6547) * fix(xray): refuse a config the running core cannot bind RestartXray stopped a working core before handing it a config whose listens collide, so the failed bind exited the whole process (main/run.go:94) and the one-second watchdog retried it in a loop: every protocol down, cause only in the logs. The save-time port guards cannot cover this -- SetInboundEnable, the AmneziaWG relay created on the first peer, template and bridge edits all reach a colliding config with no guard on that path. Probe the generated config at the single restart funnel instead. Collisions the running core already serves are excused, so an established setup is never refused by a static read being wrong about it, and the port-bucketed pass costs nothing on a clean config. * fix(xray): surface a refused config and re-key the bind excuse set Round-1 findings on this PR. Refusing the swap left the running core on its previous config with nothing but a log line to show for it, so the status response now carries the reason while the core runs and the overview marks it; the node list picks the same field up through that response. The excuse set is keyed on the two listens, the port and the shared transports instead of the tag pair, so a pair whose listen moves onto the other's address is refused again, while the same two sockets stay excused however the generator orders them. TestBindConflicts/excused_pair_whose_listen_changed_into_a_real_collision fails without the key change -- watched red first.
2026-09-15 05:27:28 -07:00
// heldBack is why the running core still serves the previous config.
heldBack string
}
func (s *xrayLifecycle) snapshot() (*xray.Process, string) {
s.mu.RLock()
defer s.mu.RUnlock()
return s.process, s.result
}
func (s *xrayLifecycle) replace(process *xray.Process) {
s.mu.Lock()
s.process = process
s.result = ""
fix(xray): refuse a config the running core cannot bind (#6547) * fix(xray): refuse a config the running core cannot bind RestartXray stopped a working core before handing it a config whose listens collide, so the failed bind exited the whole process (main/run.go:94) and the one-second watchdog retried it in a loop: every protocol down, cause only in the logs. The save-time port guards cannot cover this -- SetInboundEnable, the AmneziaWG relay created on the first peer, template and bridge edits all reach a colliding config with no guard on that path. Probe the generated config at the single restart funnel instead. Collisions the running core already serves are excused, so an established setup is never refused by a static read being wrong about it, and the port-bucketed pass costs nothing on a clean config. * fix(xray): surface a refused config and re-key the bind excuse set Round-1 findings on this PR. Refusing the swap left the running core on its previous config with nothing but a log line to show for it, so the status response now carries the reason while the core runs and the overview marks it; the node list picks the same field up through that response. The excuse set is keyed on the two listens, the port and the shared transports instead of the tag pair, so a pair whose listen moves onto the other's address is refused again, while the same two sockets stay excused however the generator orders them. TestBindConflicts/excused_pair_whose_listen_changed_into_a_real_collision fails without the key change -- watched red first.
2026-09-15 05:27:28 -07:00
s.heldBack = ""
s.mu.Unlock()
}
fix(xray): refuse a config the running core cannot bind (#6547) * fix(xray): refuse a config the running core cannot bind RestartXray stopped a working core before handing it a config whose listens collide, so the failed bind exited the whole process (main/run.go:94) and the one-second watchdog retried it in a loop: every protocol down, cause only in the logs. The save-time port guards cannot cover this -- SetInboundEnable, the AmneziaWG relay created on the first peer, template and bridge edits all reach a colliding config with no guard on that path. Probe the generated config at the single restart funnel instead. Collisions the running core already serves are excused, so an established setup is never refused by a static read being wrong about it, and the port-bucketed pass costs nothing on a clean config. * fix(xray): surface a refused config and re-key the bind excuse set Round-1 findings on this PR. Refusing the swap left the running core on its previous config with nothing but a log line to show for it, so the status response now carries the reason while the core runs and the overview marks it; the node list picks the same field up through that response. The excuse set is keyed on the two listens, the port and the shared transports instead of the tag pair, so a pair whose listen moves onto the other's address is refused again, while the same two sockets stay excused however the generator orders them. TestBindConflicts/excused_pair_whose_listen_changed_into_a_real_collision fails without the key change -- watched red first.
2026-09-15 05:27:28 -07:00
func (s *xrayLifecycle) holdBack(reason string) {
s.mu.Lock()
s.heldBack = reason
s.mu.Unlock()
}
func (s *xrayLifecycle) heldBackReason() string {
s.mu.RLock()
defer s.mu.RUnlock()
return s.heldBack
}
func (s *xrayLifecycle) storeResult(process *xray.Process, result string) {
s.mu.Lock()
if s.process == process && s.result == "" {
s.result = result
}
s.mu.Unlock()
}
func currentXrayProcess() *xray.Process {
process, _ := xrayState.snapshot()
return process
}
// ErrXrayNotRunning reports an operation that requires a running core when xray is stopped.
var ErrXrayNotRunning = xray.ErrXrayNotRunning
2025-09-20 00:35:50 -07:00
// XrayService provides business logic for Xray process management.
// It handles starting, stopping, restarting Xray, and managing its configuration.
2023-02-09 11:18:06 -08:00
type XrayService struct {
inboundService InboundService
settingService SettingService
nodeService NodeService
xrayAPI xray.XrayAPI
2023-02-09 11:18:06 -08:00
}
2025-09-20 00:35:50 -07:00
// IsXrayRunning checks if the Xray process is currently running.
2023-02-09 11:18:06 -08:00
func (s *XrayService) IsXrayRunning() bool {
process := currentXrayProcess()
return process != nil && process.IsRunning()
2023-02-09 11:18:06 -08:00
}
refactor: focused service files, leaf subpackages, and an internal/ layout (#5167) * refactor(service): split client.go into focused files client.go had grown to 4455 lines mixing ~10 responsibilities. Split it verbatim into cohesive same-package files (no behavior change): client.go foundation: ClientService, ClientWithAttachments, ClientCreatePayload, ErrClientNotInInbound, sqlInChunk client_locks.go inbound mutation locks, delete tombstones, compactOrphans client_lookup.go read-only lookups (GetByID, List, EffectiveFlow, ...) client_link.go inbound association sync (SyncInbound, DetachInbound, ...) client_crud.go single-client CRUD + validation + protocol defaults client_inbound_apply.go low-level inbound-settings mutators + by-email setters client_bulk.go bulk attach/detach/adjust/delete/create + DelDepleted client_traffic.go traffic-reset paths client_groups.go client group management client_paging.go paged listing, filtering, sorting, summary Every declaration moved unchanged (verified: identical func/type/const/var signature set before vs after). Imports redistributed per file via goimports. go build ./..., go vet, and go test ./web/service/... all pass. * refactor(service): split inbound.go into focused files inbound.go was 4100 lines. Split it verbatim into cohesive same-package files (no behavior change): inbound.go core inbound CRUD + InboundService (keeps pkg doc) inbound_protocol.go protocol / stream capability helpers inbound_node.go node/runtime/remote coordination + online tracking inbound_traffic.go traffic accounting, reset, client stats inbound_client_ips.go per-client IP tracking inbound_clients.go client lookups within inbounds + copy-clients inbound_disable.go auto-disable invalid inbounds/clients inbound_migration.go DB migrations inbound_sublink.go subscription link providers inbound_util.go generic slice/string helpers Identical func/type/const/var signature set before vs after; package doc comment preserved on inbound.go. Imports redistributed via goimports. Build, vet, and go test ./web/service/... all pass. * refactor(service): split tgbot.go into focused files tgbot.go was 3738 lines dominated by a 1246-line answerCallback. Split it verbatim into cohesive same-package files (no behavior change): tgbot.go lifecycle, bot setup, caches, small utils tgbot_router.go incoming update / command / callback dispatch tgbot_send.go outbound messaging primitives tgbot_client.go client views, actions, subscription links tgbot_inbound.go inbound listing / pickers tgbot_report.go server usage, exhausted, online, backups, notifications Identical func/type/const/var signature set before vs after. Imports redistributed via goimports. Build, vet, and go test ./web/service/... pass. * refactor(client): dedupe single-field by-email setters ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, and ResetClientTrafficLimitByEmail shared an identical ~50-line body that resolves the inbound by email, confirms the client exists, rewrites a single-client settings payload, and delegates to UpdateInboundClient. Extract that into applyClientFieldByEmail(inboundSvc, email, mutate) and reduce each setter to a 3-line wrapper. Behavior is unchanged: same checks and error strings, same single-client payload contract, same totalGB guard. SetClientTelegramUserID (resolves by traffic id, different error text) and ToggleClientEnableByEmail/SetClientEnableByEmail (different return shape and a pre-read of the old state) intentionally keep their own bodies. * refactor(service): extract panel/ subpackage Move the panel-administration leaf services out of the flat service package into web/service/panel/ (package panel): user.go UserService (auth / 2FA / LDAP) panel.go PanelService (restart / self-update) + version helpers panel_other.go non-unix RestartPanel panel_unix.go unix RestartPanel api_token.go ApiTokenService websocket.go WebSocketService panel_test.go version/shellQuote unit tests These are leaves: they depend on core (SettingService, Release) but no core file references them, so the extraction creates no import cycle. Core references are now qualified (service.SettingService, service.Release); callers in main.go, web/web.go, and web/controller/* updated to panel.*. Build, vet, and go test ./web/... pass. * refactor(service): extract integration/ subpackage Move the external-provider integration leaves into web/service/integration/ (package integration): warp.go WarpService (Cloudflare WARP) nord.go NordService (NordVPN) custom_geo.go CustomGeoService (custom geo asset management) *_test.go custom_geo / panel-proxy tests These depend on core (SettingService, ServerService, XraySettingService) but no core file references them. xray_setting.go stays in core because it calls the unexported SettingService.saveSetting. The shared isBlockedIP SSRF helper (used by core url_safety.go and by custom_geo) now has a small copy in each package rather than being exported. Core references qualified; callers in web/web.go, web/job/*, and web/controller/* updated to integration.*. Build, vet, and go test ./web/... pass. * refactor(service): extract tgbot/ subpackage Move the Telegram bot (6 files + test) into web/service/tgbot/ (package tgbot). It is a leaf: it embeds five core services (Inbound/Client/Setting/ Server/Xray) and the core never references it, so no import cycle. To support the package boundary without changing behavior: - core exposes XrayProcess() *xray.Process so tgbot keeps calling the exact same running-process methods it used via the package-level `p`; - three core methods tgbot calls are exported: ClientService.checkIs- EnabledByEmail -> CheckIsEnabledByEmail, InboundService.getAllEmails -> GetAllEmails (callers updated in-package); - tgbot's embedded-field types and the few core type refs (Status, ClientCreatePayload, SanitizePublicHTTPURL) are now service-qualified. Callers in main.go, web/web.go, web/job/*, and web/controller/* updated to tgbot.*. Build, vet, and go test ./web/... pass. * refactor(service): extract outbound/ subpackage OutboundService (outbound.go) imports only neutral packages (config, database, model, xray) and its production code is referenced by no core or sibling service file — only by web/controller/xray_setting.go and web/job/xray_traffic_job.go. Move it to web/service/outbound/ (package outbound); no core qualification needed inside. Callers updated to outbound.*. The one coupling was a tiny pure test helper, outboundsContainTag, used by both outbound.go and the core outbound_subscription_test.go; it now has a small copy in that test file rather than being shared across the boundary. Build, vet, and go test ./web/... pass. * refactor(util): move wireguard into its own subpackage util/wireguard.go was the lone file of the root `util` package (24 lines, one exported func GenerateWireguardKeypair), while every other util concern lives in a focused subpackage (util/common, util/crypto, util/netsafe, ...). Move it to util/wireguard/ (package wireguard) for consistency; its only importer, web/service/integration/warp.go, is updated. The root `util` package no longer exists. * refactor(sub): drop redundant sub prefix from filenames Inside package sub the subXxx.go prefix just repeats the package name (like client_*.go did inside service). Rename for consistency; content and type names are unchanged: subController.go -> controller.go subService.go -> service.go subClashService.go -> clash_service.go subJsonService.go -> json_service.go (+ matching _test.go files) * refactor(controller): rename xui.go -> spa.go XUIController serves the panel's single-page-app shell; spa.go names that role plainly (the other controller files are domain-named). File rename only — the type stays XUIController. api_docs_test.go keys route base paths by filename, so its "xui.go" case is updated to "spa.go". * refactor: move backend packages under internal/ Adopt the idiomatic Go application layout: the backend packages now live under internal/ (a boundary the toolchain enforces), signalling private implementation instead of a library-style flat root. No runtime behavior changes — only import paths and a few build/config paths move. Moved: config, database, logger, mtproto, sub, util, web, xray -> internal/. main.go stays at the repo root and tools/openapigen stays under tools/ (both still import internal/* because the internal rule keys off the module root). The module path github.com/mhsanaei/3x-ui/v3 is unchanged; 149 .go files had their import prefix rewritten to .../internal/<pkg>. Couplings the Go compiler can't see, updated to the new layout: - frontend i18n imports of web/translation (react.ts, setup.components.ts) - vite outDir + eslint/tsconfig ignore globs -> internal/web/dist - Dockerfile COPY paths for web/dist and web/translation - locale.go os.DirFS("web") disk fallback -> "internal/web" - .gitignore and ci.yml go:embed stub for internal/web/dist - api_docs_test.go repo-root relative walk (one level deeper) - tools/openapigen filesystem package paths; ApiTokenView repointed to the web/service/panel subpackage and codegen regenerated (clears a stale type the ci.yml codegen check was failing on) Verified: go build/vet/test (all packages), and frontend typecheck, lint, vitest (478 tests), and production build into internal/web/dist. * fix(config): keep test runs from writing logs into the source tree GetLogFolder() returns a CWD-relative "./log" on Windows. Under `go test` the working directory is each package's own folder, so InitLogger (called by tests in web/job, web/service, xray, web/websocket) created stray log/ directories scattered through the source tree (e.g. internal/web/job/log/). Redirect to a shared temp folder when testing.Testing() reports a test run. Production behavior is unchanged: Windows still uses ./log next to the binary and Linux /var/log/x-ui. The log files were always gitignored (*.log) and never committed; this just stops the noise at the source. * docs: move subscription-template guide out of root into docs/ sub_templates/ was a top-level folder holding only a README and no actual templates (3x-ui ships none by design), referenced nowhere and unlinked from any doc — it read like an empty placeholder cluttering the repo root. Move the guide to docs/custom-subscription-templates.md (a proper docs home), reword its intro to read as documentation rather than a folder note, link it from the Features list in README.md, and drop the empty sub_templates/ folder. * fix: update stale web/ path references after the internal/ move The internal/ migration rewrote Go import paths but left some references to the old top-level layout in docs, comments, and a few runtime disk paths. Functional (dev-mode only): the disk-serving fallbacks that read the Vite build from disk when running from source still pointed at web/dist/, which moved to internal/web/dist/ — so `os.DirFS`/`os.Stat`/`os.ReadFile` in internal/web/web.go and internal/sub/{sub,controller}.go are corrected. Production was unaffected (it serves the embedded FS; verified by the Docker build), but `go run` with a live frontend build silently fell back to embed. Docs/comments: frontend/README.md, CONTRIBUTING.md, the claude-issue-bot and release workflows, the openapigen -root help text, and assorted Go comments now reference internal/web, internal/database, internal/sub, internal/xray, etc. Package-name mentions (the "web" package), root paths (main.go, frontend/, install scripts, /etc/x-ui), routes (/panel/api/xray), and the historical "web/assets no longer exists" note were intentionally left as-is. * refactor(web): remove the legacy /xui -> /panel redirect middleware RedirectMiddleware existed only for backward compatibility with the old `/xui` URL scheme (301-redirecting /xui and /xui/API to /panel and /panel/api). That cutover was long ago, so drop the middleware, its registration in initRouter, and the now-inaccurate "URL redirection" mention in the middleware package doc. Old /xui URLs now 404 like any other unknown path. HTTPS auto-redirect and auth redirects are unrelated and stay. * build: fix .dockerignore for internal/ layout and exclude runtime dir - web/dist -> internal/web/dist: the embedded frontend moved under internal/, so the stale exclude no longer matched and the locally-built dist could be sent to the build context (the frontend stage rebuilds it fresh anyway). - exclude x-ui/: the local runtime directory (SQLite db, geo .dat files, xray binaries, certs — ~150MB) was being shipped into the build context for no reason. Verified the pattern excludes only the directory and still keeps x-ui.sh, which the Dockerfile copies to /usr/bin/x-ui.
2026-06-10 06:19:22 -07:00
// XrayProcess returns the current Xray process instance (may be nil when Xray
// is not running). It exposes the lifecycle snapshot to callers outside this
// package (e.g. the tgbot subpackage).
refactor: focused service files, leaf subpackages, and an internal/ layout (#5167) * refactor(service): split client.go into focused files client.go had grown to 4455 lines mixing ~10 responsibilities. Split it verbatim into cohesive same-package files (no behavior change): client.go foundation: ClientService, ClientWithAttachments, ClientCreatePayload, ErrClientNotInInbound, sqlInChunk client_locks.go inbound mutation locks, delete tombstones, compactOrphans client_lookup.go read-only lookups (GetByID, List, EffectiveFlow, ...) client_link.go inbound association sync (SyncInbound, DetachInbound, ...) client_crud.go single-client CRUD + validation + protocol defaults client_inbound_apply.go low-level inbound-settings mutators + by-email setters client_bulk.go bulk attach/detach/adjust/delete/create + DelDepleted client_traffic.go traffic-reset paths client_groups.go client group management client_paging.go paged listing, filtering, sorting, summary Every declaration moved unchanged (verified: identical func/type/const/var signature set before vs after). Imports redistributed per file via goimports. go build ./..., go vet, and go test ./web/service/... all pass. * refactor(service): split inbound.go into focused files inbound.go was 4100 lines. Split it verbatim into cohesive same-package files (no behavior change): inbound.go core inbound CRUD + InboundService (keeps pkg doc) inbound_protocol.go protocol / stream capability helpers inbound_node.go node/runtime/remote coordination + online tracking inbound_traffic.go traffic accounting, reset, client stats inbound_client_ips.go per-client IP tracking inbound_clients.go client lookups within inbounds + copy-clients inbound_disable.go auto-disable invalid inbounds/clients inbound_migration.go DB migrations inbound_sublink.go subscription link providers inbound_util.go generic slice/string helpers Identical func/type/const/var signature set before vs after; package doc comment preserved on inbound.go. Imports redistributed via goimports. Build, vet, and go test ./web/service/... all pass. * refactor(service): split tgbot.go into focused files tgbot.go was 3738 lines dominated by a 1246-line answerCallback. Split it verbatim into cohesive same-package files (no behavior change): tgbot.go lifecycle, bot setup, caches, small utils tgbot_router.go incoming update / command / callback dispatch tgbot_send.go outbound messaging primitives tgbot_client.go client views, actions, subscription links tgbot_inbound.go inbound listing / pickers tgbot_report.go server usage, exhausted, online, backups, notifications Identical func/type/const/var signature set before vs after. Imports redistributed via goimports. Build, vet, and go test ./web/service/... pass. * refactor(client): dedupe single-field by-email setters ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, and ResetClientTrafficLimitByEmail shared an identical ~50-line body that resolves the inbound by email, confirms the client exists, rewrites a single-client settings payload, and delegates to UpdateInboundClient. Extract that into applyClientFieldByEmail(inboundSvc, email, mutate) and reduce each setter to a 3-line wrapper. Behavior is unchanged: same checks and error strings, same single-client payload contract, same totalGB guard. SetClientTelegramUserID (resolves by traffic id, different error text) and ToggleClientEnableByEmail/SetClientEnableByEmail (different return shape and a pre-read of the old state) intentionally keep their own bodies. * refactor(service): extract panel/ subpackage Move the panel-administration leaf services out of the flat service package into web/service/panel/ (package panel): user.go UserService (auth / 2FA / LDAP) panel.go PanelService (restart / self-update) + version helpers panel_other.go non-unix RestartPanel panel_unix.go unix RestartPanel api_token.go ApiTokenService websocket.go WebSocketService panel_test.go version/shellQuote unit tests These are leaves: they depend on core (SettingService, Release) but no core file references them, so the extraction creates no import cycle. Core references are now qualified (service.SettingService, service.Release); callers in main.go, web/web.go, and web/controller/* updated to panel.*. Build, vet, and go test ./web/... pass. * refactor(service): extract integration/ subpackage Move the external-provider integration leaves into web/service/integration/ (package integration): warp.go WarpService (Cloudflare WARP) nord.go NordService (NordVPN) custom_geo.go CustomGeoService (custom geo asset management) *_test.go custom_geo / panel-proxy tests These depend on core (SettingService, ServerService, XraySettingService) but no core file references them. xray_setting.go stays in core because it calls the unexported SettingService.saveSetting. The shared isBlockedIP SSRF helper (used by core url_safety.go and by custom_geo) now has a small copy in each package rather than being exported. Core references qualified; callers in web/web.go, web/job/*, and web/controller/* updated to integration.*. Build, vet, and go test ./web/... pass. * refactor(service): extract tgbot/ subpackage Move the Telegram bot (6 files + test) into web/service/tgbot/ (package tgbot). It is a leaf: it embeds five core services (Inbound/Client/Setting/ Server/Xray) and the core never references it, so no import cycle. To support the package boundary without changing behavior: - core exposes XrayProcess() *xray.Process so tgbot keeps calling the exact same running-process methods it used via the package-level `p`; - three core methods tgbot calls are exported: ClientService.checkIs- EnabledByEmail -> CheckIsEnabledByEmail, InboundService.getAllEmails -> GetAllEmails (callers updated in-package); - tgbot's embedded-field types and the few core type refs (Status, ClientCreatePayload, SanitizePublicHTTPURL) are now service-qualified. Callers in main.go, web/web.go, web/job/*, and web/controller/* updated to tgbot.*. Build, vet, and go test ./web/... pass. * refactor(service): extract outbound/ subpackage OutboundService (outbound.go) imports only neutral packages (config, database, model, xray) and its production code is referenced by no core or sibling service file — only by web/controller/xray_setting.go and web/job/xray_traffic_job.go. Move it to web/service/outbound/ (package outbound); no core qualification needed inside. Callers updated to outbound.*. The one coupling was a tiny pure test helper, outboundsContainTag, used by both outbound.go and the core outbound_subscription_test.go; it now has a small copy in that test file rather than being shared across the boundary. Build, vet, and go test ./web/... pass. * refactor(util): move wireguard into its own subpackage util/wireguard.go was the lone file of the root `util` package (24 lines, one exported func GenerateWireguardKeypair), while every other util concern lives in a focused subpackage (util/common, util/crypto, util/netsafe, ...). Move it to util/wireguard/ (package wireguard) for consistency; its only importer, web/service/integration/warp.go, is updated. The root `util` package no longer exists. * refactor(sub): drop redundant sub prefix from filenames Inside package sub the subXxx.go prefix just repeats the package name (like client_*.go did inside service). Rename for consistency; content and type names are unchanged: subController.go -> controller.go subService.go -> service.go subClashService.go -> clash_service.go subJsonService.go -> json_service.go (+ matching _test.go files) * refactor(controller): rename xui.go -> spa.go XUIController serves the panel's single-page-app shell; spa.go names that role plainly (the other controller files are domain-named). File rename only — the type stays XUIController. api_docs_test.go keys route base paths by filename, so its "xui.go" case is updated to "spa.go". * refactor: move backend packages under internal/ Adopt the idiomatic Go application layout: the backend packages now live under internal/ (a boundary the toolchain enforces), signalling private implementation instead of a library-style flat root. No runtime behavior changes — only import paths and a few build/config paths move. Moved: config, database, logger, mtproto, sub, util, web, xray -> internal/. main.go stays at the repo root and tools/openapigen stays under tools/ (both still import internal/* because the internal rule keys off the module root). The module path github.com/mhsanaei/3x-ui/v3 is unchanged; 149 .go files had their import prefix rewritten to .../internal/<pkg>. Couplings the Go compiler can't see, updated to the new layout: - frontend i18n imports of web/translation (react.ts, setup.components.ts) - vite outDir + eslint/tsconfig ignore globs -> internal/web/dist - Dockerfile COPY paths for web/dist and web/translation - locale.go os.DirFS("web") disk fallback -> "internal/web" - .gitignore and ci.yml go:embed stub for internal/web/dist - api_docs_test.go repo-root relative walk (one level deeper) - tools/openapigen filesystem package paths; ApiTokenView repointed to the web/service/panel subpackage and codegen regenerated (clears a stale type the ci.yml codegen check was failing on) Verified: go build/vet/test (all packages), and frontend typecheck, lint, vitest (478 tests), and production build into internal/web/dist. * fix(config): keep test runs from writing logs into the source tree GetLogFolder() returns a CWD-relative "./log" on Windows. Under `go test` the working directory is each package's own folder, so InitLogger (called by tests in web/job, web/service, xray, web/websocket) created stray log/ directories scattered through the source tree (e.g. internal/web/job/log/). Redirect to a shared temp folder when testing.Testing() reports a test run. Production behavior is unchanged: Windows still uses ./log next to the binary and Linux /var/log/x-ui. The log files were always gitignored (*.log) and never committed; this just stops the noise at the source. * docs: move subscription-template guide out of root into docs/ sub_templates/ was a top-level folder holding only a README and no actual templates (3x-ui ships none by design), referenced nowhere and unlinked from any doc — it read like an empty placeholder cluttering the repo root. Move the guide to docs/custom-subscription-templates.md (a proper docs home), reword its intro to read as documentation rather than a folder note, link it from the Features list in README.md, and drop the empty sub_templates/ folder. * fix: update stale web/ path references after the internal/ move The internal/ migration rewrote Go import paths but left some references to the old top-level layout in docs, comments, and a few runtime disk paths. Functional (dev-mode only): the disk-serving fallbacks that read the Vite build from disk when running from source still pointed at web/dist/, which moved to internal/web/dist/ — so `os.DirFS`/`os.Stat`/`os.ReadFile` in internal/web/web.go and internal/sub/{sub,controller}.go are corrected. Production was unaffected (it serves the embedded FS; verified by the Docker build), but `go run` with a live frontend build silently fell back to embed. Docs/comments: frontend/README.md, CONTRIBUTING.md, the claude-issue-bot and release workflows, the openapigen -root help text, and assorted Go comments now reference internal/web, internal/database, internal/sub, internal/xray, etc. Package-name mentions (the "web" package), root paths (main.go, frontend/, install scripts, /etc/x-ui), routes (/panel/api/xray), and the historical "web/assets no longer exists" note were intentionally left as-is. * refactor(web): remove the legacy /xui -> /panel redirect middleware RedirectMiddleware existed only for backward compatibility with the old `/xui` URL scheme (301-redirecting /xui and /xui/API to /panel and /panel/api). That cutover was long ago, so drop the middleware, its registration in initRouter, and the now-inaccurate "URL redirection" mention in the middleware package doc. Old /xui URLs now 404 like any other unknown path. HTTPS auto-redirect and auth redirects are unrelated and stay. * build: fix .dockerignore for internal/ layout and exclude runtime dir - web/dist -> internal/web/dist: the embedded frontend moved under internal/, so the stale exclude no longer matched and the locally-built dist could be sent to the build context (the frontend stage rebuilds it fresh anyway). - exclude x-ui/: the local runtime directory (SQLite db, geo .dat files, xray binaries, certs — ~150MB) was being shipped into the build context for no reason. Verified the pattern excludes only the directory and still keeps x-ui.sh, which the Dockerfile copies to /usr/bin/x-ui.
2026-06-10 06:19:22 -07:00
func XrayProcess() *xray.Process {
return currentXrayProcess()
refactor: focused service files, leaf subpackages, and an internal/ layout (#5167) * refactor(service): split client.go into focused files client.go had grown to 4455 lines mixing ~10 responsibilities. Split it verbatim into cohesive same-package files (no behavior change): client.go foundation: ClientService, ClientWithAttachments, ClientCreatePayload, ErrClientNotInInbound, sqlInChunk client_locks.go inbound mutation locks, delete tombstones, compactOrphans client_lookup.go read-only lookups (GetByID, List, EffectiveFlow, ...) client_link.go inbound association sync (SyncInbound, DetachInbound, ...) client_crud.go single-client CRUD + validation + protocol defaults client_inbound_apply.go low-level inbound-settings mutators + by-email setters client_bulk.go bulk attach/detach/adjust/delete/create + DelDepleted client_traffic.go traffic-reset paths client_groups.go client group management client_paging.go paged listing, filtering, sorting, summary Every declaration moved unchanged (verified: identical func/type/const/var signature set before vs after). Imports redistributed per file via goimports. go build ./..., go vet, and go test ./web/service/... all pass. * refactor(service): split inbound.go into focused files inbound.go was 4100 lines. Split it verbatim into cohesive same-package files (no behavior change): inbound.go core inbound CRUD + InboundService (keeps pkg doc) inbound_protocol.go protocol / stream capability helpers inbound_node.go node/runtime/remote coordination + online tracking inbound_traffic.go traffic accounting, reset, client stats inbound_client_ips.go per-client IP tracking inbound_clients.go client lookups within inbounds + copy-clients inbound_disable.go auto-disable invalid inbounds/clients inbound_migration.go DB migrations inbound_sublink.go subscription link providers inbound_util.go generic slice/string helpers Identical func/type/const/var signature set before vs after; package doc comment preserved on inbound.go. Imports redistributed via goimports. Build, vet, and go test ./web/service/... all pass. * refactor(service): split tgbot.go into focused files tgbot.go was 3738 lines dominated by a 1246-line answerCallback. Split it verbatim into cohesive same-package files (no behavior change): tgbot.go lifecycle, bot setup, caches, small utils tgbot_router.go incoming update / command / callback dispatch tgbot_send.go outbound messaging primitives tgbot_client.go client views, actions, subscription links tgbot_inbound.go inbound listing / pickers tgbot_report.go server usage, exhausted, online, backups, notifications Identical func/type/const/var signature set before vs after. Imports redistributed via goimports. Build, vet, and go test ./web/service/... pass. * refactor(client): dedupe single-field by-email setters ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, and ResetClientTrafficLimitByEmail shared an identical ~50-line body that resolves the inbound by email, confirms the client exists, rewrites a single-client settings payload, and delegates to UpdateInboundClient. Extract that into applyClientFieldByEmail(inboundSvc, email, mutate) and reduce each setter to a 3-line wrapper. Behavior is unchanged: same checks and error strings, same single-client payload contract, same totalGB guard. SetClientTelegramUserID (resolves by traffic id, different error text) and ToggleClientEnableByEmail/SetClientEnableByEmail (different return shape and a pre-read of the old state) intentionally keep their own bodies. * refactor(service): extract panel/ subpackage Move the panel-administration leaf services out of the flat service package into web/service/panel/ (package panel): user.go UserService (auth / 2FA / LDAP) panel.go PanelService (restart / self-update) + version helpers panel_other.go non-unix RestartPanel panel_unix.go unix RestartPanel api_token.go ApiTokenService websocket.go WebSocketService panel_test.go version/shellQuote unit tests These are leaves: they depend on core (SettingService, Release) but no core file references them, so the extraction creates no import cycle. Core references are now qualified (service.SettingService, service.Release); callers in main.go, web/web.go, and web/controller/* updated to panel.*. Build, vet, and go test ./web/... pass. * refactor(service): extract integration/ subpackage Move the external-provider integration leaves into web/service/integration/ (package integration): warp.go WarpService (Cloudflare WARP) nord.go NordService (NordVPN) custom_geo.go CustomGeoService (custom geo asset management) *_test.go custom_geo / panel-proxy tests These depend on core (SettingService, ServerService, XraySettingService) but no core file references them. xray_setting.go stays in core because it calls the unexported SettingService.saveSetting. The shared isBlockedIP SSRF helper (used by core url_safety.go and by custom_geo) now has a small copy in each package rather than being exported. Core references qualified; callers in web/web.go, web/job/*, and web/controller/* updated to integration.*. Build, vet, and go test ./web/... pass. * refactor(service): extract tgbot/ subpackage Move the Telegram bot (6 files + test) into web/service/tgbot/ (package tgbot). It is a leaf: it embeds five core services (Inbound/Client/Setting/ Server/Xray) and the core never references it, so no import cycle. To support the package boundary without changing behavior: - core exposes XrayProcess() *xray.Process so tgbot keeps calling the exact same running-process methods it used via the package-level `p`; - three core methods tgbot calls are exported: ClientService.checkIs- EnabledByEmail -> CheckIsEnabledByEmail, InboundService.getAllEmails -> GetAllEmails (callers updated in-package); - tgbot's embedded-field types and the few core type refs (Status, ClientCreatePayload, SanitizePublicHTTPURL) are now service-qualified. Callers in main.go, web/web.go, web/job/*, and web/controller/* updated to tgbot.*. Build, vet, and go test ./web/... pass. * refactor(service): extract outbound/ subpackage OutboundService (outbound.go) imports only neutral packages (config, database, model, xray) and its production code is referenced by no core or sibling service file — only by web/controller/xray_setting.go and web/job/xray_traffic_job.go. Move it to web/service/outbound/ (package outbound); no core qualification needed inside. Callers updated to outbound.*. The one coupling was a tiny pure test helper, outboundsContainTag, used by both outbound.go and the core outbound_subscription_test.go; it now has a small copy in that test file rather than being shared across the boundary. Build, vet, and go test ./web/... pass. * refactor(util): move wireguard into its own subpackage util/wireguard.go was the lone file of the root `util` package (24 lines, one exported func GenerateWireguardKeypair), while every other util concern lives in a focused subpackage (util/common, util/crypto, util/netsafe, ...). Move it to util/wireguard/ (package wireguard) for consistency; its only importer, web/service/integration/warp.go, is updated. The root `util` package no longer exists. * refactor(sub): drop redundant sub prefix from filenames Inside package sub the subXxx.go prefix just repeats the package name (like client_*.go did inside service). Rename for consistency; content and type names are unchanged: subController.go -> controller.go subService.go -> service.go subClashService.go -> clash_service.go subJsonService.go -> json_service.go (+ matching _test.go files) * refactor(controller): rename xui.go -> spa.go XUIController serves the panel's single-page-app shell; spa.go names that role plainly (the other controller files are domain-named). File rename only — the type stays XUIController. api_docs_test.go keys route base paths by filename, so its "xui.go" case is updated to "spa.go". * refactor: move backend packages under internal/ Adopt the idiomatic Go application layout: the backend packages now live under internal/ (a boundary the toolchain enforces), signalling private implementation instead of a library-style flat root. No runtime behavior changes — only import paths and a few build/config paths move. Moved: config, database, logger, mtproto, sub, util, web, xray -> internal/. main.go stays at the repo root and tools/openapigen stays under tools/ (both still import internal/* because the internal rule keys off the module root). The module path github.com/mhsanaei/3x-ui/v3 is unchanged; 149 .go files had their import prefix rewritten to .../internal/<pkg>. Couplings the Go compiler can't see, updated to the new layout: - frontend i18n imports of web/translation (react.ts, setup.components.ts) - vite outDir + eslint/tsconfig ignore globs -> internal/web/dist - Dockerfile COPY paths for web/dist and web/translation - locale.go os.DirFS("web") disk fallback -> "internal/web" - .gitignore and ci.yml go:embed stub for internal/web/dist - api_docs_test.go repo-root relative walk (one level deeper) - tools/openapigen filesystem package paths; ApiTokenView repointed to the web/service/panel subpackage and codegen regenerated (clears a stale type the ci.yml codegen check was failing on) Verified: go build/vet/test (all packages), and frontend typecheck, lint, vitest (478 tests), and production build into internal/web/dist. * fix(config): keep test runs from writing logs into the source tree GetLogFolder() returns a CWD-relative "./log" on Windows. Under `go test` the working directory is each package's own folder, so InitLogger (called by tests in web/job, web/service, xray, web/websocket) created stray log/ directories scattered through the source tree (e.g. internal/web/job/log/). Redirect to a shared temp folder when testing.Testing() reports a test run. Production behavior is unchanged: Windows still uses ./log next to the binary and Linux /var/log/x-ui. The log files were always gitignored (*.log) and never committed; this just stops the noise at the source. * docs: move subscription-template guide out of root into docs/ sub_templates/ was a top-level folder holding only a README and no actual templates (3x-ui ships none by design), referenced nowhere and unlinked from any doc — it read like an empty placeholder cluttering the repo root. Move the guide to docs/custom-subscription-templates.md (a proper docs home), reword its intro to read as documentation rather than a folder note, link it from the Features list in README.md, and drop the empty sub_templates/ folder. * fix: update stale web/ path references after the internal/ move The internal/ migration rewrote Go import paths but left some references to the old top-level layout in docs, comments, and a few runtime disk paths. Functional (dev-mode only): the disk-serving fallbacks that read the Vite build from disk when running from source still pointed at web/dist/, which moved to internal/web/dist/ — so `os.DirFS`/`os.Stat`/`os.ReadFile` in internal/web/web.go and internal/sub/{sub,controller}.go are corrected. Production was unaffected (it serves the embedded FS; verified by the Docker build), but `go run` with a live frontend build silently fell back to embed. Docs/comments: frontend/README.md, CONTRIBUTING.md, the claude-issue-bot and release workflows, the openapigen -root help text, and assorted Go comments now reference internal/web, internal/database, internal/sub, internal/xray, etc. Package-name mentions (the "web" package), root paths (main.go, frontend/, install scripts, /etc/x-ui), routes (/panel/api/xray), and the historical "web/assets no longer exists" note were intentionally left as-is. * refactor(web): remove the legacy /xui -> /panel redirect middleware RedirectMiddleware existed only for backward compatibility with the old `/xui` URL scheme (301-redirecting /xui and /xui/API to /panel and /panel/api). That cutover was long ago, so drop the middleware, its registration in initRouter, and the now-inaccurate "URL redirection" mention in the middleware package doc. Old /xui URLs now 404 like any other unknown path. HTTPS auto-redirect and auth redirects are unrelated and stay. * build: fix .dockerignore for internal/ layout and exclude runtime dir - web/dist -> internal/web/dist: the embedded frontend moved under internal/, so the stale exclude no longer matched and the locally-built dist could be sent to the build context (the frontend stage rebuilds it fresh anyway). - exclude x-ui/: the local runtime directory (SQLite db, geo .dat files, xray binaries, certs — ~150MB) was being shipped into the build context for no reason. Verified the pattern excludes only the directory and still keeps x-ui.sh, which the Dockerfile copies to /usr/bin/x-ui.
2026-06-10 06:19:22 -07:00
}
2025-09-20 00:35:50 -07:00
// GetXrayErr returns the error from the Xray process, if any.
2023-02-09 11:18:06 -08:00
func (s *XrayService) GetXrayErr() error {
process := currentXrayProcess()
if process == nil {
2023-02-09 11:18:06 -08:00
return nil
}
err := process.GetErr()
if err == nil {
return nil
}
if runtime.GOOS == "windows" && err.Error() == "exit status 1" {
// exit status 1 on Windows means that Xray process was killed
// as we kill process to stop in on Windows, this is not an error
return nil
}
return err
2023-02-09 11:18:06 -08:00
}
fix(xray): refuse a config the running core cannot bind (#6547) * fix(xray): refuse a config the running core cannot bind RestartXray stopped a working core before handing it a config whose listens collide, so the failed bind exited the whole process (main/run.go:94) and the one-second watchdog retried it in a loop: every protocol down, cause only in the logs. The save-time port guards cannot cover this -- SetInboundEnable, the AmneziaWG relay created on the first peer, template and bridge edits all reach a colliding config with no guard on that path. Probe the generated config at the single restart funnel instead. Collisions the running core already serves are excused, so an established setup is never refused by a static read being wrong about it, and the port-bucketed pass costs nothing on a clean config. * fix(xray): surface a refused config and re-key the bind excuse set Round-1 findings on this PR. Refusing the swap left the running core on its previous config with nothing but a log line to show for it, so the status response now carries the reason while the core runs and the overview marks it; the node list picks the same field up through that response. The excuse set is keyed on the two listens, the port and the shared transports instead of the tag pair, so a pair whose listen moves onto the other's address is refused again, while the same two sockets stay excused however the generator orders them. TestBindConflicts/excused_pair_whose_listen_changed_into_a_real_collision fails without the key change -- watched red first.
2026-09-15 05:27:28 -07:00
// GetHeldBackConfig returns why the running core still serves its previous
// config, or "" when the pending config was applied.
func (s *XrayService) GetHeldBackConfig() string {
return xrayState.heldBackReason()
}
2025-09-20 00:35:50 -07:00
// GetXrayResult returns the result string from the Xray process.
2023-02-09 11:18:06 -08:00
func (s *XrayService) GetXrayResult() string {
process, cachedResult := xrayState.snapshot()
if cachedResult != "" {
return cachedResult
2023-02-09 11:18:06 -08:00
}
if process == nil || process.IsRunning() {
2023-02-09 11:18:06 -08:00
return ""
}
result := process.GetResult()
if runtime.GOOS == "windows" && result == "exit status 1" {
// exit status 1 on Windows means that Xray process was killed
// as we kill process to stop in on Windows, this is not an error
return ""
}
xrayState.storeResult(process, result)
2023-02-09 11:18:06 -08:00
return result
}
2025-09-20 00:35:50 -07:00
// GetXrayVersion returns the version of the running Xray process.
2023-02-09 11:18:06 -08:00
func (s *XrayService) GetXrayVersion() string {
process := currentXrayProcess()
if process == nil {
2023-02-09 11:18:06 -08:00
return "Unknown"
}
return process.GetXrayVersion()
2023-02-09 11:18:06 -08:00
}
2023-02-18 04:37:32 -08:00
2025-09-20 00:35:50 -07:00
// RemoveIndex removes an element at the specified index from a slice.
// Returns a new slice with the element removed.
func RemoveIndex(s []any, index int) []any {
2023-02-09 11:18:06 -08:00
return append(s[:index], s[index+1:]...)
}
2025-09-20 00:35:50 -07:00
// GetXrayConfig retrieves and builds the Xray configuration from settings and inbounds.
2023-02-09 11:18:06 -08:00
func (s *XrayService) GetXrayConfig() (*xray.Config, error) {
templateConfig, err := s.settingService.GetXrayConfigTemplate()
if err != nil {
return nil, err
}
xrayConfig := &xray.Config{}
err = json.Unmarshal([]byte(templateConfig), xrayConfig)
if err != nil {
return nil, err
}
xrayConfig.LogConfig = resolveXrayLogPaths(xrayConfig.LogConfig)
xrayConfig.API = ensureAPIServices(xrayConfig.API)
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
xrayConfig.Policy = ensureStatsPolicy(xrayConfig.Policy)
xrayConfig.RouterConfig = stripDisabledRules(xrayConfig.RouterConfig)
// A pasted or restored template can carry pre-#6258 XHTTP keys, a pre-26.10.10 xdns mask or a
// pre-26.9.9 quicParams.udpHop, which the core ignores, or a udphop not last, which it cannot dial.
feat(xray): update xray-core to v26.10.10 and adapt panel Move the three Xray binary pins (DockerInit.sh, both release.yml fetches) to v26.10.10, in lockstep with the xtls/xray-core module that 82c77111 already bumped to 701af60772cd. xdns finalmask (#7090): a domain now takes a "names" list instead of a single "name", and a resolver is an "addrs" list of udp:// or tcp:// URLs instead of {type, settings:{addr}}. The loader ignores the old keys rather than failing, so every mask the 26.9.30 seeder wrote would load with no domain and no resolver and the tunnel would carry nothing. maskcompat.UpgradeLegacyXdns now lifts both the pre-26.9.30 string lists and the 26.9.30 objects straight to names/addrs, dropping entries the old loader refused. A new seeder (XdnsFinalmaskNamesAddrsFix) reruns it over every stored finalmask, and the save path, GetXrayConfig, both link importers and the mask editor's mount lift go through it. Template outbounds now get the build-time heal too: a 26.9.30 template pasted after the seeder passes the save check, since the core loads that shape without error. The editor edits names and addrs as tag lists and no longer requires record types, which the core now defaults (TXT on a client, A/CNAME/TXT/AAAA on a server). The wire format did not change, but clients on an older core cannot read the new config shape. Lua scripts (#6823): dns.script, routing.script and dns servers[].id are new. The xray page parsed the dns block with a strict schema, so loading it dropped script and id and the next save deleted them; the schema now keeps both and the DNS server modal carries the id through an edit. The hot routing apply built routing in-process, and that build now resolves the script file from the panel's working directory, so a script beside the xray binary failed every rule change into a full restart and broke the node bridge. RoutingService cannot change the script anyway, so the panel no longer resolves it there. TLS client roots (#7105): the core now verifies against its bundled Mozilla roots instead of the OS store unless tlsSettings.useSystemCA is set. The outbound TLS form gets a switch for it, so an outbound to a server signed by a privately installed CA can opt back in.
2026-10-10 22:55:12 -07:00
xrayConfig.OutboundConfigs = healOutboundStreams(xrayConfig.OutboundConfigs,
liftXhttpSessionIDKeys, upgradeStreamLegacyXdns, upgradeStreamLegacyUDPHop, normalizeStreamUDPHop)
feat(xray): update xray-core to v26.10.21 and adapt panel Bump xtls/xray-core to 1b1647cd9ddf (v26.10.21) and move the three binary pins (DockerInit.sh, both release.yml fetches) in lockstep. The v26.10.10 release was pulled upstream, so the old pins already 404. XDRIVE (#7182): the core now runs XDRIVE only for VLESS inside VLESS encryption, with no flow, and an outbound only over Mux; it refuses the whole config otherwise. The panel offers XDRIVE only for VLESS, refuses to save an inbound without decryption, never emits a client flow on an XDRIVE stream (config, JSON subscription, forms), and turns Mux on when an outbound picks XDRIVE. An inbound or template outbound stored before the upgrade that the core would refuse is left out of the generated config with a warning, so it no longer stops every other inbound. The service "Google Drive" is now "google-drive" (old name read as a deprecated alias); saves, the schema and both form adapters store the new name. XDRIVE and MASQUE gain a client xmux; both schemas keep it instead of stripping it on an inbound save. Lua scripts (#7124, #7136): the core refuses a routing script beside rules or domainStrategy, and a dns script beside the fallback/parallel switches or a domain-scoped server. The panel always keeps its api routing rule, so a template save now refuses any routing script and a dns script in a refused combination, and a stored one is dropped from the generated config so the core still starts. Noise masks gain a Lua script that the loader compiles inside Build: the in-process build (template save check, hot add) now resolves a relative script the way the xray child does, beside the xray binary, instead of failing it. The mask editor gets a Script field that replaces the noise list. XDNS (#7140): a resolver takes https:// DoH addresses with worker, tlsConfig and quicConfig. The mask editor registered only addrs, so a save dropped the rest; it now edits all four. Finalmask filter (#7206): a new TCP/UDP mask screens an inbound's source IPs. The inbound mask editor offers it with its allow and block lists; it is a no-op on a dialer and an older client core refuses the type, so the share link and JSON subscription drop it like the Go link already does.
2026-10-22 08:10:11 -07:00
xrayConfig.OutboundConfigs = dropXDriveOutboundsRefusedByCore(xrayConfig.OutboundConfigs)
feat(amneziawg): add AmneziaWG as an outbound protocol (#6320) * feat(amneziawg): add AmneziaWG as an outbound protocol - AmneziaWG outbound protocol end-to-end: config schema, socks bridge, netstack, panel UI - Route amneziawg outbounds to HTTP probe in TCP mode (backend + frontend classifiers) with pinning test - Add 2-minute idle read deadline to pumpUDPEgress to reap idle egress sessions - Require SOCKS5 username/password auth on the egress server (reject NO-AUTH with 0xFF) with test - Bound the egress TCP tunnel dial with portForwardDialTimeout (10s), matching portfwd.go - Resolve UDP domain targets off the association's reader loop via deliverUDPDatagram; race-safe getOrDial starts the reply pump at session creation; client passed by value into resolver goroutines (pinned by TestEgressUDPDatagramDomainInterleavedClients) - Reconcile early-returns on an empty desired set and closes the egress listener; EgressBasePort (64900) is reserved against local inbound port conflicts like the internal API port, with pinning tests for both the port reservation (TestCheckPortConflict_EgressPortBlockedLocal) and the Reconcile empty-desired Close/Listen lifecycle (TestOutboundManagerReconcileEmptyDesiredClosesEgress) - Eliminate acceptLoop shutdown race by validating listener != nil and registering to tracked under s.mu before wg.Add; bound pre-auth handshake with deadline (pinned by TestEgressServerCloseDuringConcurrentAccepts) - Support AAAA and dual-stack domain resolution in tunnel DNS resolver with v6 default fallback (DefaultTunnelDNSServerV6); add DNS field to frontend protocol form; avoid unneeded cache flushes on unchanged SetStack ticks * fix(amneziawg): resolve IPv6-only DNS default fallback and validate required keys - Default to IPv6 tunnel DNS on IPv6-only outbounds with blank dns - Require non-empty secretKey and peer publicKey in ValidateAmneziaWGOutbound - Add end-to-end IPv6 tunnel domain resolution test and test empty key rejection - Trim comment blocks exceeding 2 lines across modified files - Fix Storybook test execution on environments with POSIX locale Co-Authored-By: Claude Code <noreply@anthropic.com> --------- Co-authored-by: rqzbeh <rqzbeh@users.noreply.github.com> Co-authored-by: Claude Code <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-09-10 05:50:48 -07:00
// Bridge amneziawg outbounds before anything else reads OutboundConfigs;
// the core has no amneziawg proxy and would reject the raw entry.
if err := transformAmneziaWGOutbounds(xrayConfig); err != nil {
return nil, err
}
if xray.KernelTunBlocked() {
xrayConfig.OutboundConfigs = forceUserspaceWireGuardTun(xrayConfig.OutboundConfigs)
}
2023-02-09 11:18:06 -08:00
_, _, _ = s.inboundService.AddTraffic(nil, nil)
2023-02-09 11:18:06 -08:00
inbounds, err := s.inboundService.GetAllInbounds()
if err != nil {
return nil, err
}
excludedClients, excludedInbounds := 0, 0
2023-02-09 11:18:06 -08:00
for _, inbound := range inbounds {
if !inbound.Enable {
continue
}
if inbound.NodeID != nil {
continue
}
Feature/tuic v5 (#6337) * Feat(tuic): Implement native TUIC v5 protocol support via Rust sidecar daemon - Add internal/tuic package for official tuic-server sidecar lifecycle management, configuration generation, and graceful process control - Bridge decrypted TUIC QUIC traffic into loopback Xray SOCKS5 inbounds (63200+id) for traffic accounting, statistics, and routing rules - Implement periodic reconciliation job (cadence @every 10s) and immediate runtime synchronization on inbound/client mutations - Add TUIC inbound & multi-user client settings (UUID + Password authentication) in Web UI with SNI auto-fill and panel certificate loader - Integrate tuic:// subscription links and Clash.Meta (Mihomo) proxy generation for TUIC - Update install.sh to automatically download and install official tuic-server release for x86_64, aarch64, and armv7 - Add full localization for TUIC protocol across all 13 supported languages * Feat(install): Support custom repository and branch in install and update scripts * Ci(release): Enable publish-dev for feature branch and workflow dispatch * Feat(sub): Add TUIC to subscription resolution and client QR config generator - Add 'tuic' to getInboundsBySubId SQL allowlist to resolve TUIC inbounds in subscriptions and sub links - Enhance buildTuicProxy in Clash subscription generator with robust host and credentials resolution - Add tuicConfig.ts to generate standalone Clash/Mihomo YAML configuration - Add dedicated TUIC Config tab in ClientQrModal with QR code and .yaml download button - Add localization keys for TUIC config across all 13 supported languages * Fix(tuic): Exclude TUIC from native Xray inbounds and strip udp_relay_mode from server config - Exclude model.TUIC from native Xray inbounds in GetXrayConfig to prevent Xray startup failure - Remove udp_relay_mode from tuic-server JSON configuration builder - Update install.sh to install tuic-server binary to both xui_folder/bin and /usr/local/bin * Fix(install): Fallback to dev-latest when releases/latest is not present on fork * Feat(tuic): Add real-time online status and LastOnline tracking for TUIC clients - Track client activity by mapping client UUID in tuic-server logs to email - Integrate TUIC active clients into XrayTrafficJob to refresh local online clients - Bump LastOnline timestamp in database and broadcast live online status over WebSocket * Feat(tuic): Implement real-time traffic statistics and live speed reporting for TUIC - Collect precise I/O traffic deltas for tuic-server child processes via /proc/<pid>/io - Aggregate and attribute TUIC traffic deltas per client in tuic Manager - Integrate TUIC traffic deltas into XrayTrafficJob to update database and broadcast live speed * Feat(tuic): Finalize TUIC v5 integration with 1:1 traffic counting and orphan process cleanup - Use exact 1:1 byte delta accounting from /proc/<pid>/io - Add killStrayTuicProcesses to terminate orphan sidecars on panel startup - Fully integrate TUIC with subscriptions, live speed meter, and all 13 locales * Feat(frontend): Polish TUIC UI, support bulk operations, and update translations - Align TUIC inbound certificate form with standard 3X-UI layout (Set Default Cert, Clear) - Remove extra subtitle hint text from TUIC inbound form fields - Support TUIC in client bulk attach/detach and bulk add modals - Add TUIC badge color to client info modal, clients table, and host list - Update password tooltip across all 13 locales to include TUIC - Remove obsolete dead translation keys across all 13 locales * Chore(ci): Finalize TUIC v5 bundling across release workflow, Docker, and scripts * Feat(openapi): Update OpenAPI generator and schemas for TUIC types * Fix(backend): Address core review findings for TUIC types, port checks, and xray bridge * Refactor(traffic): Isolate proc reading with build tags and decouple TUIC metering into TuicJob * Feat(client): Add TuicServer to InboundOption, fix config export and clean share links * Fix(frontend): Register TUIC in multi-user helpers, tracked protocols, and tag derivation * Chore(openapi): Re-generate OpenAPI specification and sync Zod schemas * Chore(scripts): Add Alpine musl binaries, 386 and Windows packaging, and anchor pkill * Fix(review): Remove stale import, correct binary names, switch to musl, and drop unreachable relay gate * Feat(frontend): Show share link in Inbound Info and display UDP tag for TUIC * Docs: Add TUIC v5 configuration guide and link specifications * Docs(tuic): Correct Clash Meta configuration parameter to reduce-rtt * Fix(tuic): Generate client credentials on copy, enforce ID/password validation, and add i386 to DockerInit * Fix(tuic): drop unused relay, fix traffic accounting, and honor host endpoints - Drop unused loopback SOCKS relay and eliminate port collision with AmneziaWG - Correct inbound traffic calculation without double-counting - Drop heuristic client traffic division while retaining online tracking - Support externalProxy host fan-out and conditional parameters in share links - Scope orphan process termination to managed config directory * Fix(tuic): enforce client quotas, decouple Xray restart, and sync openapi schemas - Regenerate OpenAPI, Zod schemas, and TypeScript types without route_through_xray - Populate clientTraffics in TuicJob to enforce client quotas and first-use expiry - Split process I/O delta into up and down in Process.CollectTraffic - Remove SetNeedRestart from updateTuicInbound to prevent Xray session drops - Use InstanceFromInbound for default ALPN and UDP relay mode in tuic:// share links - Support allow_insecure on externalProxy host endpoints without parameter collision * Fix(tuic): attribute client traffic only on single-user inbounds and sync link defaults - Attribute I/O deltas to the client only when the inbound has exactly one configured client, avoiding false billing and disablings on multi-user inbounds - Aggregate client traffic by email in TuicJob so clients on multiple inbounds don't lose deltas - Match frontend genTuicLink defaults for alpn and udp_relay_mode with backend subscription links * Fix(tuic): gate client traffic by total sidecar clients and require client email * Fix(tuic): enforce inbound-only traffic limits and disable client totalGB * fix(tuic): restore delayed start, remove client totalGB rejection, and document linux-only limits * fix(tuic): anchor pkill, fix io baseline/split, escape yaml, and deduplicate start errors * fix(tuic): prevent traffic double-counting, ensure info log level for delayed start, and broaden pkill matching * fix(tuic): address review round 11 findings - internal/sub/json_service: skip tuic protocol in json subscription to prevent direct routing leak - internal/sub/clash_service: honor externalProxy/host row allowInsecure, sni, and alpn in buildTuicProxy - internal/web/runtime: decouple tuic inbound add/delete from xray restart - internal/tuic/config: restore user log-level options (warn, error) without forced info clamp - frontend/src/lib/xray/inbound-link: fix duplicate remark suffix and apply externalProxy TLS overrides - frontend/src/schemas/protocols/stream/external-proxy: propagate allowInsecure through host mapping - tests: add coverage for json sub skip, clash proxy overrides, and link generation * fix(tuic): meter inbound traffic through a UDP relay and bracket IPv6 binds Review repairs on the TUIC v5 sidecar integration: - Inbound traffic was read from the sidecar's /proc/<pid>/io rchar, but the kernel only counts read()/write() there and tuic-server moves its sockets with recvfrom/recvmmsg/sendmmsg/sendto, so an inbound's up/down stayed at 0 forever and inbound total limits never tripped (measured: 12 MiB relayed, rchar delta 0). The panel now owns the inbound's public UDP port with a small relay and runs tuic-server behind it on a loopback port, counting up/down exactly on every OS. tuic-server therefore logs 127.0.0.1 as every client's address; per-client attribution stays unsupported since QUIC is opaque. - Instance.BindTo formatted an IPv6 listen address as ":::8443", which tuic-server rejects with "invalid socket address syntax", so an inbound listening on "::" or any IPv6 literal never started. It now uses net.JoinHostPort; IPv4 output is unchanged. - The log level is passed to the sidecar as chosen. Online status, last-online and delayed start are read from its Info lines, so the Log Level field now says that Warn and Error switch them off for the inbound, and the docs say the same. - Drop two frontend tests that only exercised a getter and a set lookup, and strip the trailing blank line that made gofumpt fail on two of the new Go test files. * fix(tuic): harden tag updates, runtime routing, and relay stability --------- Co-authored-by: poise52 <equipoise52@gmail.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-09-12 01:15:48 -07:00
if inbound.Protocol == model.MTProto || inbound.Protocol == model.AmneziaWG || inbound.Protocol == model.TUIC {
feat(mtproto): add MTProto (FakeTLS) protocol via managed mtg sidecar (#5076) * feat(mtproto): add MTProto (FakeTLS) protocol via managed mtg sidecar Xray-core has no mtproto proxy, so mtproto inbounds run as standalone mtg (9seconds/mtg) sidecar processes managed by the panel — one per inbound — and are excluded from the generated Xray config entirely. - model: MTProto protocol constant, validator, and FakeTLS secret helpers (GenerateFakeTLSSecret/HealMtprotoSecret) - mtproto package: per-inbound mtg process manager with reconcile, graceful stop, and best-effort Prometheus traffic scraping - runtime: delegate mtproto inbounds to the mtg manager instead of the Xray gRPC API; skip mtproto when building the Xray config - web: boot reconcile + StopAll wiring, periodic reconcile/traffic job, port-conflict transport, secret healing on inbound add/update - sub: tg:// proxy share-link generation - frontend: protocol option, Zod schema, Protocol tab (FakeTLS domain + regenerable secret), info-modal link, and i18n - provisioning: fetch mtg v2.2.8 in install.sh, DockerInit.sh, and the Linux + Windows release workflows * fix * fix * fix: address Copilot review comments on mtproto PR - web/web.go: create NewMtprotoJob once and reuse for cron + initial run - mtproto/manager.go: StopAll cleans up per-inbound config files on shutdown - mtproto/manager.go: CollectTraffic releases mutex before HTTP scrapes to avoid blocking Ensure/Reconcile/Remove during network I/O - database/model/model.go: panic on crypto/rand failure in mtprotoRandomMiddle instead of silently producing a weak all-zero secret - install.sh: fix chmod to handle renamed bin/mtg-linux-arm on armv5/v6/v7
2026-06-08 05:28:19 -07:00
continue
}
if err := normalizeVlessFallbacks(inbound); err != nil {
return nil, err
}
if err := validateVlessManagedFallbacks(nil, inbound); err != nil {
return nil, err
}
settings := map[string]any{}
_ = json.Unmarshal([]byte(inbound.Settings), &settings)
flowAllowed := !inbound.DisableFlow &&
inboundCanEnableTlsFlow(string(inbound.Protocol), inbound.StreamSettings, inbound.Settings)
var wireguardClientsByEmail map[string]model.Client
if inbound.Protocol == model.WireGuard {
inboundClients, _ := ParseInboundSettingsClients(inbound.Settings)
if len(inboundClients) > 0 {
wireguardClientsByEmail = make(map[string]model.Client, len(inboundClients))
for _, client := range inboundClients {
wireguardClientsByEmail[strings.ToLower(strings.TrimSpace(client.Email))] = client
}
}
}
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
dbClients, listErr := s.inboundService.clientService.ListForInbound(nil, inbound.Id)
if listErr != nil {
return nil, listErr
}
clientStats := inbound.ClientStats
enableMap := make(map[string]bool, len(clientStats))
for _, clientTraffic := range clientStats {
enableMap[clientTraffic.Email] = clientTraffic.Enable
}
finalClients := make([]any, 0, len(dbClients))
var wgPeers []any
excludedBefore := excludedClients
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
for i := range dbClients {
c := dbClients[i]
if enable, exists := enableMap[c.Email]; exists && !enable {
excludedClients++
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
continue
2023-02-09 11:18:06 -08:00
}
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
if !c.Enable {
continue
}
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
flow := c.Flow
if flow == "xtls-rprx-vision-udp443" {
flow = "xtls-rprx-vision"
}
if !flowAllowed {
feat(inbound): DisableFlow — opt an inbound out of auto XTLS Vision (#5689) (#5698) * feat(inbound): add DisableFlow to opt an inbound out of auto XTLS Vision Adds an inbound-level DisableFlow flag so operators can suppress automatic xtls-rprx-vision injection on a specific inbound even when its transport is flow-capable — e.g. a tunneled/CDN-fronted XHTTP+vlessenc inbound where Vision is not wanted, while keeping it on the same client's Reality inbounds. When set, the inbound reports tlsFlowCapable=false, the write path clamps each attached client's flow to empty (so flow_override stores ""), and share links/subscriptions never carry the flow for it. The flag is panel-only metadata and is never sent to xray. Closes part of #5689. * feat(inbound): DisableFlow toggle in the inbound form (frontend) Wire the DisableFlow field through the form schema + adapters and add a VLESS-gated switch in the inbound form, plus en-US strings. tsc --noEmit and eslint pass. * fix(inbound): honor DisableFlow in all emitters + on toggle; regen OpenAPI Addresses review on #5690: - Clash (clash_service.go) and JSON (json_service.go) subscription emitters now also skip the flow for a DisableFlow inbound — previously only the raw share-link path was gated, so those two still advertised it (blocking 1). - UpdateInbound now strips any flow already stored on a DisableFlow inbound's clients (settings.clients[].flow + client_inbounds.flow_override) so xray and the subscription agree; otherwise toggling DisableFlow on an existing Vision client left xray expecting a flow the client no longer sends. - Regenerated the OpenAPI + zod/types/examples artifacts for the new field and added an example tag (blocking 2; make gen-check is clean). - Added Clash + JSON DisableFlow suppression tests alongside the raw-link one. * fix(inbound): make DisableFlow durable, clamp on create, guard live config Addresses the review + completeness audit on #5690: - UpdateInbound now persists inbound.DisableFlow onto the saved row. It was only read to branch strip-vs-restore, so toggling the flag on an existing inbound never stuck and MigrationRestoreVisionFlow re-injected the flow — the exact #5689 path (editing a multi-inbound client's inbound) self-reverted. - DBInbound (frontend) declares + initializes disableFlow so ObjectUtil .cloneProps carries the API value through; the edit Switch previously always read false and re-saving silently reverted the opt-out. - AddInbound strips client flow (settings + parsed clients) when DisableFlow is set, so a created-disabled inbound never persists a flow xray would expect. - GetXrayConfig forces flow="" for DisableFlow inbounds (VLESS + Trojan) as defense-in-depth, keeping the live config and the subscription in agreement. - genTrojanLink share link honors DisableFlow too. - Drop the dead explicit flow_override clear in UpdateInbound (SyncInbound rebuilds it from the stripped settings). - Clear disableFlow in the inbound form when switching to a non-VLESS protocol. - Add disableFlow/disableFlowHelp to the remaining 12 locales. Tests: stripClientFlows unit cases; DB-backed AddInbound clamp; UpdateInbound persist+strip+resist-restore regression (fails without the persist fix); frontend DBInbound + adapter round-trip (fails without the model field). * style(inbound): drop // line comments per repo CLAUDE.md The DisableFlow work followed the surrounding code's commenting style; the repo CLAUDE.md forbids // line comments in committed Go/TS. Remove the comments I added (Go + frontend + tests) and regenerate OpenAPI/schemas, which drops the generated field descriptions sourced from the Go doc comments. No behavior change; full go test (service+sub, CGO) + frontend typecheck/vitest green; golangci-lint clean on the changed files. * fix(runtime): propagate disableFlow to nodes Preserve the inbound DisableFlow flag when syncing inbounds across nodes and when recreating central records from remote traffic snapshots. This keeps multi-node deployments from reintroducing VLESS Vision flow in node configs and share links, and updates the related tests to cover the wired field and VLESS JSON generation.
2026-08-15 14:09:16 -07:00
flow = ""
}
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
entry := map[string]any{"email": c.Email}
switch inbound.Protocol {
case model.VLESS:
if c.ID != "" {
entry["id"] = c.ID
}
if flow != "" {
entry["flow"] = flow
}
if c.Reverse != nil {
entry["reverse"] = c.Reverse
}
case model.VMESS:
if c.ID != "" {
entry["id"] = c.ID
}
if c.Security != "" {
entry["security"] = c.Security
}
case model.Trojan:
if c.Password != "" {
entry["password"] = c.Password
}
if flow != "" {
entry["flow"] = flow
}
case model.Shadowsocks:
if c.Password != "" {
entry["password"] = c.Password
}
feat: complete Zod migration of frontend + bulk client batching (#4599) * feat(frontend): add Zod runtime validation at API boundary Introduces Zod 4 schemas for response validation on the three highest-traffic endpoints (server/status, nodes/list, setting/all) and a Zod->AntD form rule adapter, replacing the duplicated per-file ApiMsg<T> interfaces. Validation runs safeParse with console.warn + raw-payload fallback so backend drift never breaks the UI for users. Login form switches to schema-driven rules as the proof-of-life for the adapter. Class-based models stay untouched; remaining query/mutation hooks and form modals will migrate in follow-ups. * feat(frontend): extend Zod validation to remaining query/mutation hooks Adds Zod schemas for client/inbound/xray/node-probe endpoints and wires useNodeMutations, useClients, useInbounds, useXraySetting, useDatepicker through parseMsg. Drops the duplicated per-file ApiMsg<T> interfaces and the local ClientRecord / OutboundTrafficRow / XraySettingsValue / DefaultsPayload declarations in favour of schema-inferred types re-exported from the new src/schemas/ modules. API boundary now validates: clients list/paged, clients onlines, clients lastOnline, clients get/hydrate, inbounds slim, inbounds get, inbounds options, defaultSettings, xray config, xray outbounds traffic, xray testOutbound, xray getXrayResult, getDefaultJsonConfig, nodes probe, nodes test. Mutation responses that consume obj (bulkAdjust, delDepleted, nodes probe / test) get response validation; pass-through mutations stay agnostic. NodeFormModal type-aligned to Msg<ProbeResult>. * fix(frontend): allow null slices in client/summary schemas Go's encoding/json emits nil []T as null, not []. The initial ClientPageResponseSchema and ClientHydrateSchema rejected null inboundIds / summary.online / summary.depleted / etc., causing [zod] warnings on every empty list. Add nullableStringArray / nullableNumberArray helpers that accept null and transform to [] so consuming code keeps seeing arrays. Mark ClientRecord.traffic and .reverse nullable too (reverse is explicitly null in MarshalJSON when storage is empty). * fix(vite): treat /panel/xray as SPA page, not API root The dev-server bypass classified /panel/xray as an API path because the PANEL_API_PREFIXES matcher did `stripped === prefix.replace(/\/$/, '')`, which made the bare path collide with the SPA route of the same name (see web/controller/xui.go: g.GET("/xray", a.panelSPA)). On reload, /panel/xray got proxied to the Go backend instead of being served by Vite. The backend returned the embedded built index.html with hashed asset names that the dev server doesn't have, so every asset 404'd. Prefix-only match for trailing-slash entries fixes it: panel/xray/... still routes to the API, but panel/xray itself reaches the SPA branch. * feat(frontend): drive form validation from Zod schemas NodeFormModal — full conversion to AntD Form.useForm with antdRule on every required field. Inline field errors replace the single 'fillRequired' toast. testConnection now runs validateFields(['address','port']) before sending. ClientFormModal and ClientBulkAddModal — minimal conversion: keep the existing useState-driven controlled-component pattern, but replace the hand-rolled `if (!form.x)` checks with schema.safeParse(form). The schema is the single source of truth for required-ness and types; ClientCreateFormSchema layers on the create-only `inboundIds.min(1)` rule. New schemas (in src/schemas/): NodeFormSchema (node.ts) ClientFormSchema / ClientCreateFormSchema (client.ts) ClientBulkAddFormSchema (client.ts) Other 16+ form modals stay on the current pattern — the antdRule adapter ships from the first Zod pass for opportunistic migration as forms are touched. * chore(frontend): silence swagger-ui-react peer-dep warnings on React 19 swagger-ui-react@5.32.6 bundles three deps whose declared peer ranges predate React 19: react-copy-to-clipboard@5.1.0 (peer 15-18) react-debounce-input@3.3.0 (peer 15-18, unmaintained) react-inspector@6.0.2 (peer 16-18) For the first two, the actual code is React-19 compatible - only the metadata is stale. Resolve via npm overrides: - react-copy-to-clipboard bumped to ^5.1.1 (peer is open-ended >=15.3.0 in that release). - react-inspector bumped to ^9.0.0 (^8 was a broken publish per its own deprecation notice). - react-debounce-input is wedged on 3.3.0 with no maintained successor on npm. Use the nested-override syntax to satisfy its react peer: "react-debounce-input": { "react": "^19.0.0" } That tells npm to use our React 19 for the package's peer dependency, which silences the warning without changing the package version. * fix(vite): bypass es-toolkit CJS shim for recharts deep imports The Nodes page (and any other recharts-using route) crashed in dev and prod with TypeError: require_isUnsafeProperty is not a function. Root cause: es-toolkit's package.json exports './compat/*' only via a default condition pointing at the CJS shims under compat/<name>.js. Those shims use a require_X.Y access pattern that Vite's optimizer (Rolldown in Vite 8) and the production Rolldown build both mishandle, losing the named-export accessor and calling the namespace object as a function. recharts imports a dozen of these subpaths with default- import syntax, so every chart path tripped the bug. The matching ESM build at dist/compat/<category>/<name>.mjs is fine, but it only carries a named export. Recharts uses default imports. Plug a small Rollup-compatible plugin (enforce: 'pre') in front of the resolver: any 'es-toolkit/compat/<name>' request becomes a virtual module that imports the named symbol from the right .mjs file and re-exports it as both default and named. The plugin is registered as a top-level plugin (for the prod build) and via the new Vite 8 optimizeDeps.rolldownOptions.plugins (for the dev pre-bundler), so both pipelines pick it up consistently. * feat(frontend): migrate five secondary form modals to Zod schemas Apply the schema + safeParse-on-submit pattern (introduced for ClientFormModal / ClientBulkAddModal) to five more forms: - ClientBulkAdjustModal: ClientBulkAdjustFormSchema enforces 'at least one of addDays / addGB is non-zero' via .refine(), replacing the ad-hoc days+gb check. - BalancerFormModal: BalancerFormSchema covers tag and selector required-ness; the duplicate-tag check stays inline since it needs the otherTags prop. Per-field validateStatus now reads from the parsed issues map. - RuleFormModal: RuleFormSchema captures the form shape (no required fields - every property is optional by design). safeParse short- circuits if anything is structurally wrong. - CustomGeoFormModal: CustomGeoFormSchema folds the regex alias rule and the http(s) URL validation (including URL parse) into the schema, replacing a 20-line validate() function. - TwoFactorModal: TotpCodeSchema (z.string().regex(/^\d{6}$/)) drives both the disabled-state of the OK button and the safeParse gate before the TOTP comparison. Schemas live alongside the matching API schemas: - ClientBulkAdjustFormSchema in schemas/client.ts - BalancerFormSchema / RuleFormSchema / CustomGeoFormSchema in schemas/xray.ts - TotpCodeSchema in schemas/login.ts (next to LoginFormSchema) No UX change for valid inputs. * feat(frontend): block invalid settings saves with Zod pre-save check Tighten AllSettingSchema with the actual valid ranges and patterns: - webPort / subPort / ldapPort: integer 1-65535 - pageSize: integer 1-1000 - sessionMaxAge: integer >= 1 - tgCpu: integer 0-100 (percentage) - subUpdates: integer 1-168 (hours) - expireDiff / trafficDiff / ldapDefault*: non-negative integers - webBasePath / subPath / subJsonPath / subClashPath: must start with / The existing useAllSettings save path runs AllSettingSchema.partial() through safeParse and logs drift without blocking. SettingsPage now adds a stronger gate before the mutation: run the full schema against the draft and, on failure, surface the first issue (field path + message) via the existing messageApi.error so the user actually sees what's wrong instead of silently sending bad data to the backend. Use cases caught: port out of range, negative quota, sub path missing leading slash, page size set to 0, tgCpu > 100. * feat(frontend): schema-guard Inbound and Outbound form submits The two largest forms in the panel send to the backend without ever checking their own port range or required-ness. Schema-gate the top-level fields so obviously bad payloads stop at the client. InboundFormModal: InboundFormSchema (port 1-65535 int, non-empty protocol, the rest of the keys present) runs as a safeParse just before the HttpUtil.post in submit(). The 2000+ lines of protocol- specific subform code stay untouched - that's a separate effort and the existing per-protocol logic (e.g. canEnableStream, isFallbackHost) already gates most of the structural correctness. OutboundFormModal: OutboundTagSchema (trim + min 1) replaces the hand-rolled `if (!ob.tag?.trim()) messageApi.error('Tag is required')` check. The duplicateTag check stays inline because it needs the existingTags prop. Both schemas emit i18n keys for messages with a defaultValue fallback, matching the pattern in BalancerFormModal and SettingsPage. * feat(backend): gate request bodies with go-playground/validator Add a generic BindAndValidate helper in web/middleware that wraps gin's content-aware binder with an explicit validator.Struct call and emits a structured `entity.Msg{Obj: ValidationPayload{Issues...}}` on failure so the frontend can map each issue to an i18n key. Tag the user-facing fields on model.Inbound, model.Node, and entity.AllSetting with the range/enum constraints they were previously relying on hand-rolled CheckValid logic (or nothing) to enforce, and wire the helper into the inbound/node/settings controllers that bind those structs directly. Promotes validator/v10 from indirect to direct require, plus six unit tests covering valid payloads, range violations, enum violations, malformed JSON, in-place binding, and JSON-only strict mode. This is PR1 of a planned end-to-end Zod rollout — controllers using local form structs (custom_geo, setEnable, fallbacks, client) keep their existing handling and will be migrated as their schemas firm up. * feat(codegen): Go-first tool emitting Zod schemas and TS types Add tools/openapigen — a single-binary Go program that walks the exported structs in database/model, web/entity, and xray via go/parser and emits two committed artifacts under frontend/src/generated: - zod.ts shared Zod schemas keyed off `validate:` tags (ports get .min(1).max(65535), Inbound.protocol becomes a z.enum, Node.scheme too, etc.) - types.ts plain TS interfaces inferred from the same walk, so consumers can import Inbound without dragging Zod along The walker flattens embedded structs (AllSettingView.AllSetting), honors json:"-" and omitempty, and accepts per-struct overrides so the JSON-string-inside-JSON columns (Inbound.Settings/StreamSettings/ Sniffing, ClientRecord.Reverse, InboundClientIps.Ips) render as z.unknown() instead of leaking the DB-storage type into the API contract. Type aliases like model.Protocol are emitted as TS aliases and Zod schemas in their own right. Wires `npm run gen:zod` in frontend/package.json so the generator can be re-run without leaving the frontend tree. The existing openapi.json build (gen:api) is left alone for now; migrating the OpenAPI surface to this generator is a follow-up. PR2 of the planned Zod end-to-end rollout. * refactor(frontend): tighten HttpUtil generics from any to unknown Switch the class-level default on Msg<T> and the per-method defaults on HttpUtil.get/post/postWithModal from `any` to `unknown`, so callers that don't pass an explicit T get a narrowed response that must be schema- checked or type-cast before its shape is trusted. Drops the four file-level eslint-disable comments these defaults required. Fixes the nine direct `.obj.field` consumers that surfaced (IndexPage, XrayMetricsModal, NordModal, WarpModal, LogModal, VersionModal, XrayLogModal, CustomGeoSection) by giving each call site the explicit T it should have had from the start — typically a small ad-hoc shape, sometimes a string for the JSON-text-in-Msg.obj pattern used by NordModal/WarpModal/Xray nord/warp endpoints. PR3 of the planned Zod end-to-end rollout — schemas/inbound.ts and schemas/client.ts loose() removal stays parked until the protocol schemas land in Phase 3 to avoid silently dropping fields. * feat(frontend): protocol-leaf Zod schemas with discriminated unions Stand up schemas/primitives (Port, Flow, Protocol, Sniffing) and per-protocol leaf schemas for all 10 inbound and 13 outbound xray protocols. The leaves omit any inner `protocol` literal — the discriminator lives at the parent level so consumers narrow on `.protocol` without redundant projection. Wire shape is preserved per protocol: vmess outbound stays in `vnext[]`, trojan and shadowsocks outbound in `servers[]`, vless outbound flat, http/socks outbound in `servers[].users[]`. Cross-protocol atoms (port, flow, sniffing dest, protocol enum) live in primitives. Protocol-specific enums (vmess security, ss method/network, hysteria version, freedom domain strategy, dns rule action) stay with their leaves. Tagged-wrapper `z.discriminatedUnion('protocol', [...])` composes both InboundSettingsSchema and OutboundSettingsSchema; existing class-based models in src/models/ are untouched and will be retired in Step 3 once the golden-file safety net is in place. * feat(frontend): stream and security Zod families with discriminated unions Stand up the remaining Step 2 families. NetworkSettingsSchema is a 6-branch DU on `network` covering tcp/kcp/ws/grpc/httpupgrade/xhttp, with asymmetric per-network wire keys (tcpSettings, wsSettings, ...) preserved exactly so fixtures round-trip byte-identical. SecuritySettingsSchema is a 3-branch DU on `security` covering none/tls/reality. TLS certs use a file-vs-inline union; uTLS fingerprints are shared between TLS and Reality via a single primitive enum. Hysteria-as-network, finalmask, and sockopt are not in the plan's Step 2 inventory and are deferred to Step 6 (Tighten) - they're orthogonal extras on the stream root, not network-discriminated branches. Resolves a Security identifier collision in protocols/index.ts by re-exporting the type alias as SecurityKind (the `Security` name is taken by the namespace re-export). * test(frontend): vitest harness with golden-file fixtures for inbound protocols Stand up Phase 3 safety net before the models/ rewrite. The harness loads JSON fixtures via Vite's import.meta.glob, parses each through InboundSettingsSchema (the tagged-wrapper DU), and snapshots the canonical parsed shape. Snapshots stay byte-stable across the upcoming class-to- pure-function extraction, catching any normalization drift. Six representative inbound fixtures cover the high-traffic protocols: vless, vmess, trojan, shadowsocks (2022-blake3 multi-user), wireguard, hysteria2. Stream and security branches plus the remaining protocols (http, mixed, tunnel, hysteria) follow in subsequent turns. Uses /// <reference types="vite/client" /> instead of @types/node so we avoid pulling in another type package; import.meta.glob is enough to walk the fixtures directory at compile time. Adds vitest 4.1.7 as the only new dev dependency. test/test:watch scripts land in package.json; a standalone vitest.config.ts keeps the production vite.config.js (which reads from sqlite via DatabaseSync) out of the test runner. * test(frontend): broaden golden coverage to remaining inbounds + stream + security DUs Round out Step 3b. Four more inbound fixtures complete the protocol set (http with two accounts, mixed with socks-style auth, tunnel with a port map, hysteria v1). Two parallel test files cover the other DUs: stream.test.ts walks tcp/ws/grpc fixtures through NetworkSettingsSchema, and security.test.ts walks none/tls/reality through SecuritySettingsSchema. Snapshot count is now 16 across three test files. The reality fixture locks in the array form of serverNames/shortIds (the panel class stores them comma-joined internally but they ship as arrays on the wire). The TLS fixture pins the file-vs-inline cert DU on the file branch. Stream coverage for httpupgrade/xhttp/kcp and security mixed-with-stream combos follow in the next turn, alongside the shadow harness. * test(frontend): shadow-parse harness asserting legacy class and Zod converge Add Step 3c's safety net: for every inbound golden fixture, run the raw payload through both pipelines — legacy: Inbound.Settings.fromJson(protocol, raw.settings).toJson() zod: InboundSettingsSchema.parse(raw).settings — canonicalize each (recursively sort keys, drop empty arrays / null / undefined), and assert byte-equality. This locks the wire shape across the upcoming class-to-pure-function extraction in Step 3d. Any normalization drift introduced by the rewrite trips an assertion here before it can reach users. Two ergonomic wrinkles handled inline: - The legacy class lumps hysteria + hysteria2 onto a single HysteriaSettings (no hysteria2 case in the dispatch table); the test routes hysteria2 fixtures through the HYSTERIA branch. - Empty arrays in Zod's output (e.g. fallbacks: [] from a .default([])) are treated as equivalent to the legacy class's omit-when-empty behavior. Same wire state, different syntactic surface. All 26 tests across 4 test files pass on first run. * refactor(frontend): extract toHeaders + toV2Headers to lib/xray/headers.ts First Step 3d extraction. The XrayCommonClass static helpers toHeaders/toV2Headers are pure data shape conversions with no class hierarchy needs, so they move to a standalone module that callers can import without dragging in models/inbound.ts. The new module exports HeaderEntry + V2HeaderMap as named types so consumers stop reaching into the legacy class for type shapes. A new test file (headers.test.ts) asserts byte-equality with the legacy XrayCommonClass.toHeaders / .toV2Headers across 18 cases — null / undefined / primitive inputs, single-string headers, array-valued headers, duplicate names, empty-name and empty-value filtering, both arr=true (TCP request/response shape) and arr=false (WS / xHTTP / sockopt shape). Drift between the legacy and new impls fails these tests, so the follow-up call-site swap stays safe. Callers (TcpStreamSettings, WsStreamSettings, HTTPUpgradeStreamSettings, TunnelSettings, etc.) still go through XrayCommonClass for now — those swaps land alongside class-method extractions in subsequent turns. Suite is now 44 tests across 5 files; typecheck + lint clean. * refactor(frontend): extract createDefault*Client factories to lib/xray Next Step 3d slice. Five plain-object factories — Vless, Vmess, Trojan, Shadowsocks, Hysteria — replace the legacy `new Inbound.<Protocol>Settings.<Protocol>(...)` constructor chain and the ClientBase XrayCommonClass machinery. Each factory takes an optional seed; missing random fields (id, password, auth, email, subId) fall through to RandomUtil at call time. Forms can hand-pick a UUID; tests pass deterministic seeds so the suite never touches window.crypto. Tests double-verify each factory: a snapshot locks the exact shape, and the matching Zod ClientSchema.parse(out) must equal `out` — no missing defaults, no stray fields, type-narrowed end-to-end. Discovered: VmessClientSchema and VlessClientSchema enforce z.uuid() format, so the test seeds use real-shape UUIDs. Suite: 49 tests across 6 files; typecheck + lint clean. Outbound and inbound-settings factories follow in subsequent turns alongside the toShareLink extraction. * refactor(frontend): add createDefault*InboundSettings factories for all 10 protocols Round out Step 3d's settings factory set. Ten plain-object factories (vless / vmess / trojan / shadowsocks / hysteria / hysteria2 / http / mixed / tunnel / wireguard) replace the legacy `new Inbound.<X>Settings(protocol)` constructors. Each returns a Zod- parsable wire shape with schema defaults applied — no class instance. Forms (Step 4) and InboundsPage clone (Step 5) call these factories directly once the swap lands. Three factories take a seed for random fields: - shadowsocks: method-dependent password length via RandomUtil.randomShadowsocksPassword(method) - hysteria: explicit `version` override (defaults to 2, matching the legacy panel constructor — v1 is opt-in) - wireguard: secretKey from Wireguard.generateKeypair().privateKey Tests double-verify each factory the same way as the client factories: snapshot the shape, then Zod parse round-trip to confirm no missing defaults or stray fields. Suite: 59 tests across 6 files; typecheck + lint clean. Outbound factories and the toShareLink extraction follow next. * refactor(frontend): add getHeaderValue wire-shape lookup to lib/xray/headers Tiny piece of the toShareLink scaffold. The legacy Inbound.getHeader(obj, name) iterated the panel's internal HeaderEntry[] form; the new getHeaderValue reads the Record<string, string|string[]> map our Zod schemas store on the wire. Case-insensitive, returns '' on miss to match the legacy fallback so link-generator call sites stay simple. For repeated-name maps (TCP/WS-style string[] values) the first value wins — matches the legacy iteration order so the share URL's Host hint stays deterministic. Five unit tests cover undefined/null/empty inputs, case folding, string-valued and array-valued matches, empty-array edge case, and missing-key fallback. Suite: 64 tests across 6 files; typecheck + lint clean. This unblocks the next slice: per-protocol link generators (genVmessLink etc.) take a typed inbound + client and call getHeaderValue against the ws/httpupgrade/xhttp/tcp.request header maps. * feat(frontend): stream extras + full InboundSchema with DU intersection Step 3d's last scaffolding piece before link generators. Three new stream-extras schemas land alongside the network/security DUs: - finalmask: TcpMask[] + UdpMask[] + QuicParams. Mask `settings` stays record<string, unknown> for now — there are 13 UDP mask types and 3 TCP mask types with distinct per-type setting shapes, and modeling them all as DUs would dwarf the rest of stream/ without buying anything the shadow harness doesn't already catch. Tightened in Step 6. - sockopt: 17 socket-tuning knobs (TCP keepalive, TFO, mark, tproxy, mptcp, dialer proxy, IPv6-only, congestion). `interfaceName` field matches the panel class naming; serializers rename to `interface` on the wire. - external-proxy: rows ship per inbound describing edge fronts (CDN mirrors). Used by link generators to fan out share URLs. schemas/api/inbound.ts composes the top-level wire shape with intersection-of-DUs: StreamSettingsSchema = NetworkSettingsSchema .and(SecuritySettingsSchema) .and(StreamExtrasSchema) InboundSchema = InboundCoreSchema.and(InboundSettingsSchema) A fixture (vless-ws-tls.json) exercises the full shape — protocol DU, network DU, security DU, and TLS cert file branch in one round trip. The snapshot pins the canonical parsed form so the upcoming link extractor consumes typed input with no class hierarchy underneath. Suite: 65 tests across 7 files; typecheck + lint clean. Zod 4 intersection-of-DUs works. * refactor(frontend): extract genVmessLink to lib/xray/inbound-link.ts First link generator to leave the class hierarchy. genVmessLink takes a typed Inbound + client args and returns the base64-encoded vmess:// URL. Internal helpers (buildXhttpExtra, applyXhttpExtraToObj, applyFinalMaskToObj, applyExternalProxyTLSObj, serializeFinalMask, hasShareableFinalMaskValue, externalProxyAlpn) port across from XrayCommonClass — same logic, rewritten to read the Zod schemas' Record<string, string> headers instead of the legacy HeaderEntry[]. Parity test (inbound-link.test.ts) loads each vmess fixture in golden/fixtures/inbound-full, parses it with InboundSchema for the new pure fn AND constructs LegacyInbound.fromJson(raw) for the class method, then asserts the URLs match byte-for-byte. Drift between the two impls fails here before the call sites in pages/inbounds/* get swapped. Adds a small test setup file that aliases globalThis.window to globalThis so Base64.encode's window.btoa works under Node — keeps the test env at 'node' and avoids pulling jsdom as a new dep. A first vmess-tcp-tls full-inbound fixture pins the round-trip path. Suite: 67 tests across 8 files; typecheck + lint clean. Five more link generators (vless/trojan/ss/hysteria/wireguard) plus the orchestrator (toShareLink, genAllLinks) follow in subsequent turns. * test(frontend): refresh inbound-full snapshot with vmess-tcp-tls fixture * refactor(frontend): extract genVlessLink to lib/xray/inbound-link Second link generator. genVlessLink builds the vless://<uuid>@<host>:<port>?<query>#<remark> share URL from a typed Inbound + client args, dispatching on streamSettings.network for the network-specific knobs and on streamSettings.security for the TLS/Reality knobs. Three param-style helpers move alongside the obj- style ones already in this file: - applyXhttpExtraToParams — writes path/host/mode/x_padding_bytes and the JSON extra blob into URLSearchParams - applyFinalMaskToParams — writes the fm payload when shareable - applyExternalProxyTLSParams — overrides sni/fp/alpn when an external proxy entry is supplied and security is tls A vless-tcp-reality fixture lands alongside the existing vless-ws-tls one, so the parity test now exercises both security branches. Discovered a latent legacy bug while writing parity: the old class stored realitySettings.serverNames as a comma-joined string and gated SNI on `!ObjectUtil.isArrEmpty(serverNames)`, which always returns true for strings — so SNI was never written into Reality share URLs. Existing clients rely on the omission (they pull SNI from realitySettings.target instead). We preserve the omission here to keep this extraction byte-stable; an inline comment marks the spot for a separate intentional fix. Suite: 70 tests across 8 files; typecheck + lint clean. * refactor(frontend): extract genTrojanLink + genShadowsocksLink to lib/xray Third and fourth link generators. genTrojanLink mirrors genVlessLink's shape (URLSearchParams + network/security branches + remark hash) minus the encryption/flow VLESS-isms. genShadowsocksLink shares the same query construction but base64-encodes the userinfo portion as method:password or method:settingsPw:clientPw depending on whether SS-2022 is in single-user or multi-user mode. Three reusable helpers move out of the per-protocol functions: - writeNetworkParams: the per-network switch that all param-style links share (tcp http header / kcp mtu+tti / ws path+host / grpc serviceName+authority / httpupgrade / xhttp extras) - writeTlsParams: fingerprint/alpn/ech/sni - writeRealityParams: pbk/sid/spx/pqv (preserves the SNI-omission legacy parity quirk noted in the genVlessLink commit) genVmessLink stays with its inline switch — it builds a JSON obj instead of URLSearchParams and has per-network quirks (kcp emits mtu+tti at the obj root, grpc maps multiMode to obj.type='multi') that don't factor cleanly through the shared writer. Two new full-inbound fixtures (trojan-ws-tls, shadowsocks-tcp-2022) plus matching parity tests bring the suite to 74 tests across 8 files; typecheck + lint clean. * refactor(frontend): extract genHysteriaLink + Wireguard link/config to lib/xray Fifth and sixth link generators. genHysteriaLink builds the v1/v2 share URL (scheme picked from settings.version), copying TLS knobs into the query, surfacing the salamander obfs password from finalmask.udp[type=salamander] when present, and writing the broader finalmask payload under `fm` like the other links. Legacy parity note: the old genHysteriaLink read stream.tls.settings.allowInsecure, which isn't a field on TlsStreamSettings.Settings — the guard always evaluated false and the `insecure` param never made it into the URL. We omit it here to stay byte-stable. genWireguardLink and genWireguardConfig take a typed WireguardInboundSettings + peer index and: - link: wireguard://<peerPriv>@host:port?publickey=&address=&mtu=#remark - config: the .conf text WireGuard clients consume directly Both derive the server pubKey from settings.secretKey via Wireguard.generateKeypair at call time — Zod stores only secretKey on the wire (pubKey is computed). The Wireguard utility is pure JS (X25519 over Float64Array), so it runs fine under node + the window polyfill we added with the vmess extraction. Two new full-inbound fixtures (hysteria-v1-tls, wireguard-server) plus matching parity tests bring the suite to 78 tests across 8 files; typecheck + lint clean. Hysteria2 (protocol literal) parity stays deferred — the legacy class has no HYSTERIA2 dispatch case, so it can't round-trip a hysteria2 fixture without a protocol remap. Same trick the shadow harness uses; revisit in the orchestrator commit. * refactor(frontend): extract share-link orchestrator to lib/xray/inbound-link Last slice of Step 3d. Five orchestrator exports compose the per- protocol generators into the public surface the panel consumes: - resolveAddr(inbound, hostOverride, fallbackHostname): picks the address that goes into share/sub URLs. Browser `location.hostname` is no longer a hidden dependency — callers pass it in (or any other fallback they want). - getInboundClients(inbound): protocol-aware clients accessor. Mirrors the legacy `Inbound.clients` getter, including the SS quirk where 2022-blake3-chacha20 single-user inbounds report null (no client loop) and everything else returns the clients array. - genLink: per-protocol dispatcher matching legacy Inbound.genLink. - genAllLinks: per-client fanout. Builds the remarkModel-formatted remark (separator + 'i'/'e'/'o' field picker) and iterates streamSettings.externalProxy when present. - genInboundLinks: top-level \r\n-joined link block. Loops per client for clientful protocols, single-shots SS for non-multi-user, and delegates to genWireguardConfigs for wireguard. Returns '' for http/mixed/tunnel (no share URL at all). Plus genWireguardLinks / genWireguardConfigs fanouts which iterate peers and append index-suffixed remarks. Parity test exercises every full-inbound fixture against legacy Inbound.genInboundLinks. Skips hysteria2 (no legacy dispatch case; that bridge belongs in a separate intentional commit alongside the form modal swap). Suite: 89 tests across 8 files; typecheck + lint clean. Next: Step 4 form modal migrations. Forms can now drop `new Inbound.Settings.getSettings(protocol)` in favor of the createDefault*InboundSettings factories, and InboundsPage clone can swap to genInboundLinks. Models/ deletion follows in Step 5 once all call sites are off the class. * refactor(frontend): swap InboundsPage clone fallback off Inbound.Settings.getSettings First Step 4 call-site swap. createDefaultInboundSettings(protocol) lands in lib/xray/inbound-defaults — a protocol-aware dispatch over the 10 per-protocol settings factories already in this module. Returns a Zod- parsable plain object instead of a class instance, so callers that just need the wire-shape JSON can drop the class hierarchy without touching the broader form modals. InboundsPage's clone path used Inbound.Settings.getSettings(p).toString() as the fallback when settings JSON parsing failed. That's now createDefaultInboundSettings + JSON.stringify, with a final '{}' guard for unknown protocols (legacy returned null and .toString() crashed — we just emit empty settings instead). The Inbound import on this file is now unused and removed. The 2 remaining getSettings call sites in InboundFormModal aren't safe to swap in isolation — the form mutates the returned class instance through methods like .addClient() and .toJson() across ~2000 lines of JSX. Those land with the full Pattern A rewrite of InboundFormModal, which the plan budgets at multiple days on its own. Suite: 89 tests across 8 files; typecheck + lint clean. * refactor(frontend): lift Protocols + TLS_FLOW_CONTROL consts to schemas/primitives Step 4b. The Protocols and TLS_FLOW_CONTROL enums on models/inbound.ts were dragging five page files into that 3,300-line module just to read literal string constants. Lifting them to schemas/primitives lets those pages drop the @/models/inbound import entirely. - schemas/primitives/protocol.ts now exports a Protocols const map alongside the existing ProtocolSchema. TUN stays in the const for parity (legacy panel deployments may have saved TUN inbounds) even though the Go validator no longer accepts it as a new write. - schemas/primitives/flow.ts now exports TLS_FLOW_CONTROL. The empty-string default isn't keyed because the legacy never had a NONE entry — call sites compare against the two real flow values. Updated five consumers: - useInbounds.ts: TRACKED_PROTOCOLS now annotated readonly string[] so .includes(string) keeps narrowing through the array literal - QrCodeModal.tsx, InboundInfoModal.tsx: Protocols - ClientFormModal.tsx, ClientBulkAddModal.tsx: TLS_FLOW_CONTROL Suite: 89 tests across 8 files; typecheck + lint clean. models/inbound.ts is now imported by: - InboundFormModal.tsx (heavy use of Inbound class + getSettings) - test/inbound-link.test.ts + test/shadow.test.ts + test/headers.test.ts (intentional — these are parity tests against the legacy class) OutboundFormModal still imports from models/outbound. Both form modals are the multi-day Pattern A rewrites the plan scopes separately. * refactor(frontend): lift OutboundProtocols + OutboundDomainStrategies to schemas/primitives Moves the two outbound-side consts out of models/outbound.ts and into schemas/primitives/outbound-protocol.ts. Renames the export to OutboundProtocols to disambiguate from the inbound Protocols const (different key casing — PascalCase vs ALL CAPS — and partly different member set, so they cannot share a single const). OutboundsTab.tsx keeps its 15+ Protocols.X call sites by aliasing the import. FinalMaskForm.tsx and BasicsTab.tsx swap directly. Drops a stale `as string[]` cast in BasicsTab that no longer fits the new readonly-tuple typing. After this commit only the two big form modals (InboundFormModal/OutboundFormModal) plus three intentional parity tests still import from @/models/. * refactor(frontend): lift outbound option dictionaries to schemas/primitives Adds schemas/primitives/options.ts with UTLS_FINGERPRINT, ALPN_OPTION, SNIFFING_OPTION, USERS_SECURITY, MODE_OPTION (all identical between models/inbound.ts and models/outbound.ts) plus the outbound-only WireguardDomainStrategy, Address_Port_Strategy, and DNSRuleActions. OutboundFormModal now pulls 9 consts from primitives. Only `Outbound` (the class) and `SSMethods` (whose inbound/outbound versions diverge by 2 legacy aliases — keep the picker open for the Pattern A rewrite) still come from @/models/outbound. Drops three stale `as string[]` casts on what are now readonly tuples. * refactor(frontend): swap InboundFormModal option dicts to schemas/primitives Extends primitives/options.ts with the five inbound-only option dicts (TLS_VERSION_OPTION, TLS_CIPHER_OPTION, USAGE_OPTION, DOMAIN_STRATEGY_OPTION, TCP_CONGESTION_OPTION) and lifts InboundFormModal off @/models/inbound for 10 of its 12 imports. Only the Inbound class and SSMethods (inbound vs outbound versions diverge by 2 entries) still come from @/models/. Widens NODE_ELIGIBLE_PROTOCOLS Set element type to string since the new primitives const exposes a narrow literal union that `.has(arbitraryString)` would otherwise reject. * feat(frontend): InboundFormValues schema for Pattern A rewrite Foundation for the InboundFormModal rewrite. Mirrors the wire Inbound shape (intersection of core fields + protocol settings DU + stream/security DUs) plus the DB-side fields (up/down/total/trafficReset/nodeId/...) that flow through DBInbound rather than the xray config slice. InboundStreamFormSchema is exported separately so individual sub-form sections can rule against just the stream portion when needed. FallbackRowSchema is co-located here even though fallbacks save via a distinct endpoint after the main POST — they belong to the same form state from the user's perspective. No modal changes in this commit. Foundation only; subsequent turns swap the modal's `inboundRef`/`dbFormRef` mutable-class state for Form.useForm<InboundFormValues>(). * feat(frontend): adapter between raw inbound rows and InboundFormValues Adds lib/xray/inbound-form-adapter.ts with rawInboundToFormValues and formValuesToWirePayload. The pair is the data boundary the upcoming Pattern A modal will use: it consumes the DB row shape (settings et al. as string OR object — coerced internally), hands the modal typed InboundFormValues, and on submit reverses the trip to a wire payload with the three JSON-stringified slices the Go endpoints expect. No dependency on the legacy Inbound/DBInbound classes — the coerce step is inlined so the adapter survives the eventual models/ deletion. Adds 10 Vitest cases covering string vs object inputs, the optional streamSettings/nodeId fields, trafficReset coercion, and a raw-to-payload -to-raw round-trip equality. * feat(frontend): protocol capability predicates as pure functions Adds lib/xray/protocol-capabilities.ts with the seven predicates the modals call: canEnableTls, canEnableReality, canEnableTlsFlow, canEnableStream, canEnableVisionSeed, isSS2022, isSSMultiUser. Each takes a minimal slice of an InboundFormValues, no class instance. The legacy isSSMultiUser returns true on non-shadowsocks protocols too (method getter resolves to "" which != blake3-chacha20-poly1305). The new function preserves this quirk and documents it inline; callers all narrow on protocol === shadowsocks before checking, so the surprising return value never surfaces. Parity harness in test/protocol-capabilities.test.ts crosses each of the 10 golden fixtures with 14 stream configurations (network × security) and asserts each predicate matches the legacy class method — 140 cases, all green. * feat(frontend): outbound settings factories + dispatcher Adds lib/xray/outbound-defaults.ts parallel to inbound-defaults.ts: 13 createDefault*OutboundSettings factories (one per outbound protocol) plus the createDefaultOutboundSettings(protocol) dispatcher mirroring Outbound.Settings.getSettings's contract — non-null on each known protocol, null otherwise. The factory output matches the legacy `new Outbound.<X>Settings()` start state: required-by-schema fields the user fills in via the form (address, port, password, id, peer publicKey/endpoint) come back as empty stubs. Wireguard alone seeds secretKey via the X25519 generator; the rest expose blank fields. This is the same behavior the OutboundFormModal relies on for protocol-change resets. Shadowsocks defaults to 2022-blake3-aes-128-gcm rather than the legacy undefined — the Select snaps to the first option anyway, so the coherent default keeps the modal from rendering an empty picker. Tests cover three layers: - exact-shape snapshots per factory (13 cases) - Zod schema acceptance after sensible stub fill-in (13 cases) - dispatcher non-null per known protocol + null for the unknown (14 cases) * feat(frontend): InboundFormModal.new.tsx skeleton (Pattern A) First commit of the sibling-file modal rewrite. The new modal mounts Form.useForm<InboundFormValues>, hydrates via rawInboundToFormValues on open (edit) or buildAddModeValues (add), runs validateFields + safeParse on submit, and posts the formValuesToWirePayload result. No tabs yet — the modal body shows a WIP placeholder. The file is not imported anywhere; the existing InboundFormModal.tsx remains the one InboundsPage renders. Build, lint, and 280 tests stay green. Subsequent commits add the basic / sniffing / protocol / stream / security / advanced / fallbacks sections; the atomic import swap in InboundsPage.tsx lands last. * feat(frontend): basic tab on InboundFormModal.new.tsx (Pattern A) First real section of the sibling-file rewrite. Wires AntD Form.Items to InboundFormValues paths for the basic tab — enable, remark, deployTo (when protocol is node-eligible), protocol, listen, port, totalGB, trafficReset, expireDate. The port input gets a per-field antdRule against InboundFormBaseSchema.shape.port — the spec's Pattern A reference. The intersection-typed InboundFormSchema has no .shape accessor, so per-field rules pull from the underlying ZodObject components. totalGB and expireDate are bytes/timestamp on the wire but a GB number / dayjs picker in the UI. Both use shouldUpdate-closure children that read form state and call setFieldValue on user input — no transient form-only fields, no DU-shape surprises at submit time. Protocol-change cascade lives in Form's onValuesChange: pick a new protocol and the settings DU branch is reset to createDefaultInboundSettings(next); a non-node-eligible protocol also clears nodeId. Modal still renders a single-tab Tabs container. Sniffing tab is next. * feat(frontend): sniffing tab on InboundFormModal.new.tsx (Pattern A) Second section of the sibling-file rewrite. Wires the six sniffing sub-fields to nested form paths ['sniffing', 'enabled'], ['sniffing', 'destOverride'], etc. Uses Form.useWatch on the enabled flag to drive conditional rendering of the dependent fields — the same gate the legacy modal expressed via `ib.sniffing.enabled &&`. Checkbox.Group renders one Checkbox per SNIFFING_OPTION entry. The two exclusion lists use Select mode="tags" so the user can paste comma- separated IP/CIDR or domain rules. No transient form state, no class methods — every field maps directly to a wire-shape path in InboundFormValues. Protocol tab is next. * feat(frontend): protocol tab VLESS auth on InboundFormModal.new.tsx Adds the protocol tab to the sibling-file rewrite — currently only the VLESS section, which lays out decryption/encryption inputs and the three buttons that drive them: Get New x25519, Get New mlkem768, Clear. getNewVlessEnc + clearVlessEnc are ported from the legacy modal as pure setFieldValue paths into ['settings', 'decryption'] / ['settings', 'encryption'] — no class methods, no inboundRef. The matchesVlessAuth helper mirrors the legacy fuzzy label-matching so the backend response shape stays the only source of truth. selectedVlessAuth derives the displayed auth label from the encryption string via Form.useWatch — same heuristic as the legacy modal (.length > 300 → mlkem768, otherwise x25519). Tab spread is conditional: the protocol tab only appears when protocol === 'vless' right now. As more protocol sections land (shadowsocks, http/mixed, tunnel, tun, wireguard) the condition will widen to cover each one. * feat(frontend): protocol tab Shadowsocks section (Pattern A) Adds the Shadowsocks sub-form: method picker (from SSMethodSchema's seven schema-aligned options), conditional password input gated on isSS2022, network picker (tcp/udp/tcp,udp), ivCheck toggle. Method change cascades through the Select's onChange — regenerating the inbound-level password via RandomUtil.randomShadowsocksPassword. The shadowsockses[] multi-user list reset is deferred until the clients-management section lands. Uses isSS2022 from lib/xray/protocol-capabilities to gate the password field exactly the way the legacy modal did — keeps the form behavior identical without referencing the legacy class. SSMethodSchema.options drives the Select rather than the legacy SSMethods const (which the inbound modal pulled from models/inbound.ts). This commits to the schema-aligned 7-entry list for inbound; the outbound divergence (9 entries with legacy aliases) is still pending in OutboundFormModal — defer the UX decision to that rewrite. * feat(frontend): protocol tab HTTP and Mixed sections (Pattern A) Adds the HTTP and Mixed sub-forms. Both share an accounts list — first Form.List usage in the rewrite. Each row binds via [field.name, 'user'] / [field.name, 'pass'] under the parent ['settings', 'accounts'] path, so the wire shape stays exactly what HttpInboundSettingsSchema and MixedInboundSettingsSchema validate. HTTP-only: allowTransparent Switch. Mixed-only: auth Select (noauth/password), udp Switch, conditional ip Input gated on the udp value via Form.useWatch. Tab visibility widens to include http + mixed alongside vless + shadowsocks. The string cast on the includes-check keeps the frozen Protocols const's narrow union from rejecting the broader protocol string at the call site. * feat(frontend): protocol tab Tunnel section (Pattern A) Adds the Tunnel sub-form: rewriteAddress + rewritePort, allowedNetwork picker (tcp/udp/tcp,udp), Form.List-driven portMap with name/value pairs, and the followRedirect Switch. portMap is the second Form.List in the rewrite — same shape as the HTTP/Mixed accounts list but with name/value rather than user/pass. The wire shape stays `settings.portMap: { name, value }[]` exactly. Tab visibility widens to Tunnel. * feat(frontend): protocol tab TUN section (Pattern A) Adds the TUN sub-form: interface name, MTU, four primitive-array Form.Lists (gateway, dns, autoSystemRoutingTable), userLevel, autoOutboundsInterface. Primitive Form.Lists bind each row's Input directly to `field.name` (no inner key) — distinct from the object-row Form.Lists that bind to `[field.name, 'fieldKey']`. The Form.useWatch('protocol') return type comes from the schema's protocol enum which excludes 'tun' (TUN is in the legacy Protocols const for data parity but never accepted by the wire validator). Cast to string at the source so per-section comparisons against Protocols.TUN typecheck. Why: legacy DB rows with protocol === 'tun' still need to render; widening here keeps reads from rejecting them. Tab visibility widens to TUN. * feat(frontend): protocol tab Wireguard section (Pattern A) Adds the Wireguard sub-form: server secretKey input with regen icon, derived disabled public-key display, mtu, noKernelTun toggle, and a Form.List of peers — each peer having its own privateKey (regen icon), publicKey, preSharedKey, allowedIPs (nested Form.List for the string array), keepAlive. pubKey is purely derived (computed via Wireguard.generateKeypair from the watched secretKey) and is NOT stored in the form value — the schema omits it from the wire shape on purpose. The disabled display shows the live derivation without polluting form state. regenInboundWg generates a fresh keypair and writes only the secretKey path; pubKey re-derives automatically. regenWgPeerKeypair writes both privateKey and publicKey at the peer's path index. The preSharedKey wire-shape name is used instead of the legacy class's internal psk — matches WireguardInboundPeerSchema. Tab visibility widens to Wireguard. * feat(frontend): stream tab skeleton with TCP + KCP (Pattern A) Opens the stream tab on the sibling-file rewrite. Tab visibility is driven by canEnableStream from lib/xray/protocol-capabilities — same gate the legacy modal used, now schema-aware. Transmission picker (network select) is hidden for HYSTERIA since that protocol's network is implicit. onNetworkChange clears any stale per-network settings keys (tcpSettings/kcpSettings/...) and seeds an empty object for the new branch so AntD Form.Items don't read from undefined nested paths. TCP section: acceptProxyProtocol Switch (literal-true-optional on the wire — the form stores true/false but Zod's strip behavior keeps false-as-omission round-trips clean) plus an HTTP-camouflage toggle that flips header.type between 'none' and 'http'. The full HTTP camouflage request/response sub-form lands in a follow-up commit. KCP section: six numeric knobs (mtu, tti, upCap, downCap, cwndMultiplier, maxSendingWindow). WS / gRPC / HTTPUpgrade / XHTTP / external-proxy / sockopt / hysteria stream / FinalMaskForm hookup all still pending. * feat(frontend): stream tab WS + gRPC + HTTPUpgrade sections (Pattern A) Adds the three medium-complexity network branches to the stream tab. Plain Form.Item paths into the corresponding *Settings keys — no Form.List wrappers since these schemas don't have arrays at the top level. WS: acceptProxyProtocol, host, path, heartbeatPeriod gRPC: serviceName, authority, multiMode HTTPUpgrade: acceptProxyProtocol, host, path Header editing is deferred to a later commit — WsHeaderMap is a Record<string,string> on the wire, V2HeaderMap a Record<string,string[]>, and the form needs an array-of-{name,value} UI that converts on edit. Worth building once and reusing across WS, HTTPUpgrade, XHTTP, TCP request/response, and Hysteria masquerade headers. XHTTP + external-proxy + sockopt + hysteria stream + finalmask hookup still pending. * feat(frontend): stream tab XHTTP section (Pattern A) XHTTP is the heaviest network branch — 19 fields rendered conditionally on mode, xPaddingObfsMode, and the three *Placement selectors. Each gates its dependent field set via Form.useWatch. Field structure mirrors the legacy XHTTPStreamSettings form 1:1: - mode picker (auto / packet-up / stream-up / stream-one) - packet-up adds scMaxBufferedPosts + scMaxEachPostBytes; stream-up adds scStreamUpServerSecs - serverMaxHeaderBytes, xPaddingBytes, uplinkHTTPMethod (with the packet-up gate on the GET option) - xPaddingObfsMode unlocks xPadding{Key,Header,Placement,Method} - sessionPlacement / seqPlacement each unlock their respective Key field when set to anything other than 'path' - packet-up mode additionally unlocks uplinkDataPlacement, and that in turn unlocks uplinkDataKey when the placement is not 'body' - noSSEHeader Switch at the tail XHTTP headers editor still pending (same WsHeaderMap as WS — will be unified in the header-editor extraction commit). * feat(frontend): stream tab external-proxy + sockopt sections (Pattern A) External Proxy: Switch driven by externalProxy array length. Toggling on seeds one row with the window hostname + the inbound's current port; toggling off clears the array. Each row is a Form.List item with forceTls/dest/port/remark inline, and a nested SNI/Fingerprint/ALPN row that conditionally renders on forceTls === 'tls' via a shouldUpdate-closure that watches the per-row forceTls path. Sockopt: Switch driven by whether the sockopt object exists in form state. Toggling on calls SockoptStreamSettingsSchema.parse({}) so every default the schema declares (mark=0, tproxy='off', domainStrategy='UseIP', tcpcongestion='bbr', etc.) flows into the form; toggling off sets to undefined. Renders the seventeen sockopt fields directly bound to ['streamSettings', 'sockopt', X] paths. Option lists pull from the primitives const dictionaries (UTLS_FINGERPRINT, ALPN_OPTION, DOMAIN_STRATEGY_OPTION, TCP_CONGESTION_OPTION) rather than the schema's .options to keep one source of truth for UI label strings. * feat(frontend): security tab base + TLS section (Pattern A) Adds the security tab to the sibling-file rewrite. Visibility is paired with the stream tab — both gated on canEnableStream. The security selector is itself disabled when canEnableTls is false, and the reality option only appears when canEnableReality is true, mirroring the legacy modal's Radio.Group guards. onSecurityChange clears the previous branch's *Settings key and seeds the new branch from the schema's parsed defaults (the same trick the sockopt toggle uses). The security selector itself is rendered via a shouldUpdate closure so the on-change handler can write the cleaned streamSettings shape atomically without racing AntD's per-field sync. TLS section: serverName (the wire field — the legacy class calls it sni internally), cipherSuites (with the 13 named suites from TLS_CIPHER_OPTION), min/max version pair, uTLS fingerprint, ALPN multi-select, plus the three policy Switches. TLS certificates list, ECH controls, the full Reality sub-form, and the four API-call buttons (genRealityKeypair / genMldsa65 / getNewEchCert / randomizers) land in a follow-up commit. * feat(frontend): security tab Reality + ECH + mldsa65 controls (Pattern A) Adds the Reality sub-form and the four API-call buttons that drive the server-generated material: - genRealityKeypair calls /panel/api/server/getNewX25519Cert and writes the result into ['streamSettings', 'realitySettings', 'privateKey'] and the nested settings.publicKey path. - genMldsa65 calls /panel/api/server/getNewmldsa65 for the post-quantum seed/verify pair. - getNewEchCert calls /panel/api/server/getNewEchCert with the current serverName and writes echServerKeys + settings.echConfigList. - randomizeRealityTarget seeds target + serverNames from the random reality-targets pool. - randomizeShortIds calls RandomUtil.randomShortIds (comma-joined string) and splits into the schema's string[] form. Reality fields are bound directly to schema paths — show/xver/target, maxTimediff, min/max ClientVer, the settings.{publicKey, fingerprint, spiderX, mldsa65Verify} nested subtree, plus the array fields (serverNames, shortIds) rendered as Select mode="tags" since both ship as string[] on the wire. TLS certificates list (Form.List with the useFile DU) still pending — that's a chunky sub-form on its own. * feat(frontend): security tab TLS certificates list (Pattern A) Closes out the security tab: a Form.List of certificates that toggles between TlsCertFileSchema (certificateFile + keyFile string paths) and TlsCertInlineSchema (certificate + key as string arrays per the wire shape) via a per-row useFile boolean. useFile is a transient form-only field — not part of TlsCertSchema. Zod's default-strip behavior drops it during InboundFormSchema parse on submit, leaving only the matching wire branch's keys populated. Whichever side the user wasn't on stays empty, so Zod's union picks the populated branch. For inline certs the TextAreas use normalize + getValueProps to convert between the wire-side string[] and the multi-line text the user types. Each line becomes one array element, matching the legacy class's `cert.split('\n')` toJson convention. Per-row buildChain is conditionally rendered when usage === 'issue' — a shouldUpdate-closure watches the specific path so the toggle re-renders inline without listening to unrelated form changes. Security tab is now functionally complete. Advanced JSON tab, Fallbacks card, and the atomic swap in InboundsPage are next. * feat(frontend): advanced JSON tab on InboundFormModal.new.tsx (Pattern A) Adds the advanced JSON tab. Each sub-tab (settings / streamSettings / sniffing) renders an AdvancedSliceEditor — a small CodeMirror-backed JsonEditor that holds a local text buffer and forwards parsed JSON to form state on every valid edit. Invalid JSON sits silently in the local buffer; once the user finishes balancing braces / quoting, the next valid parse pushes through to the form. No stamping ref, no apply-on-tab-switch ceremony — the form is the single source of truth. The buffer seeds once from form state on mount. The Modal's destroyOnHidden means each open is a fresh editor instance, so external form mutations during a single open session can't desync the editor either. The streamSettings sub-tab is omitted when streamEnabled is false (matching the legacy modal's behavior for protocols like Http / Mixed that have no stream layer). * feat(frontend): fallbacks card on InboundFormModal.new.tsx (Pattern A) Adds the fallbacks card rendered inside the protocol tab whenever the current values describe a fallback host — VLESS or Trojan on tcp with tls or reality security. The protocol tab visibility widens to include Trojan in that exact case (it has no other protocol sub-form). Fallbacks live in a useState alongside the form rather than inside form values, mirroring the legacy modal: fallbacks save via a distinct endpoint (/panel/api/inbounds/{id}/fallbacks) after the main inbound POST, not as part of the inbound payload. loadFallbacks runs on open for edit-mode VLESS/Trojan; saveFallbacks runs after a successful POST inside the submit handler. Each row: child picker (filtered down to other inbounds), then four inline edits for SNI / ALPN / path / xver. Add adds an empty row; delete pulls the row from state. Quick-Add-All, the rederive-from-child helper, and the per-row up/down movers are deferred — the basic add/edit/remove cycle is what the modal actually needs to function. * feat(frontend): atomic swap InboundFormModal to Pattern A Deletes the 2261-line class-mutation modal and renames the 1900-line sibling rewrite into its place. InboundsPage.tsx already imports the file by path so no consumer change is needed — the swap is one file delete plus one file rename. Build, lint, and 280 tests stay green. What the new modal covers end-to-end: - Basic (enable / remark / nodeId / protocol / listen / port / totalGB / trafficReset / expireDate) - Sniffing (enabled / destOverride / metadataOnly / routeOnly / ipsExcluded / domainsExcluded) - Protocol per DU branch: VLESS (decryption/encryption + buttons), Shadowsocks (method/password/network/ivCheck), HTTP + Mixed (accounts list + per-protocol toggles), Tunnel (rewrite + portMap + followRedirect), TUN (interface/mtu + four primitive lists + userLevel/autoInterface), Wireguard (secretKey + derived pubKey + peers list with nested allowedIPs) - Stream per network: TCP base, KCP, WS, gRPC, HTTPUpgrade, XHTTP (the 22-field one), plus external-proxy and sockopt extras - Security: TLS (SNI/cipher/version/uTLS/ALPN/policy switches + certificates list with file/inline toggle + ECH controls), Reality (every field + the four API-call buttons), none - Advanced JSON (settings / streamSettings / sniffing live editors that round-trip into form state on every valid parse) - Fallbacks (load on open for VLESS/Trojan TLS-or-Reality TCP hosts; save through the secondary endpoint after the main POST succeeds) Known regressions vs the legacy modal, all reachable via Advanced JSON until backfilled in follow-up commits: - Hysteria stream sub-form (masquerade / udpIdleTimeout / version) — schema gap; the existing inbound DU has no hysteria stream branch - FinalMaskForm hookup — the component is still class-shape coupled - HeaderMapEditor — TCP request/response headers, WS / HTTPUpgrade / XHTTP headers, Hysteria masquerade headers all need a shared editor - TCP HTTP camouflage request/response body (version, method, path list, headers, status, reason) — only the on/off toggle is wired - Fallbacks polish — up/down move, quick-add-all, rederive-from-child, the per-row advanced-toggle / proxy-tag chips No reference to @/models/inbound's Inbound class anywhere in the new modal — only @/models/dbinbound (out of scope) and @/models/reality-targets (out of scope). The protocol-capabilities predicates and the rawInboundToFormValues + formValuesToWirePayload adapters carry every behavior the class used to provide. * fix(frontend): finish InboundFormModal rename after atomic swap The atomic-swap commit landed the new file but the exported function was still named InboundFormModalNew. Rename to match the file. * feat(frontend): outbound form schema + wire adapter foundation Lay the groundwork for OutboundFormModal's Pattern A rewrite: - schemas/forms/outbound-form.ts: discriminated-union form values across all 12 outbound protocols, with flat per-protocol settings shapes that match the legacy class fields (vmess vnext / trojan-ss-socks-http servers / wireguard csv address-reserved all flattened). - lib/xray/outbound-form-adapter.ts: rawOutboundToFormValues converts wire-shape outbound JSON to typed form values; formValuesToWirePayload re-nests on submit. Replaces the Outbound.fromJson/toJson dependency the modal currently has on the legacy class hierarchy. - test/outbound-form-adapter.test.ts: 15 round-trip cases covering each protocol's wire quirks (vmess vnext flatten, vless reverse-wrap, wireguard csv↔array, blackhole response wrap, DNS rule normalization, mux gating). * feat(frontend): OutboundFormModal.new.tsx skeleton (Pattern A) Sibling .new.tsx file with the Modal shell, Tabs (Basic/JSON), Form.useForm hydration via rawOutboundToFormValues, and the submit pipeline that calls formValuesToWirePayload before onConfirm. Tag uniqueness check is wired in. Protocol-specific sub-forms, stream, security, sockopt, and mux sections are deferred to subsequent commits — accessible via the JSON tab in the meantime. The InboundsPage continues to render the legacy modal until the atomic swap at the end. Also: rawOutboundToFormValues now returns streamSettings as undefined when the wire payload omits it, so Form.useForm doesn't receive a value that does not match the NetworkSettings discriminated union. * feat(frontend): OutboundFormModal.new.tsx vmess/vless/trojan/ss sections - Shared connect-target sub-block (address + port) for the six protocols whose form schema carries them flat at settings root. - VMess: id + security Select (USERS_SECURITY). - VLESS: id + encryption + flow + reverseTag (reverse-sniffing slice and Vision testpre/testseed come in a later commit). - Trojan: password. - Shadowsocks: password + method Select (SSMethodSchema) + UoT switch + UoT version. onValuesChange cascade: when the user picks a different protocol, the adapter re-seeds the settings sub-object to the new protocol's defaults so leftover fields from the previous protocol do not bleed through. * feat(frontend): OutboundFormModal.new.tsx socks/http/hysteria/loopback/blackhole/wireguard sections - SOCKS / HTTP: user + pass at settings root. - Hysteria: read-only version=2 (the actual transport knobs live on stream.hysteria, added with the stream tab). - Loopback: inboundTag. - Blackhole: response type Select with empty/none/http options. - Wireguard: address (csv) + secretKey (with regenerate icon) + derived pubKey + domain strategy + MTU + workers + no-kernel-tun + reserved (csv) + peers Form.List with nested allowedIPs sub-list. Wireguard regenerate icon uses Wireguard.generateKeypair() and writes both keys to the form via setFieldValue — preserves the legacy UX of the SyncOutlined inline-icon next to the privateKey label. * feat(frontend): OutboundFormModal.new.tsx DNS + Freedom + VLESS reverse-sniffing - DNS: rewriteNetwork (udp/tcp Select) + rewriteAddress + rewritePort + userLevel + rules Form.List (action/qtype/domain). - Freedom: domainStrategy + redirect + Fragment Switch with conditional 4-field sub-block (legacy 'enable Fragment' UX preserved — Switch sets all four fields to populated defaults, off-state empties them all out so the adapter strips them on submit) + Noises Form.List (rand/base64/ str/hex types, packet/delay/applyTo per row) + Final Rules Form.List with conditional block-delay sub-field. - VLESS reverse-sniffing slice: rendered only when reverseTag is set (matches the legacy modal's nested conditional). All six fields wired to the form state with appropriate widgets (Switch / Select multi / Select tags). * feat(frontend): OutboundFormModal.new.tsx stream tab (TCP/KCP/WS/gRPC/HTTPUpgrade) Wire the stream sub-form into the Pattern A modal: - newStreamSlice(network) helper bootstraps the per-network DU branch with Xray defaults (mtu=1350, tti=20, uplinkCapacity=5, etc.). - streamSettings is seeded once when the protocol supports streams but the form has no slice yet (new outbound + protocol switch). - onNetworkChange swaps the sub-key and preserves security when the new network still supports it, else snaps back to 'none'. - Per-network sub-forms wired: TCP: HTTP camouflage Switch (sets header.type = 'http' / 'none') KCP: 6 numeric tuning fields WS: host + path + heartbeat gRPC: service name + authority + multi-mode switch HTTPUpgrade: host + path XHTTP: host + path + mode + padding bytes (advanced fields via JSON) Security radio, TLS/Reality sub-forms, sockopt, and mux still pending. * feat(frontend): OutboundFormModal.new.tsx security tab (TLS + Reality + Flow) - onSecurityChange cascade: swaps tlsSettings/realitySettings sub-key matching the DU branch, seeding the new sub-form with empty/default fields so the UI does not reference undefined values. - Flow Select rendered when canEnableTlsFlow is true (VLESS + TCP + TLS/Reality). Moved from the basic VLESS section so it only appears in the relevant security context — matches the legacy modal UX. - Security Radio (none / TLS / Reality) gated by canEnableTls and canEnableReality pure-function predicates from lib/xray/protocol-capabilities. - TLS sub-form: 6 outbound-specific fields (SNI/uTLS/ALPN/ECH/ verifyPeerCertByName/pinnedPeerCertSha256) matching the legacy TlsStreamSettings flat shape (no certificates list — outbound is client-side). - Reality sub-form: 6 fields (SNI/uTLS/shortId/spiderX/publicKey/ mldsa65Verify). publicKey + mldsa65Verify get TextAreas to handle the long base64 strings. * feat(frontend): OutboundFormModal.new.tsx sockopt + mux sections - Sockopts: Switch toggles streamSettings.sockopt between undefined and a populated default object (17 fields with sane bbr/UseIP defaults). Only the 8 most-used fields are rendered (dialer proxy, domain strategy, keep alive interval, TFO, MPTCP, penetrate, mark, interface). The remaining sockopt knobs (acceptProxyProtocol, tcpUserTimeout, tcpcongestion, tcpKeepAliveIdle, tcpMaxSeg, tcpWindowClamp, V6Only, trustedXForwardedFor, tproxy) are still in the wire payload — edit them via the JSON tab. - Mux: gated by isMuxAllowed(protocol, flow, network) — VMess/VLESS/ Trojan/SS/HTTP/SOCKS, no flow set, no xhttp transport. Sub-fields (concurrency / xudpConcurrency / xudpProxyUDP443) only render when enabled is true. - Sockopt section visible only when streamAllowed AND network is set — non-stream protocols (freedom/blackhole/dns/loopback) still edit sockopt via the JSON tab. * feat(frontend): atomic swap OutboundFormModal to Pattern A Delete the legacy 1473-line class-based OutboundFormModal.tsx and replace it with the new Pattern A modal (Form.useForm + antdRule + per-protocol discriminated-union form values + wire adapter). Net diff: legacy file gone, function renamed from OutboundFormModalNew to OutboundFormModal so the existing OutboundsTab import resolves unchanged. What is migrated: - All 12 protocols (vmess/vless/trojan/ss/socks/http/wireguard/ hysteria/freedom/blackhole/dns/loopback) - Stream tab with TCP/KCP/WS/gRPC/HTTPUpgrade + partial XHTTP - Security tab with TLS + Reality + Flow gating - Sockopt + Mux sections (gated by isMuxAllowed) - JSON tab with bidirectional bridge to form state - Tag uniqueness check - VLESS reverse-sniffing slice - Freedom fragment/noises/finalRules - DNS rewrite + rules list - Wireguard peers + nested allowedIPs sub-list - Wireguard secret/public key regeneration Deferred to follow-up commits (still accessible via the JSON tab): - XHTTP advanced fields (xmux, sequence/session placement, padding obfs) - Hysteria stream transport sub-form - TCP HTTP camouflage host/path body - WS/HTTPUpgrade/XHTTP headers map editor - Remaining sockopt knobs (tcpUserTimeout, tcpcongestion, tcpKeepAliveIdle, tcpMaxSeg, tcpWindowClamp, V6Only, trustedXForwardedFor, tproxy, acceptProxyProtocol) - VLESS Vision testpre/testseed - Reality API helpers (random target, x25519/mldsa65 generate-import) - Link import (vmess:// vless:// etc → outbound) - FinalMaskForm hookup (deferred from inbound rewrite too) * test(frontend): convert legacy-class parity tests to snapshot baselines With the inbound/outbound modal rewrites complete, the cross-check against the legacy Inbound class has served its purpose. The new pure-function / Zod-schema paths are the source of truth for production code; the parity assertions were the migration safety net. Convert the three parity test files to snapshot-based regression tests: - headers.test.ts: toHeaders + toV2Headers run against snapshots captured at the close of the migration (when both new and legacy were verified byte-equal). - protocol-capabilities.test.ts: 140 cases (10 fixtures × 14 stream shapes) snapshot the predicate-result tuple. Was: parity vs legacy Inbound.canEnableX() class methods. - inbound-link.test.ts: per-protocol genXxxLink + genInboundLinks orchestrator output is snapshotted. Was: byte-equality vs legacy Inbound.genXxxLink() methods. Also delete shadow.test.ts — its purpose was a dual-parse drift detector (Inbound.Settings.fromJson vs InboundSettingsSchema.parse). inbound-full.test.ts already snapshots the Zod parse output, which covers the same ground without the legacy dependency. models/inbound.ts and models/outbound.ts stay in the tree for now — DBInbound still consumes Inbound via its toInbound() method, and DBInbound migration is out of scope per the migration spec ('Do NOT migrate Status, DBInbound, or AllSetting...'). No production page imports from @/models/inbound or @/models/outbound directly anymore. * chore(frontend): enforce no-explicit-any: error + add typecheck/test to CI Step 7 of the Zod migration: lock the migration's gains in place via lint + CI enforcement. - eslint.config.js: `@typescript-eslint/no-explicit-any` set to error. Verified locally — zero violations in src/, with the only file-level disables being src/models/inbound.ts and src/models/outbound.ts (kept for DBInbound's toInbound() consumer; their migration is out of spec scope). - .github/workflows/ci.yml: add Typecheck and Test steps to the frontend job, between Lint and Build. PRs now have to pass tsc --noEmit and the full vitest suite (285 tests + 172 snapshots) before build runs. Migration scoreboard (vs the spec): Step 1 primitives + barrels done Step 2 protocol leaf + DUs done Step 3 pure-fn extraction done Step 4 form modals -> Pattern A done (Inbound + Outbound) Step 5 delete models/ files DEFERRED (DBInbound still uses Inbound; spec marks DBInbound migration out of scope) Step 6 tighten .loose() / unknown DEFERRED (invasive, separate PR) Step 7 lint + CI enforcement done (this commit) Production code paths now have no direct dependency on the legacy Inbound or Outbound classes. * feat(frontend): OutboundFormModal deferred features (Vision seed / TCP host+path / WG pubKey derive) Three small wins from the post-atomic-swap deferred list: - VLESS Vision testpre + testseed: shown only when flow === 'xtls-rprx-vision' (mirrors the legacy canEnableVisionSeed gate). testseed binds to a Select mode='tags' with a normalize() that coerces strings to positive integers and drops invalid entries. - TCP HTTP camouflage host + path: when the TCP HTTP camouflage Switch is on, surface two inputs that read/write directly into streamSettings.tcpSettings.header.request.headers.Host and .path. Both fields are string[] on the wire; normalize + getValueProps translate to/from comma-joined strings in the UI (one entry per host or path the user wants camouflaged). - Wireguard pubKey auto-derive: Form.useWatch on settings.secretKey + useEffect that runs Wireguard.generateKeypair(secret).publicKey on every change and writes the result into the disabled pubKey display field. Matches the legacy modal's per-keystroke derive. * feat(frontend): symmetric TCP HTTP host/path + extra sockopt knobs OutboundFormModal: - Sockopt section gains 5 common-but-rarely-tweaked knobs: acceptProxyProtocol, tproxy (off/redirect/tproxy), tcpcongestion (bbr/cubic/reno), V6Only, tcpUserTimeout. The remaining sockopt fields (tcpKeepAliveIdle, tcpMaxSeg, tcpWindowClamp, trustedXForwardedFor) are still edit-via-JSON; they are deeply tunable and not commonly touched. InboundFormModal: - TCP HTTP camouflage gains host + path inputs symmetric to the outbound side. Switch ON seeds request with sensible defaults (version 1.1, method GET, path ['/'], empty headers). The two inputs use the same normalize/getValueProps comma-string ↔ string[] dance the outbound side uses, so the wire shape stays identical to what xray-core expects. * feat(frontend): HeaderMapEditor reusable component + wire WS/HTTPUpgrade headers Add a single reusable header-map editor that handles the two wire shapes Xray uses: - v1: { name: 'value' } — used by WS / HTTPUpgrade / Hysteria masquerade. One value per name. - v2: { name: ['value1', 'value2'] } — used by TCP HTTP camouflage. Each header can repeat (RFC 7230 §3.2.2). Internal state is always a flat list of {name, value} rows regardless of mode; conversion to/from the wire shape happens at the value / onChange boundary so consumers bind straight to a Form.Item with no extra transforms. Wired into: - InboundFormModal: WS Headers, HTTPUpgrade Headers - OutboundFormModal: WS Headers, HTTPUpgrade Headers XHTTP headers are already in a list-of-rows wire shape (different from these two), so they keep their bespoke editor. Hysteria masquerade is still deferred until the Hysteria stream sub-form lands. * feat(frontend): Hysteria stream sub-form (schema branch + outbound UI) Add the 7th branch to NetworkSettingsSchema for Hysteria transport. schemas/protocols/stream/hysteria.ts: - HysteriaStreamSettingsSchema covers the full wire shape: version=2, auth, congestion (''|'brutal'), up/down bandwidth strings, optional udphop sub-object for port-hopping, receive-window tuning fields, maxIdleTimeout, keepAlivePeriod, disablePathMTUDiscovery. schemas/protocols/stream/index.ts: - NetworkSchema gains 'hysteria'. - NetworkSettingsSchema gains the 7th branch { network: 'hysteria', hysteriaSettings: HysteriaStreamSettingsSchema }. OutboundFormModal.tsx: - NETWORK_OPTIONS keeps the 6 standard transports for non-hysteria protocols; when protocol === 'hysteria', a 7th option is appended (matches the legacy [...NETWORKS, 'hysteria'] gate). - newStreamSlice handles the 'hysteria' case with sensible defaults matching the legacy HysteriaStreamSettings constructor. - New sub-form when network === 'hysteria': 8 common fields (auth, congestion, up, down, udphop Switch + 3 nested fields when on, maxIdleTimeout, keepAlivePeriod, disablePathMTUDiscovery). - Receive-window tuning fields are still edit-via-JSON (rarely touched + would clutter the form). * feat(frontend): fallbacks polish — move up/down + Add all button Two small UX wins on the InboundFormModal Fallbacks card: - Per-row Move up / Move down buttons (ArrowUp/Down icons) that swap adjacent indices. Order survives reloads via sortOrder (rebuilt from index on save). First row's Up button + last row's Down button are disabled. - 'Add all' button next to 'Add fallback' that one-shot inserts a fresh row for every eligible inbound (every option in fallbackChildOptions) not already wired up. Disabled when every eligible inbound is already covered. Convenient for operators running catch-all routing across every host on the panel. * feat(frontend): XHTTP advanced fields on outbound modal Replace the 'edit via JSON' deferred-features hint with the full XHTTP sub-form matching the legacy modal's XhttpFields helper. schemas/protocols/stream/xhttp.ts: - New XHttpXmuxSchema: 6 connection-multiplexing knobs (maxConcurrency, maxConnections, cMaxReuseTimes, hMaxRequestTimes, hMaxReusableSecs, hKeepAlivePeriod). - XHttpStreamSettingsSchema gains 5 outbound-only fields and one UI-only toggle: scMinPostsIntervalMs, uplinkChunkSize, noGRPCHeader, xmux, enableXmux. outbound-form-adapter.ts: - New stripUiOnlyStreamFields() drops xhttpSettings.enableXmux on the way to wire so the panel never embeds the UI toggle into the saved config. xray-core ignores unknown fields anyway, but the panel reads back its own emitted JSON, so a clean wire shape matters. OutboundFormModal.tsx: - Headers editor (HeaderMapEditor v1) for xhttpSettings.headers. - Padding obfs Switch + 4 conditional fields (key/header/placement/ method) when on. - Uplink HTTP method Select with GET disabled outside packet-up. - Session placement + session key (key shown when placement != path). - Sequence placement + sequence key (same pattern). - packet-up mode: scMinPostsIntervalMs, scMaxEachPostBytes, uplink data placement + key + chunk size (key/chunk-size shown when placement != body). - stream-up / stream-one mode: noGRPCHeader Switch. - XMUX Switch + 6 nested fields when on. * feat(frontend): inbound TCP HTTP camouflage response fields + request headers Complete the TCP HTTP camouflage UI on the inbound side. Already there from the previous symmetric host/path commit: - Request host (string[] via comma-string) - Request path (string[] via comma-string) This commit adds: - Request headers (V2 map: name -> string[]) via HeaderMapEditor. - Response version (defaults to '1.1' when camouflage toggles on). - Response status (defaults to '200'). - Response reason (defaults to 'OK'). - Response headers (V2 map) via HeaderMapEditor. The HTTP camouflage Switch seeds both request and response sub-objects on toggle-on so xray-core sees a valid TcpHeader.http shape from the first save. Without the response seed, partial fills would emit a schema-incomplete response block that xray-core might reject. * feat(frontend): link import on outbound modal (vmess/vless/trojan/ss/hy2) The legacy outbound modal could import a vmess://, vless://, trojan://, ss://, or hysteria2:// share link via a Convert button on the JSON tab. Restore that UX with a focused pure-function parser. lib/xray/outbound-link-parser.ts: - parseVmessLink: base64 JSON, maps net/tls + per-network params onto the discriminated stream branch. - parseVlessLink: standard URL with type/security/sni/pbk/sid/fp/flow query params, dispatches transport via buildStream + applies security params via applySecurityParams. - parseTrojanLink: same URL pattern, defaults security to tls. - parseShadowsocksLink: both modern (base64 userinfo@host:port) and legacy (base64 of whole thing) ss:// formats. - parseHysteria2Link: accepts both hysteria2:// and hy2:// schemes, uses the hysteria stream branch with version=2 + TLS h3. - parseOutboundLink dispatcher returns the first non-null parser result, or null when no scheme matches. test/outbound-link-parser.test.ts: - 13 cases covering happy paths for each protocol family plus malformed input, ss:// dual-format handling, hy2:// alias. OutboundFormModal.tsx: - Import button on the JSON tab Input.Search; on success, parsed payload flows through rawOutboundToFormValues, the form is reset, and we switch back to the Basic tab. - Tag is preserved when the parsed link does not carry one. Out of scope: advanced fields the legacy parser handled (xmux, padding obfs, reality short IDs, finalmask from fm= param). Power users can finish the import in the form after the basics land. * feat(frontend): inbound Hysteria stream sub-form (auth + udpIdleTimeout + masquerade) Restore the inbound side of Hysteria stream configuration that was previously hidden — the legacy modal exposed these knobs but the Pattern A rewrite gated them out. schemas/protocols/stream/hysteria.ts: - HysteriaMasqueradeSchema covers the inbound-only masquerade wire shape: type ('proxy'|'file'|'string'), dir, url, rewriteHost, insecure, content, headers, statusCode. The three masquerade types cover the spectrum: reverse-proxy upstream, serve static files, or return a fixed string body. - HysteriaStreamSettingsSchema gains 3 inbound-side optional fields: protocol, udpIdleTimeout, masquerade. Outbound side is untouched (the legacy class accepted both wire shapes via the same struct). InboundFormModal.tsx: - New hysteria stream sub-form section in streamTab, gated by protocol === HYSTERIA. Fields: version (disabled, locked to 2), auth, udpIdleTimeout, masquerade Switch + nested type-Select with three conditional sub-blocks (proxy URL+rewriteHost+insecure, file dir, string statusCode+body+headers). - onValuesChange cascade: switching TO hysteria seeds streamSettings with the hysteria branch (forcing network='hysteria' + TLS); switching AWAY from hysteria snaps back to TCP so the standard network selector has a valid starting point. masquerade headers use the HeaderMapEditor v1 component. * feat(frontend): complete outbound sockopt section with remaining knobs Add the four remaining SockoptStreamSettings fields that were edit-via-JSON-only after the initial outbound modal rewrite: - TCP keep-alive idle (s) — tcpKeepAliveIdle, time before sending the first probe on an idle TCP connection. - TCP max segment — tcpMaxSeg, override the default MSS. - TCP window clamp — tcpWindowClamp, cap the TCP receive window. - Trusted X-Forwarded-For — trustedXForwardedFor, list of trusted proxy hostnames/CIDRs whose XFF headers Xray will honor. The outbound sockopt section now exposes all 17 SockoptStreamSettings fields from the schema. The InboundFormModal's sockopt section has its own field list (closer to the legacy class) and is unchanged. * feat(frontend): outbound TCP HTTP camouflage parity with inbound Add method/version inputs, request header map, and full response sub-section (version/status/reason/headers) to OutboundFormModal so the outbound side can configure the same HTTP-1.1 obfuscation knobs the inbound side already exposed. * feat(frontend): round-trip XHTTP advanced fields in outbound link parser Pick up xPaddingBytes, scMaxEachPostBytes, scMinPostsIntervalMs, uplinkChunkSize, and noGRPCHeader from both vmess:// JSON and the URL query-param parsers (vless/trojan). The advanced xmux/padding-obfs/ reality-shortId knobs still wait on a follow-up; this slice unblocks the common case where a phone-issued xhttp link carries non-default padding or post sizes. * feat(frontend): round-trip XHTTP padding-obfs + remaining advanced knobs Extract the XHTTP key-mapping into typed string/number/bool key arrays applied by both the URL query-param branch and the vmess JSON branch. The parser now covers xPaddingObfsMode + xPaddingKey/Header/Placement/ Method, sessionKey/seqKey/uplinkData{Placement,Key}, noSSEHeader, scMaxBufferedPosts, scStreamUpServerSecs, serverMaxHeaderBytes, and uplinkHTTPMethod alongside the previous five XHTTP fields. Two new round-trip tests cover the padding-obfs surface on both link forms. * feat(frontend): FinalMaskForm rewrite to Pattern A + wire into both modals Rewrite FinalMaskForm.tsx from a class-coupled component (mutated stream.finalmask.tcp[] via .addTcpMask/.delTcpMask methods, notified parent via onChange callback) into a Pattern A sub-form: takes a NamePath base, a FormInstance, and the surrounding network/protocol, then composes Form.List + Form.Item at absolute paths under that base. All array structures use nested Form.List — tcp/udp mask arrays, the clients/servers groups in header-custom (Form.List of Form.List of ItemEditor), and the noise list. Type Selects use onChange to reset the settings sub-object via form.setFieldValue, mirroring the legacy changeMaskType behavior. The kcp.mtu side effect on xdns type change is preserved. Wired into both InboundFormModal and OutboundFormModal stream tabs, placed after the sockopt section. The component is the first Pattern A consumer of nested Form.List inside another Form.List, so it stands as the reference for future nested-array sub-forms. * docs(frontend): record FinalMaskForm rewrite + hookup in status doc Mainline migration goal — replace class-based xray models with Zod schemas as the single source of truth + drive all forms through AntD `Form.useForm` + `antdRule(schema.shape.X)` — is complete. Remaining items are incremental polish. * fix(frontend): Phase 2 Inbound form reactivity bugs (B1-B9, consolidated) A run of resets dropped the per-bug commits 1401d833 / 5b1ae450 / 5bce0dc5 / 4007eec7. Re-landing all fixes against the same files in one commit to avoid another rebase-style drop. B1 — Transmission Select / External Proxy + Sockopt switches didn't react after click. AntD 6.4.3 Form.useWatch on nested paths doesn't re-fire reliably after `setFieldValue('streamSettings', cleaned)` on the parent. Bound Transmission via `name={['streamSettings', 'network']}` and wrapped the two switches in `<Form.Item shouldUpdate>` blocks that read state via getFieldValue. B2 — Security regressed from `Radio.Group buttonStyle="solid"` to a Select dropdown, and disable state didn't refresh because tlsAllowed/ realityAllowed were derived at the top of the component. Restored Radio.Button group and moved canEnableTls/canEnableReality evaluation inside the shouldUpdate render prop. B3 — Advanced tab "All" sub-tab was missing. Added it as the first item with a new AdvancedAllEditor that round-trips top-level fields + the three nested slices on edit. B4 — Advanced tab title/subtitle and per-section help text were gone. Wrapped the Tabs in the existing `.advanced-shell` / `.advanced-panel` structure and restored the `.advanced-editor-meta` help under each sub-tab using existing i18n keys. B5 — TLS / Reality sub-forms didn't render when selecting tls or reality on the Security tab. The `{security === 'tls' && ...}` and `{security === 'reality' && ...}` conditionals used a stale top-level useWatch value. Wrapped both in <Form.Item shouldUpdate> blocks that read `security` via getFieldValue. B6 — Advanced JSON editors stale after Stream/Sniffing changes. The editors seeded text via lazy useState and AntD Tabs renders all panes upfront, so the Advanced tab was already mounted with stale data. Both AdvancedSliceEditor and AdvancedAllEditor now subscribe via Form.useWatch and re-sync the text buffer when the watched JSON differs from a lastEmitRef (the serialization at the moment of our own last accepted write). User typing doesn't trigger re-sync because setFieldValue updates lastEmitRef too. (A prior attempt added `destroyOnHidden` to the outer Tabs but broke conditional tab items when the unmounted Form.Item for `protocol` lost its value — abandoned in favor of useWatch reactivity.) B7 — HeaderMapEditor + button did nothing. addRow() appended a blank {name:'', value:''} row, but commit() filtered it via rowsToMap before reaching the form, so AntD saw no change and didn't re-render. The editor now keeps a local rows state so blank rows survive during editing; only filled rows are emitted to onChange. B9 — Sniffing destOverride defaults (HTTP/TLS/QUIC/FAKEDNS) were not pre-checked on a fresh Add Inbound. buildAddModeValues() seeded sniffing: {} which left destOverride undefined. Now seeds with SniffingSchema.parse({}) so the Zod defaults populate. * fix(frontend): FinalMaskForm TCP Mask sub-forms + Advanced JSON wrap (B10/B11) B10 — FinalMaskForm TCP Mask: after adding a mask and picking a Type (Fragment/Header Custom/Sudoku), the type-specific sub-forms didn't render. TcpMaskItem read `type` via Form.useWatch on a path inside Form.List, which doesn't re-fire reliably in AntD 6.4.3 — same root cause as the earlier B1/B2/B5 reactivity issues. Replaced with a <Form.Item shouldUpdate> wrapper that reads `type` via getFieldValue inside the render prop. B11 — Advanced sub-tabs (settings / streamSettings / sniffing) showed just the inner value (e.g. `{clients:[],decryption:"none",...}`), but the legacy modal wrapped each slice with its key envelope (e.g. `{settings:{...}}`) so the JSON matches the wire shape's slice and round-trips cleanly from copy-pasted inbound configs. Added a `wrapKey` prop to AdvancedSliceEditor that wraps/unwraps the value on render/write; the three sub-tabs now pass settings / streamSettings / sniffing as their wrapKey. * fix(frontend): import InboundFormModal.css so layout classes apply (B12) The file InboundFormModal.css existed but was never imported, so every class in it had no effect — including: - .vless-auth-state — the "Selected: <auth>" caption next to the X25519/ ML-KEM/Clear button row stayed inline next to Clear instead of display:block beneath the row - .advanced-shell / .advanced-panel — the Advanced tab's header / panel framing was missing - .advanced-editor-meta — the per-section help text under each Advanced sub-tab had no spacing - .wg-peer — wireguard peer rows had no top margin Add a side-effect import of the CSS file at the top of the modal. No other change needed; the legacy modal must have either imported it or had a global import that the new modal didn't inherit. * fix(frontend): FinalMaskForm relative paths + network-switch defaults (B13/B14) B13 — FinalMaskForm used absolute paths like ['streamSettings', 'finalmask', 'tcp', 0, 'type'] for Form.Item names inside Form.List render props. AntD's Form.List prefixes Form.Item names with the list's own name, so the actual storage path became ['streamSettings', 'finalmask', 'tcp', 'streamSettings', 'finalmask', 'tcp', 0, 'type'] — total nonsense. Symptoms: Type Select didn't show the 'fragment' default after add(), and the sub-form for the picked type never rendered (Fragment/Sudoku/HeaderCustom). Rewrote FinalMaskForm to use RELATIVE names inside every Form.List context (TCP/UDP outer list + nested clients/servers/noise inner lists). Added a `listPath` prop on the items so the shouldUpdate guard and the side-effect setFieldValue calls (resetting `settings` when type changes) can still address the absolute path; the displayed Form.Items use the relative form (`[fieldName, 'type']`). Replaced top-level Form.useWatch on nested paths with <Form.Item shouldUpdate> blocks reading via getFieldValue, same pattern as the earlier B5 fix — Form.useWatch on paths inside Form.List doesn't re-fire reliably in AntD 6.4.3. B14 — Switching network (KCP, WS, gRPC, XHTTP, ...) seeded the new XSettings blob as `{}` so every field showed as empty. The legacy `newStreamSlice` populated mtu=1350, tti=20, etc. Restored those defaults in onNetworkChange and seeded the initial tcpSettings.header in buildAddModeValues so even the default TCP state shows the HTTP-camouflage Switch in the correct off state instead of an undefined header object. * fix(frontend): inbound TCP HTTP camouflage drops request fields + KCP UI field rename (B15/B16) B15 — Inbound TCP HTTP camouflage exposed Host / Path / Method / Version / request-headers inputs. Per Xray docs (https://xtls.github.io/config/transports/raw.html#httpheaderobject), the `request` object is honored only by outbound proxies; the inbound listener reads `response`. Those inputs were writing dead data the server ignored. Removed them from the inbound modal; only Response {version, status, reason, headers} remain. The toggle still seeds an empty request object so the wire shape stays valid against the schema. B16 — KCP Uplink / Downlink inputs bound to non-existent form fields `upCap` / `downCap`, while the schema (and wire) use `uplinkCapacity` / `downlinkCapacity`. Renamed the Form.Items to the schema names so defaults populate and saves persist. Also corrected newStreamSlice('kcp') to seed the four KCP defaults (uplinkCapacity / downlinkCapacity / cwndMultiplier / maxSendingWindow) — the missing two were why "CWND Multiplier" and "Max Sending Window" still showed empty after switching to KCP. * fix(frontend): seed full Zod-schema defaults for stream slices + QUIC params (B17) XHTTP showed blank Selects for Session Placement / Sequence Placement / Padding Method / Uplink HTTP Method (and several other knobs). Those fields have a literal "" (empty string) value in the schema, which the Select renders as "Default (path)" / "Default (repeat-x)" / etc. The form field was `undefined`, not `""`, so the Select showed blank instead of the labelled default option. newStreamSlice in InboundFormModal hand-rolled per-network seed objects with only a handful of fields. Replaced with {Tcp,Kcp,Ws,Grpc,HttpUpgrade,XHttp}StreamSettingsSchema.parse({}) so every default declared in the schema populates the form on network switch. Same change in buildAddModeValues for the initial TCP state. QUIC Params (FinalMaskForm) had the same shape on a smaller scale — defaultQuicParams() only seeded congestion + debug + udpHop. The schema's other fields are .optional() (no Zod default) so a schema parse won't help. Hard-coded the xray-core / hysteria recommended values (maxIdleTimeout 30, keepAlivePeriod 10, brutalUp/Down 0, maxIncomingStreams 1024, four window sizes) so the InputNumber controls render with usable starting values instead of blank. * fix(frontend): forceRender all tabs so fields register at modal open (B18) AntD Tabs with the `items` API lazy-mounts inactive tab panes by default. The Form.Items inside an unvisited tab never register, so: - Form.useWatch on a parent path (e.g. 'sniffing') returns a partial view containing only registered children. Until the user clicked the Sniffing tab, Advanced > Sniffing JSON showed `{sniffing: {}}` instead of the full default object set by setFieldsValue. - After visiting the Sniffing tab once, the `sniffing.enabled` Form.Item registered, so useWatch suddenly returned `{enabled: false}` — still partial, because the rest of the sniffing children only register when their Form.Items mount in conditional sub-sections. Setting `forceRender: true` on every tab item forces all tab panes to mount at modal open. Every Form.Item registers immediately; the watch result reflects the full form value seeded by buildAddModeValues. This also likely resolves the earlier "Invalid discriminator value" error on submit, which surfaced when streamSettings had an unregistered security field whose Form.Item hadn't mounted yet. * refactor(frontend): align hysteria with new docs + drop hysteria2 protocol Phase 2 smoke fixes on the Inbound add flow surfaced that hysteria2 was modeled as a separate top-level protocol when it's really just hysteria v2. The xray transports/hysteria.html docs also pin the hysteria stream to a minimal shape (version/auth/udpIdleTimeout/masquerade) — the previous schema carried legacy congestion/up/down/udphop/window knobs that aren't part of the wire contract. Hysteria2 removal: - Drop 'hysteria2' from ProtocolSchema enum and Protocols const - Drop hysteria2 branches from inbound/outbound discriminated unions - Drop createDefaultHysteria2InboundSettings / OutboundSettings - Delete schemas/protocols/inbound/hysteria2.ts and outbound/hysteria2.ts - Drop hysteria2 case in getInboundClients / genLink (fell through to the hysteria handler anyway) - Update client form modals' MULTI_CLIENT_PROTOCOLS sets - Remove hysteria2-basic fixture + snapshot entries (14 capability cases, 1 protocols fixture, 1 inbound-defaults factory) - Keep parseHysteria2Link() outbound parser since hysteria2:// is the share-link URI prefix for hysteria v2 Hysteria stream alignment with xtls docs: - HysteriaStreamSettingsSchema reduced to version/auth/udpIdleTimeout/ masquerade per transports/hysteria.html - Masquerade type adds '' (default 404 page) and defaults to it - Outbound form drops Congestion/Upload/Download/UDP hop/Max idle/ Keep alive/Disable Path MTU controls and the receive-window note - newStreamSlice('hysteria') in OutboundFormModal mirrors the trimmed shape; outbound-link-parser emits the trimmed shape too - InboundFormModal Masquerade Select gains the default option New TUN inbound schema: - Add schemas/protocols/inbound/tun.ts with name/mtu/gateway/dns/ userLevel/autoSystemRoutingTable/autoOutboundsInterface - Wire into ProtocolSchema enum, InboundSettingsSchema discriminated union, createDefaultInboundSettings dispatcher Other Phase 2 smoke fixes folded in: - Tunnel portMap UI swaps Form.List for HeaderMapEditor v1 — wire shape is Record<string,string> and the List was producing arrays - Hysteria onValuesChange seeds full TLS schema defaults + one empty certificate row (Cipher Suites/Min/Max Version/uTLS/ALPN were undefined before) - HTTP/Mixed accounts Add button auto-fills user/pass with RandomUtil.randomLowerAndNum - Hysteria security tab gates the 'none' radio out — TLS only - Hysteria stream tab drops the inbound Auth password field (xray inbound auth is per-user via 'users', not stream-level) - Reality onSecurityChange auto-randomizes target/serverNames/ shortIds and fetches an X25519 keypair - Tag and DB-side fields (up/down/total/expiryTime/ lastTrafficResetTime/clientStats/security) gain hidden Form.Items so validateFields keeps them in the wire payload (rc-component form strips unregistered fields) - WireGuard inbound auto-seeds one peer with generated keypair, allowedIPs ['10.0.0.2/32'], keepAlive 0 — matches legacy - WireGuard peer rows separated by Divider with the Peer N title and a small inline remove button (titlePlacement="center") * refactor(frontend): retire class-based xray models (Step 5) Delete models/inbound.ts (3,359 lines) and outbound.ts (2,405). The Inbound/Outbound classes and ~50 sub-classes are replaced by Zod-typed data + pure functions in lib/xray/*. Consumer migration off dbInbound.toInbound(): - useInbounds: isSSMultiUser({protocol, settings}) directly - QrCodeModal: genWireguardConfigs/Links/AllLinks from lib/xray - InboundList: derives tags from streamSettings raw fields - InboundsPage: clone via raw JSON, fallback projection via schema-shape stream object, exports via genInboundLinks - InboundInfoModal: builds an InboundInfo facade locally from raw streamSettings (host/path/serverName/serviceName per network), canEnableTlsFlow + isSS2022 from lib/xray New helper: lib/xray/inbound-from-db.ts exposes inboundFromDb(raw) converting a raw DBInbound row into a schema-typed Inbound for the link-generation orchestrators. DBInbound trimmed: drops toInbound, isMultiUser, hasLink, genInboundLinks, _cachedInbound. Imports Protocols from @/schemas/primitives now that ./inbound is gone. Bundled Phase 2 fixes: - Outbound modal: Form.useWatch with preserve: true so the stream block doesn't gate itself out when network is unmounted - Inbound form adapter: pruneEmpty preserves empty objects; per-protocol client field projection via Zod safeParse; sniffing collapse to {enabled:false} - useClients invalidateAll also invalidates inbounds.root() - IndexPage Config modal top/maxHeight polish Tests: 283/283 pass. typecheck/lint clean. * fix(frontend): inboundFromDb fills Zod defaults for stream + settings Smoke-testing the new inboundFromDb helper surfaced two regressions that the strict lib/xray link generators expose when fed raw DB streamSettings without per-network sub-keys. 1. genVlessLink / genTrojanLink crash on `stream.tcpSettings.header` when streamSettings lacks `tcpSettings` (true for slim list rows and for handcrafted minimal-JSON inbounds). The legacy Inbound.fromJson chain populated TcpStreamSettings via its own constructor; the new helper now does the same by parsing the raw <network>Settings sub-object through the matching Zod schema and merging schema defaults onto whatever the DB stored. 2. genVlessLink writes `encryption=undefined` into the share URL when settings lacks the `encryption: 'none'` literal that vless wire JSON normally carries. Fixed by running raw settings through InboundSettingsSchema.safeParse() to populate per-protocol defaults (encryption, decryption, fallbacks, etc.) the same way the legacy class fromJson chain did. Same pattern applied to security branch (tls/realitySettings). Tests: src/test/inbound-from-db.test.ts covers - JSON-string / object / empty settings coercion - genInboundLinks vless (TCP/none, with encryption=none) - genWireguardConfigs + genWireguardLinks peer fanout - genAllLinks trojan with TLS sub-defaults applied - protocol-capability helpers with raw shapes - getInboundClients across vless/SS-single/non-client protocols 296/296 pass. * fix(frontend): QUIC udpHop.interval is a range string, not a number (B19) User report: "streamSettings.finalmask.quicParams.udpHop.interval: Invalid input: expected string, received number". Three-part fix: - FinalMaskForm: Hop Interval input changed from InputNumber to Input with "e.g. 5-10" placeholder. xray-core spec says interval is a range string like '5-10' (seconds between min-max hops), not a single number. - FinalMaskForm: defaultQuicParams() seeds interval: '5-10' instead of the broken `interval: 5`. - QuicUdpHopSchema: preprocess coerces number → string for legacy DB rows that were written by the now-fixed buggy UI. Stops the load-time validation crash on existing inbounds. Tests still 296/296. * fix(frontend): outbound link parser handles extra/fm/x_padding_bytes (B20) User-reported vless share link with full xhttp + reality + finalmask config failed to round-trip on outbound import. The inbound link generator emits three payloads the outbound parser was ignoring: 1. `extra=<json>` — bundles advanced xhttp knobs (xPaddingBytes, scMaxEachPostBytes, scMinPostsIntervalMs, padding-obfs keys, etc.). applyXhttpStringFromParams now JSON.parses this and merges the fields into xhttpSettings via the same JSON-branch logic used by vmess. 2. `x_padding_bytes=<range>` — snake_case alias the inbound emits alongside the camelCase form. Now applied before camelCase so explicit `xPaddingBytes` URL params still win. 3. `fm=<json>` — full finalmask object including quicParams.udpHop and tcp/udp mask arrays. New applyFinalMaskParam attaches the decoded object to streamSettings.finalmask. Wired into both parseVlessLink and parseTrojanLink. Tests: - Real B20 link parses with xhttp + reality + finalmask all populated - Precedence: camelCase URL > extra JSON > snake_case alias > default - Malformed extra JSON falls through without crashing the parser 300/300 pass. * fix(frontend): Outbound submit crash on non-mux protocols + tab a11y (B21) Two issues surfaced on Outbound save: 1. Crash: `Cannot read properties of undefined (reading 'enabled')` at formValuesToWirePayload. The modal hides the Mux switch entirely for non-stream protocols (dns/freedom/blackhole/loopback) and for stream protocols when isMuxAllowed gates it out (xhttp, vless+flow). With the field never registered, validateFields() returns no `mux` key — `values.mux.enabled` then dereferences undefined. Fix: optional chain `values.mux?.enabled` so missing mux skips the mux clause silently. Documented why mux can be absent. 2. Chrome a11y warning: "Blocked aria-hidden on an element because its descendant retained focus" — when the user has an input focused inside one Tab panel and switches to another tab, AntD marks the outgoing panel aria-hidden while focus is still inside. The browser warns, but the focused control is now invisible to AT users. Fix: blur the active element before setActiveKey in onTabChange. * fix(frontend): blur active element on every tab switch path (B21 follow-up) The previous B21 patch only blurred on user-initiated tab clicks via onTabChange. Two other paths still set activeKey while a JSON-tab input retained focus: - importLink: after a successful share-link parse, setActiveKey('1') switched to the form tab while the user's focus was still on the Input.Search they just pressed Enter in. Chrome logged the same "Blocked aria-hidden" warning because the panel they were leaving became aria-hidden synchronously, with their input still focused. - onTabChange entering the JSON tab: also did a bare setActiveKey with no blur, so going from a focused form input INTO the JSON tab could trip the warning in reverse. Fix: centralized switchTab(key) that blurs document.activeElement sync before calling setActiveKey. Every internal tab transition (importLink, onTabChange both directions) now routes through it. The single setActiveKey('1') in the open-modal useEffect is left as a plain setter because there's no focused input at modal-open time. * refactor(frontend): extract fillStreamDefaults to shared helper Move the network/security schema-default filler out of inbound-from-db.ts into stream-defaults.ts so other consumers can reuse it without dragging in the DBInbound-specific code path. * fix(frontend): derive QUIC/UDP-hop switch state from data presence (B22) The QUIC Params and UDP Hop toggles previously persisted as separate boolean flags (enableQuicParams / hasUdpHop) which weren't part of the xray wire format and weren't restored when a config was pasted into the modal. Use data presence as the single source of truth: the switch is on iff the corresponding sub-object exists. Switching off clears it back to undefined. * fix(frontend): xhttp form binding + drop empty strings from JSON (B23) uplinkHTTPMethod was wrapped Form.Item -> Form.Item(shouldUpdate) -> Select, which broke AntD's value/onChange injection (AntD only clones the immediate child). Restructured so shouldUpdate is the outer wrapper and Form.Item(name) directly wraps the Select. Also drop empty-string fields from xhttpSettings in the wire payload — fields like uplinkHTTPMethod, sessionPlacement, seqPlacement, xPaddingKey default to '' meaning "use server default", so they shouldn't appear in JSON as "field": "". Adds placeholder text to the 3 xhttp Selects so the form reflects the current value after JSON paste. * feat(frontend): align finalmask + sockopt with xray docs, add golden fixtures Schema fixes per https://xtls.github.io/config/transports/finalmask.html and https://xtls.github.io/config/transports/sockopt.html: finalmask: - QuicCongestionSchema: remove non-doc 'cubic', keep reno/bbr/brutal/force-brutal - Add BbrProfileSchema (conservative/standard/aggressive) and bbrProfile field - brutalUp/brutalDown: number -> string per docs (units like '60 mbps') - Tighten ranges: maxIdleTimeout 4-120, keepAlivePeriod 2-60, maxIncomingStreams min 8 - UdpMaskTypeSchema: add missing 'sudoku' - udpHop.interval stays as preprocessed string-range per intentional B19 divergence sockopt: - tcpFastOpen: boolean -> union(boolean, number) per docs (number tunes queue size) - mark: drop min(0) (can be any int) - domainStrategy default: 'UseIP' -> 'AsIs' per docs - tcpKeepAlive Interval/Idle defaults: 0/300 -> 45/45 per docs (outbound) - Add AddressPortStrategySchema enum (7 values) + addressPortStrategy field - Add HappyEyeballsSchema (tryDelayMs/prioritizeIPv6/interleave/maxConcurrentTry) - Add CustomSockoptSchema (system/type/level/opt/value) + customSockopt array Bug fixes: - options.ts: Address_Port_Strategy values were lowercase ('srvportonly'); xray-core requires camelCase ('SrvPortOnly'). Fixed all 6 entries. - OutboundFormModal: domainStrategy Select was mistakenly populated from ADDRESS_PORT_STRATEGY_OPTIONS; now uses DOMAIN_STRATEGY_OPTION. - OutboundFormModal: inline sockopt defaults (hardcoded {acceptProxyProtocol: false, domainStrategy: 'UseIP', ...}) replaced with SockoptStreamSettingsSchema.parse({}) so schema is the single source. Form additions (both InboundFormModal + OutboundFormModal): - Address+port strategy Select - Happy Eyeballs Switch + sub-form (tryDelayMs/prioritizeIPv6/interleave/maxConcurrentTry) - Custom sockopt Form.List (system/type/level/opt/value) - FinalMaskForm: BBR Profile Select (visible when congestion='bbr'), Brutal Up/Down placeholders updated to string format Golden fixtures (8 new + 4 xhttp extras): - finalmask/{tcp-mask, udp-mask, quic-params, combined}.json — cover all TCP mask types, 7 UDP mask types including new sudoku, full QUIC params shape - sockopt/{defaults, tcp-tuning, tproxy, full}.json — full sockopt knobs - stream/xhttp-{basic, extra-padding, extra-placement, extra-tuning}.json — cover the extra-blob fields bundled into share-link extra=<json> Tests now at 312 (up from 300); typecheck/lint clean. * feat(frontend): migrate DNS + Routing to Zod, align with xray docs Adds first-class Zod schemas for the xray-core DNS block and routing sub-objects (Balancer, Rule) matching the documented shape at https://xtls.github.io/config/dns.html and https://xtls.github.io/config/routing.html, then wires the DnsServerModal and BalancerFormModal up to those schemas. schemas/dns.ts (new): - DnsQueryStrategySchema enum (UseIP/UseIPv4/UseIPv6/UseSystem) - DnsHostsSchema record(string -> string | string[]) - DnsServerObjectInnerSchema + DnsServerObjectSchema (with preprocess to migrate legacy `expectIPs` -> `expectedIPs` alias) - DnsServerEntrySchema = string | DnsServerObject (xray accepts both) - DnsObjectSchema with all documented fields and defaults schemas/routing.ts (new): - RuleProtocolSchema enum (http/tls/quic/bittorrent) - RuleWebhookSchema (url/deduplication/headers) - RuleObjectSchema covering every documented field (domain/ip/port/ sourcePort/localPort/network/sourceIP/localIP/user/vlessRoute/ inboundTag/protocol/attrs/process/outboundTag/balancerTag/ruleTag/ webhook) with type=literal('field').default('field') - BalancerStrategyTypeSchema enum (random/roundRobin/leastPing/leastLoad) - BalancerCostObjectSchema {regexp,match,value} - BalancerStrategySettingsSchema (expected/maxRTT/tolerance/baselines/costs) - BalancerStrategySchema + BalancerObjectSchema schemas/xray.ts: - routing.rules: was loose 3-field object, now z.array(RuleObjectSchema) - routing.balancers: was z.array(z.unknown()), now z.array(BalancerObjectSchema) - dns: was 2-field loose, now full DnsObjectSchema - BalancerFormSchema: strategy now BalancerStrategyTypeSchema (enum) instead of z.string(); fallbackTag defaults to ''; settings? added for leastLoad DnsServerModal (full Pattern A rewrite): - useState/DnsForm interface -> Form.useForm<DnsServerForm>() - manual domain/expectedIP/unexpectedIP list -> Form.List - antdRule on address/port/timeoutMs for inline validation - preserves legacy collapse-to-bare-string behavior on submit BalancerFormModal: - Adds conditional leastLoad sub-form (Expected/MaxRTT/Tolerance/ Baselines/Costs) wired to BalancerStrategySettingsSchema - Strategy options derived from schema enum - Cost rows with regexp/literal switch + match + value - required prop on Tag and Selector for red asterisk visual BalancersTab: - BalancerRecord interface -> type alias to BalancerObject - onConfirm now propagates strategy.settings to wire when leastLoad - Removes useMemo wrapping `columns` array. The memo had deps [t, isMobile] (with an eslint-disable) so the column render functions kept their original closure over `openEdit`. Once a balancer was created and the user clicked the edit button, the stale openEdit fired with empty `rows`, so rows[idx] was undefined and the modal opened blank. Columns are cheap to rebuild each render, so dropping the memo is the right fix. DnsTab + RoutingTab: switch ad-hoc interfaces to schema-derived types. translations (en-US, fa-IR): add the previously-missing pages.xray.balancerTagRequired and pages.xray.balancerSelectorRequired keys so antdRule surfaces a real message instead of the raw i18n key. * test(frontend): golden fixtures for DNS, Balancer, Rule schemas Adds JSON fixtures under golden/fixtures/{dns,dns-server,balancer,rule} plus three vitest files that parse them through the new schemas and snapshot the result. dns/: minimal (servers as strings) + full (every top-level field plus hosts with geosite/domain/full prefixes and 5 mixed string/object servers covering fakedns, localhost, https://, tcp://, quic+local://). dns-server/: full (every DnsServerObject field) + legacy-expectips (asserts the z.preprocess that migrates the legacy `expectIPs` key into the canonical `expectedIPs`). balancer/: random-minimal (default strategy by omission), roundrobin, leastping, leastload-full (covers all StrategySettings fields and both regexp=true|false costs). rule/: minimal, full (exercises every RuleObject field including localPort, localIP, process aliases like `self/`, all four protocol enum values, ip negation `!geoip:`, attrs with regexp value, and the WebhookObject with deduplication+headers), balancer-routed (uses balancerTag instead of outboundTag), port-number (port as a number to prove the union(number,string) accepts both). * fix(frontend): serialize bulk client delete + drop deprecated Alert.message useClients.removeMany was firing all DELETEs in parallel via Promise.all. The 3x-ui backend mutates a single config JSON per request (read / modify / write), so 20 concurrent deletes raced on the same file: every request reported success, but only the last writer's copy stuck — about half the selected clients reappeared after the toast. Replace the parallel fan-out with a sequential for-of loop so each delete sees the committed state of the previous one. The trade-off is total latency (20 * ~250ms = ~5s) which is the correct behavior until the backend grows a proper /bulkDel endpoint. Also rename the Alert `message` prop to `title` in ClientBulkAdjustModal to clear the AntD v6 deprecation warning. * feat(clients): server-side bulk create/delete with per-inbound batching Replace the panel-side fan-out (Promise.all of single /add and /del calls) that raced on the shared inbound config and capped throughput at roughly one round-trip per client. New endpoints batch the work on the server: - POST /panel/api/clients/bulkDel { emails, keepTraffic } - POST /panel/api/clients/bulkCreate [ {client, inboundIds}, ... ] BulkDelete groups emails by inbound and performs a single read-modify-write per inbound (one JSON parse, one marshal, one Save) instead of N. Per-row DB cleanups (ClientInbound, ClientTraffic, InboundClientIps, ClientRecord) are batched with WHERE...IN queries. Per-email failures are reported via Skipped[] and processing continues. BulkCreate iterates payloads sequentially through the same Create path single-add uses, so heterogeneous batches (different inboundIds, plans) remain valid in one round-trip. Frontend bulkDelete/bulkCreate hooks parse the new response shape ({ deleted|created, skipped[] }) and the bulk-add modal now posts a single request instead of fanning out emails. * perf(clients): batch BulkAdjust per inbound, skip no-op xray calls on local Same per-inbound batching strategy as BulkDelete. The previous code called Update once per email, which itself looped through each inbound the client belonged to — reparsing the same settings JSON, calling RemoveUser+AddUser on xray, and running SyncInbound for every single email. For 200 emails in one inbound that's 200 JSON read/write cycles and 400 xray runtime calls. The new BulkAdjust groups emails by inbound and per inbound: - locks once, reads settings JSON once - mutates expiryTime/totalGB in place for every target client - writes the inbound and runs SyncInbound once ClientTraffic rows are updated with a single per-email query at the end (values differ per client so they can't be folded into one statement). For local-node inbounds the xray runtime calls are skipped entirely. The AddUser payload only contains email/id/security/flow/auth/password/ cipher — none of which change in an adjust — so RemoveUser+AddUser was a no-op that briefly flapped active users. Limit enforcement is driven by the panel's traffic loop reading ClientTraffic, not by xray-core. For remote-node inbounds rt.UpdateUser is preserved so the remote panel receives the new totals/expiry. Skip+report semantics match BulkDelete: any per-email error leaves that email's record/traffic untouched and is returned in Skipped[]. * refactor(backend): retire hysteria2 as a top-level protocol Hysteria v2 is not a separate xray protocol — it is plain "hysteria" with streamSettings.version = 2. The frontend already dropped hysteria2 from the protocol enum in 5a90f7e3; the backend was still carrying the literal as a compat alias. Removed: - model.Hysteria2 constant - model.IsHysteria helper (only callers were buildProxy + genHysteriaLink) - TestIsHysteria - "hysteria2" from the Inbound.Protocol validate oneof enum - All `case model.Hysteria, model.Hysteria2:` and `case "hysteria", "hysteria2":` branches across client.go, inbound.go, outbound.go, xray.go, port_conflict.go, xray/api.go, subService.go, subJsonService.go, subClashService.go - Stale #4081 comments Kept (correctly — these are client-side URI/config schemes that are independent of the xray protocol type): - hysteria2:// share-link URI in subService.genHysteriaLink - "hysteria2" Clash proxy type in subClashService.buildHysteriaProxy - Comments referring to Hysteria v2 as a transport version Note: this change does not include a DB migration. Existing rows with protocol = 'hysteria2' will fall through to the default switch arms after upgrade. A separate `UPDATE inbounds SET protocol = 'hysteria' WHERE protocol = 'hysteria2'` is required for installs that still hold legacy data. * refactor(frontend): retire all AntD + Zod deprecations Swept the codebase for @deprecated APIs using a one-off type-aware ESLint config (eslint.deprecated.config.js) and fixed every hit: - 78 instances of `<Select.Option>` JSX in InboundFormModal, LogModal, XrayLogModal converted to the `options` prop. - Zod's `z.ZodTypeAny` (deprecated for `z.ZodType` in zod v4) replaced in _envelope.ts, zodForm.ts, zodValidate.ts, and inbound-form-adapter.ts. - Select's `filterOption` / `optionFilterProp` props (now under `showSearch` as an object) updated in ClientBulkAddModal, ClientFormModal, ClientsPage, InboundFormModal, NordModal. - `Input.Group compact` swapped for `Space.Compact` in FinalMaskForm. - Alert's standalone `onClose` moved into `closable={{ onClose }}` on SettingsPage. - `document.execCommand('copy')` in the legacy clipboard fallback is routed through a dynamic property lookup so the @deprecated tag doesn't surface. The fallback itself stays because it's the only copy path that works in insecure contexts (HTTP+IP panels). The dropped ClientFormModal.css was already unimported. eslint.deprecated.config.js loads the type-aware ruleset and turns everything off except `@typescript-eslint/no-deprecated`, so future scans are a single command: npx eslint --config eslint.deprecated.config.js src Not wired into `npm run lint` because typed linting roughly triples the run time. Verified clean: typecheck, lint, and the deprecated scan all 0 warnings. * feat(clients): show comment under email in the Client column The clients table's Client cell already stacks email + subId; add the admin comment as a third muted line so notes like "VIP" or "friend of X" are visible in the list view without opening the info modal. Renders only when set, so rows without a comment look unchanged. * docs(frontend): refresh README + simplify deprecated-scan config README rewrite reflects the post-Zod-migration state: - 3 Vite entries (index/login/subpage), not "one per panel route" - New folders: schemas/, lib/xray/, generated/, test/, layouts/ - Scripts table covers test/gen:api/gen:zod alongside the existing dev/build/lint/typecheck - New sections on the Zod schema tree, the three validation layers, the unified Form.useForm + antdRule pattern, and the golden fixture testing setup - "Adding a new page" updated to reflect that most additions are just react-router entries in routes.tsx, not new Vite bundles - Explicit note that `@deprecated` in the prose is a JSDoc tag, not a shell command — comes with the exact one-line npx invocation eslint.deprecated.config.js trimmed: dropping the recommendedTypeChecked spread + the ~28 rule overrides that came with it. The config now wires the @typescript-eslint and react-hooks plugins manually and enables exactly one rule (`@typescript-eslint/no-deprecated`). 45 lines → 30, same output: zero false-positives, zero noise, zero deprecations on the current tree. * chore(frontend): bump deps + refresh lockfile `npm update` within the existing semver ranges, plus a Vite bump the user explicitly accepted: - vite 8.0.13 → 8.0.14 (exact pin kept) - dayjs 1.11.20 → 1.11.21 - i18next 26.2.0 → 26.3.0 - typescript-eslint 8.59.4 → 8.60.0 - @rc-component/table + a handful of other transitive antd deps resolved to newer patch versions in the lockfile The earlier 8.0.13 pin was carried over from an esbuild dep-optimizer regression that broke vue-i18n in Vite 8.0.14 dev mode. This codebase uses react-i18next, doesn't hit the same chunking edge case, and `npm run dev` was smoked clean on 8.0.14 before accepting the bump. * feat(clients): compact link + inbound rows in the info modal and table ClientInfoModal — Copy URL section reskinned: - Each link is a single row: [PROTOCOL] [remark] [copy] [QR] instead of a card with the raw 200-char URL printed inline - Remark is parsed per-protocol — VMess pulls it from the base64-JSON `ps` field, the rest from the `#fragment` - The row title strips the client email suffix so the same string isn't repeated three times in the modal; the QR popover still uses the full remark (it's the QR's own name for the download file) - QR button opens an inline Popover with the existing QrPanel, size 220, destroyed on close - Subscription section uses the same row layout (SUB / JSON tags, clickable subId, copy + QR actions) - New per-protocol Tag colors so the protocol is identifiable at a glance ClientInfoModal — Attached inbounds + ClientsPage table column: - Chip format changed from `${remark} (${proto}:${port})` to just `${proto}:${port}` — when an admin attaches 5 inbounds to one client the remark was repeated 5 times and wrapped onto two lines - Only the first inbound chip is shown; the rest collapse into a `+N` chip that opens a Popover with the full list (remark included). INBOUND_CHIP_LIMIT = 1 - Per-protocol Tag colors - Tooltip on each chip shows the full `${remark} (${proto}:${port})` - Table column pinned to width: 170 so the row doesn't reserve the old 300px of whitespace next to the compact chip Comment row in the info table is always shown now (renders `-` when unset) so the layout doesn't jump per-client. VmessSecuritySchema gets a preprocess pass that maps legacy `security: ""` (persisted on pre-enum-lock VMess inbounds) back to `'auto'`. z.enum's `.default()` only fires on a missing field, not on an empty string — without this, old rows fail validation with "expected one of aes-128-gcm|chacha20-poly1305| auto|none|zero". `z.infer` is taken from the raw enum so the inferred type stays the union, not `unknown`. i18n adds a `more` key (en-US + fa-IR) used by the overflow chip label. * fix(xray): heal shadowsocks per-client method across all start paths xray-core's multi-user shadowsocks insists the per-client `method` matches the inbound's top-level cipher exactly for legacy ciphers, and is empty for 2022-blake3-*. The previous code (xray.go) copied `Client.Security` into the per-client `method` blindly, so a multi-protocol client created with the VMess default `"auto"` poisoned the SS config with `method: "auto"` → "unsupported cipher method: auto". Fix in two parts: - GetXrayConfig no longer projects `Client.Security` into the SS entry; the inbound's top-level method is now the single source of truth. - HealShadowsocksClientMethods moves to `database/model` and is invoked from `Inbound.GenXrayInboundConfig`, so the runtime add/update path (runtime.AddInbound) is normalised in addition to the full-restart path. For legacy ciphers heal now overwrites mismatched per-client methods rather than preserving them, so stale DB rows are also healed. * feat(sub): compact subscription rows with per-link email + PQ QR hide Mirror the ClientInfoModal redesign on the public SubPage so the subscription viewer reads as a tight `[PROTO] [remark] [copy] [QR]` row per link instead of raw URL cards. - subService.GetSubs now returns the per-link email list alongside the links, threaded through subController and BuildPageData into the `emails` field on subData (env.d.ts updated). Public links.go is updated to ignore the new return. - SubPage strips the client email from each row title using the matched per-link email (same trimEmail behaviour as the modal), and hides the QR button for post-quantum links (`pqv=`, `mlkem768`, `mldsa65`) since the encoded URL won't fit in a single QR. * feat(clients): hide QR for post-quantum links in client info modal Post-quantum keys (mldsa65 / ML-KEM-768) blow the encoded URL past what a single QR can hold. Detect them by the markers VLESS share links actually carry — `pqv=<base64>` for mldsa65Verify and `encryption=mlkem768x25519plus.*` for ML-KEM-768 — and drop the QR button for those rows. Copy still works. * fix(schemas): widen VLESS decryption/encryption to accept PQ values The post-quantum auth blocks (ML-KEM-768, X25519) populate `settings.decryption` / `settings.encryption` with values like `mlkem768x25519plus.<base64>` and `xchacha20-poly1305.aead.x25519`, but the schema pinned both fields to z.literal('none') so saving an inbound after picking "ML-KEM-768 auth" failed with `Invalid input: expected "none"`. Relax both fields (inbound + outbound + outbound form) to z.string().min(1) keeping the 'none' default. xray-core does its own validation server-side so a string check at the form boundary is enough. * feat(sub): clash row + reorganise SubPage around Subscription info ClientInfoModal: - Add a Clash / Mihomo row to the subscription section, gated on subClashEnable + subClashURI from /panel/setting/defaultSettings. Defaults payload schema is widened to carry subClashURI/subClashEnable. SubPage: - Drop the rectangular QR-codes header that used to sit at the very top of the card. The subscription info table now leads, followed by Divider("Copy URL") + per-protocol link rows (already converted to the compact ClientInfoModal pattern), then a new Divider("Subscription") + compact rows for the SUB / JSON / CLASH URLs with copy + QR-popover actions. The apps dropdown row remains the footer. CSS clean-up: removed the now-unused .qr-row/.qr-col/.qr-box/.qr-code rules; kept .qr-tag and trimmed the info-table top gap. Added a .sub-link-anchor underline-on-hover style for the new URL rows. * fix(sub): multi-inbound traffic + trojan/hysteria userinfo + utf-8 vmess remark Three bugs surfaced by the new SubPage and the recent client-record refactor: - xray.ClientTraffic.Email is globally unique, so a multi-inbound client has exactly one traffic row attached to whichever inbound claimed it. Iterating inbound.ClientStats per inbound dedup-locked the first lookup to zero for clients that lived under any other inbound, so the SubPage info table read 0 B for all the multi- inbound subs. Replaced appendUniqueTraffic with a single AggregateTrafficByEmails(emails) helper that runs one WHERE email IN (?) over xray.ClientTraffic and folds the rows. GetSubs / SubClashService.GetClash / SubJsonService.GetJson all share it. - Trojan and Hysteria share-links embedded the raw password/auth into the userinfo (scheme://<value>@host) without percent-encoding, so passwords containing `/` or `=` (e.g., base64-with-padding) broke popular trojan clients with parse errors. Added encodeUserinfo() that wraps url.QueryEscape and rewrites the `+` (space) back to `%20` for parity with encodeURIComponent on the frontend; applied to trojan.password and hysteria.auth. Same fix on the frontend's genTrojanLink. - VMess link remarks ride inside a base64-encoded JSON payload, but the SubPage / ClientInfoModal parser used JSON.parse(atob(body)), which treats the binary string as Latin-1 and shreds any multi-byte UTF-8 sequence. Most visible on the emoji decorations (genRemark appends 📊/⏳), so a remark like `test-1.00GB📊` rendered as `test-1.00GBð…`. Routed through Uint8Array + TextDecoder('utf-8') so multi-byte codepoints survive. * feat(settings): drop email leg from default remark model Change the default remarkModel from "-ieo" to "-io" so a freshly installed panel composes share-link remarks from the inbound name + optional extra only, leaving out the client email. Existing panels keep whatever value they have saved — only fresh installs and fallback paths (parse failure, missing setting) pick up the new default. Touched everywhere the literal "-ieo" lived: the canonical default map, the two sub-package fallback constants, the four frontend defaults (model class, link generator, two inbound modals, useInbounds hook). Two snapshot tests regenerated and one obsolete "contains email" assertion in inbound-from-db.test.ts removed. To migrate an existing panel that wants the new behaviour, edit Settings → Remark Model and remove the email leg. * feat(sub): usage summary card + remark-email on QR popover labels SubPage now opens with a clear quota panel directly under the info table: large `used / total` numbers, gradient progress bar (green ≤ 75%, orange to 90%, red above), `remained` and `%` on the foot, plus a Tag chip for unlimited subscriptions and a coloured chip for days left until expiry (blue >3d, orange ≤3d, red on expiry). Driven entirely off existing subData fields — no backend changes. While the row title in the link list stays email-stripped (default remark model omits email now), the QR popover label folds it back in so the rendered QR card identifies the client unambiguously. Tag content becomes `<rowTitle>-<email>` in both SubPage and ClientInfoModal — the encoded link itself is unchanged. SubPage section order is now: info table → usage summary → SUB / JSON / CLASH endpoints → per-protocol Copy URL rows → apps row, so the most-glanceable status sits above the fold.
2026-05-26 19:26:50 -07:00
case model.Hysteria:
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
if c.Auth != "" {
entry["auth"] = c.Auth
}
feat(xray): add the MASQUE protocol and transport xray-core v26.10.10 serves MASQUE: CONNECT-IP over HTTP/3 or HTTP/2, always behind TLS, as an inbound and an outbound protocol riding a dedicated "masque" transport. The panel could not create either. Inbound: MASQUE is a multi-user protocol whose users log in with their email and password over Basic auth. The panel keeps the client's "password" like trojan and GenXrayInboundConfig hands it to the core as "pass", so the full-config path, the live AddInbound path and node pushes all emit what the core reads; one user with an empty pass makes the core refuse the whole inbound. Live user adds build a masque.Account from either key, client-only changes diff per user so a new client does not drop every connected tunnel, and client creation, copy, validation and edits key on the password. The address pool and MTU get form fields, TLS is mandatory, and the port check follows the core's listener choice: HTTP/2 on TCP when the ALPN offers h2, HTTP/3 on UDP when it offers h3 or not h2. Nodes may host MASQUE from v3.9.1, the first panel release that knows to rename the password. MASQUE has no share-link format, so links and Clash skip it; the JSON subscription builds the client's masque outbound with its email and password, and the subscription allowlist now lets a subId see MASQUE inbounds at all. The IP-limit job, client pages and traffic tracking list the protocol. Outbound: the form edits the server, remote DNS, and the transport's host, path, Basic credentials, headers and WARP block (enrolled key, endpoint public key, tunnel addresses), which xray-core refuses next to user/pass. The outbound test probes it through the core, since its default HTTP/3 dial is UDP.
2026-10-11 01:26:21 -07:00
case model.MASQUE:
if c.Password != "" {
entry["password"] = c.Password
}
case model.WireGuard:
if inboundClient, ok := wireguardClientsByEmail[strings.ToLower(strings.TrimSpace(c.Email))]; ok {
c.AllowedIPs = inboundClient.AllowedIPs
c.PreSharedKey = inboundClient.PreSharedKey
c.PublicKey = inboundClient.PublicKey
// #6731: a nil settings keepalive must not keep the other tunnel's value.
if inboundClient.KeepAlive == nil {
c.KeepAlive = nil
} else {
keepalive := *inboundClient.KeepAlive
c.KeepAlive = &keepalive
}
}
fix(wireguard): build peers in GenXrayInboundConfig so node reconcile keeps clients (#5684) Adding a WireGuard client on the master broke every WireGuard connection on the sub-node until Xray was manually restarted on the node. Adding the same client directly on the node worked. Root cause: the panel stores WireGuard clients under the settings key `clients` (the shape every other protocol uses), but xray-core's wireguard inbound is configured with `peers`. The `clients`->`peers` conversion lived only in the full-config generation path (XrayService.GetXrayConfig), which runs on a full Xray restart. The live gRPC AddInbound path goes through (*Inbound).GenXrayInboundConfig, which passed the WireGuard settings verbatim - with `clients` and no `peers`. Why the master path broke it and the node path did not: - Adding on the node is a single safe operation: AddInboundClient -> AddUser -> AlterInbound{AddUser} -> wireguard.Server.AddUser, which appends one peer via IPC without touching the others. The inbound is local (NodeID == nil), so nothing is marked dirty and no reconcile runs. - Adding on the master does two things: it pushes the client to the node (the same safe hot-add, which succeeds), and it marks the node dirty. The reconcile then pushes panel/api/inbounds/update/:id to the node, whose InboundService.UpdateInbound applies it live via DelInbound + AddInbound (buildRuntimeInboundForAPI -> Local.AddInbound -> GenXrayInboundConfig). That re-adds the wireguard inbound with zero peers, wiping the device and dropping every connected client. A manual restart regenerated the full config, converted clients to peers, and restored them - hence "only a restart fixes it". Fix: convert WireGuard `clients` to `peers` in GenXrayInboundConfig itself, the single chokepoint for every live AddInbound (create, edit, node reconcile). WireguardClientsToPeers always rebuilds `peers` from `clients` (matching GetXrayConfig field for field) and drops the `clients` key. It does not gate on `peers` being absent: the panel seeds every WireGuard inbound with an empty `peers: []` placeholder (frontend inbound-defaults), so a "skip if peers present" guard would match that placeholder and make the conversion never run, leaving the live path emitting zero peers. The conversion stays idempotent by removing `clients`, so a second call - or an inbound with no `clients` - is a no-op, leaving the full-config path unaffected. This also fixes plain WireGuard inbound edits on a standalone panel, which went through the same peerless rebuild.
2026-07-08 15:52:03 -07:00
wgPeers = append(wgPeers, model.WireguardPeerFromClient(c))
continue
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
}
finalClients = append(finalClients, entry)
}
if excludedClients > excludedBefore {
excludedInbounds++
}
var mutated bool
if inbound.Protocol == model.WireGuard {
delete(settings, "clients")
if wgPeers == nil {
wgPeers = []any{}
}
settings["peers"] = wgPeers
mutated = true
} else {
_, hadClients := settings["clients"]
mutated = hadClients || len(finalClients) > 0
if mutated {
settings["clients"] = finalClients
}
if inbound.Protocol == model.Shadowsocks && xray.LockShadowsocks2022(settings) {
mutated = true
}
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
}
if inboundCanHostFallbacks(inbound) {
fallbacks, fbErr := s.inboundService.fallbackService.BuildFallbacksJSON(nil, inbound.Id)
if fbErr != nil {
return nil, fbErr
}
if len(fallbacks) > 0 {
generic := make([]any, 0, len(fallbacks))
for _, f := range fallbacks {
generic = append(generic, f)
}
settings["fallbacks"] = generic
mutated = true
}
}
if mutated {
modifiedSettings, err := json.MarshalIndent(settings, "", " ")
2023-02-09 11:18:06 -08:00
if err != nil {
return nil, err
}
inbound.Settings = string(modifiedSettings)
}
if len(inbound.StreamSettings) > 0 {
// Unmarshal stream JSON
var stream map[string]any
_ = json.Unmarshal([]byte(inbound.StreamSettings), &stream)
// Remove the "settings" field under "tlsSettings" and "realitySettings"
tlsSettings, ok1 := stream["tlsSettings"].(map[string]any)
realitySettings, ok2 := stream["realitySettings"].(map[string]any)
if ok1 || ok2 {
if ok1 {
delete(tlsSettings, "settings")
} else if ok2 {
delete(realitySettings, "settings")
}
}
delete(stream, "externalProxy")
fix(inbound): reject finalmask + REALITY combo (crashes Xray-core) (#5861) * fix(inbound): reject finalmask configured together with REALITY security finalmask wraps the connection before REALITY's own handshake takes over (TcpmaskManager.WrapListener -> WrapConnServer runs at Accept() time, ahead of reality.Server()). reality.Server() does an unchecked type assertion assuming a raw *net.TCPConn; with finalmask in front, that assertion panics and takes down the entire xray-core process on the very first connection to the inbound - not just that connection. Upstream (XTLS/Xray-core#6453) confirmed this will be documented as unsupported rather than made graceful, so the panel needs to stop this combination from being saved rather than relying on docs. AddInbound/UpdateInbound now reject streamSettings with security=reality and a non-empty finalmask.tcp/udp with a clear error instead of letting it reach Xray. Related: MHSanaei/3x-ui#5857 * fix(inbound): heal legacy rows and narrow the finalmask+REALITY guard Per review feedback on #5861: - Narrow the check to finalmask.tcp only. xray-core's TcpmaskManager (the thing that wraps the TCP listener ahead of REALITY's handshake, the actual cause of the panic) is only constructed when tcp masks are present; a finalmask.udp-only config never touches that accept path and doesn't reproduce the crash, so it shouldn't be rejected. Extracted the shared check into finalMaskRealityTcpMasks() so both the save-time guard and the config-build heal below use one definition of "dangerous". - Heal already-saved bad rows in GetXrayConfig(), the same way liftXhttpSessionIDKeys and HealShadowsocksClientMethods heal other legacy data at config-build time. AddInbound/UpdateInbound only cover the two save paths - a row that already carries this combination (saved before this guard existed, synced from a node, restored from a backup, or edited directly in the DB) would still crash Xray-core on the next restart without this. - Add end-to-end tests exercising AddInbound, UpdateInbound, and GetXrayConfig directly (seeding rows through the real DB) rather than only unit-testing the extracted helper in isolation, so a wiring regression in any of the three call sites gets caught.
2026-07-08 11:29:51 -07:00
// finalmask.tcp + REALITY panics Xray-core on the first connection
// (XTLS/Xray-core#6453). AddInbound/UpdateInbound reject this
// combination at save time, but a row saved before that guard
// existed (upgrade, node sync, restored backup, direct DB edit)
// would still crash Xray on the next restart without this — drop
// it here too, the same way liftXhttpSessionIDKeys and
// HealShadowsocksClientMethods heal other legacy data in place.
if len(finalMaskRealityTcpMasks(stream)) > 0 {
logger.Warningf("Inbound %q: dropping finalmask, incompatible with REALITY security (crashes Xray-core, see XTLS/Xray-core#6453)", inbound.Tag)
delete(stream, "finalmask")
}
fix(xray): reject configs xray-core refuses, and check the fixtures against it The frontend's golden fixtures are the panel's model of an xray config, but nothing ever asked xray-core whether it would accept them: the snapshots only prove the Zod schemas agree with themselves. Building every fixture through the same config builders the panel hands its config to — conf.InboundDetourConfig for the full-config and AddInbound paths, conf.RouterConfig for ApplyRoutingConfig, conf.DNSConfig for the dns section — found seven the core refuses, three of them reachable from the panel's own UI. A refusal is not scoped to one inbound: the config fails to load and every inbound stays down. Hysteria: xray-core builds version 2 only, in both the protocol settings and the transport settings, but the inbound settings schema accepted any version from 1 up and its comment claimed upstream still supported v1. Both fixtures carried version 1. The schema now pins 2, GenXrayInboundConfig heals stored rows on the way out the way it already heals shadowsocks ciphers and wireguard peers, and the share link drops the dead hysteria:// scheme — the subscription server already emitted hysteria2:// for the same inbound. XHTTP uplinkDataPlacement: both transport forms offered "query", which the core has never accepted for that field (auto and body always, cookie and header in packet-up mode). Replaced with auto, which was missing, and the default label now names auto rather than body. FinalMask items: switching an item to the rand-driven array kind wrote packet:[] next to the rand. xray-core counts an empty array as a packet and every item kind is exclusive, so noise answers "len(item.Packet) > 0 && item.Rand.To > 0" and header-custom "exactly one item kind must be set". The editor now clears the packet, and GetXrayConfig strips the residue from rows already saved with it. The remaining four were stale fixtures: an xmc mask still on the usernames shape v26.7.28 replaced with profiles, a fragment mask with no length, and header-custom and noise items passing an array to the string packet kind — all shapes the panel's own editors cannot produce. golden_fixtures_xray_test.go keeps this from drifting again: every fixture in every category is built through xray-core on each run, with a self-signed pair standing in for the deployment certificate paths, so the next core bump reports which fixture it broke.
2026-07-28 05:43:55 -07:00
dropEmptyRandPackets(stream["finalmask"])
feat(xray): update xray-core to v26.7.28 and adapt panel Bump xtls/xray-core to 5ca6f4b7d4dc (v26.7.28) and move the three binary pins (DockerInit.sh, the Linux and Windows URLs in release.yml) in lockstep so the in-process conf.Build() validation and the child binary agree. XMC finalmask (#6487) is the breaking change. The mask's `usernames` string list is gone, replaced by a required `profiles` array whose entries each need a 3-16 character [A-Za-z0-9_] username, a parseable UUID and both Mojang texture fields; the "default to Dream when empty" fallback was removed, so an xmc mask saved by an older panel now fails to build and takes the whole config down with it rather than degrading one inbound. The textures are a signed blob only Mojang's session server can issue, so a legacy username cannot be upgraded automatically. The panel now: - rejects an incomplete xmc mask at save time (AddInbound/UpdateInbound), pointing at the specific field that is missing; - drops only the offending mask when generating the core config, for rows that never went through the form (upgrade, node sync, restored backup, direct DB edit), warning which inbound lost its obfuscation instead of leaving every inbound offline; - carries legacy usernames into profile stubs in the finalmask form so the operator keeps their player names and sees exactly what still needs filling in, and edits profiles through a list editor. No destructive DB migration: unlike the removed shadowsocks ciphers there is no valid replacement to rewrite to, and dropping the mask from stored rows would discard the operator's hostname and password for config they can still repair. The generation-time strip already prevents the startup failure. Also track the core's xmux maxConnections fallback, lowered from 6 to 3 for anti-TSPU, in the fresh-XMUX seed so a new panel config matches what the core would pick on its own. TUN gained a `desc` key and random utunN naming, but the Go validator no longer accepts TUN inbounds and the panel only renders legacy saved rows, so nothing there needs adapting. The remaining commits are REALITY log-warning wording, gRPC/XHTTP localAddr accuracy and a routing tweak, none of which change the JSON config surface. Tests cross-check the panel's profile predicate against conf.XMCProfile.Build() so a future core release that tightens or relaxes the rules fails loudly rather than silently emitting configs the core refuses to start on.
2026-07-28 04:14:06 -07:00
if dropped := stripIncompleteXmcMasks(stream); dropped > 0 {
logger.Warningf("Inbound %q: dropping %d XMC finalmask mask(s) without complete Minecraft profiles — reconfigure them to restore the obfuscation (see XTLS/Xray-core#6487)", inbound.Tag, dropped)
}
feat(xray): update xray-core to v26.10.10 and adapt panel Move the three Xray binary pins (DockerInit.sh, both release.yml fetches) to v26.10.10, in lockstep with the xtls/xray-core module that 82c77111 already bumped to 701af60772cd. xdns finalmask (#7090): a domain now takes a "names" list instead of a single "name", and a resolver is an "addrs" list of udp:// or tcp:// URLs instead of {type, settings:{addr}}. The loader ignores the old keys rather than failing, so every mask the 26.9.30 seeder wrote would load with no domain and no resolver and the tunnel would carry nothing. maskcompat.UpgradeLegacyXdns now lifts both the pre-26.9.30 string lists and the 26.9.30 objects straight to names/addrs, dropping entries the old loader refused. A new seeder (XdnsFinalmaskNamesAddrsFix) reruns it over every stored finalmask, and the save path, GetXrayConfig, both link importers and the mask editor's mount lift go through it. Template outbounds now get the build-time heal too: a 26.9.30 template pasted after the seeder passes the save check, since the core loads that shape without error. The editor edits names and addrs as tag lists and no longer requires record types, which the core now defaults (TXT on a client, A/CNAME/TXT/AAAA on a server). The wire format did not change, but clients on an older core cannot read the new config shape. Lua scripts (#6823): dns.script, routing.script and dns servers[].id are new. The xray page parsed the dns block with a strict schema, so loading it dropped script and id and the next save deleted them; the schema now keeps both and the DNS server modal carries the id through an edit. The hot routing apply built routing in-process, and that build now resolves the script file from the panel's working directory, so a script beside the xray binary failed every rule change into a full restart and broke the node bridge. RoutingService cannot change the script anyway, so the panel no longer resolves it there. TLS client roots (#7105): the core now verifies against its bundled Mozilla roots instead of the OS store unless tlsSettings.useSystemCA is set. The outbound TLS form gets a switch for it, so an outbound to a server signed by a privately installed CA can opt back in.
2026-10-10 22:55:12 -07:00
// A row that skipped the save path can still carry an xdns shape older than 26.10.10.
feat(xray): update xray-core to v26.9.30 and adapt panel Bump xtls/xray-core to b26a91de4f (v26.9.30) and the three binary pins (DockerInit.sh, release.yml Linux + Windows) in lockstep. No deleted symbols; the sing and sing-shadowsocks indirect deps drop out with the SS2022 rewrite. XDNS finalmask (#6718) replaced its string lists with objects: domains are {name, types, edns0, lenLimit, labelLimit} and client resolvers {type, settings.addr}. The loader no longer parses the old lists, so a single stored xdns mask keeps the whole core from starting. The new leaf package internal/util/maskcompat converts them: "name[:type]" becomes a domain and "name[:type]+udp://addr" a domain plus a udp resolver. A bare name maps to TXT, the type legacy clients queried by default, and each converted domain keeps the 1232-byte EDNS0 the old code always used (without it the server caps answers at 512). It runs from: - the XdnsFinalmaskObjectsFix seeder, over inbound streams, hosts, the xray template, the global sub-JSON mask and cached subscription outbounds; - inbound save (normalizeStreamSettings) and GetXrayConfig, for rows that never went through the seeder; - both link importers, since fm= from an older panel carries the lists. The finalmask form edits the object shape (every key needs a registered field, or the finalmask watch drops it on save) and lifts legacy masks on open. The udp-mask golden fixture moves to the object shape, which TestGoldenStreamFixturesBuildInXray now builds through the core. The wire format changed as well, so pre-upgrade clients need the new core. WireGuard outbound (#6771) dropped settings.domainStrategy and the remoteDNS "local" mode. The endpoint lookup now follows sockopt.domainStrategy and in-tunnel targets the outbound's targetStrategy. The old key is silently ignored, which undid the IPv4-first endpoint lookup the WARP outbound depends on (#5205), and "local" now panics the core at startup because remoteDNS goes through netip.MustParseAddr. The WireguardDomainStrategyFix seeder moves a stored family preference to both keys (a value already set wins) and turns "local" into targetStrategy; the outbound form lifts legacy rows the same way and drops its select, the WARP modal writes the new placement, and the inbound form loses a field the server never read. ValidateOutboundConfig now refuses a non-IP remoteDNS entry, which conf.Build() lets through, on template save and for outbound subscriptions. Noise finalmask items accept type "exp" (#6862), a tag expression. The form offers it for noise items only: header-custom items go through the core's PraseByteSlice, which refuses it. TUN gained autoSystemDnsToGateway (Linux) and autoSystemWfpBlockLeak (Windows). Both pass through the settings schema so a value set in JSON survives the next form save. MASQUE (inbound, outbound, transport) and the XDRIVE transport are new protocols the panel does not offer yet; their new loader refusals only cover configs the panel never generates. The FakeDNS IPv6 pool default, the SS2022 rewrite (same gRPC account; emails are now deduped case-insensitively, as the panel already does), the restored udphop interval default and the rest change no panel-facing config.
2026-09-30 08:27:10 -07:00
maskcompat.UpgradeLegacyXdns(stream["finalmask"])
feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) (#5506) * feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) xray-core v26.6.22 (PR #6258) renamed the XHTTP session config keys sessionPlacement/sessionKey to sessionIDPlacement/sessionIDKey (no fallback kept in core) and added sessionIDTable (predefined charset name or literal ASCII) and sessionIDLength (range, e.g. 16-32, lower bound > 0). Panel changes: - Schema (xhttp.ts): rename the two keys, add sessionIDTable/sessionIDLength, and a z.preprocess that lifts legacy keys off stored configs so an upgraded panel never silently drops a saved session setting. - Wire normalize + share-link build/parse: rename keys, emit the two new fields, and accept legacy sessionPlacement/sessionKey from old share links. - Inbound + outbound XHTTP forms: rename field paths, add a sessionIDTable autocomplete (9 predefined tables + free ASCII) and a sessionIDLength range input shown only when a table is set, with light client validation (ASCII table, length min > 0; xray enforces the room-size minimum server-side). - Subscription (service.go) and Clash (clash_service.go) builders: emit the renamed + new keys, with a legacy fallback for not-yet-resaved inbounds. - Locales: add sessionIDTable/sessionIDLength labels + hints in all 13 files. Two sibling v26.6.22 XHTTP commits need no panel change and are covered by the core bump alone: #6332 (XHTTP/3 closes QUIC/UDP) and #6320 (udpHop honors the existing dialerProxy). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(xhttp): add Session ID Table to inbound form-blocks snapshot The new sessionIDTable input renders by default in the inbound XHTTP form, so its label joins the field-structure snapshot. sessionIDLength stays conditional (only shown when a table is set), so it does not appear here. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xhttp): migrate legacy session keys in the running xray config The Zod preprocess plus the subscription/Clash fallbacks only covered the panel UI and share-link output. The config handed to the running xray-core process is built from the raw stored streamSettings in GetXrayConfig, which did not rewrite the renamed XHTTP session keys — so a pre-upgrade inbound (or template outbound) stored with a non-default sessionPlacement was emitted unchanged and dropped by xray-core v26.6.22, until the admin re-saved it. Lift sessionPlacement/sessionKey onto sessionIDPlacement/sessionIDKey at config-generation time, in the existing inbound stream-rewrite block (next to the tls/reality/externalProxy handling) and across template outbounds. The lift is idempotent and leaves unchanged configs byte-identical so the hot-reload diff never sees a spurious change. Also tighten validateSessionIDLength to reject an inverted range (e.g. 32-16) in addition to the existing lower-bound > 0 check. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xray): avoid summed-capacity allocation in mergeSubscriptionOutbounds CodeQL go/allocation-size-overflow flagged the pre-sized make() whose capacity was a sum of three slice lengths. Grow the slice via append on a nil slice instead; same result, no overflow-prone capacity expression.
2026-06-23 08:38:16 -07:00
// xray-core v26.6.22 (#6258) renamed the XHTTP session keys and
// kept no fallback. Lift legacy sessionPlacement/sessionKey onto the
// new names here so inbounds stored before the rename keep working
// without the admin re-saving them.
liftXhttpSessionIDKeys(stream)
newStream, err := json.MarshalIndent(stream, "", " ")
if err != nil {
return nil, err
}
inbound.StreamSettings = string(newStream)
}
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
if inbound.Protocol == model.Shadowsocks {
feat: complete Zod migration of frontend + bulk client batching (#4599) * feat(frontend): add Zod runtime validation at API boundary Introduces Zod 4 schemas for response validation on the three highest-traffic endpoints (server/status, nodes/list, setting/all) and a Zod->AntD form rule adapter, replacing the duplicated per-file ApiMsg<T> interfaces. Validation runs safeParse with console.warn + raw-payload fallback so backend drift never breaks the UI for users. Login form switches to schema-driven rules as the proof-of-life for the adapter. Class-based models stay untouched; remaining query/mutation hooks and form modals will migrate in follow-ups. * feat(frontend): extend Zod validation to remaining query/mutation hooks Adds Zod schemas for client/inbound/xray/node-probe endpoints and wires useNodeMutations, useClients, useInbounds, useXraySetting, useDatepicker through parseMsg. Drops the duplicated per-file ApiMsg<T> interfaces and the local ClientRecord / OutboundTrafficRow / XraySettingsValue / DefaultsPayload declarations in favour of schema-inferred types re-exported from the new src/schemas/ modules. API boundary now validates: clients list/paged, clients onlines, clients lastOnline, clients get/hydrate, inbounds slim, inbounds get, inbounds options, defaultSettings, xray config, xray outbounds traffic, xray testOutbound, xray getXrayResult, getDefaultJsonConfig, nodes probe, nodes test. Mutation responses that consume obj (bulkAdjust, delDepleted, nodes probe / test) get response validation; pass-through mutations stay agnostic. NodeFormModal type-aligned to Msg<ProbeResult>. * fix(frontend): allow null slices in client/summary schemas Go's encoding/json emits nil []T as null, not []. The initial ClientPageResponseSchema and ClientHydrateSchema rejected null inboundIds / summary.online / summary.depleted / etc., causing [zod] warnings on every empty list. Add nullableStringArray / nullableNumberArray helpers that accept null and transform to [] so consuming code keeps seeing arrays. Mark ClientRecord.traffic and .reverse nullable too (reverse is explicitly null in MarshalJSON when storage is empty). * fix(vite): treat /panel/xray as SPA page, not API root The dev-server bypass classified /panel/xray as an API path because the PANEL_API_PREFIXES matcher did `stripped === prefix.replace(/\/$/, '')`, which made the bare path collide with the SPA route of the same name (see web/controller/xui.go: g.GET("/xray", a.panelSPA)). On reload, /panel/xray got proxied to the Go backend instead of being served by Vite. The backend returned the embedded built index.html with hashed asset names that the dev server doesn't have, so every asset 404'd. Prefix-only match for trailing-slash entries fixes it: panel/xray/... still routes to the API, but panel/xray itself reaches the SPA branch. * feat(frontend): drive form validation from Zod schemas NodeFormModal — full conversion to AntD Form.useForm with antdRule on every required field. Inline field errors replace the single 'fillRequired' toast. testConnection now runs validateFields(['address','port']) before sending. ClientFormModal and ClientBulkAddModal — minimal conversion: keep the existing useState-driven controlled-component pattern, but replace the hand-rolled `if (!form.x)` checks with schema.safeParse(form). The schema is the single source of truth for required-ness and types; ClientCreateFormSchema layers on the create-only `inboundIds.min(1)` rule. New schemas (in src/schemas/): NodeFormSchema (node.ts) ClientFormSchema / ClientCreateFormSchema (client.ts) ClientBulkAddFormSchema (client.ts) Other 16+ form modals stay on the current pattern — the antdRule adapter ships from the first Zod pass for opportunistic migration as forms are touched. * chore(frontend): silence swagger-ui-react peer-dep warnings on React 19 swagger-ui-react@5.32.6 bundles three deps whose declared peer ranges predate React 19: react-copy-to-clipboard@5.1.0 (peer 15-18) react-debounce-input@3.3.0 (peer 15-18, unmaintained) react-inspector@6.0.2 (peer 16-18) For the first two, the actual code is React-19 compatible - only the metadata is stale. Resolve via npm overrides: - react-copy-to-clipboard bumped to ^5.1.1 (peer is open-ended >=15.3.0 in that release). - react-inspector bumped to ^9.0.0 (^8 was a broken publish per its own deprecation notice). - react-debounce-input is wedged on 3.3.0 with no maintained successor on npm. Use the nested-override syntax to satisfy its react peer: "react-debounce-input": { "react": "^19.0.0" } That tells npm to use our React 19 for the package's peer dependency, which silences the warning without changing the package version. * fix(vite): bypass es-toolkit CJS shim for recharts deep imports The Nodes page (and any other recharts-using route) crashed in dev and prod with TypeError: require_isUnsafeProperty is not a function. Root cause: es-toolkit's package.json exports './compat/*' only via a default condition pointing at the CJS shims under compat/<name>.js. Those shims use a require_X.Y access pattern that Vite's optimizer (Rolldown in Vite 8) and the production Rolldown build both mishandle, losing the named-export accessor and calling the namespace object as a function. recharts imports a dozen of these subpaths with default- import syntax, so every chart path tripped the bug. The matching ESM build at dist/compat/<category>/<name>.mjs is fine, but it only carries a named export. Recharts uses default imports. Plug a small Rollup-compatible plugin (enforce: 'pre') in front of the resolver: any 'es-toolkit/compat/<name>' request becomes a virtual module that imports the named symbol from the right .mjs file and re-exports it as both default and named. The plugin is registered as a top-level plugin (for the prod build) and via the new Vite 8 optimizeDeps.rolldownOptions.plugins (for the dev pre-bundler), so both pipelines pick it up consistently. * feat(frontend): migrate five secondary form modals to Zod schemas Apply the schema + safeParse-on-submit pattern (introduced for ClientFormModal / ClientBulkAddModal) to five more forms: - ClientBulkAdjustModal: ClientBulkAdjustFormSchema enforces 'at least one of addDays / addGB is non-zero' via .refine(), replacing the ad-hoc days+gb check. - BalancerFormModal: BalancerFormSchema covers tag and selector required-ness; the duplicate-tag check stays inline since it needs the otherTags prop. Per-field validateStatus now reads from the parsed issues map. - RuleFormModal: RuleFormSchema captures the form shape (no required fields - every property is optional by design). safeParse short- circuits if anything is structurally wrong. - CustomGeoFormModal: CustomGeoFormSchema folds the regex alias rule and the http(s) URL validation (including URL parse) into the schema, replacing a 20-line validate() function. - TwoFactorModal: TotpCodeSchema (z.string().regex(/^\d{6}$/)) drives both the disabled-state of the OK button and the safeParse gate before the TOTP comparison. Schemas live alongside the matching API schemas: - ClientBulkAdjustFormSchema in schemas/client.ts - BalancerFormSchema / RuleFormSchema / CustomGeoFormSchema in schemas/xray.ts - TotpCodeSchema in schemas/login.ts (next to LoginFormSchema) No UX change for valid inputs. * feat(frontend): block invalid settings saves with Zod pre-save check Tighten AllSettingSchema with the actual valid ranges and patterns: - webPort / subPort / ldapPort: integer 1-65535 - pageSize: integer 1-1000 - sessionMaxAge: integer >= 1 - tgCpu: integer 0-100 (percentage) - subUpdates: integer 1-168 (hours) - expireDiff / trafficDiff / ldapDefault*: non-negative integers - webBasePath / subPath / subJsonPath / subClashPath: must start with / The existing useAllSettings save path runs AllSettingSchema.partial() through safeParse and logs drift without blocking. SettingsPage now adds a stronger gate before the mutation: run the full schema against the draft and, on failure, surface the first issue (field path + message) via the existing messageApi.error so the user actually sees what's wrong instead of silently sending bad data to the backend. Use cases caught: port out of range, negative quota, sub path missing leading slash, page size set to 0, tgCpu > 100. * feat(frontend): schema-guard Inbound and Outbound form submits The two largest forms in the panel send to the backend without ever checking their own port range or required-ness. Schema-gate the top-level fields so obviously bad payloads stop at the client. InboundFormModal: InboundFormSchema (port 1-65535 int, non-empty protocol, the rest of the keys present) runs as a safeParse just before the HttpUtil.post in submit(). The 2000+ lines of protocol- specific subform code stay untouched - that's a separate effort and the existing per-protocol logic (e.g. canEnableStream, isFallbackHost) already gates most of the structural correctness. OutboundFormModal: OutboundTagSchema (trim + min 1) replaces the hand-rolled `if (!ob.tag?.trim()) messageApi.error('Tag is required')` check. The duplicateTag check stays inline because it needs the existingTags prop. Both schemas emit i18n keys for messages with a defaultValue fallback, matching the pattern in BalancerFormModal and SettingsPage. * feat(backend): gate request bodies with go-playground/validator Add a generic BindAndValidate helper in web/middleware that wraps gin's content-aware binder with an explicit validator.Struct call and emits a structured `entity.Msg{Obj: ValidationPayload{Issues...}}` on failure so the frontend can map each issue to an i18n key. Tag the user-facing fields on model.Inbound, model.Node, and entity.AllSetting with the range/enum constraints they were previously relying on hand-rolled CheckValid logic (or nothing) to enforce, and wire the helper into the inbound/node/settings controllers that bind those structs directly. Promotes validator/v10 from indirect to direct require, plus six unit tests covering valid payloads, range violations, enum violations, malformed JSON, in-place binding, and JSON-only strict mode. This is PR1 of a planned end-to-end Zod rollout — controllers using local form structs (custom_geo, setEnable, fallbacks, client) keep their existing handling and will be migrated as their schemas firm up. * feat(codegen): Go-first tool emitting Zod schemas and TS types Add tools/openapigen — a single-binary Go program that walks the exported structs in database/model, web/entity, and xray via go/parser and emits two committed artifacts under frontend/src/generated: - zod.ts shared Zod schemas keyed off `validate:` tags (ports get .min(1).max(65535), Inbound.protocol becomes a z.enum, Node.scheme too, etc.) - types.ts plain TS interfaces inferred from the same walk, so consumers can import Inbound without dragging Zod along The walker flattens embedded structs (AllSettingView.AllSetting), honors json:"-" and omitempty, and accepts per-struct overrides so the JSON-string-inside-JSON columns (Inbound.Settings/StreamSettings/ Sniffing, ClientRecord.Reverse, InboundClientIps.Ips) render as z.unknown() instead of leaking the DB-storage type into the API contract. Type aliases like model.Protocol are emitted as TS aliases and Zod schemas in their own right. Wires `npm run gen:zod` in frontend/package.json so the generator can be re-run without leaving the frontend tree. The existing openapi.json build (gen:api) is left alone for now; migrating the OpenAPI surface to this generator is a follow-up. PR2 of the planned Zod end-to-end rollout. * refactor(frontend): tighten HttpUtil generics from any to unknown Switch the class-level default on Msg<T> and the per-method defaults on HttpUtil.get/post/postWithModal from `any` to `unknown`, so callers that don't pass an explicit T get a narrowed response that must be schema- checked or type-cast before its shape is trusted. Drops the four file-level eslint-disable comments these defaults required. Fixes the nine direct `.obj.field` consumers that surfaced (IndexPage, XrayMetricsModal, NordModal, WarpModal, LogModal, VersionModal, XrayLogModal, CustomGeoSection) by giving each call site the explicit T it should have had from the start — typically a small ad-hoc shape, sometimes a string for the JSON-text-in-Msg.obj pattern used by NordModal/WarpModal/Xray nord/warp endpoints. PR3 of the planned Zod end-to-end rollout — schemas/inbound.ts and schemas/client.ts loose() removal stays parked until the protocol schemas land in Phase 3 to avoid silently dropping fields. * feat(frontend): protocol-leaf Zod schemas with discriminated unions Stand up schemas/primitives (Port, Flow, Protocol, Sniffing) and per-protocol leaf schemas for all 10 inbound and 13 outbound xray protocols. The leaves omit any inner `protocol` literal — the discriminator lives at the parent level so consumers narrow on `.protocol` without redundant projection. Wire shape is preserved per protocol: vmess outbound stays in `vnext[]`, trojan and shadowsocks outbound in `servers[]`, vless outbound flat, http/socks outbound in `servers[].users[]`. Cross-protocol atoms (port, flow, sniffing dest, protocol enum) live in primitives. Protocol-specific enums (vmess security, ss method/network, hysteria version, freedom domain strategy, dns rule action) stay with their leaves. Tagged-wrapper `z.discriminatedUnion('protocol', [...])` composes both InboundSettingsSchema and OutboundSettingsSchema; existing class-based models in src/models/ are untouched and will be retired in Step 3 once the golden-file safety net is in place. * feat(frontend): stream and security Zod families with discriminated unions Stand up the remaining Step 2 families. NetworkSettingsSchema is a 6-branch DU on `network` covering tcp/kcp/ws/grpc/httpupgrade/xhttp, with asymmetric per-network wire keys (tcpSettings, wsSettings, ...) preserved exactly so fixtures round-trip byte-identical. SecuritySettingsSchema is a 3-branch DU on `security` covering none/tls/reality. TLS certs use a file-vs-inline union; uTLS fingerprints are shared between TLS and Reality via a single primitive enum. Hysteria-as-network, finalmask, and sockopt are not in the plan's Step 2 inventory and are deferred to Step 6 (Tighten) - they're orthogonal extras on the stream root, not network-discriminated branches. Resolves a Security identifier collision in protocols/index.ts by re-exporting the type alias as SecurityKind (the `Security` name is taken by the namespace re-export). * test(frontend): vitest harness with golden-file fixtures for inbound protocols Stand up Phase 3 safety net before the models/ rewrite. The harness loads JSON fixtures via Vite's import.meta.glob, parses each through InboundSettingsSchema (the tagged-wrapper DU), and snapshots the canonical parsed shape. Snapshots stay byte-stable across the upcoming class-to- pure-function extraction, catching any normalization drift. Six representative inbound fixtures cover the high-traffic protocols: vless, vmess, trojan, shadowsocks (2022-blake3 multi-user), wireguard, hysteria2. Stream and security branches plus the remaining protocols (http, mixed, tunnel, hysteria) follow in subsequent turns. Uses /// <reference types="vite/client" /> instead of @types/node so we avoid pulling in another type package; import.meta.glob is enough to walk the fixtures directory at compile time. Adds vitest 4.1.7 as the only new dev dependency. test/test:watch scripts land in package.json; a standalone vitest.config.ts keeps the production vite.config.js (which reads from sqlite via DatabaseSync) out of the test runner. * test(frontend): broaden golden coverage to remaining inbounds + stream + security DUs Round out Step 3b. Four more inbound fixtures complete the protocol set (http with two accounts, mixed with socks-style auth, tunnel with a port map, hysteria v1). Two parallel test files cover the other DUs: stream.test.ts walks tcp/ws/grpc fixtures through NetworkSettingsSchema, and security.test.ts walks none/tls/reality through SecuritySettingsSchema. Snapshot count is now 16 across three test files. The reality fixture locks in the array form of serverNames/shortIds (the panel class stores them comma-joined internally but they ship as arrays on the wire). The TLS fixture pins the file-vs-inline cert DU on the file branch. Stream coverage for httpupgrade/xhttp/kcp and security mixed-with-stream combos follow in the next turn, alongside the shadow harness. * test(frontend): shadow-parse harness asserting legacy class and Zod converge Add Step 3c's safety net: for every inbound golden fixture, run the raw payload through both pipelines — legacy: Inbound.Settings.fromJson(protocol, raw.settings).toJson() zod: InboundSettingsSchema.parse(raw).settings — canonicalize each (recursively sort keys, drop empty arrays / null / undefined), and assert byte-equality. This locks the wire shape across the upcoming class-to-pure-function extraction in Step 3d. Any normalization drift introduced by the rewrite trips an assertion here before it can reach users. Two ergonomic wrinkles handled inline: - The legacy class lumps hysteria + hysteria2 onto a single HysteriaSettings (no hysteria2 case in the dispatch table); the test routes hysteria2 fixtures through the HYSTERIA branch. - Empty arrays in Zod's output (e.g. fallbacks: [] from a .default([])) are treated as equivalent to the legacy class's omit-when-empty behavior. Same wire state, different syntactic surface. All 26 tests across 4 test files pass on first run. * refactor(frontend): extract toHeaders + toV2Headers to lib/xray/headers.ts First Step 3d extraction. The XrayCommonClass static helpers toHeaders/toV2Headers are pure data shape conversions with no class hierarchy needs, so they move to a standalone module that callers can import without dragging in models/inbound.ts. The new module exports HeaderEntry + V2HeaderMap as named types so consumers stop reaching into the legacy class for type shapes. A new test file (headers.test.ts) asserts byte-equality with the legacy XrayCommonClass.toHeaders / .toV2Headers across 18 cases — null / undefined / primitive inputs, single-string headers, array-valued headers, duplicate names, empty-name and empty-value filtering, both arr=true (TCP request/response shape) and arr=false (WS / xHTTP / sockopt shape). Drift between the legacy and new impls fails these tests, so the follow-up call-site swap stays safe. Callers (TcpStreamSettings, WsStreamSettings, HTTPUpgradeStreamSettings, TunnelSettings, etc.) still go through XrayCommonClass for now — those swaps land alongside class-method extractions in subsequent turns. Suite is now 44 tests across 5 files; typecheck + lint clean. * refactor(frontend): extract createDefault*Client factories to lib/xray Next Step 3d slice. Five plain-object factories — Vless, Vmess, Trojan, Shadowsocks, Hysteria — replace the legacy `new Inbound.<Protocol>Settings.<Protocol>(...)` constructor chain and the ClientBase XrayCommonClass machinery. Each factory takes an optional seed; missing random fields (id, password, auth, email, subId) fall through to RandomUtil at call time. Forms can hand-pick a UUID; tests pass deterministic seeds so the suite never touches window.crypto. Tests double-verify each factory: a snapshot locks the exact shape, and the matching Zod ClientSchema.parse(out) must equal `out` — no missing defaults, no stray fields, type-narrowed end-to-end. Discovered: VmessClientSchema and VlessClientSchema enforce z.uuid() format, so the test seeds use real-shape UUIDs. Suite: 49 tests across 6 files; typecheck + lint clean. Outbound and inbound-settings factories follow in subsequent turns alongside the toShareLink extraction. * refactor(frontend): add createDefault*InboundSettings factories for all 10 protocols Round out Step 3d's settings factory set. Ten plain-object factories (vless / vmess / trojan / shadowsocks / hysteria / hysteria2 / http / mixed / tunnel / wireguard) replace the legacy `new Inbound.<X>Settings(protocol)` constructors. Each returns a Zod- parsable wire shape with schema defaults applied — no class instance. Forms (Step 4) and InboundsPage clone (Step 5) call these factories directly once the swap lands. Three factories take a seed for random fields: - shadowsocks: method-dependent password length via RandomUtil.randomShadowsocksPassword(method) - hysteria: explicit `version` override (defaults to 2, matching the legacy panel constructor — v1 is opt-in) - wireguard: secretKey from Wireguard.generateKeypair().privateKey Tests double-verify each factory the same way as the client factories: snapshot the shape, then Zod parse round-trip to confirm no missing defaults or stray fields. Suite: 59 tests across 6 files; typecheck + lint clean. Outbound factories and the toShareLink extraction follow next. * refactor(frontend): add getHeaderValue wire-shape lookup to lib/xray/headers Tiny piece of the toShareLink scaffold. The legacy Inbound.getHeader(obj, name) iterated the panel's internal HeaderEntry[] form; the new getHeaderValue reads the Record<string, string|string[]> map our Zod schemas store on the wire. Case-insensitive, returns '' on miss to match the legacy fallback so link-generator call sites stay simple. For repeated-name maps (TCP/WS-style string[] values) the first value wins — matches the legacy iteration order so the share URL's Host hint stays deterministic. Five unit tests cover undefined/null/empty inputs, case folding, string-valued and array-valued matches, empty-array edge case, and missing-key fallback. Suite: 64 tests across 6 files; typecheck + lint clean. This unblocks the next slice: per-protocol link generators (genVmessLink etc.) take a typed inbound + client and call getHeaderValue against the ws/httpupgrade/xhttp/tcp.request header maps. * feat(frontend): stream extras + full InboundSchema with DU intersection Step 3d's last scaffolding piece before link generators. Three new stream-extras schemas land alongside the network/security DUs: - finalmask: TcpMask[] + UdpMask[] + QuicParams. Mask `settings` stays record<string, unknown> for now — there are 13 UDP mask types and 3 TCP mask types with distinct per-type setting shapes, and modeling them all as DUs would dwarf the rest of stream/ without buying anything the shadow harness doesn't already catch. Tightened in Step 6. - sockopt: 17 socket-tuning knobs (TCP keepalive, TFO, mark, tproxy, mptcp, dialer proxy, IPv6-only, congestion). `interfaceName` field matches the panel class naming; serializers rename to `interface` on the wire. - external-proxy: rows ship per inbound describing edge fronts (CDN mirrors). Used by link generators to fan out share URLs. schemas/api/inbound.ts composes the top-level wire shape with intersection-of-DUs: StreamSettingsSchema = NetworkSettingsSchema .and(SecuritySettingsSchema) .and(StreamExtrasSchema) InboundSchema = InboundCoreSchema.and(InboundSettingsSchema) A fixture (vless-ws-tls.json) exercises the full shape — protocol DU, network DU, security DU, and TLS cert file branch in one round trip. The snapshot pins the canonical parsed form so the upcoming link extractor consumes typed input with no class hierarchy underneath. Suite: 65 tests across 7 files; typecheck + lint clean. Zod 4 intersection-of-DUs works. * refactor(frontend): extract genVmessLink to lib/xray/inbound-link.ts First link generator to leave the class hierarchy. genVmessLink takes a typed Inbound + client args and returns the base64-encoded vmess:// URL. Internal helpers (buildXhttpExtra, applyXhttpExtraToObj, applyFinalMaskToObj, applyExternalProxyTLSObj, serializeFinalMask, hasShareableFinalMaskValue, externalProxyAlpn) port across from XrayCommonClass — same logic, rewritten to read the Zod schemas' Record<string, string> headers instead of the legacy HeaderEntry[]. Parity test (inbound-link.test.ts) loads each vmess fixture in golden/fixtures/inbound-full, parses it with InboundSchema for the new pure fn AND constructs LegacyInbound.fromJson(raw) for the class method, then asserts the URLs match byte-for-byte. Drift between the two impls fails here before the call sites in pages/inbounds/* get swapped. Adds a small test setup file that aliases globalThis.window to globalThis so Base64.encode's window.btoa works under Node — keeps the test env at 'node' and avoids pulling jsdom as a new dep. A first vmess-tcp-tls full-inbound fixture pins the round-trip path. Suite: 67 tests across 8 files; typecheck + lint clean. Five more link generators (vless/trojan/ss/hysteria/wireguard) plus the orchestrator (toShareLink, genAllLinks) follow in subsequent turns. * test(frontend): refresh inbound-full snapshot with vmess-tcp-tls fixture * refactor(frontend): extract genVlessLink to lib/xray/inbound-link Second link generator. genVlessLink builds the vless://<uuid>@<host>:<port>?<query>#<remark> share URL from a typed Inbound + client args, dispatching on streamSettings.network for the network-specific knobs and on streamSettings.security for the TLS/Reality knobs. Three param-style helpers move alongside the obj- style ones already in this file: - applyXhttpExtraToParams — writes path/host/mode/x_padding_bytes and the JSON extra blob into URLSearchParams - applyFinalMaskToParams — writes the fm payload when shareable - applyExternalProxyTLSParams — overrides sni/fp/alpn when an external proxy entry is supplied and security is tls A vless-tcp-reality fixture lands alongside the existing vless-ws-tls one, so the parity test now exercises both security branches. Discovered a latent legacy bug while writing parity: the old class stored realitySettings.serverNames as a comma-joined string and gated SNI on `!ObjectUtil.isArrEmpty(serverNames)`, which always returns true for strings — so SNI was never written into Reality share URLs. Existing clients rely on the omission (they pull SNI from realitySettings.target instead). We preserve the omission here to keep this extraction byte-stable; an inline comment marks the spot for a separate intentional fix. Suite: 70 tests across 8 files; typecheck + lint clean. * refactor(frontend): extract genTrojanLink + genShadowsocksLink to lib/xray Third and fourth link generators. genTrojanLink mirrors genVlessLink's shape (URLSearchParams + network/security branches + remark hash) minus the encryption/flow VLESS-isms. genShadowsocksLink shares the same query construction but base64-encodes the userinfo portion as method:password or method:settingsPw:clientPw depending on whether SS-2022 is in single-user or multi-user mode. Three reusable helpers move out of the per-protocol functions: - writeNetworkParams: the per-network switch that all param-style links share (tcp http header / kcp mtu+tti / ws path+host / grpc serviceName+authority / httpupgrade / xhttp extras) - writeTlsParams: fingerprint/alpn/ech/sni - writeRealityParams: pbk/sid/spx/pqv (preserves the SNI-omission legacy parity quirk noted in the genVlessLink commit) genVmessLink stays with its inline switch — it builds a JSON obj instead of URLSearchParams and has per-network quirks (kcp emits mtu+tti at the obj root, grpc maps multiMode to obj.type='multi') that don't factor cleanly through the shared writer. Two new full-inbound fixtures (trojan-ws-tls, shadowsocks-tcp-2022) plus matching parity tests bring the suite to 74 tests across 8 files; typecheck + lint clean. * refactor(frontend): extract genHysteriaLink + Wireguard link/config to lib/xray Fifth and sixth link generators. genHysteriaLink builds the v1/v2 share URL (scheme picked from settings.version), copying TLS knobs into the query, surfacing the salamander obfs password from finalmask.udp[type=salamander] when present, and writing the broader finalmask payload under `fm` like the other links. Legacy parity note: the old genHysteriaLink read stream.tls.settings.allowInsecure, which isn't a field on TlsStreamSettings.Settings — the guard always evaluated false and the `insecure` param never made it into the URL. We omit it here to stay byte-stable. genWireguardLink and genWireguardConfig take a typed WireguardInboundSettings + peer index and: - link: wireguard://<peerPriv>@host:port?publickey=&address=&mtu=#remark - config: the .conf text WireGuard clients consume directly Both derive the server pubKey from settings.secretKey via Wireguard.generateKeypair at call time — Zod stores only secretKey on the wire (pubKey is computed). The Wireguard utility is pure JS (X25519 over Float64Array), so it runs fine under node + the window polyfill we added with the vmess extraction. Two new full-inbound fixtures (hysteria-v1-tls, wireguard-server) plus matching parity tests bring the suite to 78 tests across 8 files; typecheck + lint clean. Hysteria2 (protocol literal) parity stays deferred — the legacy class has no HYSTERIA2 dispatch case, so it can't round-trip a hysteria2 fixture without a protocol remap. Same trick the shadow harness uses; revisit in the orchestrator commit. * refactor(frontend): extract share-link orchestrator to lib/xray/inbound-link Last slice of Step 3d. Five orchestrator exports compose the per- protocol generators into the public surface the panel consumes: - resolveAddr(inbound, hostOverride, fallbackHostname): picks the address that goes into share/sub URLs. Browser `location.hostname` is no longer a hidden dependency — callers pass it in (or any other fallback they want). - getInboundClients(inbound): protocol-aware clients accessor. Mirrors the legacy `Inbound.clients` getter, including the SS quirk where 2022-blake3-chacha20 single-user inbounds report null (no client loop) and everything else returns the clients array. - genLink: per-protocol dispatcher matching legacy Inbound.genLink. - genAllLinks: per-client fanout. Builds the remarkModel-formatted remark (separator + 'i'/'e'/'o' field picker) and iterates streamSettings.externalProxy when present. - genInboundLinks: top-level \r\n-joined link block. Loops per client for clientful protocols, single-shots SS for non-multi-user, and delegates to genWireguardConfigs for wireguard. Returns '' for http/mixed/tunnel (no share URL at all). Plus genWireguardLinks / genWireguardConfigs fanouts which iterate peers and append index-suffixed remarks. Parity test exercises every full-inbound fixture against legacy Inbound.genInboundLinks. Skips hysteria2 (no legacy dispatch case; that bridge belongs in a separate intentional commit alongside the form modal swap). Suite: 89 tests across 8 files; typecheck + lint clean. Next: Step 4 form modal migrations. Forms can now drop `new Inbound.Settings.getSettings(protocol)` in favor of the createDefault*InboundSettings factories, and InboundsPage clone can swap to genInboundLinks. Models/ deletion follows in Step 5 once all call sites are off the class. * refactor(frontend): swap InboundsPage clone fallback off Inbound.Settings.getSettings First Step 4 call-site swap. createDefaultInboundSettings(protocol) lands in lib/xray/inbound-defaults — a protocol-aware dispatch over the 10 per-protocol settings factories already in this module. Returns a Zod- parsable plain object instead of a class instance, so callers that just need the wire-shape JSON can drop the class hierarchy without touching the broader form modals. InboundsPage's clone path used Inbound.Settings.getSettings(p).toString() as the fallback when settings JSON parsing failed. That's now createDefaultInboundSettings + JSON.stringify, with a final '{}' guard for unknown protocols (legacy returned null and .toString() crashed — we just emit empty settings instead). The Inbound import on this file is now unused and removed. The 2 remaining getSettings call sites in InboundFormModal aren't safe to swap in isolation — the form mutates the returned class instance through methods like .addClient() and .toJson() across ~2000 lines of JSX. Those land with the full Pattern A rewrite of InboundFormModal, which the plan budgets at multiple days on its own. Suite: 89 tests across 8 files; typecheck + lint clean. * refactor(frontend): lift Protocols + TLS_FLOW_CONTROL consts to schemas/primitives Step 4b. The Protocols and TLS_FLOW_CONTROL enums on models/inbound.ts were dragging five page files into that 3,300-line module just to read literal string constants. Lifting them to schemas/primitives lets those pages drop the @/models/inbound import entirely. - schemas/primitives/protocol.ts now exports a Protocols const map alongside the existing ProtocolSchema. TUN stays in the const for parity (legacy panel deployments may have saved TUN inbounds) even though the Go validator no longer accepts it as a new write. - schemas/primitives/flow.ts now exports TLS_FLOW_CONTROL. The empty-string default isn't keyed because the legacy never had a NONE entry — call sites compare against the two real flow values. Updated five consumers: - useInbounds.ts: TRACKED_PROTOCOLS now annotated readonly string[] so .includes(string) keeps narrowing through the array literal - QrCodeModal.tsx, InboundInfoModal.tsx: Protocols - ClientFormModal.tsx, ClientBulkAddModal.tsx: TLS_FLOW_CONTROL Suite: 89 tests across 8 files; typecheck + lint clean. models/inbound.ts is now imported by: - InboundFormModal.tsx (heavy use of Inbound class + getSettings) - test/inbound-link.test.ts + test/shadow.test.ts + test/headers.test.ts (intentional — these are parity tests against the legacy class) OutboundFormModal still imports from models/outbound. Both form modals are the multi-day Pattern A rewrites the plan scopes separately. * refactor(frontend): lift OutboundProtocols + OutboundDomainStrategies to schemas/primitives Moves the two outbound-side consts out of models/outbound.ts and into schemas/primitives/outbound-protocol.ts. Renames the export to OutboundProtocols to disambiguate from the inbound Protocols const (different key casing — PascalCase vs ALL CAPS — and partly different member set, so they cannot share a single const). OutboundsTab.tsx keeps its 15+ Protocols.X call sites by aliasing the import. FinalMaskForm.tsx and BasicsTab.tsx swap directly. Drops a stale `as string[]` cast in BasicsTab that no longer fits the new readonly-tuple typing. After this commit only the two big form modals (InboundFormModal/OutboundFormModal) plus three intentional parity tests still import from @/models/. * refactor(frontend): lift outbound option dictionaries to schemas/primitives Adds schemas/primitives/options.ts with UTLS_FINGERPRINT, ALPN_OPTION, SNIFFING_OPTION, USERS_SECURITY, MODE_OPTION (all identical between models/inbound.ts and models/outbound.ts) plus the outbound-only WireguardDomainStrategy, Address_Port_Strategy, and DNSRuleActions. OutboundFormModal now pulls 9 consts from primitives. Only `Outbound` (the class) and `SSMethods` (whose inbound/outbound versions diverge by 2 legacy aliases — keep the picker open for the Pattern A rewrite) still come from @/models/outbound. Drops three stale `as string[]` casts on what are now readonly tuples. * refactor(frontend): swap InboundFormModal option dicts to schemas/primitives Extends primitives/options.ts with the five inbound-only option dicts (TLS_VERSION_OPTION, TLS_CIPHER_OPTION, USAGE_OPTION, DOMAIN_STRATEGY_OPTION, TCP_CONGESTION_OPTION) and lifts InboundFormModal off @/models/inbound for 10 of its 12 imports. Only the Inbound class and SSMethods (inbound vs outbound versions diverge by 2 entries) still come from @/models/. Widens NODE_ELIGIBLE_PROTOCOLS Set element type to string since the new primitives const exposes a narrow literal union that `.has(arbitraryString)` would otherwise reject. * feat(frontend): InboundFormValues schema for Pattern A rewrite Foundation for the InboundFormModal rewrite. Mirrors the wire Inbound shape (intersection of core fields + protocol settings DU + stream/security DUs) plus the DB-side fields (up/down/total/trafficReset/nodeId/...) that flow through DBInbound rather than the xray config slice. InboundStreamFormSchema is exported separately so individual sub-form sections can rule against just the stream portion when needed. FallbackRowSchema is co-located here even though fallbacks save via a distinct endpoint after the main POST — they belong to the same form state from the user's perspective. No modal changes in this commit. Foundation only; subsequent turns swap the modal's `inboundRef`/`dbFormRef` mutable-class state for Form.useForm<InboundFormValues>(). * feat(frontend): adapter between raw inbound rows and InboundFormValues Adds lib/xray/inbound-form-adapter.ts with rawInboundToFormValues and formValuesToWirePayload. The pair is the data boundary the upcoming Pattern A modal will use: it consumes the DB row shape (settings et al. as string OR object — coerced internally), hands the modal typed InboundFormValues, and on submit reverses the trip to a wire payload with the three JSON-stringified slices the Go endpoints expect. No dependency on the legacy Inbound/DBInbound classes — the coerce step is inlined so the adapter survives the eventual models/ deletion. Adds 10 Vitest cases covering string vs object inputs, the optional streamSettings/nodeId fields, trafficReset coercion, and a raw-to-payload -to-raw round-trip equality. * feat(frontend): protocol capability predicates as pure functions Adds lib/xray/protocol-capabilities.ts with the seven predicates the modals call: canEnableTls, canEnableReality, canEnableTlsFlow, canEnableStream, canEnableVisionSeed, isSS2022, isSSMultiUser. Each takes a minimal slice of an InboundFormValues, no class instance. The legacy isSSMultiUser returns true on non-shadowsocks protocols too (method getter resolves to "" which != blake3-chacha20-poly1305). The new function preserves this quirk and documents it inline; callers all narrow on protocol === shadowsocks before checking, so the surprising return value never surfaces. Parity harness in test/protocol-capabilities.test.ts crosses each of the 10 golden fixtures with 14 stream configurations (network × security) and asserts each predicate matches the legacy class method — 140 cases, all green. * feat(frontend): outbound settings factories + dispatcher Adds lib/xray/outbound-defaults.ts parallel to inbound-defaults.ts: 13 createDefault*OutboundSettings factories (one per outbound protocol) plus the createDefaultOutboundSettings(protocol) dispatcher mirroring Outbound.Settings.getSettings's contract — non-null on each known protocol, null otherwise. The factory output matches the legacy `new Outbound.<X>Settings()` start state: required-by-schema fields the user fills in via the form (address, port, password, id, peer publicKey/endpoint) come back as empty stubs. Wireguard alone seeds secretKey via the X25519 generator; the rest expose blank fields. This is the same behavior the OutboundFormModal relies on for protocol-change resets. Shadowsocks defaults to 2022-blake3-aes-128-gcm rather than the legacy undefined — the Select snaps to the first option anyway, so the coherent default keeps the modal from rendering an empty picker. Tests cover three layers: - exact-shape snapshots per factory (13 cases) - Zod schema acceptance after sensible stub fill-in (13 cases) - dispatcher non-null per known protocol + null for the unknown (14 cases) * feat(frontend): InboundFormModal.new.tsx skeleton (Pattern A) First commit of the sibling-file modal rewrite. The new modal mounts Form.useForm<InboundFormValues>, hydrates via rawInboundToFormValues on open (edit) or buildAddModeValues (add), runs validateFields + safeParse on submit, and posts the formValuesToWirePayload result. No tabs yet — the modal body shows a WIP placeholder. The file is not imported anywhere; the existing InboundFormModal.tsx remains the one InboundsPage renders. Build, lint, and 280 tests stay green. Subsequent commits add the basic / sniffing / protocol / stream / security / advanced / fallbacks sections; the atomic import swap in InboundsPage.tsx lands last. * feat(frontend): basic tab on InboundFormModal.new.tsx (Pattern A) First real section of the sibling-file rewrite. Wires AntD Form.Items to InboundFormValues paths for the basic tab — enable, remark, deployTo (when protocol is node-eligible), protocol, listen, port, totalGB, trafficReset, expireDate. The port input gets a per-field antdRule against InboundFormBaseSchema.shape.port — the spec's Pattern A reference. The intersection-typed InboundFormSchema has no .shape accessor, so per-field rules pull from the underlying ZodObject components. totalGB and expireDate are bytes/timestamp on the wire but a GB number / dayjs picker in the UI. Both use shouldUpdate-closure children that read form state and call setFieldValue on user input — no transient form-only fields, no DU-shape surprises at submit time. Protocol-change cascade lives in Form's onValuesChange: pick a new protocol and the settings DU branch is reset to createDefaultInboundSettings(next); a non-node-eligible protocol also clears nodeId. Modal still renders a single-tab Tabs container. Sniffing tab is next. * feat(frontend): sniffing tab on InboundFormModal.new.tsx (Pattern A) Second section of the sibling-file rewrite. Wires the six sniffing sub-fields to nested form paths ['sniffing', 'enabled'], ['sniffing', 'destOverride'], etc. Uses Form.useWatch on the enabled flag to drive conditional rendering of the dependent fields — the same gate the legacy modal expressed via `ib.sniffing.enabled &&`. Checkbox.Group renders one Checkbox per SNIFFING_OPTION entry. The two exclusion lists use Select mode="tags" so the user can paste comma- separated IP/CIDR or domain rules. No transient form state, no class methods — every field maps directly to a wire-shape path in InboundFormValues. Protocol tab is next. * feat(frontend): protocol tab VLESS auth on InboundFormModal.new.tsx Adds the protocol tab to the sibling-file rewrite — currently only the VLESS section, which lays out decryption/encryption inputs and the three buttons that drive them: Get New x25519, Get New mlkem768, Clear. getNewVlessEnc + clearVlessEnc are ported from the legacy modal as pure setFieldValue paths into ['settings', 'decryption'] / ['settings', 'encryption'] — no class methods, no inboundRef. The matchesVlessAuth helper mirrors the legacy fuzzy label-matching so the backend response shape stays the only source of truth. selectedVlessAuth derives the displayed auth label from the encryption string via Form.useWatch — same heuristic as the legacy modal (.length > 300 → mlkem768, otherwise x25519). Tab spread is conditional: the protocol tab only appears when protocol === 'vless' right now. As more protocol sections land (shadowsocks, http/mixed, tunnel, tun, wireguard) the condition will widen to cover each one. * feat(frontend): protocol tab Shadowsocks section (Pattern A) Adds the Shadowsocks sub-form: method picker (from SSMethodSchema's seven schema-aligned options), conditional password input gated on isSS2022, network picker (tcp/udp/tcp,udp), ivCheck toggle. Method change cascades through the Select's onChange — regenerating the inbound-level password via RandomUtil.randomShadowsocksPassword. The shadowsockses[] multi-user list reset is deferred until the clients-management section lands. Uses isSS2022 from lib/xray/protocol-capabilities to gate the password field exactly the way the legacy modal did — keeps the form behavior identical without referencing the legacy class. SSMethodSchema.options drives the Select rather than the legacy SSMethods const (which the inbound modal pulled from models/inbound.ts). This commits to the schema-aligned 7-entry list for inbound; the outbound divergence (9 entries with legacy aliases) is still pending in OutboundFormModal — defer the UX decision to that rewrite. * feat(frontend): protocol tab HTTP and Mixed sections (Pattern A) Adds the HTTP and Mixed sub-forms. Both share an accounts list — first Form.List usage in the rewrite. Each row binds via [field.name, 'user'] / [field.name, 'pass'] under the parent ['settings', 'accounts'] path, so the wire shape stays exactly what HttpInboundSettingsSchema and MixedInboundSettingsSchema validate. HTTP-only: allowTransparent Switch. Mixed-only: auth Select (noauth/password), udp Switch, conditional ip Input gated on the udp value via Form.useWatch. Tab visibility widens to include http + mixed alongside vless + shadowsocks. The string cast on the includes-check keeps the frozen Protocols const's narrow union from rejecting the broader protocol string at the call site. * feat(frontend): protocol tab Tunnel section (Pattern A) Adds the Tunnel sub-form: rewriteAddress + rewritePort, allowedNetwork picker (tcp/udp/tcp,udp), Form.List-driven portMap with name/value pairs, and the followRedirect Switch. portMap is the second Form.List in the rewrite — same shape as the HTTP/Mixed accounts list but with name/value rather than user/pass. The wire shape stays `settings.portMap: { name, value }[]` exactly. Tab visibility widens to Tunnel. * feat(frontend): protocol tab TUN section (Pattern A) Adds the TUN sub-form: interface name, MTU, four primitive-array Form.Lists (gateway, dns, autoSystemRoutingTable), userLevel, autoOutboundsInterface. Primitive Form.Lists bind each row's Input directly to `field.name` (no inner key) — distinct from the object-row Form.Lists that bind to `[field.name, 'fieldKey']`. The Form.useWatch('protocol') return type comes from the schema's protocol enum which excludes 'tun' (TUN is in the legacy Protocols const for data parity but never accepted by the wire validator). Cast to string at the source so per-section comparisons against Protocols.TUN typecheck. Why: legacy DB rows with protocol === 'tun' still need to render; widening here keeps reads from rejecting them. Tab visibility widens to TUN. * feat(frontend): protocol tab Wireguard section (Pattern A) Adds the Wireguard sub-form: server secretKey input with regen icon, derived disabled public-key display, mtu, noKernelTun toggle, and a Form.List of peers — each peer having its own privateKey (regen icon), publicKey, preSharedKey, allowedIPs (nested Form.List for the string array), keepAlive. pubKey is purely derived (computed via Wireguard.generateKeypair from the watched secretKey) and is NOT stored in the form value — the schema omits it from the wire shape on purpose. The disabled display shows the live derivation without polluting form state. regenInboundWg generates a fresh keypair and writes only the secretKey path; pubKey re-derives automatically. regenWgPeerKeypair writes both privateKey and publicKey at the peer's path index. The preSharedKey wire-shape name is used instead of the legacy class's internal psk — matches WireguardInboundPeerSchema. Tab visibility widens to Wireguard. * feat(frontend): stream tab skeleton with TCP + KCP (Pattern A) Opens the stream tab on the sibling-file rewrite. Tab visibility is driven by canEnableStream from lib/xray/protocol-capabilities — same gate the legacy modal used, now schema-aware. Transmission picker (network select) is hidden for HYSTERIA since that protocol's network is implicit. onNetworkChange clears any stale per-network settings keys (tcpSettings/kcpSettings/...) and seeds an empty object for the new branch so AntD Form.Items don't read from undefined nested paths. TCP section: acceptProxyProtocol Switch (literal-true-optional on the wire — the form stores true/false but Zod's strip behavior keeps false-as-omission round-trips clean) plus an HTTP-camouflage toggle that flips header.type between 'none' and 'http'. The full HTTP camouflage request/response sub-form lands in a follow-up commit. KCP section: six numeric knobs (mtu, tti, upCap, downCap, cwndMultiplier, maxSendingWindow). WS / gRPC / HTTPUpgrade / XHTTP / external-proxy / sockopt / hysteria stream / FinalMaskForm hookup all still pending. * feat(frontend): stream tab WS + gRPC + HTTPUpgrade sections (Pattern A) Adds the three medium-complexity network branches to the stream tab. Plain Form.Item paths into the corresponding *Settings keys — no Form.List wrappers since these schemas don't have arrays at the top level. WS: acceptProxyProtocol, host, path, heartbeatPeriod gRPC: serviceName, authority, multiMode HTTPUpgrade: acceptProxyProtocol, host, path Header editing is deferred to a later commit — WsHeaderMap is a Record<string,string> on the wire, V2HeaderMap a Record<string,string[]>, and the form needs an array-of-{name,value} UI that converts on edit. Worth building once and reusing across WS, HTTPUpgrade, XHTTP, TCP request/response, and Hysteria masquerade headers. XHTTP + external-proxy + sockopt + hysteria stream + finalmask hookup still pending. * feat(frontend): stream tab XHTTP section (Pattern A) XHTTP is the heaviest network branch — 19 fields rendered conditionally on mode, xPaddingObfsMode, and the three *Placement selectors. Each gates its dependent field set via Form.useWatch. Field structure mirrors the legacy XHTTPStreamSettings form 1:1: - mode picker (auto / packet-up / stream-up / stream-one) - packet-up adds scMaxBufferedPosts + scMaxEachPostBytes; stream-up adds scStreamUpServerSecs - serverMaxHeaderBytes, xPaddingBytes, uplinkHTTPMethod (with the packet-up gate on the GET option) - xPaddingObfsMode unlocks xPadding{Key,Header,Placement,Method} - sessionPlacement / seqPlacement each unlock their respective Key field when set to anything other than 'path' - packet-up mode additionally unlocks uplinkDataPlacement, and that in turn unlocks uplinkDataKey when the placement is not 'body' - noSSEHeader Switch at the tail XHTTP headers editor still pending (same WsHeaderMap as WS — will be unified in the header-editor extraction commit). * feat(frontend): stream tab external-proxy + sockopt sections (Pattern A) External Proxy: Switch driven by externalProxy array length. Toggling on seeds one row with the window hostname + the inbound's current port; toggling off clears the array. Each row is a Form.List item with forceTls/dest/port/remark inline, and a nested SNI/Fingerprint/ALPN row that conditionally renders on forceTls === 'tls' via a shouldUpdate-closure that watches the per-row forceTls path. Sockopt: Switch driven by whether the sockopt object exists in form state. Toggling on calls SockoptStreamSettingsSchema.parse({}) so every default the schema declares (mark=0, tproxy='off', domainStrategy='UseIP', tcpcongestion='bbr', etc.) flows into the form; toggling off sets to undefined. Renders the seventeen sockopt fields directly bound to ['streamSettings', 'sockopt', X] paths. Option lists pull from the primitives const dictionaries (UTLS_FINGERPRINT, ALPN_OPTION, DOMAIN_STRATEGY_OPTION, TCP_CONGESTION_OPTION) rather than the schema's .options to keep one source of truth for UI label strings. * feat(frontend): security tab base + TLS section (Pattern A) Adds the security tab to the sibling-file rewrite. Visibility is paired with the stream tab — both gated on canEnableStream. The security selector is itself disabled when canEnableTls is false, and the reality option only appears when canEnableReality is true, mirroring the legacy modal's Radio.Group guards. onSecurityChange clears the previous branch's *Settings key and seeds the new branch from the schema's parsed defaults (the same trick the sockopt toggle uses). The security selector itself is rendered via a shouldUpdate closure so the on-change handler can write the cleaned streamSettings shape atomically without racing AntD's per-field sync. TLS section: serverName (the wire field — the legacy class calls it sni internally), cipherSuites (with the 13 named suites from TLS_CIPHER_OPTION), min/max version pair, uTLS fingerprint, ALPN multi-select, plus the three policy Switches. TLS certificates list, ECH controls, the full Reality sub-form, and the four API-call buttons (genRealityKeypair / genMldsa65 / getNewEchCert / randomizers) land in a follow-up commit. * feat(frontend): security tab Reality + ECH + mldsa65 controls (Pattern A) Adds the Reality sub-form and the four API-call buttons that drive the server-generated material: - genRealityKeypair calls /panel/api/server/getNewX25519Cert and writes the result into ['streamSettings', 'realitySettings', 'privateKey'] and the nested settings.publicKey path. - genMldsa65 calls /panel/api/server/getNewmldsa65 for the post-quantum seed/verify pair. - getNewEchCert calls /panel/api/server/getNewEchCert with the current serverName and writes echServerKeys + settings.echConfigList. - randomizeRealityTarget seeds target + serverNames from the random reality-targets pool. - randomizeShortIds calls RandomUtil.randomShortIds (comma-joined string) and splits into the schema's string[] form. Reality fields are bound directly to schema paths — show/xver/target, maxTimediff, min/max ClientVer, the settings.{publicKey, fingerprint, spiderX, mldsa65Verify} nested subtree, plus the array fields (serverNames, shortIds) rendered as Select mode="tags" since both ship as string[] on the wire. TLS certificates list (Form.List with the useFile DU) still pending — that's a chunky sub-form on its own. * feat(frontend): security tab TLS certificates list (Pattern A) Closes out the security tab: a Form.List of certificates that toggles between TlsCertFileSchema (certificateFile + keyFile string paths) and TlsCertInlineSchema (certificate + key as string arrays per the wire shape) via a per-row useFile boolean. useFile is a transient form-only field — not part of TlsCertSchema. Zod's default-strip behavior drops it during InboundFormSchema parse on submit, leaving only the matching wire branch's keys populated. Whichever side the user wasn't on stays empty, so Zod's union picks the populated branch. For inline certs the TextAreas use normalize + getValueProps to convert between the wire-side string[] and the multi-line text the user types. Each line becomes one array element, matching the legacy class's `cert.split('\n')` toJson convention. Per-row buildChain is conditionally rendered when usage === 'issue' — a shouldUpdate-closure watches the specific path so the toggle re-renders inline without listening to unrelated form changes. Security tab is now functionally complete. Advanced JSON tab, Fallbacks card, and the atomic swap in InboundsPage are next. * feat(frontend): advanced JSON tab on InboundFormModal.new.tsx (Pattern A) Adds the advanced JSON tab. Each sub-tab (settings / streamSettings / sniffing) renders an AdvancedSliceEditor — a small CodeMirror-backed JsonEditor that holds a local text buffer and forwards parsed JSON to form state on every valid edit. Invalid JSON sits silently in the local buffer; once the user finishes balancing braces / quoting, the next valid parse pushes through to the form. No stamping ref, no apply-on-tab-switch ceremony — the form is the single source of truth. The buffer seeds once from form state on mount. The Modal's destroyOnHidden means each open is a fresh editor instance, so external form mutations during a single open session can't desync the editor either. The streamSettings sub-tab is omitted when streamEnabled is false (matching the legacy modal's behavior for protocols like Http / Mixed that have no stream layer). * feat(frontend): fallbacks card on InboundFormModal.new.tsx (Pattern A) Adds the fallbacks card rendered inside the protocol tab whenever the current values describe a fallback host — VLESS or Trojan on tcp with tls or reality security. The protocol tab visibility widens to include Trojan in that exact case (it has no other protocol sub-form). Fallbacks live in a useState alongside the form rather than inside form values, mirroring the legacy modal: fallbacks save via a distinct endpoint (/panel/api/inbounds/{id}/fallbacks) after the main inbound POST, not as part of the inbound payload. loadFallbacks runs on open for edit-mode VLESS/Trojan; saveFallbacks runs after a successful POST inside the submit handler. Each row: child picker (filtered down to other inbounds), then four inline edits for SNI / ALPN / path / xver. Add adds an empty row; delete pulls the row from state. Quick-Add-All, the rederive-from-child helper, and the per-row up/down movers are deferred — the basic add/edit/remove cycle is what the modal actually needs to function. * feat(frontend): atomic swap InboundFormModal to Pattern A Deletes the 2261-line class-mutation modal and renames the 1900-line sibling rewrite into its place. InboundsPage.tsx already imports the file by path so no consumer change is needed — the swap is one file delete plus one file rename. Build, lint, and 280 tests stay green. What the new modal covers end-to-end: - Basic (enable / remark / nodeId / protocol / listen / port / totalGB / trafficReset / expireDate) - Sniffing (enabled / destOverride / metadataOnly / routeOnly / ipsExcluded / domainsExcluded) - Protocol per DU branch: VLESS (decryption/encryption + buttons), Shadowsocks (method/password/network/ivCheck), HTTP + Mixed (accounts list + per-protocol toggles), Tunnel (rewrite + portMap + followRedirect), TUN (interface/mtu + four primitive lists + userLevel/autoInterface), Wireguard (secretKey + derived pubKey + peers list with nested allowedIPs) - Stream per network: TCP base, KCP, WS, gRPC, HTTPUpgrade, XHTTP (the 22-field one), plus external-proxy and sockopt extras - Security: TLS (SNI/cipher/version/uTLS/ALPN/policy switches + certificates list with file/inline toggle + ECH controls), Reality (every field + the four API-call buttons), none - Advanced JSON (settings / streamSettings / sniffing live editors that round-trip into form state on every valid parse) - Fallbacks (load on open for VLESS/Trojan TLS-or-Reality TCP hosts; save through the secondary endpoint after the main POST succeeds) Known regressions vs the legacy modal, all reachable via Advanced JSON until backfilled in follow-up commits: - Hysteria stream sub-form (masquerade / udpIdleTimeout / version) — schema gap; the existing inbound DU has no hysteria stream branch - FinalMaskForm hookup — the component is still class-shape coupled - HeaderMapEditor — TCP request/response headers, WS / HTTPUpgrade / XHTTP headers, Hysteria masquerade headers all need a shared editor - TCP HTTP camouflage request/response body (version, method, path list, headers, status, reason) — only the on/off toggle is wired - Fallbacks polish — up/down move, quick-add-all, rederive-from-child, the per-row advanced-toggle / proxy-tag chips No reference to @/models/inbound's Inbound class anywhere in the new modal — only @/models/dbinbound (out of scope) and @/models/reality-targets (out of scope). The protocol-capabilities predicates and the rawInboundToFormValues + formValuesToWirePayload adapters carry every behavior the class used to provide. * fix(frontend): finish InboundFormModal rename after atomic swap The atomic-swap commit landed the new file but the exported function was still named InboundFormModalNew. Rename to match the file. * feat(frontend): outbound form schema + wire adapter foundation Lay the groundwork for OutboundFormModal's Pattern A rewrite: - schemas/forms/outbound-form.ts: discriminated-union form values across all 12 outbound protocols, with flat per-protocol settings shapes that match the legacy class fields (vmess vnext / trojan-ss-socks-http servers / wireguard csv address-reserved all flattened). - lib/xray/outbound-form-adapter.ts: rawOutboundToFormValues converts wire-shape outbound JSON to typed form values; formValuesToWirePayload re-nests on submit. Replaces the Outbound.fromJson/toJson dependency the modal currently has on the legacy class hierarchy. - test/outbound-form-adapter.test.ts: 15 round-trip cases covering each protocol's wire quirks (vmess vnext flatten, vless reverse-wrap, wireguard csv↔array, blackhole response wrap, DNS rule normalization, mux gating). * feat(frontend): OutboundFormModal.new.tsx skeleton (Pattern A) Sibling .new.tsx file with the Modal shell, Tabs (Basic/JSON), Form.useForm hydration via rawOutboundToFormValues, and the submit pipeline that calls formValuesToWirePayload before onConfirm. Tag uniqueness check is wired in. Protocol-specific sub-forms, stream, security, sockopt, and mux sections are deferred to subsequent commits — accessible via the JSON tab in the meantime. The InboundsPage continues to render the legacy modal until the atomic swap at the end. Also: rawOutboundToFormValues now returns streamSettings as undefined when the wire payload omits it, so Form.useForm doesn't receive a value that does not match the NetworkSettings discriminated union. * feat(frontend): OutboundFormModal.new.tsx vmess/vless/trojan/ss sections - Shared connect-target sub-block (address + port) for the six protocols whose form schema carries them flat at settings root. - VMess: id + security Select (USERS_SECURITY). - VLESS: id + encryption + flow + reverseTag (reverse-sniffing slice and Vision testpre/testseed come in a later commit). - Trojan: password. - Shadowsocks: password + method Select (SSMethodSchema) + UoT switch + UoT version. onValuesChange cascade: when the user picks a different protocol, the adapter re-seeds the settings sub-object to the new protocol's defaults so leftover fields from the previous protocol do not bleed through. * feat(frontend): OutboundFormModal.new.tsx socks/http/hysteria/loopback/blackhole/wireguard sections - SOCKS / HTTP: user + pass at settings root. - Hysteria: read-only version=2 (the actual transport knobs live on stream.hysteria, added with the stream tab). - Loopback: inboundTag. - Blackhole: response type Select with empty/none/http options. - Wireguard: address (csv) + secretKey (with regenerate icon) + derived pubKey + domain strategy + MTU + workers + no-kernel-tun + reserved (csv) + peers Form.List with nested allowedIPs sub-list. Wireguard regenerate icon uses Wireguard.generateKeypair() and writes both keys to the form via setFieldValue — preserves the legacy UX of the SyncOutlined inline-icon next to the privateKey label. * feat(frontend): OutboundFormModal.new.tsx DNS + Freedom + VLESS reverse-sniffing - DNS: rewriteNetwork (udp/tcp Select) + rewriteAddress + rewritePort + userLevel + rules Form.List (action/qtype/domain). - Freedom: domainStrategy + redirect + Fragment Switch with conditional 4-field sub-block (legacy 'enable Fragment' UX preserved — Switch sets all four fields to populated defaults, off-state empties them all out so the adapter strips them on submit) + Noises Form.List (rand/base64/ str/hex types, packet/delay/applyTo per row) + Final Rules Form.List with conditional block-delay sub-field. - VLESS reverse-sniffing slice: rendered only when reverseTag is set (matches the legacy modal's nested conditional). All six fields wired to the form state with appropriate widgets (Switch / Select multi / Select tags). * feat(frontend): OutboundFormModal.new.tsx stream tab (TCP/KCP/WS/gRPC/HTTPUpgrade) Wire the stream sub-form into the Pattern A modal: - newStreamSlice(network) helper bootstraps the per-network DU branch with Xray defaults (mtu=1350, tti=20, uplinkCapacity=5, etc.). - streamSettings is seeded once when the protocol supports streams but the form has no slice yet (new outbound + protocol switch). - onNetworkChange swaps the sub-key and preserves security when the new network still supports it, else snaps back to 'none'. - Per-network sub-forms wired: TCP: HTTP camouflage Switch (sets header.type = 'http' / 'none') KCP: 6 numeric tuning fields WS: host + path + heartbeat gRPC: service name + authority + multi-mode switch HTTPUpgrade: host + path XHTTP: host + path + mode + padding bytes (advanced fields via JSON) Security radio, TLS/Reality sub-forms, sockopt, and mux still pending. * feat(frontend): OutboundFormModal.new.tsx security tab (TLS + Reality + Flow) - onSecurityChange cascade: swaps tlsSettings/realitySettings sub-key matching the DU branch, seeding the new sub-form with empty/default fields so the UI does not reference undefined values. - Flow Select rendered when canEnableTlsFlow is true (VLESS + TCP + TLS/Reality). Moved from the basic VLESS section so it only appears in the relevant security context — matches the legacy modal UX. - Security Radio (none / TLS / Reality) gated by canEnableTls and canEnableReality pure-function predicates from lib/xray/protocol-capabilities. - TLS sub-form: 6 outbound-specific fields (SNI/uTLS/ALPN/ECH/ verifyPeerCertByName/pinnedPeerCertSha256) matching the legacy TlsStreamSettings flat shape (no certificates list — outbound is client-side). - Reality sub-form: 6 fields (SNI/uTLS/shortId/spiderX/publicKey/ mldsa65Verify). publicKey + mldsa65Verify get TextAreas to handle the long base64 strings. * feat(frontend): OutboundFormModal.new.tsx sockopt + mux sections - Sockopts: Switch toggles streamSettings.sockopt between undefined and a populated default object (17 fields with sane bbr/UseIP defaults). Only the 8 most-used fields are rendered (dialer proxy, domain strategy, keep alive interval, TFO, MPTCP, penetrate, mark, interface). The remaining sockopt knobs (acceptProxyProtocol, tcpUserTimeout, tcpcongestion, tcpKeepAliveIdle, tcpMaxSeg, tcpWindowClamp, V6Only, trustedXForwardedFor, tproxy) are still in the wire payload — edit them via the JSON tab. - Mux: gated by isMuxAllowed(protocol, flow, network) — VMess/VLESS/ Trojan/SS/HTTP/SOCKS, no flow set, no xhttp transport. Sub-fields (concurrency / xudpConcurrency / xudpProxyUDP443) only render when enabled is true. - Sockopt section visible only when streamAllowed AND network is set — non-stream protocols (freedom/blackhole/dns/loopback) still edit sockopt via the JSON tab. * feat(frontend): atomic swap OutboundFormModal to Pattern A Delete the legacy 1473-line class-based OutboundFormModal.tsx and replace it with the new Pattern A modal (Form.useForm + antdRule + per-protocol discriminated-union form values + wire adapter). Net diff: legacy file gone, function renamed from OutboundFormModalNew to OutboundFormModal so the existing OutboundsTab import resolves unchanged. What is migrated: - All 12 protocols (vmess/vless/trojan/ss/socks/http/wireguard/ hysteria/freedom/blackhole/dns/loopback) - Stream tab with TCP/KCP/WS/gRPC/HTTPUpgrade + partial XHTTP - Security tab with TLS + Reality + Flow gating - Sockopt + Mux sections (gated by isMuxAllowed) - JSON tab with bidirectional bridge to form state - Tag uniqueness check - VLESS reverse-sniffing slice - Freedom fragment/noises/finalRules - DNS rewrite + rules list - Wireguard peers + nested allowedIPs sub-list - Wireguard secret/public key regeneration Deferred to follow-up commits (still accessible via the JSON tab): - XHTTP advanced fields (xmux, sequence/session placement, padding obfs) - Hysteria stream transport sub-form - TCP HTTP camouflage host/path body - WS/HTTPUpgrade/XHTTP headers map editor - Remaining sockopt knobs (tcpUserTimeout, tcpcongestion, tcpKeepAliveIdle, tcpMaxSeg, tcpWindowClamp, V6Only, trustedXForwardedFor, tproxy, acceptProxyProtocol) - VLESS Vision testpre/testseed - Reality API helpers (random target, x25519/mldsa65 generate-import) - Link import (vmess:// vless:// etc → outbound) - FinalMaskForm hookup (deferred from inbound rewrite too) * test(frontend): convert legacy-class parity tests to snapshot baselines With the inbound/outbound modal rewrites complete, the cross-check against the legacy Inbound class has served its purpose. The new pure-function / Zod-schema paths are the source of truth for production code; the parity assertions were the migration safety net. Convert the three parity test files to snapshot-based regression tests: - headers.test.ts: toHeaders + toV2Headers run against snapshots captured at the close of the migration (when both new and legacy were verified byte-equal). - protocol-capabilities.test.ts: 140 cases (10 fixtures × 14 stream shapes) snapshot the predicate-result tuple. Was: parity vs legacy Inbound.canEnableX() class methods. - inbound-link.test.ts: per-protocol genXxxLink + genInboundLinks orchestrator output is snapshotted. Was: byte-equality vs legacy Inbound.genXxxLink() methods. Also delete shadow.test.ts — its purpose was a dual-parse drift detector (Inbound.Settings.fromJson vs InboundSettingsSchema.parse). inbound-full.test.ts already snapshots the Zod parse output, which covers the same ground without the legacy dependency. models/inbound.ts and models/outbound.ts stay in the tree for now — DBInbound still consumes Inbound via its toInbound() method, and DBInbound migration is out of scope per the migration spec ('Do NOT migrate Status, DBInbound, or AllSetting...'). No production page imports from @/models/inbound or @/models/outbound directly anymore. * chore(frontend): enforce no-explicit-any: error + add typecheck/test to CI Step 7 of the Zod migration: lock the migration's gains in place via lint + CI enforcement. - eslint.config.js: `@typescript-eslint/no-explicit-any` set to error. Verified locally — zero violations in src/, with the only file-level disables being src/models/inbound.ts and src/models/outbound.ts (kept for DBInbound's toInbound() consumer; their migration is out of spec scope). - .github/workflows/ci.yml: add Typecheck and Test steps to the frontend job, between Lint and Build. PRs now have to pass tsc --noEmit and the full vitest suite (285 tests + 172 snapshots) before build runs. Migration scoreboard (vs the spec): Step 1 primitives + barrels done Step 2 protocol leaf + DUs done Step 3 pure-fn extraction done Step 4 form modals -> Pattern A done (Inbound + Outbound) Step 5 delete models/ files DEFERRED (DBInbound still uses Inbound; spec marks DBInbound migration out of scope) Step 6 tighten .loose() / unknown DEFERRED (invasive, separate PR) Step 7 lint + CI enforcement done (this commit) Production code paths now have no direct dependency on the legacy Inbound or Outbound classes. * feat(frontend): OutboundFormModal deferred features (Vision seed / TCP host+path / WG pubKey derive) Three small wins from the post-atomic-swap deferred list: - VLESS Vision testpre + testseed: shown only when flow === 'xtls-rprx-vision' (mirrors the legacy canEnableVisionSeed gate). testseed binds to a Select mode='tags' with a normalize() that coerces strings to positive integers and drops invalid entries. - TCP HTTP camouflage host + path: when the TCP HTTP camouflage Switch is on, surface two inputs that read/write directly into streamSettings.tcpSettings.header.request.headers.Host and .path. Both fields are string[] on the wire; normalize + getValueProps translate to/from comma-joined strings in the UI (one entry per host or path the user wants camouflaged). - Wireguard pubKey auto-derive: Form.useWatch on settings.secretKey + useEffect that runs Wireguard.generateKeypair(secret).publicKey on every change and writes the result into the disabled pubKey display field. Matches the legacy modal's per-keystroke derive. * feat(frontend): symmetric TCP HTTP host/path + extra sockopt knobs OutboundFormModal: - Sockopt section gains 5 common-but-rarely-tweaked knobs: acceptProxyProtocol, tproxy (off/redirect/tproxy), tcpcongestion (bbr/cubic/reno), V6Only, tcpUserTimeout. The remaining sockopt fields (tcpKeepAliveIdle, tcpMaxSeg, tcpWindowClamp, trustedXForwardedFor) are still edit-via-JSON; they are deeply tunable and not commonly touched. InboundFormModal: - TCP HTTP camouflage gains host + path inputs symmetric to the outbound side. Switch ON seeds request with sensible defaults (version 1.1, method GET, path ['/'], empty headers). The two inputs use the same normalize/getValueProps comma-string ↔ string[] dance the outbound side uses, so the wire shape stays identical to what xray-core expects. * feat(frontend): HeaderMapEditor reusable component + wire WS/HTTPUpgrade headers Add a single reusable header-map editor that handles the two wire shapes Xray uses: - v1: { name: 'value' } — used by WS / HTTPUpgrade / Hysteria masquerade. One value per name. - v2: { name: ['value1', 'value2'] } — used by TCP HTTP camouflage. Each header can repeat (RFC 7230 §3.2.2). Internal state is always a flat list of {name, value} rows regardless of mode; conversion to/from the wire shape happens at the value / onChange boundary so consumers bind straight to a Form.Item with no extra transforms. Wired into: - InboundFormModal: WS Headers, HTTPUpgrade Headers - OutboundFormModal: WS Headers, HTTPUpgrade Headers XHTTP headers are already in a list-of-rows wire shape (different from these two), so they keep their bespoke editor. Hysteria masquerade is still deferred until the Hysteria stream sub-form lands. * feat(frontend): Hysteria stream sub-form (schema branch + outbound UI) Add the 7th branch to NetworkSettingsSchema for Hysteria transport. schemas/protocols/stream/hysteria.ts: - HysteriaStreamSettingsSchema covers the full wire shape: version=2, auth, congestion (''|'brutal'), up/down bandwidth strings, optional udphop sub-object for port-hopping, receive-window tuning fields, maxIdleTimeout, keepAlivePeriod, disablePathMTUDiscovery. schemas/protocols/stream/index.ts: - NetworkSchema gains 'hysteria'. - NetworkSettingsSchema gains the 7th branch { network: 'hysteria', hysteriaSettings: HysteriaStreamSettingsSchema }. OutboundFormModal.tsx: - NETWORK_OPTIONS keeps the 6 standard transports for non-hysteria protocols; when protocol === 'hysteria', a 7th option is appended (matches the legacy [...NETWORKS, 'hysteria'] gate). - newStreamSlice handles the 'hysteria' case with sensible defaults matching the legacy HysteriaStreamSettings constructor. - New sub-form when network === 'hysteria': 8 common fields (auth, congestion, up, down, udphop Switch + 3 nested fields when on, maxIdleTimeout, keepAlivePeriod, disablePathMTUDiscovery). - Receive-window tuning fields are still edit-via-JSON (rarely touched + would clutter the form). * feat(frontend): fallbacks polish — move up/down + Add all button Two small UX wins on the InboundFormModal Fallbacks card: - Per-row Move up / Move down buttons (ArrowUp/Down icons) that swap adjacent indices. Order survives reloads via sortOrder (rebuilt from index on save). First row's Up button + last row's Down button are disabled. - 'Add all' button next to 'Add fallback' that one-shot inserts a fresh row for every eligible inbound (every option in fallbackChildOptions) not already wired up. Disabled when every eligible inbound is already covered. Convenient for operators running catch-all routing across every host on the panel. * feat(frontend): XHTTP advanced fields on outbound modal Replace the 'edit via JSON' deferred-features hint with the full XHTTP sub-form matching the legacy modal's XhttpFields helper. schemas/protocols/stream/xhttp.ts: - New XHttpXmuxSchema: 6 connection-multiplexing knobs (maxConcurrency, maxConnections, cMaxReuseTimes, hMaxRequestTimes, hMaxReusableSecs, hKeepAlivePeriod). - XHttpStreamSettingsSchema gains 5 outbound-only fields and one UI-only toggle: scMinPostsIntervalMs, uplinkChunkSize, noGRPCHeader, xmux, enableXmux. outbound-form-adapter.ts: - New stripUiOnlyStreamFields() drops xhttpSettings.enableXmux on the way to wire so the panel never embeds the UI toggle into the saved config. xray-core ignores unknown fields anyway, but the panel reads back its own emitted JSON, so a clean wire shape matters. OutboundFormModal.tsx: - Headers editor (HeaderMapEditor v1) for xhttpSettings.headers. - Padding obfs Switch + 4 conditional fields (key/header/placement/ method) when on. - Uplink HTTP method Select with GET disabled outside packet-up. - Session placement + session key (key shown when placement != path). - Sequence placement + sequence key (same pattern). - packet-up mode: scMinPostsIntervalMs, scMaxEachPostBytes, uplink data placement + key + chunk size (key/chunk-size shown when placement != body). - stream-up / stream-one mode: noGRPCHeader Switch. - XMUX Switch + 6 nested fields when on. * feat(frontend): inbound TCP HTTP camouflage response fields + request headers Complete the TCP HTTP camouflage UI on the inbound side. Already there from the previous symmetric host/path commit: - Request host (string[] via comma-string) - Request path (string[] via comma-string) This commit adds: - Request headers (V2 map: name -> string[]) via HeaderMapEditor. - Response version (defaults to '1.1' when camouflage toggles on). - Response status (defaults to '200'). - Response reason (defaults to 'OK'). - Response headers (V2 map) via HeaderMapEditor. The HTTP camouflage Switch seeds both request and response sub-objects on toggle-on so xray-core sees a valid TcpHeader.http shape from the first save. Without the response seed, partial fills would emit a schema-incomplete response block that xray-core might reject. * feat(frontend): link import on outbound modal (vmess/vless/trojan/ss/hy2) The legacy outbound modal could import a vmess://, vless://, trojan://, ss://, or hysteria2:// share link via a Convert button on the JSON tab. Restore that UX with a focused pure-function parser. lib/xray/outbound-link-parser.ts: - parseVmessLink: base64 JSON, maps net/tls + per-network params onto the discriminated stream branch. - parseVlessLink: standard URL with type/security/sni/pbk/sid/fp/flow query params, dispatches transport via buildStream + applies security params via applySecurityParams. - parseTrojanLink: same URL pattern, defaults security to tls. - parseShadowsocksLink: both modern (base64 userinfo@host:port) and legacy (base64 of whole thing) ss:// formats. - parseHysteria2Link: accepts both hysteria2:// and hy2:// schemes, uses the hysteria stream branch with version=2 + TLS h3. - parseOutboundLink dispatcher returns the first non-null parser result, or null when no scheme matches. test/outbound-link-parser.test.ts: - 13 cases covering happy paths for each protocol family plus malformed input, ss:// dual-format handling, hy2:// alias. OutboundFormModal.tsx: - Import button on the JSON tab Input.Search; on success, parsed payload flows through rawOutboundToFormValues, the form is reset, and we switch back to the Basic tab. - Tag is preserved when the parsed link does not carry one. Out of scope: advanced fields the legacy parser handled (xmux, padding obfs, reality short IDs, finalmask from fm= param). Power users can finish the import in the form after the basics land. * feat(frontend): inbound Hysteria stream sub-form (auth + udpIdleTimeout + masquerade) Restore the inbound side of Hysteria stream configuration that was previously hidden — the legacy modal exposed these knobs but the Pattern A rewrite gated them out. schemas/protocols/stream/hysteria.ts: - HysteriaMasqueradeSchema covers the inbound-only masquerade wire shape: type ('proxy'|'file'|'string'), dir, url, rewriteHost, insecure, content, headers, statusCode. The three masquerade types cover the spectrum: reverse-proxy upstream, serve static files, or return a fixed string body. - HysteriaStreamSettingsSchema gains 3 inbound-side optional fields: protocol, udpIdleTimeout, masquerade. Outbound side is untouched (the legacy class accepted both wire shapes via the same struct). InboundFormModal.tsx: - New hysteria stream sub-form section in streamTab, gated by protocol === HYSTERIA. Fields: version (disabled, locked to 2), auth, udpIdleTimeout, masquerade Switch + nested type-Select with three conditional sub-blocks (proxy URL+rewriteHost+insecure, file dir, string statusCode+body+headers). - onValuesChange cascade: switching TO hysteria seeds streamSettings with the hysteria branch (forcing network='hysteria' + TLS); switching AWAY from hysteria snaps back to TCP so the standard network selector has a valid starting point. masquerade headers use the HeaderMapEditor v1 component. * feat(frontend): complete outbound sockopt section with remaining knobs Add the four remaining SockoptStreamSettings fields that were edit-via-JSON-only after the initial outbound modal rewrite: - TCP keep-alive idle (s) — tcpKeepAliveIdle, time before sending the first probe on an idle TCP connection. - TCP max segment — tcpMaxSeg, override the default MSS. - TCP window clamp — tcpWindowClamp, cap the TCP receive window. - Trusted X-Forwarded-For — trustedXForwardedFor, list of trusted proxy hostnames/CIDRs whose XFF headers Xray will honor. The outbound sockopt section now exposes all 17 SockoptStreamSettings fields from the schema. The InboundFormModal's sockopt section has its own field list (closer to the legacy class) and is unchanged. * feat(frontend): outbound TCP HTTP camouflage parity with inbound Add method/version inputs, request header map, and full response sub-section (version/status/reason/headers) to OutboundFormModal so the outbound side can configure the same HTTP-1.1 obfuscation knobs the inbound side already exposed. * feat(frontend): round-trip XHTTP advanced fields in outbound link parser Pick up xPaddingBytes, scMaxEachPostBytes, scMinPostsIntervalMs, uplinkChunkSize, and noGRPCHeader from both vmess:// JSON and the URL query-param parsers (vless/trojan). The advanced xmux/padding-obfs/ reality-shortId knobs still wait on a follow-up; this slice unblocks the common case where a phone-issued xhttp link carries non-default padding or post sizes. * feat(frontend): round-trip XHTTP padding-obfs + remaining advanced knobs Extract the XHTTP key-mapping into typed string/number/bool key arrays applied by both the URL query-param branch and the vmess JSON branch. The parser now covers xPaddingObfsMode + xPaddingKey/Header/Placement/ Method, sessionKey/seqKey/uplinkData{Placement,Key}, noSSEHeader, scMaxBufferedPosts, scStreamUpServerSecs, serverMaxHeaderBytes, and uplinkHTTPMethod alongside the previous five XHTTP fields. Two new round-trip tests cover the padding-obfs surface on both link forms. * feat(frontend): FinalMaskForm rewrite to Pattern A + wire into both modals Rewrite FinalMaskForm.tsx from a class-coupled component (mutated stream.finalmask.tcp[] via .addTcpMask/.delTcpMask methods, notified parent via onChange callback) into a Pattern A sub-form: takes a NamePath base, a FormInstance, and the surrounding network/protocol, then composes Form.List + Form.Item at absolute paths under that base. All array structures use nested Form.List — tcp/udp mask arrays, the clients/servers groups in header-custom (Form.List of Form.List of ItemEditor), and the noise list. Type Selects use onChange to reset the settings sub-object via form.setFieldValue, mirroring the legacy changeMaskType behavior. The kcp.mtu side effect on xdns type change is preserved. Wired into both InboundFormModal and OutboundFormModal stream tabs, placed after the sockopt section. The component is the first Pattern A consumer of nested Form.List inside another Form.List, so it stands as the reference for future nested-array sub-forms. * docs(frontend): record FinalMaskForm rewrite + hookup in status doc Mainline migration goal — replace class-based xray models with Zod schemas as the single source of truth + drive all forms through AntD `Form.useForm` + `antdRule(schema.shape.X)` — is complete. Remaining items are incremental polish. * fix(frontend): Phase 2 Inbound form reactivity bugs (B1-B9, consolidated) A run of resets dropped the per-bug commits 1401d833 / 5b1ae450 / 5bce0dc5 / 4007eec7. Re-landing all fixes against the same files in one commit to avoid another rebase-style drop. B1 — Transmission Select / External Proxy + Sockopt switches didn't react after click. AntD 6.4.3 Form.useWatch on nested paths doesn't re-fire reliably after `setFieldValue('streamSettings', cleaned)` on the parent. Bound Transmission via `name={['streamSettings', 'network']}` and wrapped the two switches in `<Form.Item shouldUpdate>` blocks that read state via getFieldValue. B2 — Security regressed from `Radio.Group buttonStyle="solid"` to a Select dropdown, and disable state didn't refresh because tlsAllowed/ realityAllowed were derived at the top of the component. Restored Radio.Button group and moved canEnableTls/canEnableReality evaluation inside the shouldUpdate render prop. B3 — Advanced tab "All" sub-tab was missing. Added it as the first item with a new AdvancedAllEditor that round-trips top-level fields + the three nested slices on edit. B4 — Advanced tab title/subtitle and per-section help text were gone. Wrapped the Tabs in the existing `.advanced-shell` / `.advanced-panel` structure and restored the `.advanced-editor-meta` help under each sub-tab using existing i18n keys. B5 — TLS / Reality sub-forms didn't render when selecting tls or reality on the Security tab. The `{security === 'tls' && ...}` and `{security === 'reality' && ...}` conditionals used a stale top-level useWatch value. Wrapped both in <Form.Item shouldUpdate> blocks that read `security` via getFieldValue. B6 — Advanced JSON editors stale after Stream/Sniffing changes. The editors seeded text via lazy useState and AntD Tabs renders all panes upfront, so the Advanced tab was already mounted with stale data. Both AdvancedSliceEditor and AdvancedAllEditor now subscribe via Form.useWatch and re-sync the text buffer when the watched JSON differs from a lastEmitRef (the serialization at the moment of our own last accepted write). User typing doesn't trigger re-sync because setFieldValue updates lastEmitRef too. (A prior attempt added `destroyOnHidden` to the outer Tabs but broke conditional tab items when the unmounted Form.Item for `protocol` lost its value — abandoned in favor of useWatch reactivity.) B7 — HeaderMapEditor + button did nothing. addRow() appended a blank {name:'', value:''} row, but commit() filtered it via rowsToMap before reaching the form, so AntD saw no change and didn't re-render. The editor now keeps a local rows state so blank rows survive during editing; only filled rows are emitted to onChange. B9 — Sniffing destOverride defaults (HTTP/TLS/QUIC/FAKEDNS) were not pre-checked on a fresh Add Inbound. buildAddModeValues() seeded sniffing: {} which left destOverride undefined. Now seeds with SniffingSchema.parse({}) so the Zod defaults populate. * fix(frontend): FinalMaskForm TCP Mask sub-forms + Advanced JSON wrap (B10/B11) B10 — FinalMaskForm TCP Mask: after adding a mask and picking a Type (Fragment/Header Custom/Sudoku), the type-specific sub-forms didn't render. TcpMaskItem read `type` via Form.useWatch on a path inside Form.List, which doesn't re-fire reliably in AntD 6.4.3 — same root cause as the earlier B1/B2/B5 reactivity issues. Replaced with a <Form.Item shouldUpdate> wrapper that reads `type` via getFieldValue inside the render prop. B11 — Advanced sub-tabs (settings / streamSettings / sniffing) showed just the inner value (e.g. `{clients:[],decryption:"none",...}`), but the legacy modal wrapped each slice with its key envelope (e.g. `{settings:{...}}`) so the JSON matches the wire shape's slice and round-trips cleanly from copy-pasted inbound configs. Added a `wrapKey` prop to AdvancedSliceEditor that wraps/unwraps the value on render/write; the three sub-tabs now pass settings / streamSettings / sniffing as their wrapKey. * fix(frontend): import InboundFormModal.css so layout classes apply (B12) The file InboundFormModal.css existed but was never imported, so every class in it had no effect — including: - .vless-auth-state — the "Selected: <auth>" caption next to the X25519/ ML-KEM/Clear button row stayed inline next to Clear instead of display:block beneath the row - .advanced-shell / .advanced-panel — the Advanced tab's header / panel framing was missing - .advanced-editor-meta — the per-section help text under each Advanced sub-tab had no spacing - .wg-peer — wireguard peer rows had no top margin Add a side-effect import of the CSS file at the top of the modal. No other change needed; the legacy modal must have either imported it or had a global import that the new modal didn't inherit. * fix(frontend): FinalMaskForm relative paths + network-switch defaults (B13/B14) B13 — FinalMaskForm used absolute paths like ['streamSettings', 'finalmask', 'tcp', 0, 'type'] for Form.Item names inside Form.List render props. AntD's Form.List prefixes Form.Item names with the list's own name, so the actual storage path became ['streamSettings', 'finalmask', 'tcp', 'streamSettings', 'finalmask', 'tcp', 0, 'type'] — total nonsense. Symptoms: Type Select didn't show the 'fragment' default after add(), and the sub-form for the picked type never rendered (Fragment/Sudoku/HeaderCustom). Rewrote FinalMaskForm to use RELATIVE names inside every Form.List context (TCP/UDP outer list + nested clients/servers/noise inner lists). Added a `listPath` prop on the items so the shouldUpdate guard and the side-effect setFieldValue calls (resetting `settings` when type changes) can still address the absolute path; the displayed Form.Items use the relative form (`[fieldName, 'type']`). Replaced top-level Form.useWatch on nested paths with <Form.Item shouldUpdate> blocks reading via getFieldValue, same pattern as the earlier B5 fix — Form.useWatch on paths inside Form.List doesn't re-fire reliably in AntD 6.4.3. B14 — Switching network (KCP, WS, gRPC, XHTTP, ...) seeded the new XSettings blob as `{}` so every field showed as empty. The legacy `newStreamSlice` populated mtu=1350, tti=20, etc. Restored those defaults in onNetworkChange and seeded the initial tcpSettings.header in buildAddModeValues so even the default TCP state shows the HTTP-camouflage Switch in the correct off state instead of an undefined header object. * fix(frontend): inbound TCP HTTP camouflage drops request fields + KCP UI field rename (B15/B16) B15 — Inbound TCP HTTP camouflage exposed Host / Path / Method / Version / request-headers inputs. Per Xray docs (https://xtls.github.io/config/transports/raw.html#httpheaderobject), the `request` object is honored only by outbound proxies; the inbound listener reads `response`. Those inputs were writing dead data the server ignored. Removed them from the inbound modal; only Response {version, status, reason, headers} remain. The toggle still seeds an empty request object so the wire shape stays valid against the schema. B16 — KCP Uplink / Downlink inputs bound to non-existent form fields `upCap` / `downCap`, while the schema (and wire) use `uplinkCapacity` / `downlinkCapacity`. Renamed the Form.Items to the schema names so defaults populate and saves persist. Also corrected newStreamSlice('kcp') to seed the four KCP defaults (uplinkCapacity / downlinkCapacity / cwndMultiplier / maxSendingWindow) — the missing two were why "CWND Multiplier" and "Max Sending Window" still showed empty after switching to KCP. * fix(frontend): seed full Zod-schema defaults for stream slices + QUIC params (B17) XHTTP showed blank Selects for Session Placement / Sequence Placement / Padding Method / Uplink HTTP Method (and several other knobs). Those fields have a literal "" (empty string) value in the schema, which the Select renders as "Default (path)" / "Default (repeat-x)" / etc. The form field was `undefined`, not `""`, so the Select showed blank instead of the labelled default option. newStreamSlice in InboundFormModal hand-rolled per-network seed objects with only a handful of fields. Replaced with {Tcp,Kcp,Ws,Grpc,HttpUpgrade,XHttp}StreamSettingsSchema.parse({}) so every default declared in the schema populates the form on network switch. Same change in buildAddModeValues for the initial TCP state. QUIC Params (FinalMaskForm) had the same shape on a smaller scale — defaultQuicParams() only seeded congestion + debug + udpHop. The schema's other fields are .optional() (no Zod default) so a schema parse won't help. Hard-coded the xray-core / hysteria recommended values (maxIdleTimeout 30, keepAlivePeriod 10, brutalUp/Down 0, maxIncomingStreams 1024, four window sizes) so the InputNumber controls render with usable starting values instead of blank. * fix(frontend): forceRender all tabs so fields register at modal open (B18) AntD Tabs with the `items` API lazy-mounts inactive tab panes by default. The Form.Items inside an unvisited tab never register, so: - Form.useWatch on a parent path (e.g. 'sniffing') returns a partial view containing only registered children. Until the user clicked the Sniffing tab, Advanced > Sniffing JSON showed `{sniffing: {}}` instead of the full default object set by setFieldsValue. - After visiting the Sniffing tab once, the `sniffing.enabled` Form.Item registered, so useWatch suddenly returned `{enabled: false}` — still partial, because the rest of the sniffing children only register when their Form.Items mount in conditional sub-sections. Setting `forceRender: true` on every tab item forces all tab panes to mount at modal open. Every Form.Item registers immediately; the watch result reflects the full form value seeded by buildAddModeValues. This also likely resolves the earlier "Invalid discriminator value" error on submit, which surfaced when streamSettings had an unregistered security field whose Form.Item hadn't mounted yet. * refactor(frontend): align hysteria with new docs + drop hysteria2 protocol Phase 2 smoke fixes on the Inbound add flow surfaced that hysteria2 was modeled as a separate top-level protocol when it's really just hysteria v2. The xray transports/hysteria.html docs also pin the hysteria stream to a minimal shape (version/auth/udpIdleTimeout/masquerade) — the previous schema carried legacy congestion/up/down/udphop/window knobs that aren't part of the wire contract. Hysteria2 removal: - Drop 'hysteria2' from ProtocolSchema enum and Protocols const - Drop hysteria2 branches from inbound/outbound discriminated unions - Drop createDefaultHysteria2InboundSettings / OutboundSettings - Delete schemas/protocols/inbound/hysteria2.ts and outbound/hysteria2.ts - Drop hysteria2 case in getInboundClients / genLink (fell through to the hysteria handler anyway) - Update client form modals' MULTI_CLIENT_PROTOCOLS sets - Remove hysteria2-basic fixture + snapshot entries (14 capability cases, 1 protocols fixture, 1 inbound-defaults factory) - Keep parseHysteria2Link() outbound parser since hysteria2:// is the share-link URI prefix for hysteria v2 Hysteria stream alignment with xtls docs: - HysteriaStreamSettingsSchema reduced to version/auth/udpIdleTimeout/ masquerade per transports/hysteria.html - Masquerade type adds '' (default 404 page) and defaults to it - Outbound form drops Congestion/Upload/Download/UDP hop/Max idle/ Keep alive/Disable Path MTU controls and the receive-window note - newStreamSlice('hysteria') in OutboundFormModal mirrors the trimmed shape; outbound-link-parser emits the trimmed shape too - InboundFormModal Masquerade Select gains the default option New TUN inbound schema: - Add schemas/protocols/inbound/tun.ts with name/mtu/gateway/dns/ userLevel/autoSystemRoutingTable/autoOutboundsInterface - Wire into ProtocolSchema enum, InboundSettingsSchema discriminated union, createDefaultInboundSettings dispatcher Other Phase 2 smoke fixes folded in: - Tunnel portMap UI swaps Form.List for HeaderMapEditor v1 — wire shape is Record<string,string> and the List was producing arrays - Hysteria onValuesChange seeds full TLS schema defaults + one empty certificate row (Cipher Suites/Min/Max Version/uTLS/ALPN were undefined before) - HTTP/Mixed accounts Add button auto-fills user/pass with RandomUtil.randomLowerAndNum - Hysteria security tab gates the 'none' radio out — TLS only - Hysteria stream tab drops the inbound Auth password field (xray inbound auth is per-user via 'users', not stream-level) - Reality onSecurityChange auto-randomizes target/serverNames/ shortIds and fetches an X25519 keypair - Tag and DB-side fields (up/down/total/expiryTime/ lastTrafficResetTime/clientStats/security) gain hidden Form.Items so validateFields keeps them in the wire payload (rc-component form strips unregistered fields) - WireGuard inbound auto-seeds one peer with generated keypair, allowedIPs ['10.0.0.2/32'], keepAlive 0 — matches legacy - WireGuard peer rows separated by Divider with the Peer N title and a small inline remove button (titlePlacement="center") * refactor(frontend): retire class-based xray models (Step 5) Delete models/inbound.ts (3,359 lines) and outbound.ts (2,405). The Inbound/Outbound classes and ~50 sub-classes are replaced by Zod-typed data + pure functions in lib/xray/*. Consumer migration off dbInbound.toInbound(): - useInbounds: isSSMultiUser({protocol, settings}) directly - QrCodeModal: genWireguardConfigs/Links/AllLinks from lib/xray - InboundList: derives tags from streamSettings raw fields - InboundsPage: clone via raw JSON, fallback projection via schema-shape stream object, exports via genInboundLinks - InboundInfoModal: builds an InboundInfo facade locally from raw streamSettings (host/path/serverName/serviceName per network), canEnableTlsFlow + isSS2022 from lib/xray New helper: lib/xray/inbound-from-db.ts exposes inboundFromDb(raw) converting a raw DBInbound row into a schema-typed Inbound for the link-generation orchestrators. DBInbound trimmed: drops toInbound, isMultiUser, hasLink, genInboundLinks, _cachedInbound. Imports Protocols from @/schemas/primitives now that ./inbound is gone. Bundled Phase 2 fixes: - Outbound modal: Form.useWatch with preserve: true so the stream block doesn't gate itself out when network is unmounted - Inbound form adapter: pruneEmpty preserves empty objects; per-protocol client field projection via Zod safeParse; sniffing collapse to {enabled:false} - useClients invalidateAll also invalidates inbounds.root() - IndexPage Config modal top/maxHeight polish Tests: 283/283 pass. typecheck/lint clean. * fix(frontend): inboundFromDb fills Zod defaults for stream + settings Smoke-testing the new inboundFromDb helper surfaced two regressions that the strict lib/xray link generators expose when fed raw DB streamSettings without per-network sub-keys. 1. genVlessLink / genTrojanLink crash on `stream.tcpSettings.header` when streamSettings lacks `tcpSettings` (true for slim list rows and for handcrafted minimal-JSON inbounds). The legacy Inbound.fromJson chain populated TcpStreamSettings via its own constructor; the new helper now does the same by parsing the raw <network>Settings sub-object through the matching Zod schema and merging schema defaults onto whatever the DB stored. 2. genVlessLink writes `encryption=undefined` into the share URL when settings lacks the `encryption: 'none'` literal that vless wire JSON normally carries. Fixed by running raw settings through InboundSettingsSchema.safeParse() to populate per-protocol defaults (encryption, decryption, fallbacks, etc.) the same way the legacy class fromJson chain did. Same pattern applied to security branch (tls/realitySettings). Tests: src/test/inbound-from-db.test.ts covers - JSON-string / object / empty settings coercion - genInboundLinks vless (TCP/none, with encryption=none) - genWireguardConfigs + genWireguardLinks peer fanout - genAllLinks trojan with TLS sub-defaults applied - protocol-capability helpers with raw shapes - getInboundClients across vless/SS-single/non-client protocols 296/296 pass. * fix(frontend): QUIC udpHop.interval is a range string, not a number (B19) User report: "streamSettings.finalmask.quicParams.udpHop.interval: Invalid input: expected string, received number". Three-part fix: - FinalMaskForm: Hop Interval input changed from InputNumber to Input with "e.g. 5-10" placeholder. xray-core spec says interval is a range string like '5-10' (seconds between min-max hops), not a single number. - FinalMaskForm: defaultQuicParams() seeds interval: '5-10' instead of the broken `interval: 5`. - QuicUdpHopSchema: preprocess coerces number → string for legacy DB rows that were written by the now-fixed buggy UI. Stops the load-time validation crash on existing inbounds. Tests still 296/296. * fix(frontend): outbound link parser handles extra/fm/x_padding_bytes (B20) User-reported vless share link with full xhttp + reality + finalmask config failed to round-trip on outbound import. The inbound link generator emits three payloads the outbound parser was ignoring: 1. `extra=<json>` — bundles advanced xhttp knobs (xPaddingBytes, scMaxEachPostBytes, scMinPostsIntervalMs, padding-obfs keys, etc.). applyXhttpStringFromParams now JSON.parses this and merges the fields into xhttpSettings via the same JSON-branch logic used by vmess. 2. `x_padding_bytes=<range>` — snake_case alias the inbound emits alongside the camelCase form. Now applied before camelCase so explicit `xPaddingBytes` URL params still win. 3. `fm=<json>` — full finalmask object including quicParams.udpHop and tcp/udp mask arrays. New applyFinalMaskParam attaches the decoded object to streamSettings.finalmask. Wired into both parseVlessLink and parseTrojanLink. Tests: - Real B20 link parses with xhttp + reality + finalmask all populated - Precedence: camelCase URL > extra JSON > snake_case alias > default - Malformed extra JSON falls through without crashing the parser 300/300 pass. * fix(frontend): Outbound submit crash on non-mux protocols + tab a11y (B21) Two issues surfaced on Outbound save: 1. Crash: `Cannot read properties of undefined (reading 'enabled')` at formValuesToWirePayload. The modal hides the Mux switch entirely for non-stream protocols (dns/freedom/blackhole/loopback) and for stream protocols when isMuxAllowed gates it out (xhttp, vless+flow). With the field never registered, validateFields() returns no `mux` key — `values.mux.enabled` then dereferences undefined. Fix: optional chain `values.mux?.enabled` so missing mux skips the mux clause silently. Documented why mux can be absent. 2. Chrome a11y warning: "Blocked aria-hidden on an element because its descendant retained focus" — when the user has an input focused inside one Tab panel and switches to another tab, AntD marks the outgoing panel aria-hidden while focus is still inside. The browser warns, but the focused control is now invisible to AT users. Fix: blur the active element before setActiveKey in onTabChange. * fix(frontend): blur active element on every tab switch path (B21 follow-up) The previous B21 patch only blurred on user-initiated tab clicks via onTabChange. Two other paths still set activeKey while a JSON-tab input retained focus: - importLink: after a successful share-link parse, setActiveKey('1') switched to the form tab while the user's focus was still on the Input.Search they just pressed Enter in. Chrome logged the same "Blocked aria-hidden" warning because the panel they were leaving became aria-hidden synchronously, with their input still focused. - onTabChange entering the JSON tab: also did a bare setActiveKey with no blur, so going from a focused form input INTO the JSON tab could trip the warning in reverse. Fix: centralized switchTab(key) that blurs document.activeElement sync before calling setActiveKey. Every internal tab transition (importLink, onTabChange both directions) now routes through it. The single setActiveKey('1') in the open-modal useEffect is left as a plain setter because there's no focused input at modal-open time. * refactor(frontend): extract fillStreamDefaults to shared helper Move the network/security schema-default filler out of inbound-from-db.ts into stream-defaults.ts so other consumers can reuse it without dragging in the DBInbound-specific code path. * fix(frontend): derive QUIC/UDP-hop switch state from data presence (B22) The QUIC Params and UDP Hop toggles previously persisted as separate boolean flags (enableQuicParams / hasUdpHop) which weren't part of the xray wire format and weren't restored when a config was pasted into the modal. Use data presence as the single source of truth: the switch is on iff the corresponding sub-object exists. Switching off clears it back to undefined. * fix(frontend): xhttp form binding + drop empty strings from JSON (B23) uplinkHTTPMethod was wrapped Form.Item -> Form.Item(shouldUpdate) -> Select, which broke AntD's value/onChange injection (AntD only clones the immediate child). Restructured so shouldUpdate is the outer wrapper and Form.Item(name) directly wraps the Select. Also drop empty-string fields from xhttpSettings in the wire payload — fields like uplinkHTTPMethod, sessionPlacement, seqPlacement, xPaddingKey default to '' meaning "use server default", so they shouldn't appear in JSON as "field": "". Adds placeholder text to the 3 xhttp Selects so the form reflects the current value after JSON paste. * feat(frontend): align finalmask + sockopt with xray docs, add golden fixtures Schema fixes per https://xtls.github.io/config/transports/finalmask.html and https://xtls.github.io/config/transports/sockopt.html: finalmask: - QuicCongestionSchema: remove non-doc 'cubic', keep reno/bbr/brutal/force-brutal - Add BbrProfileSchema (conservative/standard/aggressive) and bbrProfile field - brutalUp/brutalDown: number -> string per docs (units like '60 mbps') - Tighten ranges: maxIdleTimeout 4-120, keepAlivePeriod 2-60, maxIncomingStreams min 8 - UdpMaskTypeSchema: add missing 'sudoku' - udpHop.interval stays as preprocessed string-range per intentional B19 divergence sockopt: - tcpFastOpen: boolean -> union(boolean, number) per docs (number tunes queue size) - mark: drop min(0) (can be any int) - domainStrategy default: 'UseIP' -> 'AsIs' per docs - tcpKeepAlive Interval/Idle defaults: 0/300 -> 45/45 per docs (outbound) - Add AddressPortStrategySchema enum (7 values) + addressPortStrategy field - Add HappyEyeballsSchema (tryDelayMs/prioritizeIPv6/interleave/maxConcurrentTry) - Add CustomSockoptSchema (system/type/level/opt/value) + customSockopt array Bug fixes: - options.ts: Address_Port_Strategy values were lowercase ('srvportonly'); xray-core requires camelCase ('SrvPortOnly'). Fixed all 6 entries. - OutboundFormModal: domainStrategy Select was mistakenly populated from ADDRESS_PORT_STRATEGY_OPTIONS; now uses DOMAIN_STRATEGY_OPTION. - OutboundFormModal: inline sockopt defaults (hardcoded {acceptProxyProtocol: false, domainStrategy: 'UseIP', ...}) replaced with SockoptStreamSettingsSchema.parse({}) so schema is the single source. Form additions (both InboundFormModal + OutboundFormModal): - Address+port strategy Select - Happy Eyeballs Switch + sub-form (tryDelayMs/prioritizeIPv6/interleave/maxConcurrentTry) - Custom sockopt Form.List (system/type/level/opt/value) - FinalMaskForm: BBR Profile Select (visible when congestion='bbr'), Brutal Up/Down placeholders updated to string format Golden fixtures (8 new + 4 xhttp extras): - finalmask/{tcp-mask, udp-mask, quic-params, combined}.json — cover all TCP mask types, 7 UDP mask types including new sudoku, full QUIC params shape - sockopt/{defaults, tcp-tuning, tproxy, full}.json — full sockopt knobs - stream/xhttp-{basic, extra-padding, extra-placement, extra-tuning}.json — cover the extra-blob fields bundled into share-link extra=<json> Tests now at 312 (up from 300); typecheck/lint clean. * feat(frontend): migrate DNS + Routing to Zod, align with xray docs Adds first-class Zod schemas for the xray-core DNS block and routing sub-objects (Balancer, Rule) matching the documented shape at https://xtls.github.io/config/dns.html and https://xtls.github.io/config/routing.html, then wires the DnsServerModal and BalancerFormModal up to those schemas. schemas/dns.ts (new): - DnsQueryStrategySchema enum (UseIP/UseIPv4/UseIPv6/UseSystem) - DnsHostsSchema record(string -> string | string[]) - DnsServerObjectInnerSchema + DnsServerObjectSchema (with preprocess to migrate legacy `expectIPs` -> `expectedIPs` alias) - DnsServerEntrySchema = string | DnsServerObject (xray accepts both) - DnsObjectSchema with all documented fields and defaults schemas/routing.ts (new): - RuleProtocolSchema enum (http/tls/quic/bittorrent) - RuleWebhookSchema (url/deduplication/headers) - RuleObjectSchema covering every documented field (domain/ip/port/ sourcePort/localPort/network/sourceIP/localIP/user/vlessRoute/ inboundTag/protocol/attrs/process/outboundTag/balancerTag/ruleTag/ webhook) with type=literal('field').default('field') - BalancerStrategyTypeSchema enum (random/roundRobin/leastPing/leastLoad) - BalancerCostObjectSchema {regexp,match,value} - BalancerStrategySettingsSchema (expected/maxRTT/tolerance/baselines/costs) - BalancerStrategySchema + BalancerObjectSchema schemas/xray.ts: - routing.rules: was loose 3-field object, now z.array(RuleObjectSchema) - routing.balancers: was z.array(z.unknown()), now z.array(BalancerObjectSchema) - dns: was 2-field loose, now full DnsObjectSchema - BalancerFormSchema: strategy now BalancerStrategyTypeSchema (enum) instead of z.string(); fallbackTag defaults to ''; settings? added for leastLoad DnsServerModal (full Pattern A rewrite): - useState/DnsForm interface -> Form.useForm<DnsServerForm>() - manual domain/expectedIP/unexpectedIP list -> Form.List - antdRule on address/port/timeoutMs for inline validation - preserves legacy collapse-to-bare-string behavior on submit BalancerFormModal: - Adds conditional leastLoad sub-form (Expected/MaxRTT/Tolerance/ Baselines/Costs) wired to BalancerStrategySettingsSchema - Strategy options derived from schema enum - Cost rows with regexp/literal switch + match + value - required prop on Tag and Selector for red asterisk visual BalancersTab: - BalancerRecord interface -> type alias to BalancerObject - onConfirm now propagates strategy.settings to wire when leastLoad - Removes useMemo wrapping `columns` array. The memo had deps [t, isMobile] (with an eslint-disable) so the column render functions kept their original closure over `openEdit`. Once a balancer was created and the user clicked the edit button, the stale openEdit fired with empty `rows`, so rows[idx] was undefined and the modal opened blank. Columns are cheap to rebuild each render, so dropping the memo is the right fix. DnsTab + RoutingTab: switch ad-hoc interfaces to schema-derived types. translations (en-US, fa-IR): add the previously-missing pages.xray.balancerTagRequired and pages.xray.balancerSelectorRequired keys so antdRule surfaces a real message instead of the raw i18n key. * test(frontend): golden fixtures for DNS, Balancer, Rule schemas Adds JSON fixtures under golden/fixtures/{dns,dns-server,balancer,rule} plus three vitest files that parse them through the new schemas and snapshot the result. dns/: minimal (servers as strings) + full (every top-level field plus hosts with geosite/domain/full prefixes and 5 mixed string/object servers covering fakedns, localhost, https://, tcp://, quic+local://). dns-server/: full (every DnsServerObject field) + legacy-expectips (asserts the z.preprocess that migrates the legacy `expectIPs` key into the canonical `expectedIPs`). balancer/: random-minimal (default strategy by omission), roundrobin, leastping, leastload-full (covers all StrategySettings fields and both regexp=true|false costs). rule/: minimal, full (exercises every RuleObject field including localPort, localIP, process aliases like `self/`, all four protocol enum values, ip negation `!geoip:`, attrs with regexp value, and the WebhookObject with deduplication+headers), balancer-routed (uses balancerTag instead of outboundTag), port-number (port as a number to prove the union(number,string) accepts both). * fix(frontend): serialize bulk client delete + drop deprecated Alert.message useClients.removeMany was firing all DELETEs in parallel via Promise.all. The 3x-ui backend mutates a single config JSON per request (read / modify / write), so 20 concurrent deletes raced on the same file: every request reported success, but only the last writer's copy stuck — about half the selected clients reappeared after the toast. Replace the parallel fan-out with a sequential for-of loop so each delete sees the committed state of the previous one. The trade-off is total latency (20 * ~250ms = ~5s) which is the correct behavior until the backend grows a proper /bulkDel endpoint. Also rename the Alert `message` prop to `title` in ClientBulkAdjustModal to clear the AntD v6 deprecation warning. * feat(clients): server-side bulk create/delete with per-inbound batching Replace the panel-side fan-out (Promise.all of single /add and /del calls) that raced on the shared inbound config and capped throughput at roughly one round-trip per client. New endpoints batch the work on the server: - POST /panel/api/clients/bulkDel { emails, keepTraffic } - POST /panel/api/clients/bulkCreate [ {client, inboundIds}, ... ] BulkDelete groups emails by inbound and performs a single read-modify-write per inbound (one JSON parse, one marshal, one Save) instead of N. Per-row DB cleanups (ClientInbound, ClientTraffic, InboundClientIps, ClientRecord) are batched with WHERE...IN queries. Per-email failures are reported via Skipped[] and processing continues. BulkCreate iterates payloads sequentially through the same Create path single-add uses, so heterogeneous batches (different inboundIds, plans) remain valid in one round-trip. Frontend bulkDelete/bulkCreate hooks parse the new response shape ({ deleted|created, skipped[] }) and the bulk-add modal now posts a single request instead of fanning out emails. * perf(clients): batch BulkAdjust per inbound, skip no-op xray calls on local Same per-inbound batching strategy as BulkDelete. The previous code called Update once per email, which itself looped through each inbound the client belonged to — reparsing the same settings JSON, calling RemoveUser+AddUser on xray, and running SyncInbound for every single email. For 200 emails in one inbound that's 200 JSON read/write cycles and 400 xray runtime calls. The new BulkAdjust groups emails by inbound and per inbound: - locks once, reads settings JSON once - mutates expiryTime/totalGB in place for every target client - writes the inbound and runs SyncInbound once ClientTraffic rows are updated with a single per-email query at the end (values differ per client so they can't be folded into one statement). For local-node inbounds the xray runtime calls are skipped entirely. The AddUser payload only contains email/id/security/flow/auth/password/ cipher — none of which change in an adjust — so RemoveUser+AddUser was a no-op that briefly flapped active users. Limit enforcement is driven by the panel's traffic loop reading ClientTraffic, not by xray-core. For remote-node inbounds rt.UpdateUser is preserved so the remote panel receives the new totals/expiry. Skip+report semantics match BulkDelete: any per-email error leaves that email's record/traffic untouched and is returned in Skipped[]. * refactor(backend): retire hysteria2 as a top-level protocol Hysteria v2 is not a separate xray protocol — it is plain "hysteria" with streamSettings.version = 2. The frontend already dropped hysteria2 from the protocol enum in 5a90f7e3; the backend was still carrying the literal as a compat alias. Removed: - model.Hysteria2 constant - model.IsHysteria helper (only callers were buildProxy + genHysteriaLink) - TestIsHysteria - "hysteria2" from the Inbound.Protocol validate oneof enum - All `case model.Hysteria, model.Hysteria2:` and `case "hysteria", "hysteria2":` branches across client.go, inbound.go, outbound.go, xray.go, port_conflict.go, xray/api.go, subService.go, subJsonService.go, subClashService.go - Stale #4081 comments Kept (correctly — these are client-side URI/config schemes that are independent of the xray protocol type): - hysteria2:// share-link URI in subService.genHysteriaLink - "hysteria2" Clash proxy type in subClashService.buildHysteriaProxy - Comments referring to Hysteria v2 as a transport version Note: this change does not include a DB migration. Existing rows with protocol = 'hysteria2' will fall through to the default switch arms after upgrade. A separate `UPDATE inbounds SET protocol = 'hysteria' WHERE protocol = 'hysteria2'` is required for installs that still hold legacy data. * refactor(frontend): retire all AntD + Zod deprecations Swept the codebase for @deprecated APIs using a one-off type-aware ESLint config (eslint.deprecated.config.js) and fixed every hit: - 78 instances of `<Select.Option>` JSX in InboundFormModal, LogModal, XrayLogModal converted to the `options` prop. - Zod's `z.ZodTypeAny` (deprecated for `z.ZodType` in zod v4) replaced in _envelope.ts, zodForm.ts, zodValidate.ts, and inbound-form-adapter.ts. - Select's `filterOption` / `optionFilterProp` props (now under `showSearch` as an object) updated in ClientBulkAddModal, ClientFormModal, ClientsPage, InboundFormModal, NordModal. - `Input.Group compact` swapped for `Space.Compact` in FinalMaskForm. - Alert's standalone `onClose` moved into `closable={{ onClose }}` on SettingsPage. - `document.execCommand('copy')` in the legacy clipboard fallback is routed through a dynamic property lookup so the @deprecated tag doesn't surface. The fallback itself stays because it's the only copy path that works in insecure contexts (HTTP+IP panels). The dropped ClientFormModal.css was already unimported. eslint.deprecated.config.js loads the type-aware ruleset and turns everything off except `@typescript-eslint/no-deprecated`, so future scans are a single command: npx eslint --config eslint.deprecated.config.js src Not wired into `npm run lint` because typed linting roughly triples the run time. Verified clean: typecheck, lint, and the deprecated scan all 0 warnings. * feat(clients): show comment under email in the Client column The clients table's Client cell already stacks email + subId; add the admin comment as a third muted line so notes like "VIP" or "friend of X" are visible in the list view without opening the info modal. Renders only when set, so rows without a comment look unchanged. * docs(frontend): refresh README + simplify deprecated-scan config README rewrite reflects the post-Zod-migration state: - 3 Vite entries (index/login/subpage), not "one per panel route" - New folders: schemas/, lib/xray/, generated/, test/, layouts/ - Scripts table covers test/gen:api/gen:zod alongside the existing dev/build/lint/typecheck - New sections on the Zod schema tree, the three validation layers, the unified Form.useForm + antdRule pattern, and the golden fixture testing setup - "Adding a new page" updated to reflect that most additions are just react-router entries in routes.tsx, not new Vite bundles - Explicit note that `@deprecated` in the prose is a JSDoc tag, not a shell command — comes with the exact one-line npx invocation eslint.deprecated.config.js trimmed: dropping the recommendedTypeChecked spread + the ~28 rule overrides that came with it. The config now wires the @typescript-eslint and react-hooks plugins manually and enables exactly one rule (`@typescript-eslint/no-deprecated`). 45 lines → 30, same output: zero false-positives, zero noise, zero deprecations on the current tree. * chore(frontend): bump deps + refresh lockfile `npm update` within the existing semver ranges, plus a Vite bump the user explicitly accepted: - vite 8.0.13 → 8.0.14 (exact pin kept) - dayjs 1.11.20 → 1.11.21 - i18next 26.2.0 → 26.3.0 - typescript-eslint 8.59.4 → 8.60.0 - @rc-component/table + a handful of other transitive antd deps resolved to newer patch versions in the lockfile The earlier 8.0.13 pin was carried over from an esbuild dep-optimizer regression that broke vue-i18n in Vite 8.0.14 dev mode. This codebase uses react-i18next, doesn't hit the same chunking edge case, and `npm run dev` was smoked clean on 8.0.14 before accepting the bump. * feat(clients): compact link + inbound rows in the info modal and table ClientInfoModal — Copy URL section reskinned: - Each link is a single row: [PROTOCOL] [remark] [copy] [QR] instead of a card with the raw 200-char URL printed inline - Remark is parsed per-protocol — VMess pulls it from the base64-JSON `ps` field, the rest from the `#fragment` - The row title strips the client email suffix so the same string isn't repeated three times in the modal; the QR popover still uses the full remark (it's the QR's own name for the download file) - QR button opens an inline Popover with the existing QrPanel, size 220, destroyed on close - Subscription section uses the same row layout (SUB / JSON tags, clickable subId, copy + QR actions) - New per-protocol Tag colors so the protocol is identifiable at a glance ClientInfoModal — Attached inbounds + ClientsPage table column: - Chip format changed from `${remark} (${proto}:${port})` to just `${proto}:${port}` — when an admin attaches 5 inbounds to one client the remark was repeated 5 times and wrapped onto two lines - Only the first inbound chip is shown; the rest collapse into a `+N` chip that opens a Popover with the full list (remark included). INBOUND_CHIP_LIMIT = 1 - Per-protocol Tag colors - Tooltip on each chip shows the full `${remark} (${proto}:${port})` - Table column pinned to width: 170 so the row doesn't reserve the old 300px of whitespace next to the compact chip Comment row in the info table is always shown now (renders `-` when unset) so the layout doesn't jump per-client. VmessSecuritySchema gets a preprocess pass that maps legacy `security: ""` (persisted on pre-enum-lock VMess inbounds) back to `'auto'`. z.enum's `.default()` only fires on a missing field, not on an empty string — without this, old rows fail validation with "expected one of aes-128-gcm|chacha20-poly1305| auto|none|zero". `z.infer` is taken from the raw enum so the inferred type stays the union, not `unknown`. i18n adds a `more` key (en-US + fa-IR) used by the overflow chip label. * fix(xray): heal shadowsocks per-client method across all start paths xray-core's multi-user shadowsocks insists the per-client `method` matches the inbound's top-level cipher exactly for legacy ciphers, and is empty for 2022-blake3-*. The previous code (xray.go) copied `Client.Security` into the per-client `method` blindly, so a multi-protocol client created with the VMess default `"auto"` poisoned the SS config with `method: "auto"` → "unsupported cipher method: auto". Fix in two parts: - GetXrayConfig no longer projects `Client.Security` into the SS entry; the inbound's top-level method is now the single source of truth. - HealShadowsocksClientMethods moves to `database/model` and is invoked from `Inbound.GenXrayInboundConfig`, so the runtime add/update path (runtime.AddInbound) is normalised in addition to the full-restart path. For legacy ciphers heal now overwrites mismatched per-client methods rather than preserving them, so stale DB rows are also healed. * feat(sub): compact subscription rows with per-link email + PQ QR hide Mirror the ClientInfoModal redesign on the public SubPage so the subscription viewer reads as a tight `[PROTO] [remark] [copy] [QR]` row per link instead of raw URL cards. - subService.GetSubs now returns the per-link email list alongside the links, threaded through subController and BuildPageData into the `emails` field on subData (env.d.ts updated). Public links.go is updated to ignore the new return. - SubPage strips the client email from each row title using the matched per-link email (same trimEmail behaviour as the modal), and hides the QR button for post-quantum links (`pqv=`, `mlkem768`, `mldsa65`) since the encoded URL won't fit in a single QR. * feat(clients): hide QR for post-quantum links in client info modal Post-quantum keys (mldsa65 / ML-KEM-768) blow the encoded URL past what a single QR can hold. Detect them by the markers VLESS share links actually carry — `pqv=<base64>` for mldsa65Verify and `encryption=mlkem768x25519plus.*` for ML-KEM-768 — and drop the QR button for those rows. Copy still works. * fix(schemas): widen VLESS decryption/encryption to accept PQ values The post-quantum auth blocks (ML-KEM-768, X25519) populate `settings.decryption` / `settings.encryption` with values like `mlkem768x25519plus.<base64>` and `xchacha20-poly1305.aead.x25519`, but the schema pinned both fields to z.literal('none') so saving an inbound after picking "ML-KEM-768 auth" failed with `Invalid input: expected "none"`. Relax both fields (inbound + outbound + outbound form) to z.string().min(1) keeping the 'none' default. xray-core does its own validation server-side so a string check at the form boundary is enough. * feat(sub): clash row + reorganise SubPage around Subscription info ClientInfoModal: - Add a Clash / Mihomo row to the subscription section, gated on subClashEnable + subClashURI from /panel/setting/defaultSettings. Defaults payload schema is widened to carry subClashURI/subClashEnable. SubPage: - Drop the rectangular QR-codes header that used to sit at the very top of the card. The subscription info table now leads, followed by Divider("Copy URL") + per-protocol link rows (already converted to the compact ClientInfoModal pattern), then a new Divider("Subscription") + compact rows for the SUB / JSON / CLASH URLs with copy + QR-popover actions. The apps dropdown row remains the footer. CSS clean-up: removed the now-unused .qr-row/.qr-col/.qr-box/.qr-code rules; kept .qr-tag and trimmed the info-table top gap. Added a .sub-link-anchor underline-on-hover style for the new URL rows. * fix(sub): multi-inbound traffic + trojan/hysteria userinfo + utf-8 vmess remark Three bugs surfaced by the new SubPage and the recent client-record refactor: - xray.ClientTraffic.Email is globally unique, so a multi-inbound client has exactly one traffic row attached to whichever inbound claimed it. Iterating inbound.ClientStats per inbound dedup-locked the first lookup to zero for clients that lived under any other inbound, so the SubPage info table read 0 B for all the multi- inbound subs. Replaced appendUniqueTraffic with a single AggregateTrafficByEmails(emails) helper that runs one WHERE email IN (?) over xray.ClientTraffic and folds the rows. GetSubs / SubClashService.GetClash / SubJsonService.GetJson all share it. - Trojan and Hysteria share-links embedded the raw password/auth into the userinfo (scheme://<value>@host) without percent-encoding, so passwords containing `/` or `=` (e.g., base64-with-padding) broke popular trojan clients with parse errors. Added encodeUserinfo() that wraps url.QueryEscape and rewrites the `+` (space) back to `%20` for parity with encodeURIComponent on the frontend; applied to trojan.password and hysteria.auth. Same fix on the frontend's genTrojanLink. - VMess link remarks ride inside a base64-encoded JSON payload, but the SubPage / ClientInfoModal parser used JSON.parse(atob(body)), which treats the binary string as Latin-1 and shreds any multi-byte UTF-8 sequence. Most visible on the emoji decorations (genRemark appends 📊/⏳), so a remark like `test-1.00GB📊` rendered as `test-1.00GBð…`. Routed through Uint8Array + TextDecoder('utf-8') so multi-byte codepoints survive. * feat(settings): drop email leg from default remark model Change the default remarkModel from "-ieo" to "-io" so a freshly installed panel composes share-link remarks from the inbound name + optional extra only, leaving out the client email. Existing panels keep whatever value they have saved — only fresh installs and fallback paths (parse failure, missing setting) pick up the new default. Touched everywhere the literal "-ieo" lived: the canonical default map, the two sub-package fallback constants, the four frontend defaults (model class, link generator, two inbound modals, useInbounds hook). Two snapshot tests regenerated and one obsolete "contains email" assertion in inbound-from-db.test.ts removed. To migrate an existing panel that wants the new behaviour, edit Settings → Remark Model and remove the email leg. * feat(sub): usage summary card + remark-email on QR popover labels SubPage now opens with a clear quota panel directly under the info table: large `used / total` numbers, gradient progress bar (green ≤ 75%, orange to 90%, red above), `remained` and `%` on the foot, plus a Tag chip for unlimited subscriptions and a coloured chip for days left until expiry (blue >3d, orange ≤3d, red on expiry). Driven entirely off existing subData fields — no backend changes. While the row title in the link list stays email-stripped (default remark model omits email now), the QR popover label folds it back in so the rendered QR card identifies the client unambiguously. Tag content becomes `<rowTitle>-<email>` in both SubPage and ClientInfoModal — the encoded link itself is unchanged. SubPage section order is now: info table → usage summary → SUB / JSON / CLASH endpoints → per-protocol Copy URL rows → apps row, so the most-glanceable status sits above the fold.
2026-05-26 19:26:50 -07:00
if healed, ok := model.HealShadowsocksClientMethods(inbound.Settings); ok {
Feat/multi inbound clients (#4469) * feat(clients): add shadow tables for first-class client promotion Introduces three new GORM-backed tables (clients, client_inbounds, inbound_fallback_children) and a populate-only seeder that backfills them from each inbound's existing settings.clients JSON. Duplicate emails across inbounds auto-merge under one client row, with each field conflict logged. Existing services are unchanged and continue reading from settings.clients — this commit is groundwork only. * feat(clients): make clients+client_inbounds the runtime source of truth Adds ClientService.SyncInbound that reconciles the new tables from each inbound's clients list whenever existing service paths mutate settings.clients. Wires it into AddInbound, UpdateInbound, AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, DelDepletedClients, autoRenewClients, and the timestamp-backfill path in adjustTraffics, plus DetachInbound on DelInbound. GetXrayConfig now builds settings.clients from the new tables before writing config.json, and getInboundsBySubId joins through them instead of JSON_EACH on settings JSON. Live Xray config and subscription endpoints are now driven by the relational view; settings.clients JSON stays in step as a side effect of every write. * feat(clients): add top-level Clients tab and CRUD API Adds /panel/api/clients endpoints (list, get, add, update, del, attach, detach) backed by ClientService methods that orchestrate the per-inbound Add/Update/Del flows so a single client row is created once and attached to many inbounds in one operation. The frontend gains a dedicated Clients page (frontend/clients.html + src/pages/clients/) with an AntD table, multi-inbound attach modal, and full CRUD. Axios interceptor learns to honour Content-Type: application/json so the JSON endpoints work alongside the legacy form-encoded ones. The legacy per-inbound client modal stays untouched in this PR — both flows now write to the same source of truth. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): add Port-with-Fallback inbound type Adds a new "portfallback" protocol that emits as a VLESS-TLS inbound under the hood but is paired with a sidecar table of child inbounds. Panel auto-builds settings.fallbacks at Xray-config-gen time from the sidecar — each child's listen+port becomes the fallback dest, with SNI/ALPN/path/xver match criteria pulled from the row. No more typing loopback ports by hand or keeping settings.fallbacks in sync. Backend: new FallbackService (Get/SetChildren, BuildFallbacksJSON); two new routes (GET/POST /panel/api/inbounds/:id/fallbackChildren); xray.GetXrayConfig injects fallbacks for PortFallback inbounds; the inbound model emits protocol="vless" so Xray accepts the config. Frontend: PORTFALLBACK joins the protocol dropdown; selecting it shows the standard VLESS controls plus a Fallback Children table (inbound picker + per-row SNI/ALPN/path/xver). Children are loaded on edit and replaced atomically on save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reset Traffic, QR Code, Info actions + Online/Remaining columns The Clients page table gains: - Online column — green/grey tag driven by /panel/api/inbounds/onlines, polled every 10s. - Remaining column — bytes-remaining tag, coloured green/orange/red against quota, purple infinity when unlimited. - Action icons per row: QR, Info, Reset traffic, Edit, Delete. ClientInfoModal shows the full client detail (uuid/password/auth, traffic ↑/↓ + remaining + all-time, expiry absolute + relative, attached inbounds chip list, online + last-online). ClientQrModal fetches links for the client's subId via /panel/api/inbounds/getSubLinks/:subId and renders each one through the existing QrPanel component. Reset Traffic confirms then calls the existing per-inbound endpoint on the client's first attached inbound (the traffic row is keyed on email globally, so any attached inbound resets the shared counter). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): expose Attached inbounds in edit mode The multi-select was gated on add-only, so editing a client had no way to change which inbounds it belonged to. The picker now shows in both modes, and on submit the modal diffs the picked set against the original attachedIds — additions go through the /attach endpoint, removals through /detach, both after the field update lands so the new attachments get the latest values. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): unbreak template parsing + stale i18n keys - InboundFormModal: split the multi-line help string in the PortFallback section onto one line — Vue's template parser was bailing on Unterminated string constant because a single-quoted literal spanned two lines inside a {{ }} interpolation. - ClientInfoModal: t('disable') was missing at the root level, so vue-i18n returned the key path literally. Use t('disabled') which exists. - Linter cleanup elsewhere: pages.client.* references renamed to pages.clients.* to match the merged i18n block; whitespace normalisation in a few unrelated Vue templates. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * 1 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(traffic): drop all-time traffic tracking Removes the AllTime field from Inbound and ClientTraffic and migrates existing DBs by dropping the all_time columns on startup. The counter duplicated up+down without adding signal, and the per-event accumulator ran on every traffic write. Frontend: drop the All-time column from the inbound list and the client-row table, the All-time row from the client info modal, and the All-Time Total Usage tile from the inbounds summary card. The allTimeTraffic/allTimeTrafficUsage i18n keys are removed across every locale. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): mobile cards, multi-select, bulk add Adds the same row-card layout the inbounds page uses on mobile: the table is suppressed under the mobile breakpoint and each client renders as a compact card with a status dot, email, Info button, Enable switch, and overflow menu. All the per-client detail (traffic, remaining, expiry, attached inbounds, flow, created/updated, URL, subscription) opens through the existing info modal. Multi-select with bulk delete wires AntD row-selection on desktop and a per-card checkbox on mobile; a Delete (N) button appears in the toolbar when anything is selected. Bulk add reuses the five email-generation modes from the inbound bulk modal but takes a multi-inbound picker so one bulk run can attach to several inbounds at once. Submits client-by-client through the existing /panel/api/clients/add endpoint. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): remove legacy per-inbound client UI Now that clients live as first-class rows attached to one or many inbounds, the per-inbound client UI on the inbounds page is dead weight — every client action either has a global equivalent on the Clients page or makes no sense in a many-to-many world. Deletes ClientFormModal, ClientBulkModal, CopyClientsModal, and ClientRowTable from inbounds/. Strips the matching emits, refs, handlers, and dropdown menu items from InboundList and InboundsPage, and removes the dead mobile expand-chevron state and the desktop expanded-row plumbing that drove the inline client table. The InboundFormModal Clients tab still works in add-mode (one inline client at inbound creation) — that flow goes through ClientService. SyncInbound on save and remains useful. Fixes a stray "</a-dropdown>" left over by an earlier toolbar edit in ClientsPage that broke the template parser. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Delete depleted action Mirrors the legacy delDepletedClients action that lived under the inbounds page, but as a first-class /panel/api/clients/delDepleted endpoint backed by ClientService. The new path goes through ClientService.Delete for each depleted email, so the new clients + client_inbounds + xray_client_traffic tables stay consistent. Adds a danger-styled toolbar button on the Clients page (next to Reset all client traffic) with a confirm dialog and a toast reporting the deleted count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): move every client-shaped endpoint off /inbounds onto /clients After the multi-inbound client migration, client state belongs to the client API surface, not the inbound one. Twelve routes that were crammed under /panel/api/inbounds/* now live where they belong, under /panel/api/clients/*. Moved (route, handler, doc): POST /clientIps/:email POST /clearClientIps/:email POST /onlines POST /lastOnline POST /updateClientTraffic/:email POST /resetAllClientTraffics/:id POST /delDepletedClients/:id POST /:id/resetClientTraffic/:email GET /getClientTraffics/:email GET /getClientTrafficsById/:id GET /getSubLinks/:subId GET /getClientLinks/:id/:email Their /clients/* counterparts are: POST /clients/clientIps/:email POST /clients/clearClientIps/:email POST /clients/onlines POST /clients/lastOnline POST /clients/updateTraffic/:email POST /clients/resetTraffic/:email (email-only, fans out) GET /clients/traffic/:email GET /clients/traffic/byId/:id GET /clients/subLinks/:subId GET /clients/links/:id/:email per-inbound resetAllClientTraffics and delDepletedClients are dropped entirely — the Clients page already exposes global Reset All Traffic and Delete depleted actions, and per-inbound resets are meaningless once a client can be attached to many inbounds. ClientService.ResetTrafficByEmail is the new email-only reset path: it looks up every inbound the client is attached to and pushes the counter reset + Xray re-add through inboundService.ResetClientTraffic for each one, so depleted users come back online instantly. Frontend callers (ClientsPage, useClients, ClientQrModal, ClientInfoModal, InboundInfoModal, InboundsPage, useInbounds) all switched to the new paths. The Inbounds page drops its per-inbound "Reset client traffic" and "Delete depleted clients" dropdown items — users do those at the client level now. api-docs is rebuilt to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): switch tgbot + ldap callers to ClientService Adds two thin helpers to ClientService (CreateOne, DetachByEmail) and rewrites tgbot.SubmitAddClient and ldap_sync_job to call ClientService directly. Removes the JSON-blob payloads (BuildJSONForProtocol output for add, clientsToJSON/clientToJSON helpers) that callers previously fed to InboundService.AddInboundClient/DelInboundClient. ldap_sync_job.batchSetEnable now loops InboundService.SetClientEnableByEmail per email instead of trying to coerce AddInboundClient into doing the update — the old path would have failed duplicate-email validation for existing clients anyway. The legacy InboundService.AddInboundClient/UpdateInboundClient/ DelInboundClient methods stay in place; they are now only used internally by ClientService Create/Update/Delete/Attach. Inlining + deleting them follows in a separate commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(service): move all client mutation methods to ClientService Moves the client mutation surface out of InboundService and into ClientService. These methods all operate on a single client (identity fields, traffic limits, expiry, ip limit, enable state, telegram tg id) and didn't belong on the inbound aggregate. Moved (12 methods): AddInboundClient, UpdateInboundClient, DelInboundClient, DelInboundClientByEmail, checkEmailsExistForClients, SetClientTelegramUserID, checkIsEnabledByEmail, ToggleClientEnableByEmail, SetClientEnableByEmail, ResetClientIpLimitByEmail, ResetClientExpiryTimeByEmail, ResetClientTrafficLimitByEmail. Each method now takes an explicit *InboundService for the helpers that legitimately stay on InboundService (GetInbound, GetClients, runtimeFor, AddClientStat / UpdateClientStat / DelClientStat, DelClientIPs / UpdateClientIPs, emailUsedByOtherInbounds, getAllEmailSubIDs, GetClientInboundByEmail / GetClientInboundByTrafficID, GetClientTrafficByEmail). Stays on InboundService: ResetClientTrafficByEmail and ResetClientTraffic(id, email) — these mutate xray_client_traffic rows, not client identity, so they're inbound-side bookkeeping. Callers updated: tgbot (6 calls), ldap_sync_job (1 call), InboundService internal (writeBackClientSubID, CopyInboundClients, AddInbound's email-uniqueness check), ClientService Create/Update/ Delete/Attach/Detach. Also removes a dead resetAllClientTraffics controller handler whose route was already gone after the previous /clients API migration. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): finish migrating to ClientService + tidy IP routes Two related cleanups in the new /clients surface: 1. Move ResetAllClientTraffics (bulk-reset of xray_client_traffic + last_traffic_reset_time, with node-runtime propagation) from InboundService to ClientService. PeriodicTrafficResetJob now holds a clientService and calls j.clientService.ResetAllClientTraffics(&j.inboundService, id). The last client-mutation method on InboundService is gone. 2. Shorten redundantly-named routes/handlers under /panel/api/clients: - /clientIps/:email -> /ips/:email (handler getIps) - /clearClientIps/:email -> /clearIps/:email (handler clearIps) The "client" prefix was redundant inside the clients namespace. Frontend (InboundInfoModal) and api-docs updated to match. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds,clients): clean up inbound modal + enrich client modal Inbound modal rework (InboundFormModal.vue + inbound.js): - Drop the embedded Client subform in the Protocol tab. Multi-inbound clients are managed exclusively from the Clients page now; a fresh inbound is created with zero clients (settings constructors default to []) and the user attaches clients afterwards. - Hide the Protocol tab entirely when it has nothing to render (VMESS, Trojan without fallbacks, Hysteria). Auto-switches active tab to Basic when the tab disappears while focused. - Move the Security section (Security selector + TLS block with certs and ECH + Reality block) out of the Stream tab into its own Security tab, sharing the canEnableStream gate. Client modal additions (ClientFormModal.vue + ClientBulkAddModal.vue): - Flow select (xtls-rprx-vision / -udp443) appears only when the panel actually has a Vision-capable inbound (VLESS or PortFallback on TCP with TLS or Reality). Hidden otherwise, and cleared when it disappears. - IP Limit input is disabled when the panel-level ipLimitEnable setting is off, fetched into useClients alongside subSettings and threaded through ClientsPage to both modals. - Edit modal now shows an "IP Log" section listing IPs that have connected with the client's credentials, with refresh and clear buttons (calls the renamed /panel/api/clients/ips and /clearIps endpoints). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(inbounds): drop manual Fallbacks UI from inbound modal The PortFallback protocol type now covers the common VLESS-master-plus-children case with auto-wired dests, so the manual Fallbacks editor (showFallbacks block in the Protocol tab) is mostly redundant. Removed: - the v-if="showFallbacks" template block (SNI/ALPN/Path/dest/PROXY rows) - the showFallbacks computed - the addFallback / delFallback helpers - the .fallbacks-header / .fallbacks-title styles - the showFallbacks gate from hasProtocolTabContent (so Trojan-over-TCP no longer shows an empty Protocol tab) Power users who need a non-inbound fallback dest (nginx, static site) can still author settings.fallbacks via the Advanced JSON tab. * feat(clients,inbounds): move search/filter to Clients page + small fixes Search/filter relocation: - Remove the search/filter toolbar (search switch + filter radio + protocol/node selects + the visibleInbounds projection + inboundsFilterState localStorage + filter CSS + the SearchOutlined/ FilterOutlined/ObjectUtil/Inbound imports it required) from InboundList. The filters were all client-oriented buckets bolted onto the inbound row. - Add a search/filter toolbar to ClientsPage with the same shape: switch between deep-text search and bucket filter (active / deactive / depleted / expiring / online) + protocol filter that matches clients attached to at least one inbound with the chosen protocol. State persists in clientsFilterState localStorage. filteredClients drives both the desktop table and the mobile card list, and select-all / allSelected / someSelected only span the visible subset. - useClients now also fetches expireDiff and trafficDiff from /panel/setting/defaultSettings (used to detect the expiring bucket); ClientsPage threads them into the client-bucket helper. Loose fixes folded in: - Add Client: email field is auto-filled with a random handle on open, matching uuid/subId/password/auth. - Inbound clone: parse and reuse the source settings JSON (with clients reset to []) instead of building a fresh defaulted Settings, so VLESS Encryption/Decryption and other non-client fields survive the clone. - en-US.json: add the ipLog string used by the edit-client modal. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): add Reverse tag field for VLESS-attached clients Mirrors the Flow field's pattern: a Reverse tag input appears in the Add/Edit Client modal whenever at least one selected inbound is VLESS or PortFallback. The value rides over the wire as client.reverse = { tag: '...' } so it lands directly in model.Client's *ClientReverse field; an empty value omits the reverse key entirely. On edit the field is hydrated from props.client.reverse?.tag, and the showReverseTag watcher clears the field if the user drops the last VLESS-like inbound from the selection. * fix(xray): emit only protocol-relevant fields per client entry The Xray config synthesizer was writing every identifier field (id, password, flow, auth, security/method, reverse) on every client entry regardless of the inbound's protocol. Xray ignores unknown fields, so the config worked, but it diverged from the spec and leaked secrets across protocols when one client was attached to multiple inbounds — a VLESS inbound's generated config carried the same client's Trojan password and Hysteria auth alongside its uuid. Switch on inbound.Protocol when building each entry: - VLESS / PortFallback: id, flow, reverse - VMess: id, security - Trojan: password, flow - Shadowsocks: password, method - Hysteria / Hysteria2: auth email is emitted for every protocol. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): restore auto-disable kick under new schema disableInvalidClients still resolved (inbound_tag, email) pairs via JSON_EACH(inbounds.settings.clients), which is empty after migrating to the clients + client_inbounds tables. Result: xrayApi.RemoveUser never ran for depleted clients, clients.enable stayed true so the UI showed them as active, and only xray_client_traffic.enable got flipped - making "Restart Xray After Auto Disable" only half-work. Resolve the targets via a JOIN through the new schema, flip clients.enable so the Clients page reflects the state, and drop the legacy JSON write-back plus the subId cascade workaround (email is unique now). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): live WebSocket updates + Ended status surfacing ClientsPage now subscribes to traffic / client_stats / invalidate WebSocket events instead of polling /onlines every 10s. Per-row traffic counters refresh in place, online state stays current, and list-level mutations elsewhere trigger a refresh. The client roll-up summary moves from InboundsPage to ClientsPage where it belongs, restructured into six labeled stat tiles (Total / Online / Ended / Expiring / Disabled / Active) with email popovers on the ones with issues. Auto-disabled clients (traffic exhausted or expiry passed) now classify as 'depleted' even though clients.enable=false, so they show up under the Ended filter and render a red Ended tag instead of looking indistinguishable from an operator-disabled row. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): per-node client roll-up and panel version Added transient inboundCount / clientCount / onlineCount / depletedCount fields to model.Node, populated by NodeService.GetAll via aggregated queries (one join across inbounds + client_inbounds, one over client_traffics intersected with the in-memory online emails). The Nodes list renders these as colored chips on a new "Clients" column so an operator can see at a glance how many users each node carries and how many are currently online or depleted. Also exposes the remote panel's version. The central panel adds panelVersion to its /api/server/status payload (sourced from config.GetVersion). Probe reads that field and persists it on the node row, mirroring how xrayVersion already flows. NodesPage gets a new column next to Xray Version, in both desktop and mobile views, with English and Persian strings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): stop node sync from resurrecting deleted clients Several related issues around node-managed clients: - Remote runtime: drop the per-inbound resetAllClientTraffics path and point traffic/onlines/lastOnline fetches at the new /panel/api/clients/* routes. - Delete from master: always push the updated inbound to the node even when the client was already disabled or depleted, so the node actually loses the user instead of silently keeping it. - setRemoteTraffic: mirror remote clients into the central tables only on first discovery of a node inbound. Matched inbounds let the master own the join table, so a stale snap can no longer re-create a ClientRecord (and join row) for a client that was just deleted on the master. - ClientService.Delete: route through submitTrafficWrite so deletes serialize with node traffic merges, and switch the final ClientRecord delete to an explicit Where("id = ?") clause. - setRemoteTraffic UNIQUE-constraint fix: use clause.OnConflict on inserts and email-keyed UPDATEs for client_traffics, so mirroring a snap doesn't trip the unique email index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(clients): switch client API endpoints from id to email All client-scoped routes now use the unique email as the path key (get, update, del, attach, detach, links). Email is the stable, protocol-independent identifier — UUIDs don't exist for trojan or shadowsocks, and internal numeric ids leaked panel implementation detail into the public API. Removed the redundant /traffic/byId/:id endpoint (covered by /traffic/:email) and collapsed /links/:id/:email into /links/:email, which now returns links across every attached inbound for the client. Frontend selection, bulk delete, and toggle state are now keyed by email as well, dropping the id→email lookup workaround. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(server): move cached state and helpers into ServerService ServerController had grown to hold its own status cache, version-list TTL cache, history-bucket whitelist, and the loop that drove all three — concerns that belong in the service layer. Pull them out: - lastStatus + the @2s refresh become ServerService.RefreshStatus and ServerService.LastStatus; the controller's cron now just orchestrates the cross-service side effects (xrayMetrics sample, websocket broadcast). - The 15-minute Xray-versions cache (with stale-on-error fallback) moves into ServerService.GetXrayVersionsCached, collapsing the controller handler to a single call. - The freedom/blackhole outbound-tag walk used by /xraylogs becomes ServerService.GetDefaultLogOutboundTags. - The allowed-history-bucket whitelist moves to package-level service.IsAllowedHistoryBucket, so both NodeController and ServerController validate against the same list. Net result: web/controller/server.go drops from 458 to 365 lines and contains only HTTP wiring + presentation-y side effects. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(api): emit JSON-text columns as nested objects Inbound, ClientRecord, and InboundClientIps store settings / streamSettings / sniffing / reverse / ips as JSON-text in the DB. The API was passing that text through verbatim, so every consumer had to JSON.parse a string inside a string. Add MarshalJSON / UnmarshalJSON so the wire format is a real nested object, while still accepting the legacy escaped-string shape on write. Frontend dbinbound.js gets a matching coerceInboundJsonField helper for the same dual-shape read path, and inbound.js toJson stops emitting empty/placeholder fields (externalProxy [], sniffing destOverride when disabled, etc.) so the new normalised JSON stays terse. api-docs and the inbound-clone path are updated to the new shape. Controller route lists are regrouped so all GETs sit above POSTs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): include inboundIds and traffic in /clients/list ClientRecord got its own MarshalJSON in the previous commit, and ClientWithAttachments embeds it to add inboundIds and traffic. Go promotes the embedded MarshalJSON to the outer struct, so the encoder was calling ClientRecord.MarshalJSON for the whole value and silently dropping the extras. The frontend reads row.inboundIds / row.traffic from /clients/list, so attached inbounds didn't render and newly added clients looked like they hadn't saved. Add an explicit MarshalJSON on ClientWithAttachments that splices the extras in. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): gate IP Log on ipLimitEnable + clean access-log dropdown Legacy panel hid the IP Log section when access logging was off; the Vue 3 migration left it gated on isEdit only, so the section showed even when xray's access log was 'none' and nothing was being recorded. Restore the ipLimitEnable gate on the edit modal's IP Log form-item. While here, clean up the Xray Settings access-log dropdown: previously two 'none' entries appeared (an empty value labelled with t('none') and the literal 'none' from the options array). Drop the empty option for access log (the literal 'none' covers it) and relabel the empty option for error log / mask address to t('empty') so they're distinguishable. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(nodes): route per-client ops through node clients API + orphan sweep Adds Runtime methods AddClient, UpdateUser, and DeleteUser so master mutates clients on a node via /panel/api/clients/{add,update,del} rather than pushing the whole inbound. The previous rt.UpdateInbound path made the node DelInbound+AddInbound on every single-client change, briefly cycling every other user on the same inbound. DelInbound no longer filters by enable=true, so a disabled node inbound actually gets removed from the node instead of being resurrected by the next snap. setRemoteTrafficLocked now sweeps any ClientRecord with zero ClientInbound rows after SyncInbound rebuilds the attachments, which is how a node-side delete propagates back to master instead of leaving a detached ghost. ClientService.Delete tombstones the email first so a snap arriving mid-delete can't re-create the record. WebSocket broadcasts an "invalidate(clients)" message on every client mutation so the Clients page refreshes without manual reload. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(balancers): allow fallback on all strategies + feed burstObservatory from random/roundRobin Drops the random/roundRobin gate on the Fallback field in BalancerFormModal so every strategy can pick a fallback outbound. syncObservatories now feeds burstObservatory from leastLoad + random + roundRobin balancers (was leastLoad only), matching how leastPing feeds observatory. Fix the JsonEditor "Unexpected end of JSON input" that appeared when switching a balancer between leastPing and another strategy: the obsView watcher was gated on showObsEditor (a boolean OR of the two flags) and missed the case where one observatory swapped for the other in the same tick. Watch the individual flags instead so obsView flips to the surviving editor and the getter stops pointing at a deleted key. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): use sortedInbounds for mobile empty-state check InboundList referenced an undefined visibleInbounds in the mobile card list's empty-state guard, throwing "Cannot read properties of undefined (reading 'length')" and breaking the entire mobile render. * feat(clients): sortable table columns Adds the same sortState / sortableCol / sortFns pattern InboundList uses, wrapping filteredClients in sortedClients so sort composes with the existing search/filter pipeline. Sortable: enable, email, inboundIds (attachment count), traffic, remaining, expiryTime; actions and online stay unsorted. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(shadowsocks): generate valid ss2022 keys and per-client method for legacy ciphers The Add Client flow on shadowsocks inbounds was producing xray configs that failed to start: - 2022-blake3-* ciphers need a base64-encoded key of an exact byte length per cipher. fillProtocolDefaults was assigning a uuid-style string, which xray rejects as "bad key". Now the password is generated (or replaced if invalid) via random.Base64Bytes(n) sized to the chosen cipher. - Legacy ciphers (aes-256-gcm, chacha20-*, xchacha20-*) require a per-client method field in multi-user mode; model.Client has no Method, so settings.clients was stored without one and xray failed with "unsupported cipher method:". applyShadowsocksClientMethod now injects the top-level method into each client on add/update, and healShadowsocksClientMethods backfills it at xray-config-build time so existing inbounds heal on the next start. - xray/api.go ssCipherType switch was missing aes-256-gcm, which fell through to ss2022 path. - SSMethods dropdown now offers aes-256-gcm. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): preserve ClientRecord on inbound delete + filter Attached inbounds to multi-client protocols Replace the global orphan sweep in setRemoteTrafficLocked with a per-inbound diff cleanup: only delete a ClientRecord whose email disappeared from a snap-tracked inbound (i.e. a node-side delete). Inbounds that vanished entirely from the snap (e.g. admin deleted the inbound on master) aren't iterated, so a client whose last attachment came from that inbound is now left alone instead of being deleted alongside the inbound. ClientFormModal and ClientBulkAddModal now filter the Attached inbounds dropdown to protocols that actually support multiple clients: shadowsocks, vless, vmess, trojan, hysteria, hysteria2, and portfallback (which routes through VLESS settings). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): make empty-state text readable on dark/ultra themes The "No clients yet" empty state had a hardcoded black color (rgba(0,0,0,0.45)) that vanished against the dark backgrounds. Drop the inline color, let it inherit from the AntD theme, and fade with opacity like the mobile card empty state already does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(clients): client-first tgbot add flow, tgId field, lightweight inbound options - tgbot: drop legacy per-protocol Add Client UI in favour of a client-first multi-inbound flow. New BuildClientDraftMessage / getInboundsAttachPicker let an admin pick one or more inbounds and submit a single client; per- protocol secrets are now generated server-side via fillProtocolDefaults. Drops awaiting_id/awaiting_password_tr/awaiting_password_sh state cases and add_client_ch_default_id/pass_tr/pass_sh/flow callbacks. Adds a setTGUser button + awaiting_tg_id state so the bot can set Client.TgID during Add. - clients UI: add Telegram user ID input to ClientFormModal (0 = none). Hide IP Limit field entirely when ipLimitEnable is off — disabled fields still take layout space, this collapses Auth(Hysteria) to full width. - inbounds API: new GET /panel/api/inbounds/options that returns just {id, remark, protocol, port, tlsFlowCapable}. Used by the clients page pickers so the dropdown payload stays small on panels with thousands of clients (drops settings JSON, clientStats, streamSettings). Server-side TlsFlowCapable mirrors Inbound.canEnableTlsFlow so the modal no longer needs to parse streamSettings client-side. - clientInfoMsg now shows attached inbound remarks, and getInboundUsages reports the attached client count per inbound. - api-docs: document the new /options endpoint and add tgId / flow to the clients add/update bodies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): keep Node column visible for node-attached inbounds The Node column was bound to hasActiveNode, so disabling every node hid the column even when inbounds were still attached to those nodes — the admin lost the visual cue that those inbounds belonged to a node and would come back when it was re-enabled. Combine hasActiveNode with a new hasNodeAttachedInbound check (any dbInbound with nodeId != null) so the column survives node-disable. * fix(api-docs): accept functional-component icons in EndpointSection AntD-Vue icons (SafetyCertificateOutlined, etc.) are functional components, so the icon prop's type: Object validator was rejecting them with a "Expected Object, got Function" warning at runtime. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * test: cover crypto, random, netsafe, sub helpers, xray equals, websocket hub, node service Adds ~110 unit tests across previously untested packages. Focus on pure-logic and concurrency surfaces where regressions would silently affect users: - util/crypto, util/random: password hashing round-trip, ss2022 key generation, alphabet/length invariants. - util/netsafe: IsBlockedIP edge cases, NormalizeHost validation, SSRF guard with AllowPrivate context bypass. - util/common, util/json_util: traffic formatter, Combine nil-skip, RawMessage empty-as-null and copy-on-unmarshal. - sub: splitLinkLines, searchKey/searchHost, kcp share fields, finalmask normalization, buildVmessLink round-trip. - xray: Config.Equals and InboundConfig.Equals field-by-field, getRequiredUserString/getOptionalUserString type checks. - web/websocket: hub registration, throttling, slow-client eviction, nil-receiver safety, concurrent register/unregister. - web/service: NodeService.normalize validation, normalizeBasePath, HeartbeatPatch.ToUI mapping. - web/job: atomicBool concurrent set/takeAndReset semantics. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(clients): replace English fallbacks with proper translation keys Pulls every hard-coded English label/title in the Clients page and its four modals through the i18n layer so localized panels stop leaking English. New keys live under pages.clients (auth, hysteriaAuth, uuid, flow, flowNone, reverseTag, reverseTagPlaceholder, telegramId, telegramIdPlaceholder, created, updated, ipLimit) plus refresh at the root and toasts.bulkDeletedMixed / bulkCreatedMixed for partial-failure toasts. Also switches the add-client modal's primary button from "Add" to "Create" for consistency with other create flows. The bulk-add Random/Random+Prefix/... email-method options stay hard-coded by request - they're identifier-shaped strings. * i18n: backfill 99 missing keys across all 12 non-English locales Brings every translation file up to parity with en-US.json so the Clients page, the fallback-children inbound section, the new refresh verb, the Nodes panel-version label and a handful of older holes stop falling through to the English fallback. New strings span: - pages.clients.* (labels, confirmations, toasts, emailMethods) - pages.inbounds.portFallback.* (Reality fallback inbound section) - pages.nodes.panelVersion, menu.clients, refresh Technical identifiers (Auth, UUID, Flow, Reverse tag) are intentionally left untranslated since they correspond to xray-core field names. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: drop stale pages.client block duplicated in every non-English locale Every non-English locale carried a pages.client (singular) section with 30 entries that duplicated pages.clients (plural). The plural namespace is what the Vue code actually consumes; the singular one was dead weight from an older rename that never got cleaned up in the non-English files. Removing it brings every locale to exactly 984 keys, matching en-US.json. * chore: apply modernize analyzer fixes across codebase Mechanical replacements suggested by golang.org/x/tools/.../modernize: strings.Cut/CutPrefix/SplitSeq, slices.Contains, maps.Copy, min(), range-over-int, new(expr), strings.Builder for hot += loops, reflect.TypeFor[T](), sync.WaitGroup.Go(), drop legacy //+build lines. * feat(database): add PostgreSQL as an optional backend alongside SQLite Lets operators with large client counts or multi-node setups pick PostgreSQL at install time without breaking the existing SQLite default. Backend is selected at runtime via XUI_DB_TYPE/XUI_DB_DSN, a small dialect layer keeps the five JSON_EXTRACT/JSON_EACH queries portable, and a new `x-ui migrate-db` subcommand copies SQLite data into PostgreSQL in FK-aware order. * fix(inbounds): gate node selector to multi-node-capable protocols Hide the Deploy-To selector and clear nodeId when switching to a protocol that can't run on a remote node. Also: - subs: return 404 (not 400) when subId matches no inbounds, so VPN clients distinguish "deleted/unknown" from a server error - hysteria link gen: use the inbound's resolved address so node-managed inbounds advertise the node host instead of the central panel - shadowsocks: default network to 'tcp' (udp was causing issues for some clients on first-create) - vite dev proxy: rewrite migrated-route bypass against the live base path instead of a hardcoded single-segment regex Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(clients): bulk add/delete correctness + perf, working pagination, delayed-start in form Bulk add/delete were serial on the frontend (one toast per call, N round-trips) and the backend race exposed by parallelizing them lost client attachments and hit UNIQUE constraint failed on client_inbounds. The single add/edit modal also had no Start-After-First-Use option, and the table never showed the delayed duration. Backend (web/service/client.go): - Per-inbound mutex on Add/Update/Del InboundClient so concurrent writers on the same inbound don't lose the read-modify-write of settings JSON. - SyncInbound skips create+join when the email is tombstoned so a concurrent maintenance pass (adjustTraffics, autoRenewClients, markClientsDisabledIn- Settings) that did a stale RMW can't resurrect a just-deleted client with a fresh id. - compactOrphans sweeps settings.clients entries whose ClientRecord no longer exists, applied in Add/DelInboundClient + DelInboundClientByEmail so each user-initiated mutation self-heals the inbound's settings. - DelInboundClient uses Pluck instead of First for the stats lookup so a missing row doesn't abort the delete with a noisy ErrRecordNotFound log. Frontend: - HttpUtil.{get,post} accept a silent option that suppresses the auto-toast. - ClientBulkAddModal fires creates in parallel + silent + one summary toast. - useClients.removeMany runs deletes in parallel + silent and refreshes once; ClientsPage bulk delete uses it and shows one aggregate toast. - useClients.applyInvalidate debounces 200 ms so the burst of N WebSocket invalidate events from the backend collapses into a single refresh. - ClientsPage pagination is reactive (paginationState ref + tablePagination computed); onTableChange persists page-size and page changes. - ClientFormModal gains a Start-After-First-Use switch + Duration days input alongside the existing Expiry Date picker; on edit-mode open a negative expiryTime is decoded back to delayed mode + days; on submit the payload sends -86400000 * days or the absolute timestamp. - ClientsPage table shows the delayed-start duration (blue tag Nd, tooltip Start After First Use: Nd) instead of infinity. - Telegram ID field in the form is hidden when /panel/setting/defaultSettings reports tgBotEnable=false; Comment then fills the row. - Form row 3 collapses UUID (span 12) + Total GB (span 8) + Limit IP (span 4) when ipLimitEnable is on, else UUID + Total GB at 12/12. - useInbounds.rollupClients counts only clients with a matching clientStats row, so orphans in settings.clients no longer inflate the inbound's count. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(windows): clean shutdown, working panel restart, harden kernel32 load Three Windows-specific issues addressed: 1. Orphaned xray-windows-amd64 after VS Code debugger stop. Delve's "Stop" sends TerminateProcess to the Go binary, which is uncatchable — our signal handlers never run, so xrayService.StopXray() is skipped and xray is left dangling. Spawn xray as a child of a Job Object with JOB_OBJECT_LIMIT_KILL_ON_JOB_CLOSE so the OS kills xray when our handle to the job is closed (which happens even on TerminateProcess). Also trap os.Interrupt in main so Ctrl+C in the terminal runs the graceful path. 2. /panel/setting/restartPanel logged "failed to send SIGHUP signal: not supported by windows" because Windows can't deliver arbitrary signals. Add a restart hook in web/global; main registers it to push SIGHUP into its own signal channel, and RestartPanel calls the hook before falling back to the (Unix-only) signal path. Same restart-loop code runs in both cases. 3. util/sys/sys_windows.go now uses windows.NewLazySystemDLL so the kernel32.dll resolve is pinned to %SystemRoot%\System32 (prevents DLL hijacking by a planted DLL next to the binary). Local filetime type replaced with windows.Filetime, and the unreliable syscall.GetLastError() fallback replaced with a type assertion on the errno captured at call time. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(sys): correct CPU/connection accounting on linux + darwin util/sys/sys_linux.go: - GetTCPCount/GetUDPCount were counting the column header row in /proc/net/{tcp,udp}[6] as a connection, inflating the reported total by 1 per non-empty file (so the panel status line always showed 2 more connections than actually existed). Replace getLinesNum + safeGetLinesNum with a single bufio.Scanner-based countConnections that skips the header. - CPUPercentRaw now opens HostProc("stat") instead of a hardcoded /proc/stat so HOST_PROC overrides apply, matching the connection counters in the same file. - Simplify CPU field unpacking: pad nums to 8 once instead of guarding every assignment with a len check. util/sys/sys_darwin.go: - Fix swapped idle/intr indices on kern.cp_time. BSD CPUSTATES order is user, nice, sys, intr, idle (CP_INTR=3, CP_IDLE=4) — gopsutil's cpu_darwin_nocgo.go reads the same layout. The previous code used out[3] as idle and out[4] as intr, so busy = total - dIdle was actually subtracting interrupt time, making the panel report CPU usage close to 100% on macOS regardless of actual load. - Collapse the per-field delta math into a single loop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(xray): rotate crash reports into log folder, prevent overwrites writeCrashReport had two flaws: it wrote to the bin folder (alongside the xray binary) which conflates artifacts, and the second-precision timestamp meant a tight restart-loop crash burst overwrote prior reports. Write to the log folder with nanosecond precision and keep the last 10 reports. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(inbounds): drop unreleased portfallback protocol The Port-with-Fallback inbound (commit 62fd9f9d) was confusing as a standalone protocol — fallbacks belong on a regular VLESS/Trojan TCP-TLS inbound, the way Xray models them natively. Rip out the entire feature cleanly (no migration needed since it was never released): protocol constant, fallback children DB table, FallbackService, 2 API endpoints, all UI rows, related translations and api-docs. A native fallback flow attached to VLESS/Trojan TCP-TLS/Reality will land in a follow-up commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): native fallbacks on VLESS/Trojan TCP-TLS, with working child links A VLESS or Trojan inbound on TCP with TLS or Reality can now act as a fallback master: pick existing inbounds as children and the panel auto- fills the SNI / ALPN / path / xver routing fields from each child's transport, auto-builds settings.fallbacks at config-gen time, and rewrites the child's client-share link so it advertises the master's reachable endpoint and TLS state instead of the child's loopback listen. Layout matches the Xray All-in-One Nginx example: master at :443 with clients + TLS, each child on 127.0.0.1 with its own transport+clients. Order matters (Xray walks fallbacks top-to-bottom) — reorder via the per-row up/down arrows. Path / SNI / ALPN are exposed under a per-row Edit toggle for the rare cases where the auto-derivation needs overriding; otherwise just pick a child and you're done. Backend: new InboundFallback table + FallbackService (GetByMaster / SetByMaster / GetParentForChild / BuildFallbacksJSON); two routes (GET / POST /panel/api/inbounds/:id/fallbacks); xray.GetXrayConfig injects settings.fallbacks for any VLESS/Trojan TCP-TLS/Reality inbound; GetInbounds annotates each child with FallbackParent so the frontend can rewrite links without an extra round-trip. Link projection covers every emission path — clients-page QR/links, per-inbound Get URL, raw subscription, sub-JSON, sub-Clash, and the inbounds-page link/info/QR — via a shared projectThroughFallbackMaster on the backend and a shared projectChildThroughMaster on the frontend that both handle the panel-tracked relationship and the legacy unix-socket (@vless-ws) convention. Strings translated into all 12 non-English locales. * docs: rewrite CONTRIBUTING with full local-dev setup The prior three-line CONTRIBUTING left newcomers guessing at every non-trivial step: which Go / Node versions, where xray comes from, why the panel goes blank when XUI_DEBUG=true is flipped on, how the Vue multi-page setup is wired, what to do on Windows when go build trips on the CGo SQLite driver. Now covers prerequisites, MinGW-w64 install on Windows (niXman builds or MSYS2), one-shot first-time setup, two frontend dev workflows with the XUI_DEBUG asset-cache gotcha called out, the architecture and conventions of the Vue side, a project-layout map, useful env vars, and the PR checklist. ---------
2026-05-19 03:16:42 -07:00
inbound.Settings = healed
}
}
2023-02-09 11:18:06 -08:00
inboundConfig := inbound.GenXrayInboundConfig()
xrayConfig.InboundConfigs = append(xrayConfig.InboundConfigs, *inboundConfig)
}
feat(xray): update xray-core to v26.10.21 and adapt panel Bump xtls/xray-core to 1b1647cd9ddf (v26.10.21) and move the three binary pins (DockerInit.sh, both release.yml fetches) in lockstep. The v26.10.10 release was pulled upstream, so the old pins already 404. XDRIVE (#7182): the core now runs XDRIVE only for VLESS inside VLESS encryption, with no flow, and an outbound only over Mux; it refuses the whole config otherwise. The panel offers XDRIVE only for VLESS, refuses to save an inbound without decryption, never emits a client flow on an XDRIVE stream (config, JSON subscription, forms), and turns Mux on when an outbound picks XDRIVE. An inbound or template outbound stored before the upgrade that the core would refuse is left out of the generated config with a warning, so it no longer stops every other inbound. The service "Google Drive" is now "google-drive" (old name read as a deprecated alias); saves, the schema and both form adapters store the new name. XDRIVE and MASQUE gain a client xmux; both schemas keep it instead of stripping it on an inbound save. Lua scripts (#7124, #7136): the core refuses a routing script beside rules or domainStrategy, and a dns script beside the fallback/parallel switches or a domain-scoped server. The panel always keeps its api routing rule, so a template save now refuses any routing script and a dns script in a refused combination, and a stored one is dropped from the generated config so the core still starts. Noise masks gain a Lua script that the loader compiles inside Build: the in-process build (template save check, hot add) now resolves a relative script the way the xray child does, beside the xray binary, instead of failing it. The mask editor gets a Script field that replaces the noise list. XDNS (#7140): a resolver takes https:// DoH addresses with worker, tlsConfig and quicConfig. The mask editor registered only addrs, so a save dropped the rest; it now edits all four. Finalmask filter (#7206): a new TCP/UDP mask screens an inbound's source IPs. The inbound mask editor offers it with its allow and block lists; it is a no-op on a dialer and an older client core refuses the type, so the share link and JSON subscription drop it like the Go link already does.
2026-10-22 08:10:11 -07:00
xrayConfig.InboundConfigs = dropInboundsRefusedByXDrive(xrayConfig.InboundConfigs)
feat: add support for subscription-based outbounds with auto-update (#5037) * feat: add support for subscription-based outbounds with auto-update - New OutboundSubscription model (full support on both SQLite and PostgreSQL) - Go subscription link parser (vmess/vless/trojan/ss/hysteria2/wireguard) matching frontend behavior - Stable tag assignment across refreshes (designed for balancer + routing use) - Runtime merge of subscription outbounds into Xray config (additive only) - Full CRUD + manual refresh + preview API - Background auto-update job (per-subscription interval) - Frontend management UI in Outbounds tab (Subscriptions drawer) + tag integration in balancers/routing rules - Proper dual-database support including CLI migration path Review & hardening notes: - Fixed merge logic bug that could drop manual outbounds - Added SSRF/private-IP protection on subscription URLs using SanitizePublicHTTPURL - Improved update interval UX (hours + minutes) - Auto-fetch on first subscription creation - Added detailed comments on tag stability strategy and balancer implications when servers are added/removed/rotated - Updated migrationModels() for CLI migrate-db support * fix: resolve frontend lint/type errors and Go build break Frontend (eslint + tsc clean): - Destructure subscriptionOutboundTags prop in RoutingTab and BalancersTab. It was declared in the interface and used in useMemo but never destructured, so it resolved as an unresolved global (react-hooks warning + tsc "Cannot find name"). The prop is passed by XrayPage, so the feature was silently inert. - OutboundsTab: remove unused useEffect import, add an OutboundSub type to replace any[] state and the any/any table render signature, type the subscriptionOutbounds cast, and replace unused catch (e) bindings with parameter-less catch. Also type HttpUtil.post as OutboundSub so r.obj?.id type-checks. Backend (go build clean): - outbound_subscription_job: websocket.MessageTypeXray is undefined; use the existing MessageTypeOutbounds since the job refreshes outbound subscriptions. * fix(xray): make outbound subscription creation work end-to-end - Correct API paths from /panel/xray/outbound-subs to /panel/api/xray/outbound-subs. The controller is mounted under /panel/api, so the old paths hit the SPA page route (GET-only) and 404'd on POST. - Send the create-subscription body as a plain object instead of URLSearchParams. The axios request interceptor serializes bodies with qs.stringify, which can't read URLSearchParams' internal storage and produced an empty body, so the backend rejected it with "subscription URL is required". - Use message.useMessage() + context holder instead of the static antd message API (resolves the "Static function can not consume context" warning), matching XrayPage's pattern. - Migrate the subscriptions Drawer to antd v6 props: width -> size, destroyOnClose -> destroyOnHidden, and Space direction -> orientation. * feat(xray): show traffic/test for subscription outbounds; harden + test the feature Display (the reported issue): - Replace the flat read-only pills with a proper read-only table (desktop) and cards (mobile) in a new SubscriptionOutbounds component, showing Address, Protocol, Traffic (matched by tag — already collected by Xray), and a Test button with Latency. No edit/delete/move (read-only). - Test subscription outbounds via the existing /testOutbound endpoint, with results keyed by tag (subscriptionTestStates + testSubscriptionOutbound in useXraySetting, wired through XrayPage). Generalize isTesting/testResult to a string|number key so the same helpers serve index- and tag-keyed states. i18n: - Replace all hardcoded English subscription strings with t() calls and add pages.xray.outboundSub.* keys to en-US.json (other locales fall back). Backend hardening + tests: - xray.go: drop the tautological `subSvc != nil` check. - outbound_subscription: re-validate every redirect hop against private/ internal addresses (CheckRedirect) and cap the redirect chain, closing an SSRF gap where only the initial host was checked. - Extract assignStableTags as a pure function and add unit tests for tag stability and SSRF rejection (the feature previously had no tests). Misc: - gofmt util/link/outbound.go (it was not gofmt-clean). * fix(xray): make outbound-subs feature pass CI (test compile, route docs, openapi) - outbound_test.go: remove unused `inner`/`lines` variables that broke the `util/link` test build (declared and not used). - Document the 7 outbound-subscription routes in endpoints.ts (list, create, update, delete, del alias, refresh, parse) so TestAPIRoutesDocumented passes. - Regenerate frontend/public/openapi.json (npm run gen) to include the new endpoints, satisfying the codegen freshness check. * feat(xray): per-subscription allow-private, gap-filled tags, UI tweaks, delete refresh Backend: - Add a per-subscription AllowPrivate flag (default off). Create/Update/refresh and the redirect check sanitize the URL with it, so localhost/LAN sources work only when explicitly opted in; the SSRF guard still blocks private targets by default. Controller reads the allowPrivate form field on create/update/parse. - Default outbound tag prefix now uses the smallest free "subN-" number instead of the auto-increment id, so deleting a subscription frees its number for reuse (a fresh start gives sub1) while staying stable per subscription. Extracted a pure defaultPrefixNumber() with unit tests. - deleteOutboundSub now signals SetToNeedRestart so xray drops the outbounds. Frontend: - "Allow private address" toggle in the add form (sends allowPrivate). - Delete now refreshes the xray view immediately (no manual page reload). - Subscriptions manager opens as a centered Modal instead of a right-side Drawer. - Move Outbounds to a top-level sidebar item under Nodes (out of Xray Configs). - Collapse WARP/NordVPN into a "more" dropdown. - Document the allowPrivate param in endpoints.ts. * i18n(xray): translate outbound-subscription UI into all locales - Translate the pages.xray.outboundSub.* strings (and allowPrivate label/hint) into all 12 non-English locales, matching each file's existing terminology. - Remove the unused outboundSub.add ("Add subscription") key from every locale. * feat(xray): subscription manager — edit, reorder/priority, status, preview, refresh-all Backend: - Per-subscription Priority + Prepend: subscriptions are ordered by Priority and placed before (Prepend) or after the manual template outbounds in the merge, so a subscription server can become the default. New Move(up/down) endpoint re-normalizes priorities; merge split into prepend/template/append. - List now returns a derived OutboundCount and orders by priority, and strips the heavy LastFetchedOutbounds/LinkIdentities blobs from the list payload. - Create/Update accept the prepend flag; new subs append at the end of priority. Frontend (Outbound Subscriptions modal): - Edit existing subscriptions (reuses the form + Update endpoint). - Inline enable/disable Switch, Status column (OK / error tooltip), Outbounds count column, per-row refresh spinner, "Refresh all" button. - Reorder (move up/down) controls + a "Before manual outbounds" toggle. - Preview button: fetch+parse a URL via /parse without saving. - Document the move route + prepend param in endpoints.ts; regenerate openapi.json. * i18n(xray): translate new subscription-manager strings into all locales Add the prepend/prependHint, preview/previewEmpty, refreshAll, statusOk and toastUpdated keys to all 12 non-English locales, matching each file's terminology. --------- Co-authored-by: MHSanaei <ho3ein.sanaei@gmail.com>
2026-06-08 09:09:53 -07:00
// Merge subscription-derived outbounds (if any) into the final outbounds array.
// These are additive: each subscription is placed before or after the template
// outbounds based on its Prepend flag, ordered by Priority. Tags assigned by the
// subscription service are kept stable across refreshes so that balancers and
// routing rules continue to work.
subSvc := &OutboundSubscriptionService{}
if prepend, appendList, err := subSvc.activeOutboundsSplit(); err == nil && (len(prepend) > 0 || len(appendList) > 0) {
mergeSubscriptionOutbounds(xrayConfig, prepend, appendList)
}
// Route opted-in local mtproto inbounds through the core's router. Each one
// gets a loopback SOCKS bridge — tagged with the inbound's own tag so it is
// matchable in routing rules — that its mtg sidecar dials Telegram through.
// Done after the subscription merge so a selected subscription outbound (or
// balancer) is a valid rule target.
for i := range inbounds {
inbound := inbounds[i]
if inbound.Protocol != model.MTProto || !inbound.Enable || inbound.NodeID != nil {
continue
}
injectMtprotoEgress(xrayConfig, inbound)
}
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
// Every AmneziaWG inbound is embedded (internal/amneziawgnet: amneziawg-go
// over a gVisor netstack, no kernel module) and relays every peer's
// decapsulated traffic into its own loopback SOCKS5 inbound, always on —
// unlike mtproto's bridge above, there's no opt-in gate here: once
// traffic is decapsulated in gVisor, Xray's own freedom outbound is the
// only way it reaches the real internet at all, not an optional extra
// hop. Whether it goes anywhere beyond Xray's default routing is up to
// whatever rules the admin adds through the stock Routing page, exactly
// like routing any other protocol.
injectAmneziawgnetSocks(xrayConfig, inbounds)
tuicInstances := configuredTuicInstances(inbounds)
if err := injectTuicRelay(xrayConfig, tuicInstances); err != nil {
return nil, err
}
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
// Restores each opted-in peer's own distinct public IPv6 source identity
// for its outbound connections — a peer that has an IPv6 address in its
// AllowedIPs, on an inbound with IPv6Enabled, gets its own freedom
// outbound bound to that exact address via sendThrough.
// internal/amneziawgnet's own Manager is responsible for actually
// aliasing that address onto the host (see v6alias.go) so the kernel
// lets Xray bind an egress socket to it at all; this call only builds
// the Xray-side outbound/routing-rule half.
injectAmneziawgV6Egress(xrayConfig, inbounds)
// Wire the panel's own HTTP traffic through the configured outbound, after
// the subscription merge so subscription outbound tags are valid targets.
if egressTag, err := s.settingService.GetPanelOutbound(); err != nil {
logger.Warning("read panelOutbound setting failed:", err)
} else if egressTag != "" {
injectPanelEgress(xrayConfig, egressTag)
}
nodes, err := s.nodeService.GetAll()
if err != nil {
logger.Warning("read nodes for egress injection failed:", err)
} else {
injectNodeEgresses(xrayConfig, nodes)
}
feat(xray): update xray-core to v26.10.21 and adapt panel Bump xtls/xray-core to 1b1647cd9ddf (v26.10.21) and move the three binary pins (DockerInit.sh, both release.yml fetches) in lockstep. The v26.10.10 release was pulled upstream, so the old pins already 404. XDRIVE (#7182): the core now runs XDRIVE only for VLESS inside VLESS encryption, with no flow, and an outbound only over Mux; it refuses the whole config otherwise. The panel offers XDRIVE only for VLESS, refuses to save an inbound without decryption, never emits a client flow on an XDRIVE stream (config, JSON subscription, forms), and turns Mux on when an outbound picks XDRIVE. An inbound or template outbound stored before the upgrade that the core would refuse is left out of the generated config with a warning, so it no longer stops every other inbound. The service "Google Drive" is now "google-drive" (old name read as a deprecated alias); saves, the schema and both form adapters store the new name. XDRIVE and MASQUE gain a client xmux; both schemas keep it instead of stripping it on an inbound save. Lua scripts (#7124, #7136): the core refuses a routing script beside rules or domainStrategy, and a dns script beside the fallback/parallel switches or a domain-scoped server. The panel always keeps its api routing rule, so a template save now refuses any routing script and a dns script in a refused combination, and a stored one is dropped from the generated config so the core still starts. Noise masks gain a Lua script that the loader compiles inside Build: the in-process build (template save check, hot add) now resolves a relative script the way the xray child does, beside the xray binary, instead of failing it. The mask editor gets a Script field that replaces the noise list. XDNS (#7140): a resolver takes https:// DoH addresses with worker, tlsConfig and quicConfig. The mask editor registered only addrs, so a save dropped the rest; it now edits all four. Finalmask filter (#7206): a new TCP/UDP mask screens an inbound's source IPs. The inbound mask editor offers it with its allow and block lists; it is a no-op on a dialer and an older client core refuses the type, so the share link and JSON subscription drop it like the Go link already does.
2026-10-22 08:10:11 -07:00
xrayConfig.RouterConfig = dropRefusedScript(xrayConfig.RouterConfig, "routing", routingScriptRefused)
xrayConfig.DNSConfig = dropRefusedScript(xrayConfig.DNSConfig, "dns", dnsScriptRefused)
if excludedClients > 0 {
logger.Infof("Xray config generated: excluded %d disabled client entries across %d inbounds", excludedClients, excludedInbounds)
}
2023-02-09 11:18:06 -08:00
return xrayConfig, nil
}
// PanelEgressInboundTag is the tag of the loopback SOCKS inbound injected into
// the generated config when a panel outbound is configured. The panel's own
// HTTP clients dial through it to egress via the chosen outbound.
const PanelEgressInboundTag = "panel-egress"
// panelEgressBasePort is the first port tried for the egress bridge; ports
// already taken by other inbounds in the generated config are skipped.
const panelEgressBasePort = 62790
// injectPanelEgress appends a loopback SOCKS inbound and routing rule only when
// outboundTag resolves in the final outbound or balancer set. Otherwise the
// entire injection is skipped. Generated state is hot-appliable and never
// modifies the stored template or restarts the core.
func injectPanelEgress(cfg *xray.Config, outboundTag string) {
for i := range cfg.InboundConfigs {
if cfg.InboundConfigs[i].Tag == PanelEgressInboundTag {
logger.Warning("panel egress: inbound tag [", PanelEgressInboundTag, "] already exists, skipping injection")
return
}
}
// The rule must exist before the inbound takes traffic, otherwise the
// bridge would silently egress through the default outbound instead.
routing := map[string]any{}
if len(cfg.RouterConfig) > 0 {
if err := json.Unmarshal(cfg.RouterConfig, &routing); err != nil {
logger.Warning("panel egress: routing section is unparsable, skipping injection:", err)
return
}
}
if !routingTargetExists(routing, cfg.OutboundConfigs, outboundTag) {
logger.Warning("panel egress: target tag [", outboundTag, "] not found, skipping injection")
return
}
rules, _ := routing["rules"].([]any)
rule := map[string]any{
"type": "field",
"inboundTag": []any{PanelEgressInboundTag},
}
// The configured tag may name a routing balancer instead of a concrete
// outbound. A field rule can target either, so emit the matching key —
// balancerTag load-balances the panel's own traffic across the balancer's
// outbounds, while a plain outbound tag keeps the original behavior.
if routingTagIsBalancer(routing, outboundTag) {
rule["balancerTag"] = outboundTag
} else {
rule["outboundTag"] = outboundTag
}
routing["rules"] = append([]any{rule}, rules...)
newRouting, err := json.Marshal(routing)
if err != nil {
logger.Warning("panel egress: failed to rebuild routing section, skipping injection:", err)
return
}
cfg.RouterConfig = json_util.RawMessage(newRouting)
used := make(map[int]struct{}, len(cfg.InboundConfigs))
for i := range cfg.InboundConfigs {
used[cfg.InboundConfigs[i].Port] = struct{}{}
}
port := panelEgressBasePort
for {
if _, taken := used[port]; !taken {
break
}
port++
}
cfg.InboundConfigs = append(cfg.InboundConfigs, xray.InboundConfig{
Listen: json_util.RawMessage(`"127.0.0.1"`),
Port: port,
Protocol: "socks",
Settings: json_util.RawMessage(`{"auth":"noauth","udp":false}`),
Tag: PanelEgressInboundTag,
})
}
func outboundTagExists(outbounds json_util.RawMessage, tag string) bool {
var parsed []struct {
Tag string `json:"tag"`
}
if tag == "" || json.Unmarshal(outbounds, &parsed) != nil {
return false
}
for _, outbound := range parsed {
if outbound.Tag == tag {
return true
}
}
return false
}
func routingTargetExists(routing map[string]any, outbounds json_util.RawMessage, tag string) bool {
return routingTagIsBalancer(routing, tag) || outboundTagExists(outbounds, tag)
}
// NodeEgressInboundTag returns the loopback SOCKS inbound tag for a given node.
func NodeEgressInboundTag(nodeID int) string {
return fmt.Sprintf("node-egress-%d", nodeID)
}
// nodeEgressBasePort is the first port tried for node egress bridges.
const nodeEgressBasePort = 62800
// injectNodeEgresses appends a loopback SOCKS inbound per enabled node that has
// an OutboundTag, and prepends a routing rule sending that inbound's traffic to
// the selected outbound tag. These bridges are hot-appliable.
func injectNodeEgresses(cfg *xray.Config, nodes []*model.Node) {
routing := map[string]any{}
if len(cfg.RouterConfig) > 0 {
if err := json.Unmarshal(cfg.RouterConfig, &routing); err != nil {
logger.Warning("node egress: routing section is unparsable, skipping injection:", err)
return
}
}
used := make(map[int]struct{}, len(cfg.InboundConfigs))
usedTags := make(map[string]struct{}, len(cfg.InboundConfigs))
for i := range cfg.InboundConfigs {
used[cfg.InboundConfigs[i].Port] = struct{}{}
usedTags[cfg.InboundConfigs[i].Tag] = struct{}{}
}
rules, _ := routing["rules"].([]any)
newRules := make([]any, 0)
for _, n := range nodes {
if !n.Enable || n.OutboundTag == "" {
continue
}
if !routingTargetExists(routing, cfg.OutboundConfigs, n.OutboundTag) {
logger.Warning("node egress: target tag [", n.OutboundTag, "] not found, skipping node [", n.Id, "]")
continue
}
tag := NodeEgressInboundTag(n.Id)
if _, exists := usedTags[tag]; exists {
logger.Warning("node egress: inbound tag [", tag, "] already exists, skipping")
continue
}
usedTags[tag] = struct{}{}
rule := map[string]any{
"type": "field",
"inboundTag": []any{tag},
}
if routingTagIsBalancer(routing, n.OutboundTag) {
rule["balancerTag"] = n.OutboundTag
} else {
rule["outboundTag"] = n.OutboundTag
}
newRules = append(newRules, rule)
port := nodeEgressBasePort + n.Id
for {
if _, taken := used[port]; !taken {
break
}
port++
}
used[port] = struct{}{}
cfg.InboundConfigs = append(cfg.InboundConfigs, xray.InboundConfig{
Listen: json_util.RawMessage(`"127.0.0.1"`),
Port: port,
Protocol: "socks",
Settings: json_util.RawMessage(`{"auth":"noauth","udp":false}`),
Tag: tag,
})
}
if len(newRules) == 0 {
return
}
routing["rules"] = append(newRules, rules...)
newRouting, err := json.Marshal(routing)
if err != nil {
logger.Warning("node egress: failed to rebuild routing section, skipping injection:", err)
return
}
cfg.RouterConfig = json_util.RawMessage(newRouting)
}
// routingTagIsBalancer reports whether tag names a balancer in the parsed
// routing section. The panel-egress rule targets a balancer via balancerTag and
// a concrete outbound via outboundTag, so the caller picks the key from this.
func routingTagIsBalancer(routing map[string]any, tag string) bool {
if tag == "" {
return false
}
balancers, ok := routing["balancers"].([]any)
if !ok {
return false
}
for _, b := range balancers {
bm, ok := b.(map[string]any)
if !ok {
continue
}
if t, ok := bm["tag"].(string); ok && t == tag {
return true
}
}
return false
}
// mtprotoEgressSocksSettings is the loopback SOCKS server a routed mtproto
// inbound exposes for its mtg sidecar to dial Telegram through. mtg makes plain
// TCP connections, so UDP is left off (matching the panel egress bridge).
const mtprotoEgressSocksSettings = `{"auth":"noauth","udp":false}`
// injectMtprotoEgress wires one routed mtproto inbound into the generated
// config after any selected outbound resolves in the final target set. Invalid
// selected targets or routing data skip the entire injection; without a selected
// outbound, the bridge retains default-route behavior. Generated state remains
// hot-appliable, leaves the stored template untouched, and never forces a full
// Xray restart. Mirrors injectPanelEgress.
func injectMtprotoEgress(cfg *xray.Config, inbound *model.Inbound) {
var parsed struct {
RouteThroughXray bool `json:"routeThroughXray"`
RouteXrayPort int `json:"routeXrayPort"`
OutboundTag string `json:"outboundTag"`
}
if err := json.Unmarshal([]byte(inbound.Settings), &parsed); err != nil {
return
}
if !parsed.RouteThroughXray || parsed.RouteXrayPort <= 0 || inbound.Tag == "" {
return
}
tag := inbound.Tag
for i := range cfg.InboundConfigs {
if cfg.InboundConfigs[i].Tag == tag {
logger.Warning("mtproto egress: inbound tag [", tag, "] already present in generated config, skipping bridge")
return
}
}
if parsed.OutboundTag != "" {
routing := map[string]any{}
if len(cfg.RouterConfig) > 0 {
if err := json.Unmarshal(cfg.RouterConfig, &routing); err != nil {
logger.Warning("mtproto egress: routing section is unparsable, skipping injection:", err)
return
}
}
if !routingTargetExists(routing, cfg.OutboundConfigs, parsed.OutboundTag) {
logger.Warning("mtproto egress: target tag [", parsed.OutboundTag, "] not found, skipping injection")
return
}
rules, _ := routing["rules"].([]any)
rule := map[string]any{
"type": "field",
"inboundTag": []any{tag},
}
if routingTagIsBalancer(routing, parsed.OutboundTag) {
rule["balancerTag"] = parsed.OutboundTag
} else {
rule["outboundTag"] = parsed.OutboundTag
}
routing["rules"] = append([]any{rule}, rules...)
newRouting, err := json.Marshal(routing)
if err != nil {
logger.Warning("mtproto egress: failed to rebuild routing section, skipping injection:", err)
return
}
cfg.RouterConfig = json_util.RawMessage(newRouting)
}
cfg.InboundConfigs = append(cfg.InboundConfigs, xray.InboundConfig{
Listen: json_util.RawMessage(`"127.0.0.1"`),
Port: parsed.RouteXrayPort,
Protocol: "socks",
Settings: json_util.RawMessage(mtprotoEgressSocksSettings),
Tag: tag,
})
}
// Peers resolve DNS inside the tunnel, so domain rules match only via sniffing; routeOnly
// keeps the dial on the peer's IP, else Telegram's FakeTLS (IP + foreign SNI) breaks.
const amneziawgEgressSniffingSettings = `{"enabled":true,"destOverride":["http","tls","quic","fakedns"],"routeOnly":true}`
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
// injectAmneziawgnetSocks gives every enabled AmneziaWG inbound with at
// least one qualifying peer its own loopback SOCKS5 inbound for the
// embedded (amneziawg-go) relay path (internal/amneziawgnet) -- always on,
// since there is no alternative datapath once traffic is decapsulated in
// gVisor: Xray's own freedom outbound is how it reaches the real internet at
// all (see internal/amneziawgnet/relay.go's doc comment, Finding 3 of the
// migration plan). Tagged with the inbound's own real tag: it's already
// selectable in the panel's stock Routing page (InboundService.GetInboundTags
// is protocol-blind), and per-inbound traffic totals
// (internal/web/service/inbound_traffic.go's addClientTraffic) match by
// exact tag -- reusing it isn't a style choice.
func injectAmneziawgnetSocks(cfg *xray.Config, inbounds []*model.Inbound) {
existingTags := make(map[string]struct{}, len(cfg.InboundConfigs))
for i := range cfg.InboundConfigs {
existingTags[cfg.InboundConfigs[i].Tag] = struct{}{}
}
for _, inbound := range inbounds {
if inbound.Protocol != model.AmneziaWG || !inbound.Enable || inbound.NodeID != nil {
continue
}
inst, ok := amneziawg.InstanceFromInbound(inbound)
if !ok {
continue
}
feat(amneziawg): add opt-in Linux kernel datapath (#6877) * feat(amneziawg): add opt-in Linux kernel datapath Add the kernel backend as an explicit per-inbound option while keeping the embedded backend as the default. Include host preflight, UI/API support, the bundled installer CLI, lifecycle and accounting integration, regression tests, and upstream-facing documentation. Kernel mode is NAT-only and rejects unsupported routing, IPv6, forwarding, and pinned-listen configurations. Local saves check the local host; remote nodes check their own host during apply. Preserve adopted interface accounting across panel restarts, retain manager state after failed teardown so the job can retry, prevent embedded startup while a kernel interface remains live, and prune collector state for removed peers. Tested the restart accounting path on Debian 13 with XanMod 7.1.11 on x86_64; Docker and cross-kernel coverage are not claimed. * fix(amneziawg): align readiness and endpoint reporting Expire endpoint sightings using the latest handshake and sweep stale addresses for the polled interface. Keep the readiness API schema, generated client types, and ready-host empty blocker list consistent; watch the nested form field and defer local readiness checks for remote targets. Document safe rollback before using a build without kernel support. Regression evidence: the idle endpoint test failed before the collector change because the old address remained visible after 31 minutes; endpoint tests pass after the fix. The form component suite passes 3 tests; restoring the invalid array watch made its off-mode severity test fail, while the corrected dot-path watch passes. Targeted Go regressions across amneziawgkern, job, and service pass. * style(amneziawg): format readiness test * test: reproduce kernel AWG client validation gaps * test(service): cover kernel AWG shared client fanout * fix(amneziawg): validate kernel client fanout and config values * test(amneziawgkern): cover owned interface lifecycle recovery * fix(amneziawgkern): recover owned interfaces on restart * test(amneziawgkern): retain visibility after recovered cleanup failure * fix(amneziawgkern): report recovered interfaces while cleanup retries * style(amneziawgkern): satisfy static analysis * test(amneziawgkern): cover long rendered peer comments * fix(amneziawgkern): accept long generated config comments * test(amneziawgkern): refuse foreign occupied interfaces Capture the current Ensure ordering failure with real temporary config paths and manager side-effect boundaries. * test(amneziawgkern): cover stale interface collisions Extend the Ensure regression across absent links, mismatched live interfaces, cached manager state, and Remove teardown. * fix(amneziawgkern): verify interface ownership before mutation Refuse to replace or tear down links unless the canonical config is a valid panel-owned file and its key matches the live interface. Preserve normal adoption and key rotation, and document migration of legacy unmarked configs. Validation: go test ./internal/amneziawgkern -count=1; go test -race ./internal/amneziawgkern -count=1; golangci-lint run ./internal/amneziawgkern/...; git diff --check. * build: update Go toolchain to 1.27.3 * test(amneziawg): reproduce unsafe key rotation recovery Keep the previous owned config authoritative until teardown succeeds, and require fresh managers to rebuild structural key changes. These regressions expose the stuck live-key/disk-key mismatch after teardown failure. * fix(amneziawg): recover interrupted key rotations Keep the old owned config on disk until a structural rebuild has confirmed the previous interface is down. Fresh managers only adopt when the live config and desired structural identity match, so retries and removals retain the key needed to prove ownership. * fix(scanner): align Go image with module version * fix(docs): pin pnpm release with bootstrap fix pnpm 12.9.1 restores a Node launcher before its postinstall replaces the target with ELF; pnpm 11 then runs that binary as JavaScript. Pin 12.10.0, which includes the shim fix, so the docs CI bootstrap can switch versions safely. * fix(xray): build core from pinned module The configured Xray release asset tag is unpublished, and the latest published binary lacks GetUsersStats used by the panel. Build the packaged core from the exact Xray module version already selected in go.mod for Docker and release targets. * fix(tests): stabilize local network fixtures Keep the TUIC stream fragmentation check on a working UDP loopback path, allocate the AmneziaWG fixture ports dynamically, and wait for a real Xray stats response before starting API mutation assertions. Recreate failed readiness channels so an early connection cannot stall the probe. Verified host and offline parallel Go suites, repeated occupied-port and UDP tests, race/shuffle runs, and golangci-lint. Production behavior and dependency versions are unchanged. * fix(i18n): add the kernel-mode AmneziaWG keys to sk-SK main gained the sk-SK locale after this branch added its kernel-mode keys to the other 13, so i18n-dead-keys.test.ts failed on the merged tree. * refactor(amneziawgkern): apply kernel interfaces with direct commands, no awg-quick awg-quick evaluates PostUp/PostDown through a root shell, so every value the panel rendered into its config was one validation slip away from root command execution. The interface is now built with direct ip/awg/nft execs from a setconf-only file: link add, awg setconf, address, mtu, an atomically replaced nftables table per inbound, and the IPv4 forwarding sysctl written directly. - Panel-owned names and paths: xui-awg<id> interfaces, xui_awg<id> NAT tables and configs under the panel's data dir, so nothing collides with an operator's own awg0/awg1 awg-quick tunnels in /etc/amnezia/amneziawg. - FORWARD accepts (comment 3x-ui:xui-awg<id>) when iptables is present: Docker and ufw set the FORWARD policy to DROP, which silently black-holed the tunnel. - A full panel stop records the final counters and tears the interfaces down, like the embedded datapath; a stopped panel enforces no quota or expiry. Crash survivors are still adopted without a rebuild. - The job reads counters before reconciling, and from every managed interface, so a client disabled for quota, or a removed inbound, is charged for its last interval instead of losing it with the dropped peer. - An unchanged reconcile tick forks no process (it probed the live key with `awg show` every 10s); readiness no longer loads the module from a GET. Verified on the real module (WSL2 6.6 kernel, v3.1.20260906): create, peer and egress changes in place, SIGKILL adoption with the same ifindex, SIGTERM teardown leaving no link, table, FORWARD rule or config behind. * fix(amneziawg): gate kernel host readiness only when a save starts the datapath Every save of a kernel-mode inbound re-ran the host preflight, so a host that lost its module (a kernel upgrade without a DKMS rebuild) refused even disabling the inbound or switching it back to embedded mode. Host blockers now apply only when the save starts the kernel datapath: a new inbound, a switch from embedded, or enabling a disabled one. Feature blockers (IPv6, forwarded ports, pinned listen) still apply to every save. * feat(nodes): refuse kernel-mode AmneziaWG pushes to nodes that predate it A node built before kernel mode drops the unknown kernelMode field and serves the inbound on the embedded datapath while the master's form shows kernel mode on. Nodes now advertise awg-kernel in X-3x-Node-Caps, and Remote refuses to push a kernel-mode inbound to a node that has not advertised it, learning the caps with an inbound list first when this Remote has not heard from it. * fix(amneziawg): pin the kernel module installer to verified commits The installer ran as root and built whatever upstream tag sorted newest at install time. It now clones fixed tags (module v3.1.20260906, tools v3.1.20260812) and refuses unless each resolves to its recorded commit. The udp_tunnel perl patch is gone: those releases carry the kernel 7.1.5 compat upstream, so the patch never matched. * fix(i18n): translate the kernel-mode AmneziaWG strings All thirteen non-English locales carried the English text verbatim. * fix(amneziawg): skip the kernel traffic flush when no kernel interface runs Shutdown listed the desired AmneziaWG inbounds from the database even on a panel with no kernel interface, an unneeded query on every stop. A server stopped before its database was opened panicked on it, which turned TestServerStopWhenXrayNotRunningSucceeds red whenever it ran first. --------- Co-authored-by: MHSanaei <ho3ein.sanaei@gmail.com>
2026-10-25 05:24:50 -07:00
// Kernel-mode inbounds forward in the kernel and never dial Xray, so a
// relay inbound for them would listen forever with nothing behind it.
if inst.KernelMode {
continue
}
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
if _, taken := existingTags[inbound.Tag]; taken {
logger.Warning("amneziawgnet socks: inbound tag [", inbound.Tag, "] already present in generated config, skipping its relay inbound")
continue
}
emails := make([]string, 0, len(inst.Peers))
for _, p := range inst.Peers {
if p.Email != "" {
emails = append(emails, p.Email)
}
}
if len(emails) == 0 {
continue
}
settings, err := amneziawgnet.SocksInboundSettings(emails, amneziawgnet.SocksPassword())
if err != nil {
logger.Warning("amneziawgnet socks: building settings for inbound [", inbound.Tag, "]: ", err)
continue
}
existingTags[inbound.Tag] = struct{}{}
cfg.InboundConfigs = append(cfg.InboundConfigs, xray.InboundConfig{
Listen: json_util.RawMessage(`"127.0.0.1"`),
Port: amneziawgnet.SOCKSPortForInbound(inbound.Id),
Protocol: "socks",
Settings: json_util.RawMessage(settings),
Sniffing: json_util.RawMessage(amneziawgEgressSniffingSettings),
Tag: inbound.Tag,
})
}
}
// Keep sniffed domains available for routing without replacing the requested destination.
const tuicEgressSniffingSettings = `{"enabled":true,"destOverride":["http","tls","quic","fakedns"],"routeOnly":true}`
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
func injectTuicRelay(cfg *xray.Config, instances []tuic.Instance) error {
existingTags := map[string]struct{}{"api": {}}
var api struct {
Tag string `json:"tag"`
}
if len(cfg.API) > 0 {
if err := json.Unmarshal(cfg.API, &api); err != nil {
return err
}
if api.Tag != "" {
existingTags[api.Tag] = struct{}{}
}
}
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
for i := range cfg.InboundConfigs {
existingTags[cfg.InboundConfigs[i].Tag] = struct{}{}
}
var userLevel uint32
policyReady := false
for _, inst := range instances {
if _, taken := existingTags[inst.Tag]; taken {
logger.Warningf("tuic: skipping inbound %d: relay tag conflicts with another inbound", inst.Id)
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
continue
}
emails := make([]string, 0, len(inst.Clients))
for _, client := range inst.Clients {
if client.Email != "" {
emails = append(emails, client.Email)
}
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
}
if !policyReady {
var err error
userLevel, err = tuicRelayPolicyLevel(cfg)
if err != nil {
return fmt.Errorf("tuic relay: policy: %w", err)
}
policyReady = true
}
slices.Sort(emails)
emails = slices.Compact(emails)
settings, err := tuic.SocksInboundSettings(emails, tuic.SocksPassword(), userLevel)
if err != nil {
return fmt.Errorf("tuic relay: inbound %d: %w", inst.Id, err)
}
existingTags[inst.Tag] = struct{}{}
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
cfg.InboundConfigs = append(cfg.InboundConfigs, xray.InboundConfig{
Listen: json_util.RawMessage(`"127.0.0.1"`),
Port: tuic.SOCKSPortForInbound(inst.Id),
Protocol: "socks",
TuicRelay: true,
TuicInboundID: inst.Id,
Settings: json_util.RawMessage(settings),
Sniffing: json_util.RawMessage(tuicEgressSniffingSettings),
Tag: inst.Tag,
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
})
}
return nil
}
// Relay authentication carries routing identity; native TUIC owns client accounting and source IPs.
// Copy session limits from level 0, but disable Xray's duplicate traffic and loopback online stats.
func tuicRelayPolicyLevel(cfg *xray.Config) (uint32, error) {
var policy map[string]json.RawMessage
if len(cfg.Policy) > 0 {
if err := json.Unmarshal(cfg.Policy, &policy); err != nil {
return 0, err
}
}
if policy == nil {
policy = make(map[string]json.RawMessage)
}
var levels map[uint32]json.RawMessage
if raw, ok := policy["levels"]; ok {
if err := json.Unmarshal(raw, &levels); err != nil {
return 0, err
}
}
if levels == nil {
levels = make(map[uint32]json.RawMessage)
}
var relayPolicy map[string]json.RawMessage
if raw, ok := levels[0]; ok {
if err := json.Unmarshal(raw, &relayPolicy); err != nil {
return 0, err
}
}
if relayPolicy == nil {
relayPolicy = make(map[string]json.RawMessage)
}
for _, key := range []string{"statsUserUplink", "statsUserDownlink", "statsUserOnline"} {
relayPolicy[key] = json.RawMessage("false")
}
level := ^uint32(0)
for {
if _, exists := levels[level]; !exists {
break
}
if level == 0 {
return 0, errors.New("no free user policy level")
}
level--
}
raw, err := json.Marshal(relayPolicy)
if err != nil {
return 0, err
}
levels[level] = raw
policy["levels"], err = json.Marshal(levels)
if err != nil {
return 0, err
}
cfg.Policy, err = json.Marshal(policy)
return level, err
feat(tuic): implement native in-process Go TUIC v5 server (#6577) * feat(tuic): implement native in-process Go TUIC v5 server - Implement native TUIC v5 protocol server on pure Go using quic-go - Bridge decrypted TCP/UDP traffic into Xray-core via loopback SOCKS5 inbound - Support full Xray routing rules (geosite/geoip) and cascading outbounds - Implement atomic per-client traffic accounting with TotalGB and ExpiryTime - Add automatic legacy cleanup for older Rust tuic-server binaries, configs, and orphaned processes - Eliminate external Rust tuic-server downloads from install/CI scripts * fix(tuic): address traffic accounting, client reload, and socket lifecycle issues * fix(service): update checkTuicSocksReverseConflict to use bindAddr for listenOverlaps * fix(tuic): resolve traffic double-accounting, UDP fragmentation, and socket lifecycle issues * feat(tuic): complete native Go integration and address audit findings - Integrate an isolated QUIC fork pinned to a specific commit - Preserve original QUIC dependencies for Xray, Hysteria and Gin - Apply BBR, CUBIC and Reno to server connections and exported client profiles - Bridge Xray BBR with correct monotonic time and congestion type conversions - Handle congestion sender recreation after PMTU changes - Update congestion control for new connections without restarting the listener - Preserve existing connections and their selected congestion controller - Apply per-inbound log levels through the shared panel logger - Add lifecycle, authentication and TCP/UDP relay events without exposing secrets - Rate-limit repeated authentication and relay warnings - Support native and QUIC UDP relay modes on the same listener - Recover UDP associations after relay worker failures - Fix TCP relay cancellation, idle shutdown and half-close handling - Close active sessions when client credentials are revoked or disabled - Track traffic by immutable client statistics IDs across email and UUID changes - Prevent ambiguous accounting and duplicate UUIDs within TUIC inbounds - Persist pending traffic in a durable shutdown journal - Replay journal batches transactionally without duplicate accounting - Report server shutdown failures through the shared logger - Preserve legacy flat and nested TUIC settings compatibility - Normalize congestion controller values consistently across backend and frontend - Preserve controller, UDP mode and SNI in client links and subscriptions - Separate client profile options from server settings in the TUIC form - Keep certificate path autofill explicit when changing client SNI - Align UDP packet size validation with protocol limits - Simplify and localize TUIC field hints and certificate autofill messages - Add controller, TCP/UDP, logging and live settings update tests - Add accounting identity, journal replay and shutdown regression tests - Add relay recovery, session revocation and legacy frontend form tests * fix(service): alias the TUIC duplicate-UUID subquery for PostgreSQL < 16 syncInboundClients runs a COUNT(*) FROM (subquery) for every client sync, whatever the protocol. PostgreSQL before 16 rejects a FROM subquery with no alias, so on the distro PostgreSQL install.sh provisions (14 on Ubuntu 22.04, 15 on Debian 12) every client add or edit failed with SQLSTATE 42601. Reproduced against postgres:15 with the new env-gated test. * fix(database): create tuic_traffic_receipts through the model migration AddTuicTrafficBatch issued CREATE TABLE IF NOT EXISTS at runtime, a schema change outside db.go. The table was invisible to allModels and migrationModels, so x-ui migrate-db dropped the receipts and a retained journal could be counted twice after a SQLite to PostgreSQL move. It is now a GORM model in both lists, and the insert uses OnConflict DoNothing. * chore(tuic): skip the ICMP-dependent relay test on Windows, drop dead collectors Go disables SIO_UDP_CONNRESET on Windows, so a dead UDP bridge never fails a read there and TestAudit3UDPAssociationMustRecoverAfterBridgeReadFailure was red on every Windows run. Server.CollectTotalTraffic and Manager.CollectTraffic had no caller. * refactor(tuic): serve TUIC on apernet/quic-go instead of a personal fork The native server depended on github.com/poise52/quic-go, a personal fork of apernet/quic-go patched only to pick the congestion controller before the handshake. That put a second QUIC/TLS stack in the binary that no upstream security fix reaches. apernet/quic-go is already in the graph through xray-core and exposes SetCongestionControl, so BBR is now installed on each accepted connection with Xray's own congestion.UseBBR; the cross-module BBR adapter is gone. apernet ships New Reno as its only built-in sender, so a cubic setting is served as new_reno server-side (clients still get cubic in their profile). The test inspectors now read the sender under congestionMutex, which the post-handshake install writes under. Linux loopback, single stream through Xray: 2428 -> 3383 Mbit/s (bbr). --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com>
2026-10-02 17:56:02 -07:00
}
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
// amneziawgV6EgressTag returns the stable, globally-unique freedom outbound
// tag for one peer's IPv6 source-identity egress. Stable across config
// regenerations (a pure function of two stable identifiers), so
// internal/xray/hot_diff.go's tag-keyed outbound/routing diffing recognizes
// "unchanged" rather than remove+recreate on every poll. The inbound.Id
// prefix is defense in depth, not load-bearing on its own: email is already
// enforced globally unique across the whole panel's client table
// (model.ClientRecord.Email has a gorm uniqueIndex) — kept anyway since it
// costs nothing and makes the tag self-describing, matching
// NodeEgressInboundTag's own style.
func amneziawgV6EgressTag(inboundID int, email string) string {
return fmt.Sprintf("amneziawg-v6-%d-%s", inboundID, email)
}
// amneziawgV6EgressSettings carries the template's private-range egress block:
// the per-peer rule is matched ahead of the routing-level geoip:private rule.
func amneziawgV6EgressSettings() map[string]any {
return map[string]any{
"finalRules": []any{
map[string]any{"action": "block", "ip": []any{"geoip:private"}},
map[string]any{"action": "allow"},
},
}
}
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
// injectAmneziawgV6Egress gives every enabled, non-node-hosted AmneziaWG
// peer with an IPv6 AllowedIPs entry its own single-purpose freedom
// outbound, bound via sendThrough to that exact address, plus a routing
// rule sending only that peer's IPv6-destined traffic through it — restoring the
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
// per-client public IPv6 identity the hard cutover temporarily dropped
// (Phase 3.5 of the migration plan). Scoped to outbound source identity
// only: it depends on internal/amneziawgnet's own alias mechanism actually
// giving the host that address at the OS level (see v6alias.go's
// V6AliasesActive, the exact same gate this function uses below) — without
// that, sendThrough fails to bind and every connection through it errors
// outright (freedom.go's dial failure); there is no fallback outbound.
//
// The routing rule matches both inboundTag and user: SocksInboundSettings
// (used by injectAmneziawgnetSocks above) already authenticates each
// connection as the peer's own email via stock SOCKS5 auth, and a stock
// Xray SOCKS5 inbound sets that connection's stats/routing identity from
// the authenticated username — so "user" reliably isolates exactly one
// peer's traffic, the same building block Finding 3 of the migration plan
// already established for per-client stats.
//
// Modeled on injectNodeEgresses (the established N-per-slice inbound+rule
// precedent, not injectAmneziawgnetSocks itself, which only ever emits a
// single inbound and never touches outbounds/routing) and
// mergeSubscriptionOutbounds's unmarshal-append-remarshal pattern for
// cfg.OutboundConfigs. Synthetic rules are prepended ahead of whatever's
// already in the routing rules array, the same pattern injectNodeEgresses/
// injectMtprotoEgress already use for their own always-must-win infra
// rules — this never touches the admin's own saved Routing-page rule
// order.
func injectAmneziawgV6Egress(cfg *xray.Config, inbounds []*model.Inbound) {
// Protocol is checked alongside Tag, not just Tag alone: a tag collision
// with some unrelated (non-socks) inbound must not be mistaken for this
// instance's own relay having been created.
liveInboundTags := make(map[string]struct{}, len(cfg.InboundConfigs))
for i := range cfg.InboundConfigs {
if cfg.InboundConfigs[i].Protocol == "socks" {
liveInboundTags[cfg.InboundConfigs[i].Tag] = struct{}{}
}
}
var existingOutbounds []any
if len(cfg.OutboundConfigs) > 0 {
if err := json.Unmarshal(cfg.OutboundConfigs, &existingOutbounds); err != nil {
logger.Warning("amneziawg v6 egress: outbounds section is unparsable, skipping injection:", err)
return
}
}
usedOutboundTags := make(map[string]struct{}, len(existingOutbounds))
for _, o := range existingOutbounds {
if m, ok := o.(map[string]any); ok {
if t, ok := m["tag"].(string); ok {
usedOutboundTags[t] = struct{}{}
}
}
}
routing := map[string]any{}
if len(cfg.RouterConfig) > 0 {
if err := json.Unmarshal(cfg.RouterConfig, &routing); err != nil {
logger.Warning("amneziawg v6 egress: routing section is unparsable, skipping injection:", err)
return
}
}
rules, _ := routing["rules"].([]any)
newRules := make([]any, 0)
newOutbounds := make([]any, 0)
for _, inbound := range inbounds {
if inbound.Protocol != model.AmneziaWG || !inbound.Enable || inbound.NodeID != nil {
continue
}
if _, live := liveInboundTags[inbound.Tag]; !live {
// The relay inbound itself wasn't created this pass (e.g. a tag
// collision inside injectAmneziawgnetSocks) -- no SOCKS5 inbound
// exists for hot_diff.go's inboundTag match to ever fire against.
continue
}
inst, ok := amneziawg.InstanceFromInbound(inbound)
if !ok || !amneziawgnet.V6AliasesActive(inst) {
continue
}
for _, p := range inst.Peers {
if p.Email == "" {
continue
}
v6 := amneziawg.FirstIPv6(p.AllowedIPs)
if v6 == "" {
continue
}
tag := amneziawgV6EgressTag(inbound.Id, p.Email)
if _, taken := usedOutboundTags[tag]; taken {
logger.Warning("amneziawg v6 egress: outbound tag [", tag, "] already exists, skipping peer [", p.Email, "]")
continue
}
usedOutboundTags[tag] = struct{}{}
newOutbounds = append(newOutbounds, map[string]any{
"tag": tag,
"protocol": "freedom",
"sendThrough": v6,
"settings": amneziawgV6EgressSettings(),
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
})
newRules = append(newRules, map[string]any{
"type": "field",
"inboundTag": []any{inbound.Tag},
"user": []any{p.Email},
"ip": []any{"::/0"},
feat(amneziawg): add native AmneziaWG protocol support (#6105) * feat(amneziawg): add native AmneziaWG protocol backend AmneziaWG (WireGuard plus DPI-resistant obfuscation) needs no Docker here — it runs as a genuine kernel interface via awg-quick/awg, managed the same way internal/mtproto manages mtg: one Inbound row is one desired Instance, and a Manager reconciles running interfaces toward the database every 10s (internal/web/job/amneziawg_job.go) plus immediately after a client edit (applyLocalAmneziaWG). Clients reuse model.Client verbatim (the same PrivateKey/PublicKey/ PreSharedKey/AllowedIPs fields WireGuard already uses), so bulk operations, the QR/share-link modal and subscriptions come from the shared inbound infrastructure instead of a parallel implementation. internal/amneziawg owns the obfuscation param generator/validator (ported from coinman-dev/3ax-ui, upgraded to AmneziaWG 2.0's S3/S4 padding and I1 signature packet) and the exec wrapper around awg-quick/awg, with fingerprint-based reconcile (noop / reload-via- syncconf / full restart) mirroring mtproto.Manager so a same-protocol edit doesn't force an unnecessary interface bounce that would drop every peer's connection. Frontend and install.sh's DKMS/awg-tools setup are tracked separately; this is backend-only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add frontend support and fix a Go->Zod generator gap Wires the amneziawg protocol through the panel UI the same way every other protocol is registered: a Zod settings schema (nested {server, clients}, matching the Go JSON exactly), the protocol enum, the inbound-form's per-protocol fields component and its tab-visibility allowlist, the default-settings factory, the client schema dispatcher, and the sniffing-capability exclusion (no Xray inbound exists for amneziawg, same as mtproto). Client key/allowedIPs fields are reused rather than duplicated: since AmneziaWG clients are wire-identical to WireGuard clients (same model.Client fields), ClientFormModal renders one shared field block for both, switching only the visible label by which protocol is active. The private-key input also gets a live public-key sync via a new useEffect, because unlike WireGuard's Xray-native inbound (which re-derives its public key at runtime and never stores one), AmneziaWG's server.publicKey is a real persisted field the Go backend reads directly — free-typing a new private key without this would silently save a mismatched keypair. Adds a downloadable per-client .conf (amneziawgConfig.ts, mirroring wireguardConfig.ts) with the obfuscation lines, and an InboundOption.AwgServer field on the Go side so the config builder gets the full server block in one round trip. Along the way, running tools/openapigen surfaced a real bug: it doesn't flatten anonymously-embedded Go structs the way encoding/json does, so ServerSettings embedding Obfuscation20 produced a Zod schema with a nested `obfuscation20` key that never matches the real wire JSON. Fixed by un-embedding (flat fields + an accessor method) and registering internal/amneziawg in the generator's own package list, which had been silently emitting a dangling schema reference. English and Russian translations are complete; the other 10 locale files still fall back to English for the new keys. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): complete frontend parity for the Inbounds list page The Clients page (form, CRUD, QR/config) already worked from the prior commit; this closes the remaining gap on the Inbounds side and in a couple of protocol allowlists that a plain search for existing wireguard/mtproto handling turned up. lib/xray/inbound-link.ts gets amneziawg-specific link/config builders (genAmneziaWGLink/genAmneziaWGConfig, plus the *s fan-out variants) mirroring the wireguard ones — AmneziaWG has no legacy peers-array to fall back to, so these read settings.clients directly and add the obfuscation lines every client must share with the server. Wired into genInboundLinks generically, and into three consumers that call the wireguard builders directly rather than through that dispatcher: QrCodeModal, InboundInfoModal, and InboundsPage's bulk export. ClientInfoModal, ClientBulkAddModal, and the bulk attach/detach modals each had their own protocol allowlist that needed amneziawg added alongside wireguard/mtproto. Two real gaps surfaced by grepping every remaining 'wireguard' / Protocols.WIREGUARD hit in frontend/src rather than trusting the checklist was exhaustive: - useInbounds.ts's TRACKED_PROTOCOLS gates the deactive/depleted/ expiring/online client counts shown per inbound on the list page; without amneziawg those counts would silently read zero. - inbound-tag.ts is an explicit client-side mirror of the Go backend's port_conflict.go (the file says so itself: "Keep in sync"). It still only special-cased wireguard for UDP, so an amneziawg inbound would have fallen through to the TCP default and disagreed with the backend's own port-conflict math. Also finishes translating the AmneziaWG UI strings into the 11 locale files that were still falling back to English (ar-EG, es-ES, fa-IR, id-ID, ja-JP, pt-BR, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW), matching en-US/ru-RU key-for-key (26 new keys, verified by count in every file). Not run anywhere: npm run typecheck / build. This machine has neither Node nor npm, so nothing here has compiled — reviewed by hand plus brace/paren balance checks and cross-referencing the generated Zod/TS types. Treat this as needing a real typecheck before shipping. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(install): note that AmneziaWG kernel module install is still manual Tracked separately (not yet ported into this script) — see coinman-dev/3ax-ui's install_amneziawg for the reference approach (ppa:amnezia/ppa). Also serves as a real, path-filter-matching change to get the previous empty commit's CI trigger to actually fire — release.yml's push trigger is paths-scoped and an empty commit changes no files, so it never matched. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): add a button to randomize obfuscation parameters Mirrors the existing key-regenerate button next to the private key field. Client-side randomization matches the ranges/constraints of GenerateObfuscation20's "default" preset (internal/amneziawg/params.go) closely enough for a form suggestion — the user can still hand-edit any field afterward. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(install): auto-install the AmneziaWG DKMS module + amneziawg-tools Ports install_amneziawg from coinman-dev/3ax-ui's install.sh, adapted to this script's broader distro coverage and NONINTERACTIVE convention: - Ubuntu/Debian/Armbian: ppa:amnezia/ppa (primary, tested path), with a reachability pre-check for the Launchpad PPA host — often blocked by hosting providers, especially Russian VPS — so a flaky network skips the feature instead of hanging apt through several retries. - Fedora/RHEL-family, Arch/Manjaro/Parch: best-effort fallback to plain wireguard-tools (+ AUR amneziawg-dkms via yay/paru when available), with a manual-install pointer. - Everything else: manual-install pointer only. Also installs ndppd and persists IPv4/IPv6 forwarding (for the future IPv6/NDP phase, not yet wired into the panel) and adds a Secure Boot warning at the end of the run, since a DKMS-built module is unsigned and won't load while it's enabled — a common trap on cloud VPS images. Never fatal: the panel installs and runs fine either way, an AmneziaWG inbound just won't bring up its tunnel until the module is present. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve all 3 real CI failures (typecheck/lint/codegen) Found by checking the fork's Actions tab after the last two pushes — the release build passed (it doesn't run these checks) but the separate CI workflow caught three real issues: - golangci-lint (noctx): every internal/amneziawg/manager.go exec.Command call is now exec.CommandContext with a 30s timeout, so a hung awg-quick/awg invocation can't block the reconcile job indefinitely (mirrors internal/mtproto/process.go's own CommandContext usage). - tsc --noEmit: frontend/src/schemas/client.ts's hand-maintained InboundOptionSchema (used by the useClients hook, separate from the auto-generated one in generated/) never got an awgServer field added when the AmneziaWG frontend work was done — every read of inbound.awgServer.* in amneziawgConfig.ts was typing as {}. Added AwgServerOptionSchema, nested (not flattened like wg*) to match what amneziawgConfig.ts already expects. Also guarded server.publicKey in inbound-link.ts's genAmneziaWGLink against the schema's optional type. - codegen staleness: frontend/public/openapi.json is produced by a Node script (gen:api) this machine can't run; hand-applied the exact diff the CI failure log already showed (amneziawg protocol enum entry, ServerSettings schema, InboundOption.awgServer, one example payload), verified as valid JSON. Also confirmed independently by this run: install_amneziawg (previous commit) installed and loaded the DKMS module successfully on both amd64 and arm64 CI runners. The two "Deploy Smoke Tests" failures are unrelated to this change — this fork has only ever published the dev-latest pre-release, and GitHub's /releases/latest API deliberately excludes pre-releases, so the smoke test's no-argument install path (which resolves "latest") has nothing to find. Not a regression; needs an actual tagged release whenever that's wanted. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2a — IPv6 support + NDP proxy Adds native dual-stack IPv6 to AmneziaWG inbounds, ported from coinman-dev/3ax-ui's approach: - ServerSettings gets ipv6Enabled/ipv6Subnet/ipv6ExternalInterface; Instance carries the server's own IPv6 address (first host of the subnet) alongside its IPv4 one. - defaultAmneziaWGClients allocates an IPv6 host address per client (second AllowedIPs entry) when the server has IPv6 enabled, reusing allocateWireguardAddress — which needed a real fix along the way: it always suffixed "/32" regardless of address family, which is wrong for an IPv6 host address (needs /128). Now family-aware. - generateServerConfig's PostUp/PostDown gains IPv6 forward-accept rules, proxy_ndp sysctl, and one `ip -6 neigh add/del proxy` entry per enabled peer with an IPv6 address — the lightweight per-client method, not the ndppd-daemon whole-subnet method (not worth the config-file-management complexity at this scale; ndppd itself is still installed by install.sh in case that changes later). - ValidateIPv6Subnet rejects a malformed subnet before save. - Frontend: ipv6Enabled/ipv6Subnet/ipv6ExternalInterface fields on the AmneziaWG inbound form, EN+RU translations, openapi.json/generated/* regenerated (the latter via `go run ./tools/openapigen`, pure Go). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): fill in IPv6 fields missed by the Phase 2a commit Two real gaps the CI caught (both new fields, both my miss): - inbound-defaults.ts's createDefaultAmneziawgInboundSettings() built a server object literal predating ipv6Enabled/ipv6Subnet/ ipv6ExternalInterface — AmneziawgServer's inferred type now requires them (zod .default() fields are non-optional post-parse), so this didn't typecheck at all. - openapi.json's ipv6Enabled property was missing the description the real generator attaches (the Go doc comment covering all three IPv6 fields is attached to the first one) — a one-line diff, but git diff --exit-code doesn't care how small. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): Phase 2b — per-client port-forwarding Admins can now set a per-client ForwardedPorts string (e.g. "80, 443, 8000-8100") that gets DNAT'd + FORWARD'd to that peer's tunnel address via iptables rules in PostUp/PostDown, ported and simplified from coinman-dev/3ax-ui's shared/portfwd. Two decisions worth flagging for future readers: - The iptables --comment tag on each rule is awg-fwd-<fnv32a(email)>, not the raw client email. Email is admin/API-supplied free text that ends up embedded in a shell-executed PostUp/PostDown line; a hash can never carry a shell metacharacter through where raw interpolation could. - The reconcile manager gained a third fingerprint (portFwdFP, next to the existing structural/peers ones). `awg syncconf` only touches the WireGuard peer table — it never re-applies PostUp/PostDown iptables rules — so a port-forward-only change has to force a full awg-quick down+up bounce, same as a structural change, rather than the lighter sync a plain peer add/remove can use. Also fixes a real pre-existing bug found while wiring up IPv6 client allocation in the previous commit's spirit: allocateWireguardAddress always suffixed "/32" regardless of address family, which produced invalid host bits for IPv6 (needs "/128"). ForwardedPorts flows through model.Client -> model.ClientRecord (gorm column wg_forwarded_ports, auto-migrated) -> ToRecord/ToClient/ MergeClientRecord, mirroring the awgServer field's earlier lesson that new fields need checking against a second, hand-maintained persistence-layer struct. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): route a client's traffic through Xray via the Routing page Every enabled AmneziaWG inbound gets its own Xray TPROXY bridge automatically, with no toggle to enable first: a loopback dokodemo-door inbound (sockopt.tproxy) tagged with the AmneziaWG inbound's own real tag, so it's already selectable in the existing Routing page's inbound-tag picker — the same trick the mtproto sidecar's own bridge already relies on (InboundService.GetInboundTags is a plain, protocol-blind SELECT over every inbound row's tag, no dedicated UI plumbing needed). internal/amneziawg's defaultPostUpDown TPROXYs every peer's traffic into that bridge unconditionally; the bridge's port is derived deterministically from the inbound's id (EgressPortForInbound) so the kernel-side reconcile loop and the Xray-config generator never need to negotiate a runtime value between them. injectAmneziawgEgress never generates a routing rule itself — whether a client's traffic goes anywhere beyond Xray's default routing is entirely up to whatever rules the admin adds through the existing Routing UI (pick the AmneziaWG inbound's tag as source, optionally a specific peer's IP via that page's own Source-IP field, and an outbound), exactly the same workflow as routing any other protocol. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): recover orphaned interfaces after an ungraceful exit Two gaps left an AmneziaWG interface stuck outside the manager's control after a crash (kill -9/OOM/panic skips StopAll): - ensureRestart's teardown was gated on the in-memory `exists` map, which is always empty on a fresh process, so a survived interface never got interfaceDown before interfaceUp tried `ip link add` against a name the kernel already had — failing forever and never populating m.ifaces, so traffic accounting silently stopped and the inbound could never be removed. Gate on isInterfaceUp instead, which checks real kernel state rather than this process's own bookkeeping. - An inbound deleted from the database entirely while the panel was down has no entry in `desired` ever again, so it never reaches the per-id cleanup loop in Reconcile (which only walks m.ifaces). Add a one-time sweepOrphansLocked scan of configDir, mirroring mtproto.Manager.sweepOrphansLocked, that tears down and removes any leftover interface/config not in the current desired set. Found by the automated review on MHSanaei/3x-ui#6105 (Finding 1). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * i18n(amneziawg): backfill IPv6/obfuscation/port-forwarding keys in 11 locales Only en-US/ru-RU ever got these 9 keys as each AmneziaWG feature landed (the regenerate-obfuscation button, then Phase 2a's IPv6 fields, then Phase 2b's per-client ForwardedPorts) — the other 11 locale files were never backfilled, so i18next has been silently falling back to English for all of them since Phase 1. Cosmetic-only (never broke anything), but now closed for every shipped locale. * fix(amneziawg): resolve 7 Medium findings from the automated PR review Each is independently reproducible; fixed together since one review pass found all of them. - manager.go: the shared "ip rule add fwmark" policy route had no existence check, so it duplicated in "ip rule show" on every interface bounce (which hostRulesFingerprint forces on any client add/remove/ re-IP). Now checked via "ip rule list | grep -q ..." first. (Finding 2) - params.go: ExternalInterface, IPv6ExternalInterface, and subnetIp/ subnetCidr are interpolated unescaped into a shell-executed PostUp/ PostDown line, but only obfuscation and the IPv6 subnet were validated before save. Added ValidateInterfaceName (a strict charset+length pattern) and ValidateSubnetIPv4 (netip.ParsePrefix), wired into normalizeAmneziaWGSettings. (Finding 3) - amneziawg_job.go: IsAwgInstalled() existed but nothing ever called it, so a host without awg/awg-quick (the Docker image, RHEL, Arch, a failed install.sh PPA step) logged a reconcile failure every 10s forever. Now checked once an inbound actually needs it, warning once instead of spamming. (Finding 4) - client_inbound_apply.go: the WireGuard/AmneziaWG credential carry-forward (added so a metadata-only client edit doesn't rotate keys) never covered ForwardedPorts, so a partial edit -- an API call or Telegram-bot toggle that omits the field -- silently wiped a client's port-forwarding spec. Carried forward and written back the same way the key fields already are. (Finding 5) - manager.go: hostRulesFingerprint keyed each peer on its IPv4 address only, and structuralFingerprint omitted IPv6Enabled/IPv6ExternalInterface entirely, so an IPv6-only change could pick the syncconf reload path (which never re-runs PostUp, leaving a stale NDP-proxy entry) or be a complete no-op. Both fingerprints now cover the IPv6 fields. (Finding 6) - port_conflict.go: the AmneziaWG egress bridge (injectAmneziawgEgress) binds 127.0.0.1:63100+id with no collision check anywhere, since it isn't a database row the ordinary port-conflict query can see -- same blind spot the reserved Xray API port already has its own check for. Added the equivalent check for the AmneziaWG bridge port. (Finding 7) - install.sh: install_amneziawg ran unconditionally for every install/ update, building a DKMS kernel module and enabling host-wide IPv4/IPv6 forwarding whether or not the feature is ever used. Gated behind a new should_install_amneziawg (XUI_INSTALL_AMNEZIAWG=true/false, or an interactive y/N prompt defaulting to no). Also replaced the deprecated apt-key adv with a dedicated keyring + signed-by= on the Debian branch, and guarded its sources.list appends against duplication on a retried install. (Finding 8) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * feat(amneziawg): make the Xray TPROXY bridge a per-inbound opt-in Addresses Finding 10 from the automated PR review: an always-on TPROXY bridge makes every AmneziaWG tunnel hard-depend on Xray being up (all traffic, including DNS, drops whenever Xray restarts), and forces a full awg-quick down+up bounce on any client add/remove/re-IP, permanently losing the syncconf fast path. Adds ServerSettings.RouteThroughXray (off by default): - defaultPostUpDown only emits the TPROXY/policy-route rules when it's on; a plain AmneziaWG tunnel now has zero Xray dependency out of the box. - structuralFingerprint covers it (toggling it changes whether PostUp/ PostDown contain any TPROXY rules at all -- structural, not a per-peer host-rule). hostRulesFingerprint's IPv4 tracking is now itself conditional on RouteThroughXray (and IPv6 tracking on IPv6Enabled), so an instance that never uses either keeps the syncconf fast path for a plain peer re-IP. - injectAmneziawgEgress only creates a bridge for inbounds that opted in; checkAmneziawgEgressConflict (the Finding-7 fix) now parses each candidate through InstanceFromInbound so a non-routed inbound's port is correctly never treated as reserved. - New inbound-level Switch in the AmneziaWG form; the actual outbound decision is still made entirely through the panel's stock Routing page, same as before -- only whether the bridge exists at all is now a choice. Translation keys added to all 13 locales in the same commit this time, not backfilled later (see Finding 9's lesson). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): resolve 4 Low findings from the automated PR review - manager.go: serverAddress assumed subnetIp always ends in ".0"; a base like "10.8.1.5" was used verbatim as the server's own address, eventually colliding with peer allocation (which starts at .2 upward). Now derives the first host of the actual subnetIp/subnetCidr network via netip, matching serverAddressV6's own approach. A /32 base (no host bits at all) is still used as-is. (Finding 12, partial -- the /16 pool-widening half of this finding only exists on the upstream-pr/amneziawg branch's merged client_wireguard.go, not here; handled separately on that branch.) - manager.go: ensureLocked carried the previous per-peer traffic counters (`last`) forward even through a full restart, but awg-quick down+up resets the kernel's own counters to zero -- the next CollectTraffic computed a large negative delta (clamped to 0), silently discarding real traffic. Extracted the decision into nextTrafficBaseline: only a reload (syncconf) preserves the baseline. (Finding 13) - portfwd.go: exported ForwardedPortsInclude; inbound_amneziawg.go's new checkForwardedPortsConflict uses it to reject, at save time, a client's forwardedPorts that would DNAT the panel's own port or another enabled inbound's port to the tunnel client -- portForwardLines has no destination restriction, so this collision was previously silent. Wired into both the single-client update path and the add-client path (client_inbound_apply.go), plus normalizeAmneziaWGSettings for the whole-inbound save path. (Finding 14) - inbound.go: InboundOption.AwgServer sent the whole ServerSettings struct including PrivateKey to GetInboundOptions callers -- a shared, admin-wide dropdown-filling endpoint the frontend's own AwgServerOptionSchema never reads that field from. Redacted it before assigning. (Finding 11) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): don't widen the peer address pool past AmneziaWG's own subnet Completes Finding 12 from the automated PR review (the serverAddress half of this finding was already fixed on main and cherry-picked here). This half is specific to this branch: allocateWireguardAddress's /16 pool-widening fallback is an independent addition from upstream's own main that this branch inherited during the cherry-pick rebase -- it doesn't exist on the fork's own main at all, so this fix can't be cherry-picked the normal way and is committed directly here. Widening is safe for WireGuard's own Xray-native inbound (AllowedIPs isn't tied to a strict kernel interface subnet), but AmneziaWG's kernel interface Address is exactly the configured subnet -- an address allocated from the containing /16 once the /24 fills up would be silently unroutable. allocateWireguardAddress now takes an explicit allowWidening bool: WireGuard's own caller passes true (unchanged behavior), AmneziaWG's passes false (fails loudly on exhaustion instead). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * docs(docker): note that AmneziaWG doesn't work in this image Investigated: the image is Alpine-based, and AmneziaWG's own packaging (DKMS module + amneziawg-tools) doesn't target Alpine/musl at all -- unlike the Debian/Ubuntu/Fedora/Arch paths install.sh already handles, there's no package to apk add even with full host network/capabilities. The panel already degrades gracefully (IsAwgInstalled() logs one warning instead of retrying forever), so no code change is needed -- just made the reason explicit at the point where a user would reach for cap_add/ network_mode to try to work around it. * fix(sub): include amneziawg inbounds in subscription links getInboundsBySubId's SQL protocol allowlist never had 'amneziawg' added, so every AmneziaWG client was silently excluded from all three subscription formats (plain/individual links, JSON, Clash) and from the Telegram bot's QR/individual-link buttons, which fetch through the same path. genAmneziaWGLink itself was already fully implemented and already wired into GetLink's dispatch switch -- it just never got a chance to run. Same bug shape as the earlier TRACKED_PROTOCOLS frontend gap: a hardcoded protocol list one entry short. Found while investigating whether the Telegram bot needed AmneziaWG- specific client-management code -- it doesn't (the bot itself is fully protocol-agnostic), but this is the actual root cause of "can't share an AmneziaWG client's config via the bot." Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(inbound): enforce node-eligibility server-side, not just in the UI Investigated multi-node interaction with AmneziaWG: the master's own reconcile (DesiredAmneziaWGInstances) and Xray config generation (injectAmneziawgEgress, the GenXrayInboundConfig protocol skip) all correctly filter on NodeID IS NULL, so a node-assigned AmneziaWG (or MTProto) inbound would never be managed by the master. But nothing stopped one from being created that way: NODE_ELIGIBLE_PROTOCOLS (frontend/src/pages/inbounds/form/InboundFormModal.tsx) only hides the node picker client-side -- a direct API call could set nodeId on an AmneziaWG inbound, which every node then reconciles as an ordinary local inbound (nodes run the identical binary, full cron suite included), leaving it running unmanaged and untracked by the master's own AmneziaWG bookkeeping. Added isNodeEligibleProtocol (inbound_protocol.go), mirroring the frontend's allowlist, and enforced it in both AddInbound (the actually exploitable path -- nodeId comes straight from the request) and UpdateInbound (defense in depth; NodeID is already restored from the stored row there before this check, so it mainly guards against a protocol change on an existing node-hosted inbound). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): allow TPROXY-marked traffic through a default-deny INPUT chain TPROXY never rewrites a packet's own destination address, only the routing decision. A default-deny firewall whose INPUT chain sanity-checks "is this destination actually local" (UFW's ufw-not-local, via addrtype --dst-type LOCAL, is a concrete example) silently drops the redirected packet before Xray's socket ever sees it -- RouteThroughXray looked fully configured (TPROXY rule present and counting, Xray listening with IP_TRANSPARENT set) yet every peer's traffic vanished with no trace on either side. Adds an idempotent, never-torn-down "iptables -I INPUT 1 -m mark --mark <fwmark> -j ACCEPT" alongside the existing shared policy route, so this works regardless of which firewall manager owns the rest of the INPUT chain. * fix(frontend): give AmneziaWG the same UDP tag and its own tag color The Inbounds list only special-cased isWireguard/isHysteria for the "UDP" network badge, so an AmneziaWG row showed just the bare protocol tag with no transport badge next to it. Added the missing isAmneziawg flag (mirrors isWireguard exactly) and wired it into the same branch. Client-row protocol-color maps in ClientsPage/HostList had no amneziawg entry, silently falling back to grey -- ClientInfoModal already had amneziawg: 'yellow' from earlier work, these two just never got it. * feat(logs): show which AmneziaWG client an access-log line belongs to The dokodemo-door TPROXY bridge every AmneziaWG peer's traffic is routed through has no per-user identity, so Xray's own access log never carries an "email:" token for these lines -- the Access Logs modal showed a blank Email column for every in-*-udp row, even though every other protocol's rows show the client normally. The peer's decapsulated tunnel IP does survive as the log's "from" address, and that IP deterministically maps to exactly one configured peer. Builds a "<inbound tag>|<ip>" -> email index from the same AmneziaWG inbounds already parsed elsewhere (amneziawg.InstanceFromInbound), and fills in Email from it whenever the raw log line didn't have one. * fix(amneziawg): enable sniffing on the TPROXY bridge Domain-based Routing rules could never match RouteThroughXray traffic: an AmneziaWG peer resolves DNS itself, through the tunnel, before ever sending a packet, so the decapsulated traffic TPROXY hands to the bridge is already a bare destination IP with no domain name attached at the network layer. Every other inbound recovers this via sniffing (confirmed working for the stock wireguard inbound, which does have it configured); the bridge never got a sniffing block at all, so only tag/IP/network-based rules could ever match it -- any domain rule above it in the list was silently unreachable. * docs: add an AmneziaWG config page and list it as a supported protocol Closes the PR checklist gap: the feature shipped with zero mention on the docs site. Mirrors reality.mdx's structure (key settings, setup steps, config excerpt) and notes the Docker/multi-node/Telegram-bot caveats the PR itself is honest about not having confirmed. * fix: address the fresh review round on PR #6105 (8 findings) 1. hostRulesFingerprint didn't account for ForwardedPorts when RouteThroughXray was off, so re-IPing a peer with port-forwarding configured left stale DNAT rules pointing at an address the next peer could be handed. 2. Server/client config values (keys, email, I1) were never validated for control characters before being written into the generated .conf; a newline could smuggle a PostUp hook into awg-quick's parser. Added ValidateConfigValue at save time and a sanitizeConfigValue backstop at render time. 3. checkForwardedPortsConflict didn't scope to node_id IS NULL, so a port used only on a different node produced a false collision; also hoisted the panel-port/inbounds lookup out of the per-client loop (portConflictContext) so N clients cost one query, not N. 4. PostDown commands were ";"-joined and abort on the first failure; appendOrTrue makes teardown best-effort so an external firewall flush can't leave DNAT rules to accumulate across bounces. 5. The "ip rule list | grep -q" existence check could SIGPIPE under pipefail and re-add a duplicate rule; switched to grep -c >/dev/null. 6. Ported the vpn:// share-link format (base64url of the plain .conf text, matching the real AmneziaVPN app) onto this branch -- it had only ever landed on our own fork's main, so this PR branch was still on the old amneziawg://+query-params scheme our own docs no longer described. Also corrected the docs' install.sh claim (opt-in/ interactive, not automatic) and stale pre-opt-in comments in route_egress.go. 7. install.sh: Arch's ndppd install used pacman -Syu (full system upgrade) instead of -Sy like every other call in the script; and should_install_amneziawg re-prompted on every `x-ui update` even when awg was already installed. 8. CollectTraffic could clobber a concurrent restart's freshly-reset (empty) traffic baseline with stale pre-restart counters, since getPeerStats runs lock-free; now checks pointer identity before writing back. sweepOrphansLocked permanently disabled itself on a transient os.ReadDir failure instead of allowing a retry. go build/vet/test and frontend typecheck/lint/build/vitest all pass. * fix(install.sh): check the live sysctl value, not sysctl.conf text Reviewer feedback (cherts, PR #6105): grepping /etc/sysctl.conf for the setting name is unreliable -- many distros split sysctl config across /etc/sysctl.d/*.conf, and /etc/sysctl.conf can be a symlink into that directory, so the check can miss an already-active setting (harmless duplicate append) or match a disabled/commented line (forwarding silently stays off). Query the live value via `sysctl -n` instead, which is accurate regardless of which file set it. Applied the same fix to both the IPv6 and IPv4 checks for consistency. * fix: update inbound_amneziawg.go to the split buildInboundForLocalRuntime Same fork-only-file blind spot as the one caught on our own main after the 3.6.0 sync: upstream split buildRuntimeInboundForAPI into buildInboundForNodePush / buildInboundForLocalRuntime (part of the node-sync client-deletion fix, 5bc81dfd), updating every call site it could see. This file doesn't exist upstream, so it kept calling the old name even after the branch merged in that commit. * fix(frontend): recognize AmneziaWG's vpn:// scheme in share-link labels The shared link-tag/label helper (used by the client info modal, QR modal, and subscription page) had no entry for the vpn:// scheme AmneziaWG links use, so it fell through to the generic fallback: a plain "Vpn" tag with no color, and an empty remark/port that made the row's title fall back to "Link N" instead of the inbound's actual name:port — unlike every other protocol, which shows its real tag and label. vpn:// links are base64url of a plain .conf text (matching the real AmneziaVPN app's own share-link format), not a structured URL, so there's no query string or #hash to read a remark/port from. Decode the payload and pull the remark/endpoint back out of the .conf text directly instead. * fix(xray): force a full restart for TPROXY inbounds, never hot-add them Real incident: an AmneziaWG inbound with RouteThroughXray enabled lost all internet on that connection after a migration. Root-caused on the live box -- iptables TPROXY counters were incrementing (packets correctly redirected to 127.0.0.1:63110), but nothing was actually listening there (ss showed nothing on that port) until a full `systemctl restart x-ui`, after which the bridge came up immediately. Xray-core's gRPC AddInbound reports success for a new sockopt.tproxy inbound (internal/amneziawg's own Xray egress bridge is the only kind this fork ever generates) but doesn't reliably bind a working listener for it outside of process startup -- the bridge silently never comes up, and RouteThroughXray traffic goes nowhere until the next full restart happens to occur for an unrelated reason. diffInbounds already has this exact defensive pattern for REALITY inbounds ("a gRPC remove+add does not reliably rebuild the REALITY authenticator"), just never extended to TPROXY, and only in the already-existing-then-changed branch -- the "brand new inbound" branch had no such guard at all, which is exactly the path a freshly-enabled RouteThroughXray bridge takes. Added inboundUsesTproxy and wired it into both branches. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): flag Xray for resync when a peer edit changes qualifying state Real production bug, root-caused on iiadmin-vps: updateAmneziaWGInbound/ AddInbound/DelInbound only ever updated the kernel interface via amneziawg.GetManager() -- they never called SetNeedRestart the way every other protocol's mutation path does (client_crud.go, inbound.go, etc. all do). injectAmneziawgEgress's TPROXY bridge inbound depends on InstanceFromInbound finding at least one qualifying peer plus RouteThroughXray, so an edit that flips that (first peer added, last one removed, RouteThroughXray toggled on) previously required a full panel restart before the bridge actually got created, with no error anywhere: the kernel interface would handshake fine, but traffic redirected into the bridge's TPROXY port went nowhere because nothing was listening there. diffInbounds/inboundUsesTproxy already correctly force a full restart for a brand new TPROXY inbound (bdee0a20) -- that part was never the bug. The gap was entirely upstream: nothing ever told Xray a resync was even needed. * fix(clients): reject AllowedIPs already used on another WireGuard/AmneziaWG inbound defaultWireguardClients/defaultAmneziaWGClients only ever checked uniqueness against their own inbound's client list, so two inbounds sharing a subnet (same protocol or not) could silently hand out or accept the same address -- the exact scenario behind a real duplicate-IP incident where a WireGuard and an AmneziaWG client both ended up on the same address. otherTunnelAllowedIPs now collects every address already claimed on every other tunnel inbound and folds it into both the auto-allocation pool and the manual-entry collision check, naming the other inbound in the error when it fires. * fix(frontend): add the missing AmneziaWG config download on the sub page The subscription page already gave WireGuard links their own "Config" block (copy/download/QR of the actual .conf, via wireguardConfigFromLink reversing the wireguard:// query params) but had no equivalent for AmneziaWG's vpn:// links -- its isWireguardLink gate never matched them, and no reverse-parse helper existed for this page specifically. Every other surface (InboundInfoModal, ClientInfoModal, ClientQrModal) already had this parity; this was the one page that didn't. Fixed by adding amneziawgConfigFromLink (inbound-link.ts), simpler than its WireGuard counterpart since a vpn:// payload already *is* the plain .conf text -- just base64url-decode it, no query-param reconstruction needed -- and wiring it into SubPage.tsx alongside the existing WireGuard block, reusing the same pages.clients.amneziaWgConfig label the other three surfaces already use. * fix(xray): force a full restart for password-auth SOCKS5 hot-apply Real production incident: editing a client under an AmneziaWG inbound left its embedded SOCKS5 relay's settings byte-different (a new account list), and Xray's gRPC remove+add hot swap silently dropped the account for a peer whose email contained non-ASCII characters -- its tunnel kept handshaking fine but all its traffic got rejected at the SOCKS5 layer, while every other peer on the same relay was unaffected. A full restart (reading the same JSON straight from disk) always produced the correct account list. socks isn't in userDiffableProtocols (that only covers vless/vmess/trojan's clients+email shape, not accounts+user), so any settings drift on this inbound fell through to the generic remove+add path. Forces a restart instead, the same defensive choice already made for REALITY and TPROXY -- scoped to auth:"password" specifically so the other, noauth SOCKS5 bridges (panel/node/mtproto egress) keep the cheaper hot path. * Fix Attach reusing one identity's address across wg/awg inbounds ClientService.Attach deliberately copies one identity's stored AllowedIPs into every WireGuard/AmneziaWG inbound it's attached to in the same call, so the same person gets the same tunnel address on every protocol they use. Its loop calls addInboundClient once per inbound, and each of those independently computes otherTunnelAllowedIPs -- so by the second inbound in the batch, the first inbound's just-written copy of this identity's own address looked like a cross-inbound collision against itself. Real production symptom this caused: detaching then re-attaching a client to both wg and awg failed with "wireguard: allowedIPs entry X is already used by a client on inbound 'awg' (#N)" -- the exact address the identity is supposed to keep, rejected as if it belonged to someone else. Add a selfEmails exclusion to otherTunnelAllowedIPs and populate it from the client(s) being processed at the one real call site. Safe unconditionally: ClientRecord.Email is globally unique, so a match can only ever be this same identity's own entry on a sibling inbound, never a genuine different client's address. Reproduced the underlying mechanism live (manual entry correctly rejected as a cross-inbound collision; fresh auto-allocation correctly avoided a used address) before writing the fix, to confirm the guard itself works and the bug is specifically in how Attach's per-inbound calls interact with it. * Attach: allocate fresh when re-attaching with no active tunnel The previous fix (82cc69f5) made Attach's own address-reuse correctly not collide with itself across inbounds -- but it still always reused an identity's stored AllowedIPs verbatim, even when that identity currently has zero WireGuard/AmneziaWG attachments at all. A real report from testing this live: an identity fully detached from both its wg and awg inbounds, then re-attached, got its old address back even though several lower addresses were free -- because nothing about being fully detached ever cleared the stored value Attach copies from. Add hasTunnelAttachment, checked once against the identity's CURRENT inbound set before Attach's loop runs: if none of its current inbounds is WireGuard/AmneziaWG, clear the stored AllowedIPs so this attach allocates fresh (matching what a brand-new client would get) instead of resurrecting an address nothing reserves anymore. Left alone when the identity already has an active tunnel elsewhere, so extending it to a second protocol still keeps a consistent address. * Fix TestOtherTunnelAllowedIPsExcludesSelfEmail's own test setup CI caught this: the "genuinely different client" (other@wg) was seeded onto the SAME inbound passed as excludeID, which otherTunnelAllowedIPs already excludes entirely regardless of the selfEmails fix -- so the assertion that its address is still reported could never have passed, proving nothing either way. Move it onto the sibling inbound alongside shared@id, which is what the test actually needs to exercise (two clients on one sibling, one excluded by email, one not). * Attach: never inherit an address that doesn't fit the target inbound hasTunnelAttachment (from the earlier fix, commit 51067f16) only asked "does this identity have ANY tunnel attachment", treating that as license to reuse its stored address verbatim on every inbound being attached. Real production case this missed: an identity's stored address came from WireGuard's own fallback subnet (10.0.0.0/24, used when that inbound has no other clients to infer a base from), then got attached to a second, AmneziaWG inbound configured for a completely different subnet (10.8.1.0/24). defaultAmneziaWGClients's already-set-AllowedIPs branch only checks for collisions, never subnet membership, so the mismatched address was accepted silently -- producing a peer that can never actually connect, since an AmneziaWG address must fall inside the kernel interface's own configured subnet to be routable at all. Add addressesFitAmneziaWGInbound, checked per inbound inside Attach's loop: if the inherited address doesn't fit the SPECIFIC inbound being attached, clear it just for that one so it gets a fresh, valid allocation instead, while other already-attached inbounds keep their existing values. WireGuard has no equivalent strict subnet requirement (allocateWireguardAddress can widen to a fallback pool for it), so this only ever constrains AmneziaWG targets. * Give WireGuard an explicit, admin-configurable subnet field WireGuard previously had no configurable subnet at all -- only an implicit one, either inferred from existing clients' own addresses (wireguardAllocationBase) or a hardcoded 10.0.0.0/24 fallback when none exist yet. AmneziaWG, by contrast, has always had a real server.subnetIp/subnetCidr field in its settings, editable in the UI. User request: give WireGuard the same treatment. Backend: explicitWireguardSubnetBase reads an optional subnetIp/ subnetCidr pair from the inbound's own settings JSON (mirroring AmneziaWG's defaultAmneziaWGSubnetBases). defaultWireguardClients checks it first; only when unset does it fall back to today's inference-from-existing-clients behavior, so an inbound saved before this field existed keeps working exactly as it always has. Frontend: subnetIp/subnetCidr added to WireguardInboundSettingsSchema and the inbound form (mirroring AmneziaWG's own field layout/labels), with a real default (10.0.0.0/24, the same value the backend already fell back to) seeded for newly created inbounds so the field starts populated and editable rather than blank. Translated across all 13 locales. This also structurally closes the class of bug fixed in 82cc69f5/291c47b3: with wg and awg subnets explicit and independently controllable, an admin who wants matching addresses across both protocols can configure them to actually agree, instead of one silently inheriting the other's incompatible range. * Split the client edit form's AllowedIPs into per-protocol fields A client attached to both WireGuard and AmneziaWG shared one AllowedIPs form field with a dynamically-switching label, so its two genuinely different addresses could never both be shown or edited correctly. Worse, Update/Create broadcast that one shared value to every attached wg/awg inbound with no subnet-fit check, so an ordinary edit save could silently overwrite one protocol's address with the other's -- the same bug class already fixed for Attach, but reachable from any client edit. model.Client gains an optional AllowedIPsByInbound map so a caller can send distinct values per inbound; Update/Create honor it and, when it's absent, clear a shared value that doesn't fit an AmneziaWG inbound's own subnet instead of writing it through. A new TunnelAllowedIPsByInbound read path feeds the real per-inbound address to the client edit form via GET, which now renders two separate, correctly-labeled fields whenever both protocols are attached (unchanged single dynamic field otherwise). * Regenerate openapi.json for the new allowedIPsByInbound field Follow-up to 878ee839: gen:zod (frontend/src/generated) was already regenerated and committed, but gen:api (frontend/public/openapi.json) wasn't, so CI's codegen drift check failed. * Fix build breakage from merging upstream main: Update() gained a limitHwid param Two of our own AllowedIPs tests (not present upstream, so the merge never flagged them as conflicting) still called the old 3-arg Update(inboundSvc, id, client) -- upstream's hardware-ID-limit feature added a required limitHwid parameter that every other caller in this package already passes. Also drop createDefaultInboundSettings from InboundsPage.tsx: the merge conflict resolution kept the import, but upstream's clone-payload refactor (buildClonePayload, inbound-clone.ts) already calls it internally now -- this file doesn't need it directly anymore. * Fix real bug: AmneziaWG clients rejected as "empty client ID" in 3 places Three switch statements on inbound.Protocol handle "wireguard" explicitly (checking client.PublicKey) but fall through to the default case for "amneziawg" (checking client.ID, which AmneziaWG clients never set -- they use PublicKey/Email like WireGuard, not the VMess/VLESS UUID field). This is what the 4 AllowedIPs tests were actually catching: UpdateInboundClient's newClientId derivation hit this same default branch, so every Update() on an AmneziaWG client returned "empty client ID" before ever reaching the AllowedIPs logic being tested. Fixed by adding "amneziawg" alongside "wireguard" in each switch: addInboundClient's per-client validation, UpdateInboundClient's newClientId derivation, and AddInbound's per-client validation (the third one wasn't hit by these tests, but has the identical bug -- creating a brand-new AmneziaWG inbound with a client attached would fail the same way). * refactor(amneziawg): rename Obfuscation20 to Obfuscation31, drop the dead mobile preset Mechanical rename ahead of the AmneziaWG 3.1 parameter work: the type, generator and prose all said 2.0, and the "mobile" generator preset was reachable only from its own test. No behavior change. * feat(amneziawg): AmneziaWG 3.1 obfuscation parameters (backend + generated schemas) Adds the 3.1 parameter surface to the inbound settings and both Go config emitters: I2-I5 signature packets, HeaderProtectionKey (base64 32-byte, shared server<->client), ContentPaddingAddition, the five handshake-timing randomization ranges (RekeyAfterTime/RekeyTimeout/RejectAfterTime/ KeepaliveTimeout/MaxHandshakeAttempts), and the RandomTrailers/ DisableCookies switches. Freshly generated sets fill everything except I2-I5 (matching Amnezia's own generator) with jittered ranges bracketing WireGuard's stock timing constants; every reject window starts >= 30s above the rekey window by construction. Empty fields stay off the wire, so blanking a field disables just that feature. Validation generalizes the H1-H4 range checker for the new uint32-range fields, requires min 1 on timers, cross-checks rekey-vs-reject, and demands a real 32-byte base64 header-protection key. The manager warns once per process when the installed awg tools predate 3.1 but an inbound uses 3.1 parameters (awg-quick rejects unknown keys with a generic error otherwise); apply still proceeds. Requires amneziawg-tools v3.1.20260812+ / module or amneziawg-go v3.1.20260814+ on the host. * feat(amneziawg): emit and randomize 3.1 parameters in the frontend Both client-config emitters (the vpn:// link builder and the clients-page .conf builder) now carry the 3.1 [Interface] lines in the same order as the Go emitters. The obfuscation randomizer moves out of InboundFormModal into a shared lib/xray/amneziawg-obfuscation.ts that also fills the new fields, and createDefaultAmneziawgInboundSettings switches from static values to that generator — a fresh inbound now really gets the unique fingerprint the docs promise instead of the same jc=5/jmin=10 set on every install. Schema parse-time defaults for the new fields stay ''/false on purpose: real values come only from the generator, so resaving an inbound never mutates its stored parameters. A new parity test pins the hand-written AmneziawgServerSchema to the generated ServerSettings key set, so a field added on one side can no longer silently vanish from configs. * feat(amneziawg): 3.1 form fields and translations Inbound form gains inputs for I2-I5, HeaderProtectionKey (filled by the existing obfuscation Regenerate button), ContentPaddingAddition, the five timing ranges, and the RandomTrailers/DisableCookies switches; the MTU input picks up the min=1 its schema already enforced. All 13 locales get the 19 new keys and drop the "2.0" branding from the s3/s4/i1 labels. * docs(amneziawg): document 3.1 parameters; install.sh kernel/version notes The AmneziaWG page's obfuscation section moves from the 2.0 to the 3.1 parameter set: table rows for I2-I5, HeaderProtectionKey, ContentPaddingAddition, the timing-randomization ranges and the RandomTrailers/DisableCookies switches, a requirements callout (tools v3.1.20260812+, module/awg-go v3.1.20260814+, Linux 6.7+ for the DKMS path), and a sample client .conf that matches what the panel actually emits (including the DNS defaults and PersistentKeepalive it always had). install.sh warns before a DKMS build on a pre-6.7 kernel and after any install that left pre-3.1 amneziawg-tools on PATH. Also updates the hosts API operation paths ({id} -> {groupId}) in the stale ru/zh/fa reference pages: syncing docs/public/openapi.json for the new AmneziaWG schema fields surfaced that rename, which had never been copied over, and the docs build fails on paths missing from the spec. * fix(amneziawg): reject control characters and canonicalize 3.1 range values Adversarial review of the 3.1 work surfaced a validation gap: base64.DecodeString silently ignores CR/LF, so a header-protection key that picked up a line wrap in transit decoded to a valid 32 bytes, passed validation, and was emitted verbatim into every client config — where the orphan second line breaks the import while the server (whose emitter strips control chars) keeps running with the correct key. The key and range validators now reject control characters outright. Also from the same review: range values are canonicalized on save ("110 - 140" -> "110-140", whitespace-only collapses to feature-off, closing a case where the server conf rendered an invalid blank-value line the client emitters omitted); the rekey/reject invariant is now enforced against WireGuard's 120s/180s defaults when only one side is set; and the structural fingerprint joins on "\n" instead of "|", which is a legal I1-I5 character and made adjacent free-text fields join-ambiguous. * fix(install): resolve latest release tag via web redirect to dodge API rate limits The non-interactive install smoke test resolved the release version through the unauthenticated GitHub API (api.github.com/.../releases/latest), which allows only 60 requests/hour per IP. The test installs twice in one run, and on shared CI runner IPs the second call gets rate-limited, returns no tag_name, and install.sh treats an empty version as fatal (exit 1) — the same "Failed to fetch x-ui version" real users hit behind CGNAT/shared addresses. resolve_latest_tag() now reads the tag from the github.com releases/latest web redirect (not subject to the API rate limit), falling back to the API only if the redirect yields nothing. Verified with the real deploy/test/smoke-noninteractive.sh (two installs, both green). * fix(amneziawg): three review findings on #6105, plus a comment trim 1. A peer's allowedIPs reached the generated .conf unvalidated and unsanitized, unlike email/publicKey/preSharedKey which normalizeAmneziaWGSettings already guards. A newline in an entry let a following "[Interface]" re-open the interface section, whose "PostUp = ..." awg-quick then runs as root on the next apply. Reproduced end to end against generateServerConfig. The save path now rejects and canonicalizes through normalizeWireguardAllowedIPs, and the render path sanitizes as a backstop for rows predating the validation (an upgrade, a restored backup, a direct DB edit). H1-H4 get the same render-time sanitize, and the two NIC name fields a plausibility check, since stripping control characters alone would still let a shell metacharacter into a root-executed PostUp line. 2. EgressPortForInbound is 63100 + inbound id, so an id past 2435 derives a port above 65535 -- and Xray rejects the whole generated config over one invalid port, taking every other protocol down with it. It now reports ok=false past the range, and both the Xray bridge and its TPROXY rules are skipped instead of emitting an impossible port. 3. The downloadable AmneziaWG .conf read ClientRecord.allowedIPs, a single shared column that holds the WireGuard address for an identity attached to both protocols -- the exact ambiguity tunnelAllowedIPs was added to resolve for the edit form. The info and QR modals already hydrate that field, so they now pass this inbound's own address to the builder. Also trims the comment blocks in the files touched here to the 2-line guidance in CLAUDE.md: internal/amneziawg alone carried 423 comment lines in over-long blocks against 118 for the comparable internal/mtproto, and is now at 110. Every non-obvious constraint is kept (the kernel S1/S2 rule, why PostDown is best-effort, why grep -c and not -q, why the fingerprints split three ways); the narration is gone. Two hot_diff.go comments pointed at an internal/amneziawgnet package and an injectAmneziawgnetSocks function that exist nowhere in the tree; the checks themselves are unchanged. * feat(logs): add an AmneziaWG log view to the overview The overview has an access-log view for Xray but nothing for AmneziaWG, so when a tunnel misbehaves there is no way to see it from the panel at all. A kernel tunnel logs no per-request lines, so the equivalent view is built from the two things it does expose: - Live per-peer activity from `awg show <iface> dump`, joined to the client email through the desired peer set: last handshake, endpoint, allowed IPs, cumulative transfer and online state, newest handshake first. - The panel's own AmneziaWG event lines (interface up/down, awg-quick failures, the pre-3.1 tools warning), which are what actually explain a peer being absent from the table. POST /panel/api/server/amneziawglogs/:count serves both, with the same count + filter contract GetXrayLogs uses, and the modal mirrors XrayLogModal's toolbar, auto-update, mobile cards and download. The action-bar button is gated on a new status.amneziawg.configured, which stays true while an inbound exists but its interface is down -- exactly when the event lines matter. Verified against a running panel: the endpoint returns the peer table and real event lines ("awg/awg-quick not found on PATH", "create config dir: permission denied"), and count and filter both narrow as documented. One of those lines surfaced a Debugf that had been rendering as "for inbound1:amneziawg:"; fixed here since it is now user-visible. * fix(amneziawg): stop double-counting a routed inbound's traffic injectAmneziawgEgress tags its Xray bridge with the AmneziaWG inbound's own tag, so the stock Routing page can target it. Xray therefore reports that bridge's bytes under the inbound's tag, and XrayTrafficJob feeds them to AddTraffic -- which accumulates -- on top of the same bytes AmneziaWGJob already reported from `awg show dump`. An inbound with routeThroughXray on counted roughly twice its real traffic, which also inflates the quota checks that read the same counters. The awg counters are the complete measure: every peer, whether or not TPROXY routed it, and the same wire bytes the per-client totals are built from, so they stay and the Xray rows are dropped. Per-client stats were never affected -- a dokodemo-door bridge has no per-user identity, so Xray emits no user>>>email rows for it. Filtering happens before every consumer, so the DB totals, the external traffic inform and the dashboard's live speed all read one source per inbound. The set of bridge tags now comes from a predicate shared with injectAmneziawgEgress itself, with a test that pins the two together -- naming one tag too few doubles the traffic again, one too many makes real traffic vanish. * fix(amneziawg): align the three .conf emitters on one peer field order The panel builds an AmneziaWG client .conf in three independent places, and they disagreed: buildAmneziaWGClientConfig put PresharedKey right after PublicKey (wg-quick(8)'s own order, and what both WireGuard emitters on the clients side already use), while genAmneziaWGConfig and the Go amneziaWGConfigText put it after Endpoint. A user comparing a subscription link against a downloaded .conf sees the difference immediately, and the generators are exactly the kind of parallel implementation CLAUDE.md warns about drifting. Moves the two outliers onto the wg-quick order. Also drops the stray trailing newline that only appeared when PersistentKeepalive was set, so a config now always ends on its last set field whichever that is -- the same shape all three emitters produce for the same client. Parsing is unaffected either way (the format is order-insensitive, and the AmneziaVPN app reads it as a flat key-value bag), so this changes only the rendered text. Adds a test on each side that pins the peer block's field order, since nothing previously asserted it. * refactor(amneziawg): switch to the embedded amneziawg-go/gVisor architecture Replaces the kernel-module (DKMS) + awg-quick + TPROXY backend with the fork's own embedded design: amneziawg-go runs in-process over a userspace gVisor netstack, and each peer's decapsulated traffic relays into its own loopback Xray SOCKS5 inbound, so Xray's native stats/sniffing/routing work for free instead of through hand-rolled bridges. No kernel module, no DKMS, no Secure Boot conflicts, works the same in a container as on bare metal. - internal/amneziawgnet: new package (Device/UAPI, gVisor netstack, TCP/UDP forwarding, SOCKS5 relay, peer identity, IPv6 host-alias egress identity, per-client port-forwarding) - amneziawg-go v3.1.20260814 + gvisor. - internal/amneziawg: keep the reusable protocol-shape types/validation (Instance/Peer/Obfuscation, InstanceFromInbound); drop the OS-shellout half (awg-quick, TPROXY policy routing, NDP proxy, peer-stats parsing). - internal/web/service: rewire the 5 integration points (job, runtime, client-apply, web shutdown, xray config) from the old manager to the new one; the AmneziaWG log view is rebuilt on the embedded Device's own UAPI dump (extended to carry endpoint/AllowedIPs) instead of `awg show dump`. - install.sh: drop DKMS/ndppd/TPROXY/Secure-Boot installer code (~250 lines) - an entire recurring class of installer fragility goes away. - frontend: drop the now-meaningless routeThroughXray toggle (the relay is always on); keep the field in the Zod schema, unexposed, so it isn't silently stripped from stored settings on next save - two regression tests deliberately depend on the Go struct still carrying it. - docs/i18n: rewrite amneziawg.mdx for the new architecture; drop the dead routeThroughXray translation keys across all 13 locales. Real production throughput (embedded core datapath, isolated bench, same box the kernel-module path was measured on): ~296 Mbit/s up, ~640 Mbit/s down, vs. 414.69 MB/s (~3.3 Gbit/s) for the kernel module on the same hardware - a real gap, tempered by this being single-stream/no-SOCKS5-hop and most VPN traffic being latency-bound rather than throughput-saturating. * fix(amneziawg): restore the branch's own Obfuscation31 shape + 2 CodeQL findings The previous push's wholesale-copy of types.go/params.go from the fork's main branch pulled in that branch's own independent (and incompatible) naming for the same AWG 3.1 feature set: Obfuscation20/GenerateObfuscation20 instead of this branch's already-shipped Obfuscation31/GenerateObfuscation31, and a missing CanonicalizeUintRange -- broke every Go CI job (the whole matrix fails to compile when any one package doesn't, which is why govulncheck/ golangci/postgres-durable-first/race all failed identically, not just go-test). Restores params.go/params_test.go verbatim from this branch's own last commit (a strict superset of validation: it already cross-checks rekey vs. reject timing windows, which the copied version never did) and folds the 3.0/3.1 fields (HeaderProtectionKey, ContentPaddingAddition, the 5 timing fields, RandomTrailers/DisableCookies) into Obfuscation31 itself, matching the original struct exactly instead of as separate top-level Instance fields. instance.go, the two amneziawgnet call sites, and 7 amneziawgnet test files updated to match. Also drops the one test (sanitizeConfigValue) that only ever served the retired kernel-module .conf writer -- correctly not ported, so the test testing it shouldn't have been copied either. Also fixes 2 CodeQL findings the same push surfaced: a clamped uint64->int64 conversion for the new log view's live byte counters (server.go), and an unneeded len+len sum feeding a slice pre-size in the v6-egress outbound merge (xray.go) -- append already grows correctly without it. * chore(amneziawg): regenerate frontend schemas for updated doc comments npm run gen was missed after the previous commit's types.go doc-comment edits (Obfuscation20 -> Obfuscation31, ValidateHeaderProtection -> ValidateObfuscation in the prose) -- openapigen bakes those comments into the generated schema's description field, so the committed frontend/src/generated/schemas.ts and openapi.json still had the old wording. codegen's git-diff-exit-code check caught it correctly. * fix(amneziawg): narrow 2 test fixtures that collided with MaxForwardedPorts TestCheckForwardedPortsConflict_CollidesWithEnabledInboundPort and ..._NoCollisionWhenPortsDontOverlap used "8000-8100"/"9000-9100" as their ForwardedPorts fixture -- 101 ports each, one over MaxForwardedPorts (100). The cap check (checkForwardedPortsConflict, added this session alongside the SOCKS-phantom-port check) fires first, so both tests got "more than 100 forwarded ports" instead of ever reaching the collision logic they're actually testing. The cap itself has its own dedicated boundary test already; these two just needed a narrower range that still covers/misses port 8080 as intended -- 8075-8085 and 9075-9085, 11 ports each. * fix(amneziawg): checkAmneziawgnetSocksConflict had no receiver in its new home My merge-conflict resolution kept this as a method call (s.checkAmneziawgnetSocksConflict) inside checkPortConflictTx, a plain function with no *InboundService receiver -- upstream's #6225 fix moved the port-conflict check out of the (s *InboundService) method and into this new tx-scoped free function, and I didn't notice the call site needed to change shape too. CI caught it immediately (undefined: s); nothing in this specific package can be locally verified past internal/database's own unrelated, pre-existing CGO build issue on this dev machine. Since the signature had to change either way, folded in the fix already flagged as a separate follow-up: checkAmneziawgnetSocksConflict now takes the caller's db handle instead of fetching its own via database.GetDB(), so it actually runs inside the same serialized transaction #6225 introduced -- previously it sat right next to that race fix without benefiting from it. * fix: address the review findings on the embedded AmneziaWG PR 5 blocking findings: - Floor S3/S4 at 12 in both obfuscation generators (Go and frontend) and reject a hand-edited value below that when HeaderProtectionKey is set -- IpcSet requires it, and ~39% of previously-generated sets violated it silently. - Guard PrivateKey/PrimaryDNS/SecondaryDNS/remark against newline injection in the AmneziaWG .conf builder (both the Go subscription-link path and the frontend downloadable-config path) -- unguarded, any of them could inject an arbitrary config line into a subscriber's client. - Bound the derived AmneziaWG SOCKS relay port to <= 65535 once an inbound's id is known, and check the reverse direction (does the relay port collide with an existing inbound's port) on both create and update -- previously only port -> relay collisions were checked, not relay -> port. - Gate injectAmneziawgV6Egress on the same V6AliasesActive predicate desiredV6Aliases already uses, so the two can't disagree about whether a peer's IPv6 identity is actually active at the OS level. 2 minor findings: - Fix the forwarded-ports cap check's off-by-one (a spec covering exactly the cap was rejected as if it were over it). - Correct docker-compose.yml's stale comment describing the retired DKMS/kernel-module architecture. * chore: retrigger CI build (armv5) failed on a transient Go module proxy network error (INTERNAL_ERROR stream reset on sagernet/sing), unrelated to this PR's changes. * docs: fix doc comments still describing the retired DKMS/awg-quick design A few doc comments (and one illustrative test log line) survived the embedded-architecture cutover unchanged and now contradict the code they sit next to: - internal/amneziawg/types.go's package comment claimed this package still owns a Manager that reconciles OS-level interfaces via awg-quick/DKMS -- that Manager was removed; the reconcile loop lives in internal/amneziawgnet now, and this package is protocol-shape-only. - internal/amneziawg/params.go's ValidateObfuscation/ValidateConfigValue comments cited "awg-quick up" / "awg-quick executes as root" as the reason to validate -- the server itself never calls awg-quick in this architecture; the same value still reaches a real rendered .conf that a client app or an admin's own awg-quick CLI applies downstream, so the validation is still warranted, just for a different consumer. Mirrored the same fix in inbound_amneziawg.go's matching comment and its test's comment. - internal/amneziawgnet/manager.go's Manager doc comments (x3) pointed readers at "internal/amneziawg.Manager" for comparison -- that type no longer exists in this diff at all. Repointed at internal/mtproto.Manager, the pattern this was actually modeled on and the one that's still real. - Swapped one test's illustrative "awg-quick up awg2 failed" log line for a message shaped like this architecture's actual amneziawgnet logging, so a reader skimming the test doesn't wonder whether the server still shells out to awg-quick. No behavior change. * fix(docs): re-run codegen for xray-settings.mdx after conflict merge The automated conflict-resolution hand-merge for this generated file was content-correct but didn't byte-match a real regen (different YAML long-string folding style). Re-ran npm run gen + docs' gen:api and kept that canonical output instead. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> * fix(amneziawg): drop the dead access-log email backfill amneziawgEmailIndex keyed peers by "<tag>|<tunnel IP>", a scheme built for the retired TPROXY bridge where the peer's decapsulated tunnel address survived as the access log's from-address. The embedded architecture relays through a loopback SOCKS5 dial, so every AmneziaWG log line's from-address is 127.0.0.1:<ephemeral> and the lookup could never match: the index was rebuilt on every log view just to miss. Remove the index, its GetXrayLogs wiring and its test. If per-line emails are wanted back, the relay would have to publish a local-port->email registry for the viewer to resolve loopback sources. * fix(api): generate AmneziaWGLogs/PeerActivity schemas instead of hand-writing them The amneziawglogs endpoint's response structs were missing from openapigen's StructAllow, so they were silently absent from every generated schema/example, the endpoints.ts entry carried a hand-written response, and AmneziaWGLogModal.tsx duplicated the shapes as local interfaces - the exact drift the allowlist rule exists to prevent. Allowlist both structs with example tags, point the endpoint at the generated schema, import the generated types in the modal, and sync docs/public/openapi.json. * chore(amneziawg): drop the unreferenced quiccapture package Nothing imports internal/amneziawg/quiccapture and no route exposes it; its package doc justifies the code as a port of frontend/src/lib/xray/i1Generators.ts, which does not exist in this repository, and promises an API round-trip that also does not exist. 1,110 lines of unreachable code with misleading provenance claims. Revert this commit to bring the package back when the live-capture I1 feature and its frontend counterpart actually land. * fix(clients): re-run cross-inbound conflict checks on the serialized writer The new client-level checks - cross-inbound AllowedIPs collisions and AmneziaWG forwardedPorts conflicts - read a fresh DB snapshot, decide, and only then enter runSerializedTx, while lockInbound only serializes writers on the SAME inbound. Two concurrent client creates on two different tunnel inbounds both passed the read and both committed, yielding two peers with one address: the exact check-then-claim race 81cfd857 (#6225) closed for AddInbound, which this PR's own checkAmneziawgnetSocksReverseConflict already cites. Keep the pre-tx pass for fail-fast UX and re-validate inside the transaction, where the single writer makes the answer authoritative. The race test drives two goroutines at two inbounds and demands exactly one winner; it fails with committed=2 when the in-tx re-check is removed. * fix(amneziawg): hot-apply depletion disables like mtproto does applyTrafficMutationBatch special-cases MTProto so a quota/expiry depletion cuts the sidecar immediately, but AmneziaWG fell through to runtime AddUser/RemoveUser - explicit no-ops for this protocol - so a depleted peer kept tunneling until the next 10s reconcile tick. Route it through applyLocalAmneziaWG, whose own contract (re-read committed settings, filter depleted clients, push to the interface) is exactly this case; the comment claiming it mirrors applyLocalMtproto is now true for the depletion path too. * fix(amneziawg): persist cleared DNS fields instead of resurrecting defaults PrimaryDNS/SecondaryDNS marshaled with omitempty, so clearing them persisted settings with no key at all - and the frontend re-parses stored settings through a Zod schema whose .default('8.8.8.8') / .default('8.8.4.4') fire on missing keys, silently repopulating the form on every load and re-persisting the defaults on the next save. Blank is a documented, meaningful state (no DNS line in client configs); drop omitempty so a cleared value survives the round-trip. The regression test normalizes a server block with cleared DNS and fails when the keys are dropped. * fix(amneziawg): accept cleared numeric obfuscation/subnet fields in the form AntD InputNumber emits null when cleared, Zod .default() only replaces undefined, and unlike wireguard.ts - whose optionalClearedInt comment documents exactly this failure mode - the AmneziaWG schema declared subnetCidr and jc/jmin/jmax/s1-s4 as bare z.number() defaults. Clearing any of the eight fields made safeParse reject the null and block the save until the user retyped a value. Absorb null into undefined while keeping each field's schema default, so a cleared field refills its documented default and legacy blobs with absent keys behave as before. * fix(amneziawg): guard the third .conf emitter against newline injection The review-round fix added the newline guard to amneziaWGConfigText (Go) and buildAmneziaWGClientConfig, but genAmneziaWGConfig in inbound-link.ts - the third of the three emitters its own comment says must not drift - still rendered privateKey/primaryDns/secondaryDns/remark unescaped, so a newline there injected a config line (e.g. a rogue PostUp) into the inbound form's downloaded .conf. Add the same guard, plus the regression tests the original fix shipped without: all four fields on the Go and both frontend emitters go red if any guard is removed. * test(amneziawg): pin the S3/S4 floors the TS drift guard claims to mirror The test's docstring says it mirrors internal/amneziawg/params_test.go, but it asserted S3>=8/S4>=4 while the Go test and both generators pin 12/12 - the floor ValidateObfuscation enforces whenever a header protection key is set, which this generator always sets. A regression narrowing the TS floors into 8-11/4-11 would have passed the drift guard and produced configs the backend rejects on save. * docs: restore the pia repo-map entry and document the AmneziaWG subsystem Merging main dropped CLAUDE.md's internal/pia/ bullet (added by #6272) while resolving the repo-map conflict - the package itself is untouched. Restore it, add the missing map entries for the two packages this branch introduces (internal/amneziawg/, internal/amneziawgnet/), bump the cron count, and give amneziawg_job its row in architecture.md's 5.4 table. * chore(amneziawg): correct comments stranded by the architecture pivot ae77c7e9's cutover to the embedded gVisor path deleted the kernel-module code but left several comments describing it in the present tense: hot_diff.go cited the removed service.amneziawgEgressStreamSettings and wrongly claimed AmneziaWG is the only sockopt.tproxy source (tunnel's TProxy mode is the live one the guard protects), socks_config.go pointed at the deleted EgressBasePort/EgressPortForInbound, manager.go referred to the deleted Manager and its fingerprinting as live code, web.go's cron registration claimed the job scrapes traffic (its own doc says it does not), and types.go capped ContentPaddingAddition at uint16 when validation and upstream both use uint32. * style(lint): satisfy gofumpt/goimports so make verify is green json_service.go's two 'Tag: "proxy"}' literals came in with main's own cc245a90 formatting commit and fail the repo's gofumpt gate for everyone; the import grouping in inbound_amneziawg.go is from the serialized-writer fix on this branch. --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-08-23 17:41:15 -07:00
"outboundTag": tag,
})
}
}
if len(newOutbounds) == 0 {
return
}
merged := make([]any, 0, len(existingOutbounds))
merged = append(merged, existingOutbounds...)
merged = append(merged, newOutbounds...)
combined, err := json.MarshalIndent(merged, "", " ")
if err != nil {
logger.Warning("amneziawg v6 egress: failed to rebuild outbounds section, skipping injection:", err)
return
}
cfg.OutboundConfigs = json_util.RawMessage(combined)
routing["rules"] = append(newRules, rules...)
newRouting, err := json.Marshal(routing)
if err != nil {
logger.Warning("amneziawg v6 egress: failed to rebuild routing section, skipping injection:", err)
return
}
cfg.RouterConfig = json_util.RawMessage(newRouting)
}
feat: add support for subscription-based outbounds with auto-update (#5037) * feat: add support for subscription-based outbounds with auto-update - New OutboundSubscription model (full support on both SQLite and PostgreSQL) - Go subscription link parser (vmess/vless/trojan/ss/hysteria2/wireguard) matching frontend behavior - Stable tag assignment across refreshes (designed for balancer + routing use) - Runtime merge of subscription outbounds into Xray config (additive only) - Full CRUD + manual refresh + preview API - Background auto-update job (per-subscription interval) - Frontend management UI in Outbounds tab (Subscriptions drawer) + tag integration in balancers/routing rules - Proper dual-database support including CLI migration path Review & hardening notes: - Fixed merge logic bug that could drop manual outbounds - Added SSRF/private-IP protection on subscription URLs using SanitizePublicHTTPURL - Improved update interval UX (hours + minutes) - Auto-fetch on first subscription creation - Added detailed comments on tag stability strategy and balancer implications when servers are added/removed/rotated - Updated migrationModels() for CLI migrate-db support * fix: resolve frontend lint/type errors and Go build break Frontend (eslint + tsc clean): - Destructure subscriptionOutboundTags prop in RoutingTab and BalancersTab. It was declared in the interface and used in useMemo but never destructured, so it resolved as an unresolved global (react-hooks warning + tsc "Cannot find name"). The prop is passed by XrayPage, so the feature was silently inert. - OutboundsTab: remove unused useEffect import, add an OutboundSub type to replace any[] state and the any/any table render signature, type the subscriptionOutbounds cast, and replace unused catch (e) bindings with parameter-less catch. Also type HttpUtil.post as OutboundSub so r.obj?.id type-checks. Backend (go build clean): - outbound_subscription_job: websocket.MessageTypeXray is undefined; use the existing MessageTypeOutbounds since the job refreshes outbound subscriptions. * fix(xray): make outbound subscription creation work end-to-end - Correct API paths from /panel/xray/outbound-subs to /panel/api/xray/outbound-subs. The controller is mounted under /panel/api, so the old paths hit the SPA page route (GET-only) and 404'd on POST. - Send the create-subscription body as a plain object instead of URLSearchParams. The axios request interceptor serializes bodies with qs.stringify, which can't read URLSearchParams' internal storage and produced an empty body, so the backend rejected it with "subscription URL is required". - Use message.useMessage() + context holder instead of the static antd message API (resolves the "Static function can not consume context" warning), matching XrayPage's pattern. - Migrate the subscriptions Drawer to antd v6 props: width -> size, destroyOnClose -> destroyOnHidden, and Space direction -> orientation. * feat(xray): show traffic/test for subscription outbounds; harden + test the feature Display (the reported issue): - Replace the flat read-only pills with a proper read-only table (desktop) and cards (mobile) in a new SubscriptionOutbounds component, showing Address, Protocol, Traffic (matched by tag — already collected by Xray), and a Test button with Latency. No edit/delete/move (read-only). - Test subscription outbounds via the existing /testOutbound endpoint, with results keyed by tag (subscriptionTestStates + testSubscriptionOutbound in useXraySetting, wired through XrayPage). Generalize isTesting/testResult to a string|number key so the same helpers serve index- and tag-keyed states. i18n: - Replace all hardcoded English subscription strings with t() calls and add pages.xray.outboundSub.* keys to en-US.json (other locales fall back). Backend hardening + tests: - xray.go: drop the tautological `subSvc != nil` check. - outbound_subscription: re-validate every redirect hop against private/ internal addresses (CheckRedirect) and cap the redirect chain, closing an SSRF gap where only the initial host was checked. - Extract assignStableTags as a pure function and add unit tests for tag stability and SSRF rejection (the feature previously had no tests). Misc: - gofmt util/link/outbound.go (it was not gofmt-clean). * fix(xray): make outbound-subs feature pass CI (test compile, route docs, openapi) - outbound_test.go: remove unused `inner`/`lines` variables that broke the `util/link` test build (declared and not used). - Document the 7 outbound-subscription routes in endpoints.ts (list, create, update, delete, del alias, refresh, parse) so TestAPIRoutesDocumented passes. - Regenerate frontend/public/openapi.json (npm run gen) to include the new endpoints, satisfying the codegen freshness check. * feat(xray): per-subscription allow-private, gap-filled tags, UI tweaks, delete refresh Backend: - Add a per-subscription AllowPrivate flag (default off). Create/Update/refresh and the redirect check sanitize the URL with it, so localhost/LAN sources work only when explicitly opted in; the SSRF guard still blocks private targets by default. Controller reads the allowPrivate form field on create/update/parse. - Default outbound tag prefix now uses the smallest free "subN-" number instead of the auto-increment id, so deleting a subscription frees its number for reuse (a fresh start gives sub1) while staying stable per subscription. Extracted a pure defaultPrefixNumber() with unit tests. - deleteOutboundSub now signals SetToNeedRestart so xray drops the outbounds. Frontend: - "Allow private address" toggle in the add form (sends allowPrivate). - Delete now refreshes the xray view immediately (no manual page reload). - Subscriptions manager opens as a centered Modal instead of a right-side Drawer. - Move Outbounds to a top-level sidebar item under Nodes (out of Xray Configs). - Collapse WARP/NordVPN into a "more" dropdown. - Document the allowPrivate param in endpoints.ts. * i18n(xray): translate outbound-subscription UI into all locales - Translate the pages.xray.outboundSub.* strings (and allowPrivate label/hint) into all 12 non-English locales, matching each file's existing terminology. - Remove the unused outboundSub.add ("Add subscription") key from every locale. * feat(xray): subscription manager — edit, reorder/priority, status, preview, refresh-all Backend: - Per-subscription Priority + Prepend: subscriptions are ordered by Priority and placed before (Prepend) or after the manual template outbounds in the merge, so a subscription server can become the default. New Move(up/down) endpoint re-normalizes priorities; merge split into prepend/template/append. - List now returns a derived OutboundCount and orders by priority, and strips the heavy LastFetchedOutbounds/LinkIdentities blobs from the list payload. - Create/Update accept the prepend flag; new subs append at the end of priority. Frontend (Outbound Subscriptions modal): - Edit existing subscriptions (reuses the form + Update endpoint). - Inline enable/disable Switch, Status column (OK / error tooltip), Outbounds count column, per-row refresh spinner, "Refresh all" button. - Reorder (move up/down) controls + a "Before manual outbounds" toggle. - Preview button: fetch+parse a URL via /parse without saving. - Document the move route + prepend param in endpoints.ts; regenerate openapi.json. * i18n(xray): translate new subscription-manager strings into all locales Add the prepend/prependHint, preview/previewEmpty, refreshAll, statusOk and toastUpdated keys to all 12 non-English locales, matching each file's terminology. --------- Co-authored-by: MHSanaei <ho3ein.sanaei@gmail.com>
2026-06-08 09:09:53 -07:00
// mergeSubscriptionOutbounds appends the subscription outbounds to the
// OutboundConfigs array of the xray config. It works on the already-unmarshaled
// template so that manually configured outbounds are never overwritten.
//
// Safety: if we cannot parse the template's outbounds array, we leave
// OutboundConfigs exactly as it came from the template (we do not inject
// subscription outbounds). This prevents us from accidentally dropping the
// user's manually configured outbounds when the template is in a weird state.
func mergeSubscriptionOutbounds(cfg *xray.Config, prepend, appendList []any) {
if len(prepend) == 0 && len(appendList) == 0 {
return
}
var templateOutbounds []any
if len(cfg.OutboundConfigs) > 0 {
if err := json.Unmarshal(cfg.OutboundConfigs, &templateOutbounds); err != nil {
// Corrupt template outbounds — do not touch the field at all.
// The user will see problems on Xray start / next save.
return
}
}
feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) (#5506) * feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) xray-core v26.6.22 (PR #6258) renamed the XHTTP session config keys sessionPlacement/sessionKey to sessionIDPlacement/sessionIDKey (no fallback kept in core) and added sessionIDTable (predefined charset name or literal ASCII) and sessionIDLength (range, e.g. 16-32, lower bound > 0). Panel changes: - Schema (xhttp.ts): rename the two keys, add sessionIDTable/sessionIDLength, and a z.preprocess that lifts legacy keys off stored configs so an upgraded panel never silently drops a saved session setting. - Wire normalize + share-link build/parse: rename keys, emit the two new fields, and accept legacy sessionPlacement/sessionKey from old share links. - Inbound + outbound XHTTP forms: rename field paths, add a sessionIDTable autocomplete (9 predefined tables + free ASCII) and a sessionIDLength range input shown only when a table is set, with light client validation (ASCII table, length min > 0; xray enforces the room-size minimum server-side). - Subscription (service.go) and Clash (clash_service.go) builders: emit the renamed + new keys, with a legacy fallback for not-yet-resaved inbounds. - Locales: add sessionIDTable/sessionIDLength labels + hints in all 13 files. Two sibling v26.6.22 XHTTP commits need no panel change and are covered by the core bump alone: #6332 (XHTTP/3 closes QUIC/UDP) and #6320 (udpHop honors the existing dialerProxy). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(xhttp): add Session ID Table to inbound form-blocks snapshot The new sessionIDTable input renders by default in the inbound XHTTP form, so its label joins the field-structure snapshot. sessionIDLength stays conditional (only shown when a table is set), so it does not appear here. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xhttp): migrate legacy session keys in the running xray config The Zod preprocess plus the subscription/Clash fallbacks only covered the panel UI and share-link output. The config handed to the running xray-core process is built from the raw stored streamSettings in GetXrayConfig, which did not rewrite the renamed XHTTP session keys — so a pre-upgrade inbound (or template outbound) stored with a non-default sessionPlacement was emitted unchanged and dropped by xray-core v26.6.22, until the admin re-saved it. Lift sessionPlacement/sessionKey onto sessionIDPlacement/sessionIDKey at config-generation time, in the existing inbound stream-rewrite block (next to the tls/reality/externalProxy handling) and across template outbounds. The lift is idempotent and leaves unchanged configs byte-identical so the hot-reload diff never sees a spurious change. Also tighten validateSessionIDLength to reject an inverted range (e.g. 32-16) in addition to the existing lower-bound > 0 check. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xray): avoid summed-capacity allocation in mergeSubscriptionOutbounds CodeQL go/allocation-size-overflow flagged the pre-sized make() whose capacity was a sum of three slice lengths. Grow the slice via append on a nil slice instead; same result, no overflow-prone capacity expression.
2026-06-23 08:38:16 -07:00
var merged []any
feat: add support for subscription-based outbounds with auto-update (#5037) * feat: add support for subscription-based outbounds with auto-update - New OutboundSubscription model (full support on both SQLite and PostgreSQL) - Go subscription link parser (vmess/vless/trojan/ss/hysteria2/wireguard) matching frontend behavior - Stable tag assignment across refreshes (designed for balancer + routing use) - Runtime merge of subscription outbounds into Xray config (additive only) - Full CRUD + manual refresh + preview API - Background auto-update job (per-subscription interval) - Frontend management UI in Outbounds tab (Subscriptions drawer) + tag integration in balancers/routing rules - Proper dual-database support including CLI migration path Review & hardening notes: - Fixed merge logic bug that could drop manual outbounds - Added SSRF/private-IP protection on subscription URLs using SanitizePublicHTTPURL - Improved update interval UX (hours + minutes) - Auto-fetch on first subscription creation - Added detailed comments on tag stability strategy and balancer implications when servers are added/removed/rotated - Updated migrationModels() for CLI migrate-db support * fix: resolve frontend lint/type errors and Go build break Frontend (eslint + tsc clean): - Destructure subscriptionOutboundTags prop in RoutingTab and BalancersTab. It was declared in the interface and used in useMemo but never destructured, so it resolved as an unresolved global (react-hooks warning + tsc "Cannot find name"). The prop is passed by XrayPage, so the feature was silently inert. - OutboundsTab: remove unused useEffect import, add an OutboundSub type to replace any[] state and the any/any table render signature, type the subscriptionOutbounds cast, and replace unused catch (e) bindings with parameter-less catch. Also type HttpUtil.post as OutboundSub so r.obj?.id type-checks. Backend (go build clean): - outbound_subscription_job: websocket.MessageTypeXray is undefined; use the existing MessageTypeOutbounds since the job refreshes outbound subscriptions. * fix(xray): make outbound subscription creation work end-to-end - Correct API paths from /panel/xray/outbound-subs to /panel/api/xray/outbound-subs. The controller is mounted under /panel/api, so the old paths hit the SPA page route (GET-only) and 404'd on POST. - Send the create-subscription body as a plain object instead of URLSearchParams. The axios request interceptor serializes bodies with qs.stringify, which can't read URLSearchParams' internal storage and produced an empty body, so the backend rejected it with "subscription URL is required". - Use message.useMessage() + context holder instead of the static antd message API (resolves the "Static function can not consume context" warning), matching XrayPage's pattern. - Migrate the subscriptions Drawer to antd v6 props: width -> size, destroyOnClose -> destroyOnHidden, and Space direction -> orientation. * feat(xray): show traffic/test for subscription outbounds; harden + test the feature Display (the reported issue): - Replace the flat read-only pills with a proper read-only table (desktop) and cards (mobile) in a new SubscriptionOutbounds component, showing Address, Protocol, Traffic (matched by tag — already collected by Xray), and a Test button with Latency. No edit/delete/move (read-only). - Test subscription outbounds via the existing /testOutbound endpoint, with results keyed by tag (subscriptionTestStates + testSubscriptionOutbound in useXraySetting, wired through XrayPage). Generalize isTesting/testResult to a string|number key so the same helpers serve index- and tag-keyed states. i18n: - Replace all hardcoded English subscription strings with t() calls and add pages.xray.outboundSub.* keys to en-US.json (other locales fall back). Backend hardening + tests: - xray.go: drop the tautological `subSvc != nil` check. - outbound_subscription: re-validate every redirect hop against private/ internal addresses (CheckRedirect) and cap the redirect chain, closing an SSRF gap where only the initial host was checked. - Extract assignStableTags as a pure function and add unit tests for tag stability and SSRF rejection (the feature previously had no tests). Misc: - gofmt util/link/outbound.go (it was not gofmt-clean). * fix(xray): make outbound-subs feature pass CI (test compile, route docs, openapi) - outbound_test.go: remove unused `inner`/`lines` variables that broke the `util/link` test build (declared and not used). - Document the 7 outbound-subscription routes in endpoints.ts (list, create, update, delete, del alias, refresh, parse) so TestAPIRoutesDocumented passes. - Regenerate frontend/public/openapi.json (npm run gen) to include the new endpoints, satisfying the codegen freshness check. * feat(xray): per-subscription allow-private, gap-filled tags, UI tweaks, delete refresh Backend: - Add a per-subscription AllowPrivate flag (default off). Create/Update/refresh and the redirect check sanitize the URL with it, so localhost/LAN sources work only when explicitly opted in; the SSRF guard still blocks private targets by default. Controller reads the allowPrivate form field on create/update/parse. - Default outbound tag prefix now uses the smallest free "subN-" number instead of the auto-increment id, so deleting a subscription frees its number for reuse (a fresh start gives sub1) while staying stable per subscription. Extracted a pure defaultPrefixNumber() with unit tests. - deleteOutboundSub now signals SetToNeedRestart so xray drops the outbounds. Frontend: - "Allow private address" toggle in the add form (sends allowPrivate). - Delete now refreshes the xray view immediately (no manual page reload). - Subscriptions manager opens as a centered Modal instead of a right-side Drawer. - Move Outbounds to a top-level sidebar item under Nodes (out of Xray Configs). - Collapse WARP/NordVPN into a "more" dropdown. - Document the allowPrivate param in endpoints.ts. * i18n(xray): translate outbound-subscription UI into all locales - Translate the pages.xray.outboundSub.* strings (and allowPrivate label/hint) into all 12 non-English locales, matching each file's existing terminology. - Remove the unused outboundSub.add ("Add subscription") key from every locale. * feat(xray): subscription manager — edit, reorder/priority, status, preview, refresh-all Backend: - Per-subscription Priority + Prepend: subscriptions are ordered by Priority and placed before (Prepend) or after the manual template outbounds in the merge, so a subscription server can become the default. New Move(up/down) endpoint re-normalizes priorities; merge split into prepend/template/append. - List now returns a derived OutboundCount and orders by priority, and strips the heavy LastFetchedOutbounds/LinkIdentities blobs from the list payload. - Create/Update accept the prepend flag; new subs append at the end of priority. Frontend (Outbound Subscriptions modal): - Edit existing subscriptions (reuses the form + Update endpoint). - Inline enable/disable Switch, Status column (OK / error tooltip), Outbounds count column, per-row refresh spinner, "Refresh all" button. - Reorder (move up/down) controls + a "Before manual outbounds" toggle. - Preview button: fetch+parse a URL via /parse without saving. - Document the move route + prepend param in endpoints.ts; regenerate openapi.json. * i18n(xray): translate new subscription-manager strings into all locales Add the prepend/prependHint, preview/previewEmpty, refreshAll, statusOk and toastUpdated keys to all 12 non-English locales, matching each file's terminology. --------- Co-authored-by: MHSanaei <ho3ein.sanaei@gmail.com>
2026-06-08 09:09:53 -07:00
merged = append(merged, prepend...)
merged = append(merged, templateOutbounds...)
merged = append(merged, appendList...)
combined, err := json.MarshalIndent(merged, "", " ")
if err != nil {
return
}
cfg.OutboundConfigs = json_util.RawMessage(combined)
}
// ensureAPIServices guarantees the gRPC services the panel depends on are
// listed in the generated config's api block: HandlerService and StatsService
// have always been required for inbound/user management and traffic polling,
// and RoutingService enables hot routing reload on templates saved before it
// was added to the default template. The stored template itself is not
// modified — only the generated runtime config.
func ensureAPIServices(api json_util.RawMessage) json_util.RawMessage {
if len(api) == 0 {
// No api block means the panel's API integration is deliberately
// disabled; don't resurrect it behind the user's back.
return api
}
var parsed map[string]any
if err := json.Unmarshal(api, &parsed); err != nil {
return api
}
services, _ := parsed["services"].([]any)
have := make(map[string]bool, len(services))
for _, svc := range services {
if name, ok := svc.(string); ok {
have[name] = true
}
}
added := false
for _, name := range []string{"HandlerService", "StatsService", "RoutingService"} {
if !have[name] {
services = append(services, name)
added = true
}
}
if !added {
return api
}
parsed["services"] = services
out, err := json.Marshal(parsed)
if err != nil {
return api
}
return out
}
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
// ensureStatsPolicy guarantees every policy level in the generated config has
// statsUserOnline enabled, so the core tracks per-email online IPs for the
// panel's online view and access-log-free IP limiting. Generated clients carry
// no explicit level, so level "0" is created when absent. The flag is panel
// infrastructure and is forced on even over an explicit false in the template,
// same as the api services above. An entirely missing or unparsable policy
// block is left alone; the stored template itself is never modified — only the
// generated runtime config.
func ensureStatsPolicy(policy json_util.RawMessage) json_util.RawMessage {
if len(policy) == 0 {
return policy
}
var parsed map[string]any
if err := json.Unmarshal(policy, &parsed); err != nil {
return policy
}
levels, _ := parsed["levels"].(map[string]any)
if levels == nil {
levels = make(map[string]any)
}
if _, ok := levels["0"]; !ok {
levels["0"] = map[string]any{}
}
changed := false
for _, raw := range levels {
level, ok := raw.(map[string]any)
if !ok {
continue
}
if enabled, ok := level["statsUserOnline"].(bool); !ok || !enabled {
level["statsUserOnline"] = true
changed = true
}
}
if !changed {
return policy
}
parsed["levels"] = levels
out, err := json.Marshal(parsed)
if err != nil {
return policy
}
return out
}
// caseVariantKeys returns every key of parsed that equals want ignoring case,
// lowest first so the fold is deterministic when several variants are present.
func caseVariantKeys(parsed map[string]any, want string) []string {
var keys []string
for key := range parsed {
if strings.EqualFold(key, want) {
keys = append(keys, key)
}
}
slices.Sort(keys)
return keys
}
func resolveXrayLogPaths(logCfg json_util.RawMessage) json_util.RawMessage {
if len(logCfg) == 0 {
return logCfg
}
var parsed map[string]any
if err := json.Unmarshal(logCfg, &parsed); err != nil {
return logCfg
}
changed := false
for _, key := range []string{"access", "error"} {
// xray-core decodes this object with encoding/json, whose case-insensitive
// field match makes "Access" reach AccessLog too — fold every variant.
variants := caseVariantKeys(parsed, key)
value, hasValue := parsed[key]
for _, variant := range variants {
if variant == key {
continue
}
if !hasValue {
value, hasValue = parsed[variant], true
}
delete(parsed, variant)
changed = true
}
v, ok := value.(string)
if !ok {
continue
}
trimmed := strings.TrimSpace(v)
if trimmed == "" || strings.EqualFold(trimmed, "none") {
if changed {
parsed[key] = v
}
continue
}
base := path.Base(filepath.ToSlash(trimmed))
if base == "" || base == "." || base == ".." || base == "/" {
continue
}
confined := filepath.Join(config.GetLogFolder(), base)
if confined == trimmed {
continue
}
parsed[key] = confined
changed = true
}
if !changed {
return logCfg
}
out, err := json.Marshal(parsed)
if err != nil {
return logCfg
}
return out
}
// stripDisabledRules removes routing rules marked `enabled: false` from the
// generated runtime config and strips panel-only keys (`enabled`, `comment`)
// from the rest, since xray-core has no such fields. The internal api rule is
// always kept (see isApiRule) so traffic stats can't be toggled off. The stored
// template is untouched — only the generated config is filtered.
func stripDisabledRules(routerCfg json_util.RawMessage) json_util.RawMessage {
if len(routerCfg) == 0 {
return routerCfg
}
var parsed map[string]any
if err := json.Unmarshal(routerCfg, &parsed); err != nil {
return routerCfg
}
rules, ok := parsed["rules"].([]any)
if !ok || len(rules) == 0 {
return routerCfg
}
var activeRules []any
changed := false
for _, rawRule := range rules {
rule, ok := rawRule.(map[string]any)
if !ok {
activeRules = append(activeRules, rawRule)
continue
}
if enabledRaw, exists := rule["enabled"]; exists {
// The internal api rule carries traffic stats and must never be
// dropped, even if it was somehow marked disabled.
enabled, ok := enabledRaw.(bool)
if ok && !enabled && !isApiRule(rule) {
changed = true
continue
}
delete(rule, "enabled")
changed = true
}
if _, exists := rule["comment"]; exists {
delete(rule, "comment")
changed = true
}
activeRules = append(activeRules, rule)
}
if !changed {
return routerCfg
}
parsed["rules"] = activeRules
out, err := json.Marshal(parsed)
if err != nil {
return routerCfg
}
return out
}
2025-09-20 00:35:50 -07:00
// GetXrayTraffic fetches the current traffic statistics from the running Xray process.
2023-02-09 11:18:06 -08:00
func (s *XrayService) GetXrayTraffic() ([]*xray.Traffic, []*xray.ClientTraffic, error) {
process := currentXrayProcess()
if process == nil || !process.IsRunning() {
err := ErrXrayNotRunning
2024-07-08 14:08:00 -07:00
logger.Debug("Attempted to fetch Xray traffic, but Xray is not running:", err)
return nil, nil, err
2023-02-09 11:18:06 -08:00
}
apiPort := process.GetAPIPort()
if err := s.xrayAPI.Init(apiPort); err != nil {
logger.Debug("Failed to initialize Xray API:", err)
return nil, nil, err
}
defer s.xrayAPI.Close()
2024-07-08 14:08:00 -07:00
traffic, clientTraffic, err := s.xrayAPI.GetTraffic()
2024-07-08 14:08:00 -07:00
if err != nil {
logger.Debug("Failed to fetch Xray traffic:", err)
return nil, nil, err
}
return traffic, clientTraffic, nil
2023-02-09 11:18:06 -08:00
}
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
// GetOnlineUsers returns connection-based online users (email + source IPs)
// from the running core's online-stats API. ok=false means the API is not
// available — xray isn't running or the core predates the online-stats RPCs —
// and callers must use the legacy traffic-delta / access-log paths. The
// capability is probed lazily per process: an Unimplemented answer pins this
// core as unsupported until the next restart, while transient errors leave the
// capability undecided so a flaky poll can't lock in legacy mode.
func (s *XrayService) GetOnlineUsers() ([]xray.OnlineUser, bool, error) {
process := currentXrayProcess()
if process == nil || !process.IsRunning() {
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
return nil, false, nil
}
if process.OnlineAPISupport() == xray.OnlineAPIUnsupported {
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
return nil, false, nil
}
if err := s.xrayAPI.Init(process.GetAPIPort()); err != nil {
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
logger.Debug("Failed to initialize Xray API:", err)
return nil, false, err
}
defer s.xrayAPI.Close()
users, err := s.xrayAPI.GetOnlineUsers()
if err != nil {
if xray.IsUnimplementedErr(err) {
process.SetOnlineAPISupport(xray.OnlineAPIUnsupported)
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
logger.Info("xray core does not support the online-stats API; falling back to traffic-delta onlines and access-log IP limit")
return nil, false, nil
}
logger.Debug("Failed to fetch Xray online users:", err)
return nil, false, err
}
if process.OnlineAPISupport() == xray.OnlineAPIUnknown {
process.SetOnlineAPISupport(xray.OnlineAPISupported)
feat(online): use xray online-stats API for onlines and access-log-free IP limit Adopt xray-core's statsUserOnline policy and GetUsersStats RPC so online detection is connection-based and IP limiting no longer requires an access log. Falls back to the legacy traffic-delta onlines and access-log parsing when the running core lacks the RPCs (Unimplemented), probed lazily per process so a panel-driven version switch re-evaluates automatically. Backend: - xray/api.go: GetOnlineUsers (one GetUsersStats call returns all online users and their source IPs) and IsUnimplementedErr. - xray/process.go: per-process OnlineAPISupport tri-state capability cache. - service/xray.go: ensureStatsPolicy injects statsUserOnline into every policy level of the generated config; XrayService.GetOnlineUsers probes and falls back. - job/xray_traffic_job.go: union API onlines into the delta-derived active set; bump last_online for idle-but-connected clients. - job/check_client_ip_job.go: API-first IP source with shared enforcement; live observations bypass the 30-min stale cutoff; access-log path unchanged for older cores. - service/setting.go: GetIpLimitEnable always true; new accessLogEnable default for features that genuinely read the access log. Frontend: - Client form split into Basic and Config tabs; IP Limit and IP Log no longer gated on access log; compact Auto Renew next to Start After First Use; tabBasic/tabConfig added to all 13 locales. - Xray logs button on the dashboard now gated on accessLogEnable.
2026-06-11 10:42:03 -07:00
logger.Info("xray core supports the online-stats API; using connection-based onlines and access-log-free IP limit")
}
return users, true, nil
}
// BalancerStatus is the live view of one balancer for the panel UI. Running
// is false when the balancer isn't present in the running core (e.g. xray is
// stopped or the balancer hasn't been saved/applied yet).
type BalancerStatus struct {
Tag string `json:"tag"`
Running bool `json:"running"`
Override string `json:"override"`
Selected []string `json:"selected"`
}
// GetBalancersStatus queries the running core for the live state of the
// given balancer tags. Per-tag failures are reported as Running=false rather
// than failing the whole call, so the UI can render saved-but-not-applied
// balancers alongside live ones.
func (s *XrayService) GetBalancersStatus(tags []string) ([]BalancerStatus, error) {
statuses := make([]BalancerStatus, 0, len(tags))
process := currentXrayProcess()
if process == nil || !process.IsRunning() {
for _, tag := range tags {
statuses = append(statuses, BalancerStatus{Tag: tag})
}
return statuses, nil
}
if err := s.xrayAPI.Init(process.GetAPIPort()); err != nil {
return nil, err
}
defer s.xrayAPI.Close()
for _, tag := range tags {
info, err := s.xrayAPI.GetBalancerInfo(tag)
if err != nil {
logger.Debug("get balancer info [", tag, "] failed:", err)
statuses = append(statuses, BalancerStatus{Tag: tag})
continue
}
statuses = append(statuses, BalancerStatus{
Tag: tag,
Running: true,
Override: info.Override,
Selected: info.Selected,
})
}
return statuses, nil
}
fix(xray): keep balancer overrides across routing reloads (#6779) * fix(xray): keep balancer overrides across routing reloads A balancer override set from the panel lived only inside the running core. Every routing hot-apply (any Xray-settings save touching rules or balancers) and every node test through an outbound calls RoutingService.AddRule with shouldAppend=false, which rebuilds all balancers, so the override vanished silently even though the API docs promised it lasted until an Xray restart. The panel now records the overrides it sets, per core process, and re-applies them after its own routing reloads. Targets are validated against the running config: the core accepts any tag and then closes every connection routed to one it lacks, so an unknown target is refused, and an override whose outbound is removed is cleared instead of black-holing the balancer. Balancer targets resolve to their _bl_ loopback from the running config rather than the saved template. Restarting Xray still clears overrides, as documented. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KrJrzn9LKaPhMq9h3Fb5TV * fix(xray): serialize overrides and the node bridge with config applies Review of #6779 found two races. Neither path took the package lock that RestartXray holds while it changes the running core: - An override landing mid hot-apply was checked against the config snapshot the apply was replacing, so it could pin a balancer to an outbound the core had just removed. OverrideBalancer now takes the lock. - The node bridge's teardown checked overrides against the snapshot it took before the probe, which can run for seconds, so it dropped an override set meanwhile onto a newly saved outbound. The same stale snapshot drove the bridge's routing restore. That restore silently reverted rule changes saved during a probe. After a restart, it also pushed the old core's rules and overrides into the new core through the shared API port. The bridge is now built and torn down under the lock, which is not held while the probe runs. At teardown it restores the routing that the core's config holds then, and it leaves alone a core that was restarted meanwhile. With every caller under the package lock, balancerOverrideSet drops its own mutex.
2026-10-12 05:59:26 -07:00
// OverrideBalancer forces balancer tag onto target in the running core; an empty
// target clears it. The panel keeps it across routing reloads until Xray restarts.
func (s *XrayService) OverrideBalancer(tag, target string) error {
fix(xray): keep balancer overrides across routing reloads (#6779) * fix(xray): keep balancer overrides across routing reloads A balancer override set from the panel lived only inside the running core. Every routing hot-apply (any Xray-settings save touching rules or balancers) and every node test through an outbound calls RoutingService.AddRule with shouldAppend=false, which rebuilds all balancers, so the override vanished silently even though the API docs promised it lasted until an Xray restart. The panel now records the overrides it sets, per core process, and re-applies them after its own routing reloads. Targets are validated against the running config: the core accepts any tag and then closes every connection routed to one it lacks, so an unknown target is refused, and an override whose outbound is removed is cleared instead of black-holing the balancer. Balancer targets resolve to their _bl_ loopback from the running config rather than the saved template. Restarting Xray still clears overrides, as documented. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KrJrzn9LKaPhMq9h3Fb5TV * fix(xray): serialize overrides and the node bridge with config applies Review of #6779 found two races. Neither path took the package lock that RestartXray holds while it changes the running core: - An override landing mid hot-apply was checked against the config snapshot the apply was replacing, so it could pin a balancer to an outbound the core had just removed. OverrideBalancer now takes the lock. - The node bridge's teardown checked overrides against the snapshot it took before the probe, which can run for seconds, so it dropped an override set meanwhile onto a newly saved outbound. The same stale snapshot drove the bridge's routing restore. That restore silently reverted rule changes saved during a probe. After a restart, it also pushed the old core's rules and overrides into the new core through the shared API port. The bridge is now built and torn down under the lock, which is not held while the probe runs. At teardown it restores the routing that the core's config holds then, and it leaves alone a core that was restarted meanwhile. With every caller under the package lock, balancerOverrideSet drops its own mutex.
2026-10-12 05:59:26 -07:00
// Serialized with config applies, so the target is checked against the config the core runs.
lock.Lock()
defer lock.Unlock()
process := currentXrayProcess()
if process == nil || !process.IsRunning() {
return ErrXrayNotRunning
}
if err := s.xrayAPI.Init(process.GetAPIPort()); err != nil {
return err
}
defer s.xrayAPI.Close()
fix(xray): keep balancer overrides across routing reloads (#6779) * fix(xray): keep balancer overrides across routing reloads A balancer override set from the panel lived only inside the running core. Every routing hot-apply (any Xray-settings save touching rules or balancers) and every node test through an outbound calls RoutingService.AddRule with shouldAppend=false, which rebuilds all balancers, so the override vanished silently even though the API docs promised it lasted until an Xray restart. The panel now records the overrides it sets, per core process, and re-applies them after its own routing reloads. Targets are validated against the running config: the core accepts any tag and then closes every connection routed to one it lacks, so an unknown target is refused, and an override whose outbound is removed is cleared instead of black-holing the balancer. Balancer targets resolve to their _bl_ loopback from the running config rather than the saved template. Restarting Xray still clears overrides, as documented. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KrJrzn9LKaPhMq9h3Fb5TV * fix(xray): serialize overrides and the node bridge with config applies Review of #6779 found two races. Neither path took the package lock that RestartXray holds while it changes the running core: - An override landing mid hot-apply was checked against the config snapshot the apply was replacing, so it could pin a balancer to an outbound the core had just removed. OverrideBalancer now takes the lock. - The node bridge's teardown checked overrides against the snapshot it took before the probe, which can run for seconds, so it dropped an override set meanwhile onto a newly saved outbound. The same stale snapshot drove the bridge's routing restore. That restore silently reverted rule changes saved during a probe. After a restart, it also pushed the old core's rules and overrides into the new core through the shared API port. The bridge is now built and torn down under the lock, which is not held while the probe runs. At teardown it restores the routing that the core's config holds then, and it leaves alone a core that was restarted meanwhile. With every caller under the package lock, balancerOverrideSet drops its own mutex.
2026-10-12 05:59:26 -07:00
return balancerOverrides.set(process, &s.xrayAPI, tag, target)
feat(balancer): add balancer-to-balancer fallback support (#5586) * feat(balancer): add balancer-to-balancer fallback support Xray does not natively support using a balancer as fallbackTag for another balancer. This feature automates the loopback workaround: when a user selects a balancer as fallback, the panel generates a loopback outbound + routing rule in the template. How it works: - User picks fallback balancer from dropdown - Panel creates loopback outbound _bl_{target} + routing rule - Balancer fallbackTag set to _bl_{target} - Traffic: Balancer A → loopback _bl_B → routing rule → Balancer B Key features: - Dedup: multiple balancers sharing same fallback reuse one loopback - DFS cycle detection at edit time and on save - Self-reference guard (cannot select own balancer) - Delete protection (blocks if used as fallback by others) - Cleans up routing rules referencing deleted balancers - Override resolves balancer tags through loopback mechanism - All live status tags resolved for display - Internal _bl_ objects filtered from Outbounds/Routing UI - Backward-compatible with old _bl_ naming format - Translations for all 13 locales * fix(review): override regression, save payload sync, i18n completeness - OverrideBalancer: only resolve to loopback when resolution succeeds, pass original target through for plain outbound tags - onSaveAll: serialize cleaned template before save to ensure the healed/cleaned config is what gets persisted - Add reservedPrefix translation key to all 12 non-English locales - Restore trailing newlines in all 13 translation JSON files * fix(test): update balancer form modal tests after cycle-detection guard The okButtonProps disabled guard (added in 56d5825c) prevents the modal from firing onOk when the form is invalid. The old tests clicked the button expecting validation errors to appear, but antd Modal never calls onOk on a disabled button — causing false failures. Rewrite to test the actual guard behavior: - Button starts disabled (empty form) - Stays disabled with tag only (selector still empty) - Stays disabled for duplicate tag - Disabled button does not trigger onConfirm --------- Co-authored-by: MHSanaei <ho3ein.sanaei@gmail.com>
2026-07-08 16:59:51 -07:00
}
// TestRoute asks the running core which outbound its router picks for the
// described connection.
func (s *XrayService) TestRoute(req xray.RouteTestRequest) (*xray.RouteTestResult, error) {
process := currentXrayProcess()
if process == nil || !process.IsRunning() {
return nil, ErrXrayNotRunning
}
if err := s.xrayAPI.Init(process.GetAPIPort()); err != nil {
return nil, err
}
defer s.xrayAPI.Close()
return s.xrayAPI.TestRoute(req)
}
// RestartXray reconciles the running Xray process with the current desired
// config. When isForce is false it first tries to apply the changes through
// the Xray gRPC API without restarting the process (inbounds, outbounds and
// routing rules/balancers are hot-reloadable); only changes the core cannot
// take at runtime — or a force request — stop and restart the process.
2023-02-09 11:18:06 -08:00
func (s *XrayService) RestartXray(isForce bool) error {
lock.Lock()
defer lock.Unlock()
logger.Debug("restart Xray, force:", isForce)
Repo-wide self-correcting audit: 54 verified bug fixes (#5970) * fix(email): resolve a name-addr smtpFrom into bare envelope address and display name The save-time validator accepts any RFC 5322 address form, so a value like '3x-ui Panel <panel(at)example.com>' passes validation, but Send and TestConnection fed that raw string to MAIL FROM, which strict servers reject with 501, and buildMessage mangled it into a quoted local part. Parse the configured sender at the point of use: the envelope gets the bare address and, when no explicit sender name is set, the display name embedded in the setting is used for the From header. * fix(email): report a missing sender address from the SMTP connection test TestConnection skipped the empty-from guard that Send enforces, so with no sender and no username configured the test issued the null reverse-path and could report success against a lenient relay while every real notification send kept failing with the missing-sender error. Guard the test path the same way and surface a dedicated translated message. * fix(sub): fall back to the raw subscription when an auto-detected format has no content With format auto-detection enabled, a client whose User-Agent matched the Clash or JSON regex was routed straight to that format handler. For a subscription whose entries convert to neither format (an MTProto-only subscription, for example) the handler returns an empty document and the request ended as 404, breaking a URL that served the raw list before the toggle. The auto-detect branches now serve the detected format only when it produces content and otherwise continue to the raw response; the explicit format endpoints keep answering 404 for empty documents. * fix(node): match prefixed central tags when filtering a selected-mode node snapshot FilterNodeSnapshot compared a node snapshot's inbound tags against the raw selected-tag list with an exact match, while its two siblings (SnapshotHasUnadoptedInbounds and the reconcile tagToCentral map) expand each selected tag to both its bare node-side form and its n<id>- prefixed central form. A panel-created node inbound is recorded in the selected list under the central prefixed tag but reported by the node under the bare tag, so the exact match dropped it from every snapshot and the orphan sweep then deleted its central row one tick after creation. Expand the allowed set with the same prefix flip the siblings use. * fix(client): refuse a bulk quota reduction that would fall to or below zero BulkAdjust clamped a client's new traffic limit with max(total+addBytes, 0). Because 0 is the unlimited sentinel, reducing a client's quota by more than it had left silently granted that client unlimited traffic. The sibling expiry branch already refuses an over-reduction; mirror it for quota so the adjustment is skipped with a clear reason instead of crossing the sentinel. * fix(client): persist a bulk adjustment's applied field even when the sibling field is skipped In a mixed BulkAdjust (both a days delta and a bytes delta), a per-field planning skip such as "unlimited expiry" or "unlimited traffic" was recorded in the same map that gated the client_traffics write. The applied field was already written to the inbound JSON and the clients table, but the enforcement row was left untouched, so the depletion job cut the client on the old limit while the panel showed the new one. Gate the traffic-row write on an actual inbound-processing failure rather than on any planning-phase skip note. * fix(inbound): always create in AddInbound instead of overwriting a row whose id was posted The add controller binds the inbound model's id form field and never clears it, and AddInbound persisted with GORM Save, which updates in place when the primary key is non-zero. A client that reused an existing id (for instance by duplicating an inbound fetched from /get and changing the port) silently overwrote that stored row instead of creating a new inbound. Zero the id at the top of AddInbound, matching how it already zeroes the client-stat ids. * fix(inbound): accept WireGuard clients when creating an inbound AddInbound's per-client validation switch had cases for every protocol except WireGuard, so a WireGuard client fell through to the default branch that requires a non-empty id. WireGuard clients are keyed by their public key and carry no id, so importing a WireGuard inbound or re-adding one to a reconciling node was rejected with "empty client ID". Add a wireguard case that validates the client key, mirroring addInboundClient. * fix(client): stop holding the inbound-lock registry mutex while waiting on one inbound lockInbound acquired the global registry mutex and then blocked on the per-inbound mutex without releasing the registry first. A slow client operation holding one inbound's mutex (for example a bulk delete pushing to an unreachable node) made the next waiter park on that inbound while still holding the registry mutex, which in turn blocked lockInbound for every other inbound — freezing client mutations panel-wide. Release the registry mutex before taking the per-inbound lock. * fix(client): honor keepTraffic when deleting a client that is attached to inbounds Delete, DeleteByEmail and BulkDelete all pass keepTraffic to their final cleanup transaction, but each called the per-inbound delete helper with a hardcoded false. That helper purges the client's traffic, IP and stat rows before the gated cleanup runs, so keepTraffic=true still destroyed all traffic history for any client actually attached to an inbound (the pinned test only covered a record with no inbound mappings). Thread the caller's keepTraffic through to the per-inbound helper at all three call sites. * fix(inbound): defer a local MTProto inbound edit's sidecar push until after commit UpdateInbound applied a local MTProto inbound change by calling the runtime UpdateInbound (which stops/starts the mtg sidecar or talks to it) from inside runSerializedTx. That runs process and network I/O on the single traffic-writer goroutine while a DB transaction is open, so a slow sidecar stalls traffic accounting and every concurrent client mutation, and a later step failing the transaction leaves the sidecar ahead of the rolled-back row. Move the push into the post-commit hook, matching the xray branch. Adds a SetLocalRuntimeOverride test seam mirroring the existing node override so the deferral is regression tested. * fix(client): delete external-link rows when bulk-deleting clients The single-client Delete path removes a client's client_external_links rows, but BulkDelete (and the DelDepleted reaper that routes through it) deleted the record, mappings and traffic while leaving the external-link rows keyed by the now-dead client id, so they accumulated as orphans. Delete them in the same cleanup transaction, keyed by client id like the single path. * fix(inbound): request an xray restart when toggling a routed MTProto inbound AddInbound, DelInbound and UpdateInbound all flag needRestart when an inbound routes MTProto through xray, so the egress SOCKS bridge is regenerated. Only SetInboundEnable's local path omitted it, so toggling a routed MTProto inbound off then on left the bridge out of the running config while the sidecar dialed its loopback port, blackholing that inbound until an unrelated restart. Flag the restart on the local enable path too. * fix(client): apply enable-by-email to every inbound a client is attached to ToggleClientEnableByEmail (Telegram bot) and SetClientEnableByEmail (LDAP sync) resolved a single inbound via the legacy client_traffics pointer and flipped enable only there. A client attached to several inbounds kept connecting through the siblings' running Xray after being disabled, and the next edit could re-enable it everywhere from a stale sibling. Route both through the applyClientFieldByEmail fan-out (the #5039 fix path) so the whole multi-inbound identity is toggled at once, dropping the circular Set/Toggle dependency. * fix(traffic): commit a traffic tick even when a best-effort maintenance helper fails addTrafficLocked stages the inbound and client deltas, then runs three helpers (auto-renew, disable depleted clients, disable depleted inbounds) that are meant to log and continue. All three reused the function-scope err that the deferred commit/rollback inspects, so the last helper's error decided the whole tick: a failure in disableInvalidInbounds rolled back the already-staged traffic while AddTraffic reported success, and because xray had already advanced its counter baseline that traffic was lost for good. Give each best-effort helper its own error variable so only a genuine staging failure rolls the tick back. * fix(traffic): re-enable clients and serialize the write in Reset All Client Traffic ClientService.ResetAllTraffics zeroed up/down but, unlike every sibling reset path, never restored enable=true, so clients that had been auto-disabled for exceeding their quota stayed cut with zero usage after a reset. It also wrote client_traffics directly on the shared DB handle instead of through the serial traffic writer, reintroducing the cross-transaction lock-order deadlock the writer exists to prevent. Restore enable and run the reset inside submitTrafficWrite within one transaction. * fix(traffic): keep node reset propagation out of the serial traffic writer ResetAllTraffics and ResetInboundTraffic performed their remote-node reset HTTP calls inside submitTrafficWrite. Each call can block up to the remote timeout, and Reset All Traffics loops every node serially, so the single traffic-writer goroutine was held for seconds — long enough that the concurrent 5s traffic poll timed out submitting its own write and dropped the deltas it had already drained from xray. Do the DB reset inside the writer, then propagate to the nodes after it returns, matching how the mtproto quota reset is already sequenced. * fix(sub): stop the subscription from 500ing on valid-but-unusual stream settings The raw share-link generators used unchecked type assertions and unguarded array indexing: an empty Reality shortIds/serverNames array (random.Num(0) panics), a tcp-http header with no request block or an empty request.path, a grpc block missing its keys, empty stream settings, and a non-string Host header all panicked mid-generation. Because getSubs loops every client's link with no recover, one such client 500s the entire subscription for everyone. The sibling JSON, Clash and frontend generators already guard these; make the raw generators match with comma-ok assertions and length checks. * fix(sub): tolerate a hysteria inbound without hysteriaSettings in the JSON subscription genHy asserted stream["hysteriaSettings"].(map[string]any) without the comma-ok form, so a hysteria inbound whose StreamSettings omit the hysteriaSettings key (a valid, representable shape the raw generator renders fine) panicked and 500ed the entire JSON subscription. Use comma-ok; the downstream reads already guard each key, so a nil map degrades gracefully. * fix(sub): emit the pinned peer cert sha256 in Clash subscriptions The Clash stream builder computed tlsSettings["pin-sha256"] from the inbound's pinnedPeerCertSha256, but applySecurity's tls case never copied it onto the proxy, so it was written with no reader and silently dropped. Clash subscribers lost certificate pinning while JSON subscribers kept it. Surface pin-sha256 on the proxy in the tls case, matching the JSON emitter. * fix(link): parse the snake_case and extra-blob xhttp fields when importing a share link The panel's share-link emitters (Go and TS) carry advanced xhttp knobs as a snake_case x_padding_bytes plus an extra=<json> payload, but the Go parser's xhttp branch read only top-level camelCase params, so importing an xhttp link via the outbound-subscription feature dropped xPaddingBytes, scMaxEachPostBytes and the rest, silently reverting them to the stream defaults and producing a non-working outbound. Mirror the TS parser: read the snake_case alias, merge the extra JSON blob, then let explicit camelCase params win. * fix(frontend): decode URL-safe base64 when parsing an imported share link Base64.decode called window.atob directly, which rejects the base64url alphabet (- and _) and unpadded input. But the panel's own share-link emitter uses Base64.encode(x, true) (URL-safe, unpadded), and real SIP002 links do too, so importing a Shadowsocks link whose method:password encodes with a - or _ threw, fell back to the raw undecoded string, and produced a wrong method and garbage password (the vmess parser shared the same limitation). Normalize base64url and re-pad before atob so decode round-trips every emitted link. * fix(link): honor the vmess ws path and hysteria2 vcn params on import Two Go/TS parser parity gaps in the outbound share-link import path: parseVmess only applied a ws link's path when the inner JSON also carried a host key, so a generator that omits host dropped the path back to the default; and parseHysteria2 hardcoded verifyPeerCertByName to empty, ignoring the vcn param the panel emits, so a hysteria2 outbound with a decoy SNI and a distinct cert name failed TLS verification after import. The TS parser handles both; make the Go parser match. * fix(ui): stop the sniffing form island from clobbering unrendered fields antd's Form.useWatch only reports registered fields, so while the sniffing toggle was off the island emitted { enabled: false } upward and replaced the full Sniffing object in form state. Saving a VLESS reverse outbound then crashed in sniffingToWire on the missing ipsExcluded array; the loopback outbound and the inbound sniffing tab shared the same hole. Watch the store with preserve: true so unrendered fields keep their values, and seed a missing value from the schema defaults instead of an empty cast. * fix(sub): drop empty remark segments instead of leaving a stray separator expandSegment dropped a "|" segment only when its tokens rendered the unlimited mark, so a segment whose only token resolved to the empty string (a client with no comment, an unlimited client's expiry date) was kept as bare decoration, leaving a trailing "|" or a dangling emoji on every share link's remark. Drop a token-bearing segment whenever none of its tokens produce a real value, while still keeping pure-literal segments. * fix(xray): keep source- and domains-scoped routing rules when an inbound is deleted removeInboundTagFromRules drops a routing rule whose inboundTag list becomes empty only if the rule has no other matcher, but routingMatcherKeys omitted xray-core's canonical source and domains keys. A rule scoped by source or domains (common in hand-authored or imported configs) therefore lost its whole body — including a security-relevant block — when its single listed inbound was deleted, instead of just having the tag trimmed. Recognize source and domains as live matchers. * fix(xray): guard RemoveUser against an uninitialized handler client Every XrayAPI handler method returns an error when HandlerServiceClient is nil, except RemoveUser, which dereferenced it directly. A depletion sweep runs Init with the port ignored and, during a restart window where the fresh process's api port is still 0, Init fails and leaves the client nil — so RemoveUser panicked (recovered by the traffic writer, but re-thrown every poll) instead of returning an error. Add the same nil guard the siblings have. * fix(xray): do not revive a manually stopped Xray on a background restart RestartXray cleared isManuallyStopped unconditionally at its top, so the @30s pending-config cron (and warp/ldap/outbound reconcile jobs) that call RestartXray(false) resurrected an Xray the admin had deliberately stopped — unlike the crash-detector, which honors the manual-stop flag. Skip a non-forced restart while the stop flag is set; only an explicit forced restart clears it. * fix(xray): retry a failed pending-restart instead of dropping the config change The 30s cron consumed the need-restart flag with IsNeedRestartAndSetFalse before calling RestartXray and only logged a failure. If RestartXray failed early (a transient GetXrayConfig DB error) the old process kept running the old config, the crash detector saw a running process and never retried, and the flag stayed cleared — so an admin's saved change silently never reached the core. Move the consume/restart/retry into ApplyPendingRestart, which re-arms the flag on failure so the next tick retries. * fix(xray): synchronize the process version and apiPort fields Start writes p.version and p.apiPort (via refreshVersion/refreshAPIPort) after flipping the process to running, while GetXrayVersion and GetAPIPort read them lock-free from the status and traffic poll goroutines. The struct mutex deliberately excluded these fields, so a restart racing a poll was a real data race — a torn read of the version string header can crash. Extend the mutex to cover version and apiPort, doing the blocking version probe before taking the lock. * fix(settings): detect a wildcard listen collision between the web and sub ports The web/sub same-port check compared the two listen addresses as raw strings, so binding both on all interfaces with different spellings (webListen 0.0.0.0 vs an empty subListen) slipped past validation and only failed at startup with an opaque bind error. Treat any wildcard listen ('', 0.0.0.0, ::) as overlapping so the clash is reported up front, while still allowing two distinct specific addresses to share a port. * fix(db): mark the IP-limit cleanup seeder done on a fresh install ResetIpLimitNoFail2ban is a one-time migration that, on a host without fail2ban, zeroes every existing client's limitIp because the limit can't be enforced. It was missing from the fresh-install fast-path seeder list, so on a brand-new DB it did not run on the first boot but fired on the second — wiping any IP limits the admin had set in between. Add it to the fast-path so a truly fresh install marks it done up front (there is nothing to clean), leaving later admin-set limits intact. * fix(security): dial outbound subscriptions through the SSRF guard The outbound-subscription fetch validated the URL host once (resolving DNS and rejecting private targets) but then fetched with a plain HTTP client that re-resolves the host at dial time, so a subscription domain the attacker controls could pass validation as a public IP and rebind to 127.0.0.1 / a cloud metadata endpoint / an internal host for the actual dial — a blind SSRF into the panel's network. Route the direct fetch (and its redirects) through netsafe.SSRFGuardedDialContext, which resolves, checks and dials the same IP atomically, carrying the subscription's AllowPrivate flag on the request context; a configured egress proxy still dials its loopback bridge unguarded. * fix(security): bound the login-limiter attempts map The login rate limiter keys its records on the caller-supplied username and only evicted a record when that exact key was revisited or the login succeeded. An unauthenticated attacker replaying one CSRF token while rotating a fresh username per request seeded a record that was never revisited, growing the map without bound until the panel OOMs. Cap the map: before inserting a new record, reclaim records whose block has lapsed and whose failures aged out, and if the map is still at the ceiling under a broad flood, drop one so memory can never grow past the cap. * fix(tgbot): require admin for privileged callbacks, not just the first switch answerCallback wraps only its first callback switch in an isAdmin guard; the second switch (server usage, inbound/online enumeration, database backup export, ban logs, mass traffic reset, client creation) ran for every caller. Telegram delivers a callback with the tapping user's id, so a non-admin who can see an admin's inline keyboard — as when the bot runs in a group — could tap Backup and receive the full database and config, or reset all traffic. Default-deny before the second switch: a non-admin may only run the per-user client_* callbacks that resolve their own data from their Telegram id. * fix(eventbus): dispatch each subscriber in its own goroutine The fan-out loop called every subscriber's handler sequentially on the single dispatch goroutine. The email and Telegram notifiers block on network I/O for tens of seconds (or minutes when the remote is slow), so one slow subscriber stalled the whole loop: the 256-slot channel then filled and Publish silently dropped later events — including high-value xray.crash and node.down notifications unrelated to the slow handler. Hand each delivered event to every handler in its own goroutine so a blocking subscriber can no longer stall delivery to the others. safeCall already recovers panics, so a detached handler cannot take down the bus. * fix(integration): cap WARP API response body size doWarpRequest read the response with an unbounded io.ReadAll, unlike the sibling NordVPN client which already caps every read at maxResponseSize. A hostile panel egress proxy or a MITM on the Cloudflare WARP endpoint could stream an arbitrarily large body and force the panel into an unbounded allocation. Wrap the body in an io.LimitReader(maxResponseSize) to match the NordVPN client. * fix(email): bound every SMTP step with a connection deadline The "starttls"/"none" transport delivered through net/smtp.SendMail, which dials with an untimed net.Dial and never sets a socket deadline. When an SMTP server accepted the TCP connection but then stalled (or was a blackhole), the caller was released by Send's 30s select, but the sender goroutine and its socket stayed blocked until the OS TCP timeout — minutes per notification, leaking a goroutine and a connection each time. sendWithTLS dialed with a timeout but likewise armed no deadline on the protocol phase, and TestConnection (called synchronously from the settings handler, with no select guard) could hang the request indefinitely. Replace SendMail with sendPlain, which dials with smtpConnectTimeout and arms conn.SetDeadline(smtpDeadline) before the greeting read, preserving SendMail's opportunistic STARTTLS upgrade. Arm the same deadline in sendWithTLS and TestConnection so every SMTP step is bounded. * fix(server): guard access-log parser against malformed lines GetXrayLogs split each Xray access-log line on whitespace and then read fixed offsets — parts[1] for the timestamp and parts[i+1] after the "from", "accepted" and "email:" markers — without checking the line had that many fields. A truncated or malformed line (the logged destination is attacker-influenced) indexed past the slice and panicked; the panel handler returned a 500 via Gin's recovery. Extract the per-line field parsing into parseAccessLogFields and length guard every positional lookup so a short line yields a partial entry instead of panicking. * fix(server): guard xray key-generator output parsing GetNewX25519Cert, GetNewmldsa65 and GetNewmlkem768 parsed xray's stdout by reading lines[0], lines[1] and each line's second colon-separated field without any length check — unlike GetNewEchCert, which already guards its line count. If the xray binary printed fewer than two lines or reformatted its labels (a version change, or a silent failure that emitted nothing), the fixed slice index panicked and the handler 500'd. Extract the shared parsing into parseXrayKeyPairOutput, which length guards the line count and each label split and returns an error instead of panicking, then route all three generators through it. * fix(tgbot): stop auto-deleted messages from resetting wizard state SendMsgToTgbotDeleteAfter spawns a goroutine that, after the display delay, deleted the transient message and then unconditionally cleared the chat's conversation state. Every caller that ends a wizard step already clears the state synchronously, so that call was redundant — and harmful: if within the delay the user advanced to the next step (a callback sets a fresh awaiting_* state), the late goroutine wiped it, and the user's next message fell through unrecognized, silently dropping their input. Move the delayed deletion into deleteMessageAfterDelay, which only removes the message and no longer touches the conversation state. Guard deleteMessageTgBot against a nil bot so the deletion path is unit-testable. * fix(frontend): refetch a fresh CSRF token on 403 instead of reusing the stale meta tag On a 403 to an unsafe method the client cleared its cached CSRF token and called ensureCsrfToken to retry. But ensureCsrfToken prefers the <meta name="csrf-token"> tag baked into the page, which the production panel always injects, so the "refresh" re-read the same stale token and the /csrf-token refetch was never reached — the retry re-sent the token that had just been rejected and the save failed with an error toast. The token lives in the session and rotates when the session is regenerated (for example re-login in another tab), leaving the tab's baked-in meta token stale. Fetch the current token straight from /csrf-token in the 403 branch so the retry uses the authoritative server value. The existing tests only passed because they strip the meta tag; the new test keeps a stale tag present. * fix(frontend): surface backend error text from failed requests HttpUtil.get/post read the thrown HttpError body as response.data.message, but the backend error envelope (entity.Msg) serializes its text as msg. On any non-2xx JSON response the real reason was therefore dropped and the operator saw only the generic "Request failed with status N" toast. Read response.data.msg first (keeping message and the native error text as fallbacks). The sibling test had pinned the wrong body shape ({ message }); correct it to the real backend shape ({ success:false, msg }) so it exercises the actual envelope. * fix(frontend): share one WebSocket connection across bridge and hooks websocketBridge.ts and useWebSocket.ts each declared their own module-scoped sharedClient plus an identical getSharedClient, so the "shared" client was not shared between them: whenever a page using useWebSocket (Clients/Inbounds) mounted alongside the always-mounted bridge, the panel opened two sockets to /ws. The server then pushed every traffic/stats/nodes/inbounds snapshot to both, doubling WebSocket bandwidth and running two independent reconnect loops, and the hook's socket was never disconnected on unmount. Hoist a single getSharedWebSocketClient into api/websocket.ts and route both the bridge and the hook through it, so exactly one connection is opened. * fix(frontend): guard the outbounds WebSocket handler against non-array payloads onOutbounds wrote the raw WebSocket payload straight into the outboundsTraffic cache, unlike the sibling onNodes/onInbounds handlers which first check Array.isArray. A malformed non-array push (for example an object) would land in the cache with staleTime Infinity; consumers that call .find()/.map() on the outbounds list would then throw and crash the Outbounds tab. Add the same Array.isArray guard so a bad push is ignored. * fix(frontend): key the node table by the computed row key, not id The desktop node table used rowKey="id", but transitive sub-nodes (the read-only rows surfaced from downstream nodes) all carry id 0, so a topology with two or more transitive rows gave React duplicate keys. antd's rowKey prop overrides the row object's own computed `key` (`t-${guid}` for transitive rows, the numeric id otherwise), so the unique key the code already builds was ignored — causing row-state/DOM mis-association on any re-render (heartbeat refetch, address-eye toggle). The mobile card path already keyed by record.key. Key the table by "key" so transitive rows get their distinct t-${guid} identity; direct nodes keep key === id, so row selection (filtered to numeric keys) is unchanged. * fix(frontend): map routing row actions through the rule's real index The routing table hides balancer-loopback rules (`_bl_*`) but keeps each visible row's original index in `key`, then handed antd's positional row index straight to edit/delete/toggle/move/drag — all of which mutate the full, unfiltered routing.rules array. Once a hidden loopback rule precedes a visible one (e.g. a balancer whose fallback is another balancer, plus any rule added afterwards), the positional index no longer matches the array index, so deleting or editing a rule silently hit the wrong one — including destroying the loopback rule that keeps the balancer alive. Add originalRuleIndex to translate a positional row index back through the row's `key`, and route every mutating handler (openEdit, confirmDelete, toggleRule, moveUp/moveDown, drag) through it. When no loopback rows are hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): map outbound row actions through the outbound's real index The outbounds table hides balancer-loopback outbounds (`_bl_*`) but keeps each visible row's original index in `key`, then passed antd's positional row index to edit/delete/move and to the per-row probe (onTest) and its result lookup — all of which address the full, unfiltered outbounds array. Once a hidden loopback outbound precedes a visible one, the positional index diverges from the array index, so deleting or editing an outbound hit the wrong one (its deletion-impact plan and removal targeting the wrong entry), and the test button probed / showed results against the wrong outbound. Add originalOutboundIndex and route the mutating handlers through it; key the probe trigger and test-result columns by record.key. With no loopback rows hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): tolerate a malformed happyEyeballs value in the Xray Basics tab BasicsTab derived directHappyEyeballs by calling HappyEyeballsSchema.parse during render, guarding only against null/non-object. A wrong-typed field (e.g. happyEyeballs.tryDelayMs as a string) or any other shape mismatch — reachable via the Complete Template JSON editor or an imported config — threw straight out of render, white-screening the default Xray landing tab. Use safeParse and fall back to null so a bad value degrades to "no override" instead of crashing the page. * fix(frontend): preserve routing-rule fields the form does not surface The rule form rebuilt the rule from a fixed literal of only the fields it edits, and RoutingTab replaces the rule wholesale on confirm. Fields the form never exposes — localPort, localIP, process, ruleTag, webhook — are in the rule schema and can arrive via the advanced JSON editor or Import Rules; opening such a rule in the form and saving silently dropped them. Carry over every key of the original rule the form does not manage before applying the form-derived fields, so an edit only touches what it surfaces. * fix(frontend): re-sync the sniffing island when its value changes externally The sniffing config editor froze its seed value at mount and only watched its own inner AntD form, never reflecting a later change to the shared RHF `sniffing` path. Because the inbound form mounts every tab with forceRender, the friendly Sniffing tab and the Advanced JSON editor are live at once: editing sniffing in the JSON editor updated the RHF value but not the frozen island, so the next interaction with the friendly tab emitted the stale value and silently discarded the JSON edit. Add an effect that pushes an external value change into the inner form, guarded by the same lastEmitted marker the emit path uses so the island never re-seeds from its own echo and no update loop forms. * fix(frontend): don't drift a client's byte quota on a no-op save The quota field shows the total in GB rounded to two decimals; editing a client and saving converted that display value straight back to bytes. A byte total not aligned to 0.01 GB — one set via the API or an import — was therefore rewritten to the rounded value on any save that never touched the field, losing a few MB each time. Add resolveTotalBytes: keep the original byte total when the displayed GB still matches it, and only re-derive from GB when the user actually changed the field. * fix(eventbus): deliver events on a bounded per-subscriber worker The previous fix dispatched each event to every subscriber with a bare `go safeCall`. That unblocked the dispatch loop, but removed the bus's backpressure: under a login-attempt flood (which both notifier subscribers process without rate-limiting) with email/Telegram enabled, every attempt spawned handler goroutines that each block on network I/O for up to ~30s, with no bound — a goroutine and outbound-connection storm. It also let a subscriber's handler run concurrently with itself, racing the Telegram notifier's lazily-cached hostname. Give each subscriber its own bounded queue drained by a single worker goroutine. Dispatch does a non-blocking send per subscriber (dropping only that subscriber's event when its queue is full), so a slow subscriber still can't stall the others, concurrency is bounded to one in-flight handler per subscriber, per-subscriber event order is preserved, and Stop again waits for in-flight handlers to finish. * fix(frontend): map outbound mobile-card actions through the real index too The desktop outbounds table was keyed by the outbound's real index, but the mobile card list was left keying the probe trigger and every test-state lookup by the positional row index. With a hidden balancer-loopback outbound present, tapping Check on a mobile card probed the wrong outbound and the Test-All results landed on the wrong card. Key onTest and the testResult/isTesting reads by record.key, matching the desktop columns. * fix(frontend): meet WCAG AA contrast on the config-block link text The Storybook accessibility test flagged the share-link <code> block: with no explicit color it inherited a muted grey that renders as #888888 on the #f8f8f8 tertiary-fill background in CI's Chromium — a 3.33:1 contrast, below the 4.5:1 AA threshold. Set the text to the theme's primary text token so the colour is explicit and high-contrast in both light and dark themes instead of depending on an inherited value that varies by browser. * style(sub): simplify a negated conjunction to satisfy staticcheck QF1001 golangci-lint (staticcheck QF1001) flagged the `!(a && b)` guard in expandSegment. Rewrite it via De Morgan's law to the equivalent `!a || !b` form so the linter passes; behavior is unchanged. * fix: close panics and races the audit's own fixes left nearby Second-pass review of the 54-commit self-correcting audit. Each item below was confirmed by reading the surrounding source (and, where practical, the pre-fix code) before being changed; regression tests are included for every behavioral fix. Concurrency: - eventbus: Bus.Subscribe called wg.Add with no synchronization against a concurrent Bus.Stop's wg.Wait, a real "WaitGroup misuse" panic risk (e.g. a Telegram-bot settings save racing panel shutdown/restart). Stop now flips a mu-guarded `stopped` flag before waiting, and Subscribe checks it under the same lock, so Add and Wait can no longer race. Security: - login_limiter: evictForRoom's fallback eviction picked an arbitrary map key, including ones still under an active cooldown - an attacker flooding /login with fresh usernames could evict their own (or anyone's) blocked record and reset the lockout. The fallback now skips actively-blocked records, only falling back to an unconditional evict if the map is somehow entirely full of active blocks (preserves the hard memory cap). Subscription-endpoint panics (reachable by any client hitting /sub): - internal/sub/service.go: applyPathAndHostParams/Obj (ws/httpupgrade/xhttp with no path settings object) and the TLS alpn readers in three places used unchecked type assertions - exactly the bug class abab7cd0 patched elsewhere in the same switch statements, just not these call sites. - internal/sub/json_service.go, clash_service.go: the externalProxy loops in the JSON and Clash generators used unchecked assertions on a legacy/admin-supplied field (missing "port", non-object entry, etc.). - internal/sub/json_service.go: realityData's shortId/serverName selection could assert a non-string array element. Other correctness: - client_traffic.go: ResetAllTraffics (touched by 3eb214d0) still skipped clearing NodeClientTraffic node-sync baselines, unlike its sibling reset paths in the same file - a node's next sync would re-add pre-reset delta on top of the freshly-zeroed counter. - inbound_traffic.go: the traffic-tick tx's Commit/Rollback errors were silently discarded; now logged so a backend-level commit failure (e.g. an aborted Postgres tx from a best-effort helper) doesn't masquerade as a successful tick. - outbound_subscription.go: the new subscriptionFetchClient doc comment was wedged between fetchAndStore's existing comment and fetchAndStore itself, leaving fetchAndStore undocumented and the comment describing the wrong function. Convention cleanup: - Removed narrative // comments added by the audit that violate this repo's no-inline-comment rule (mostly narrating the specific bug/fix rather than a lasting contract, and mostly on new Test functions, which this repo's existing tests never comment) - calibrated against this exact codebase's own pre-existing comment style so legitimate godoc-style doc comments were left alone. --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-07-16 15:33:06 -07:00
if !isForce && isManuallyStopped.Load() {
return nil
}
isManuallyStopped.Store(false)
2023-02-09 11:18:06 -08:00
xrayConfig, err := s.GetXrayConfig()
if err != nil {
if process := currentXrayProcess(); process != nil && process.IsRunning() {
xrayState.holdBack(fmt.Sprintf("config refused: %v", err))
}
2023-02-09 11:18:06 -08:00
return err
}
process := currentXrayProcess()
if process != nil && process.IsRunning() {
configUnchanged := process.GetConfig().Equals(xrayConfig)
if !isForce && configUnchanged && !isNeedXrayRestart.Load() {
logger.Debug("It does not need to restart Xray")
return repairUnavailableTuicRelays(xrayConfig, process.GetAPIPort())
2023-02-09 11:18:06 -08:00
}
fix(xray): refuse a config the running core cannot bind (#6547) * fix(xray): refuse a config the running core cannot bind RestartXray stopped a working core before handing it a config whose listens collide, so the failed bind exited the whole process (main/run.go:94) and the one-second watchdog retried it in a loop: every protocol down, cause only in the logs. The save-time port guards cannot cover this -- SetInboundEnable, the AmneziaWG relay created on the first peer, template and bridge edits all reach a colliding config with no guard on that path. Probe the generated config at the single restart funnel instead. Collisions the running core already serves are excused, so an established setup is never refused by a static read being wrong about it, and the port-bucketed pass costs nothing on a clean config. * fix(xray): surface a refused config and re-key the bind excuse set Round-1 findings on this PR. Refusing the swap left the running core on its previous config with nothing but a log line to show for it, so the status response now carries the reason while the core runs and the overview marks it; the node list picks the same field up through that response. The excuse set is keyed on the two listens, the port and the shared transports instead of the tag pair, so a pair whose listen moves onto the other's address is refused again, while the same two sockets stay excused however the generator orders them. TestBindConflicts/excused_pair_whose_listen_changed_into_a_real_collision fails without the key change -- watched red first.
2026-09-15 05:27:28 -07:00
// A config the core cannot bind never replaces one that works: its failed
// start exits the core, and the watchdog would then loop on it forever.
if conflicts := bindConflicts(xrayConfig, process.GetConfig()); len(conflicts) > 0 {
refused := fmt.Sprintf("config refused: %s", conflicts[0])
for _, conflict := range conflicts {
logger.Error("xray config refused:", conflict.String())
}
// The refusal is otherwise invisible: the operator's request
// succeeded, so the status page has to carry the stale state.
xrayState.holdBack(refused)
return fmt.Errorf("xray %s", refused)
}
if !isForce && !configUnchanged {
applied, err := s.tryHotApply(process, xrayConfig)
if err != nil {
return err
}
if applied {
logger.Info("Xray config changes applied through the core API, no restart needed")
return nil
}
}
invalidateTuicRelayReadiness(process.GetConfig())
invalidateTuicRelayReadiness(xrayConfig)
_ = process.Stop()
fix(xray): refuse a config the running core cannot bind (#6547) * fix(xray): refuse a config the running core cannot bind RestartXray stopped a working core before handing it a config whose listens collide, so the failed bind exited the whole process (main/run.go:94) and the one-second watchdog retried it in a loop: every protocol down, cause only in the logs. The save-time port guards cannot cover this -- SetInboundEnable, the AmneziaWG relay created on the first peer, template and bridge edits all reach a colliding config with no guard on that path. Probe the generated config at the single restart funnel instead. Collisions the running core already serves are excused, so an established setup is never refused by a static read being wrong about it, and the port-bucketed pass costs nothing on a clean config. * fix(xray): surface a refused config and re-key the bind excuse set Round-1 findings on this PR. Refusing the swap left the running core on its previous config with nothing but a log line to show for it, so the status response now carries the reason while the core runs and the overview marks it; the node list picks the same field up through that response. The excuse set is keyed on the two listens, the port and the shared transports instead of the tag pair, so a pair whose listen moves onto the other's address is refused again, while the same two sockets stay excused however the generator orders them. TestBindConflicts/excused_pair_whose_listen_changed_into_a_real_collision fails without the key change -- watched red first.
2026-09-15 05:27:28 -07:00
} else if conflicts := bindConflicts(xrayConfig, nil); len(conflicts) > 0 {
// Nothing is running to protect and the core is the authority on what it
// can bind: start it and let its own error name the port it lost.
logger.Warning("xray config may not start:", conflicts[0].String())
2023-02-09 11:18:06 -08:00
}
invalidateTuicRelayReadiness(xrayConfig)
process = xray.NewProcess(xrayConfig)
xrayState.replace(process)
s.xrayAPI.StatsLastValues = nil
err = process.Start()
if err != nil {
return err
}
if err := publishTuicRelayReadiness(xrayConfig); err != nil {
return err
}
return nil
2023-02-09 11:18:06 -08:00
}
fix(xray): restart when a diff strands a client's live session (#6550) * fix(xray): restart when a diff strands a client's live session Disabling or deleting a client took it out of the generated config and the hot path applied that with AlterInbound/RemoveUser, which only drops the credential (vless, vmess, trojan and shadowsocks all keep the established session running) -- so the panel showed a disabled client whose connection kept passing traffic, and the core offers no API to close one session. A diff that removes a user without re-adding the same email under the same tag is that case: honour the operator's restart-on-client-disable setting and let the caller replace the process, which is already how an auto-disabled client loses its session. An edit re-adds the email and keeps the hot path. * chore(i18n): cover manual disable and delete in the restart-setting description The setting now also decides what happens when a client is disabled or deleted by hand, so the description cannot keep naming only the automatic path. All 13 locales updated in the same commit to keep the wording consistent. * fix(xray): reach the guard from the manual switch and from every protocol Round-1 findings on this PR. The guard sat in tryHotApply, but a manual disable or delete applies through runtime.Runtime and finishes with needRestart false, so none of the three RestartXray schedulers fired and the predicate was never reached: the session in #6533 kept flowing. The apply layer now asks for the restart the setting promises when the client actually leaves the config, on the single-client update and delete paths and on bulk disable, and only for local inbounds so a node row cannot make the master restart its own core. The predicate itself could not fire for shadowsocks or hysteria either, because RemovedUsers is only produced for the protocols diffInboundUsers will diff. The diff now also compares settings.clients of an inbound present in both configs, which is the one shape every account list shares, so those protocols reach the guard through the inbound instead of through nothing. TestManualClientDisableHonoursRestartSetting fails without the apply-layer fix ("needRestart = false, want true" with the setting on) and TestHotDiffDropsUsersOnProtocolsItCannotDiff fails without the diff fix -- both watched red. The two three-line comments this PR added are back inside the cap. * docs(i18n): stop scoping restartXrayOnClientDisable to auto-disable The setting now covers a client disabled or deleted by hand as well, so its title no longer says "Auto" in all 13 locales, and the docs callouts in en, ru, zh and fa describe the same behaviour instead of the auto-only one.
2026-09-15 05:39:40 -07:00
// restartToDropClients reports whether a diff that strands clients must be
// applied by restarting instead of through the API.
func (s *XrayService) restartToDropClients(diff *xray.HotDiff) bool {
if diff == nil || !diff.DropsUsers() {
return false
}
restart, err := s.settingService.GetRestartXrayOnClientDisable()
if err != nil {
logger.Warning("get RestartXrayOnClientDisable failed:", err)
return false
}
return restart
}
// tryHotApply reconciles the running core with newCfg over gRPC; false means restart, while a TUIC
// relay error defers that update instead of restarting unrelated connections. Hold the package lock.
func (s *XrayService) tryHotApply(process *xray.Process, newCfg *xray.Config) (applied bool, applyErr error) {
defer func() {
if applyErr != nil {
s.SetToNeedReconcile()
}
}()
oldCfg := process.GetConfig()
diff, ok := xray.ComputeHotDiff(oldCfg, newCfg)
if !ok {
logger.Debug("hot apply: config change is not API-applicable, falling back to restart")
return false, nil
}
if diff.Empty() {
if err := repairUnavailableTuicRelays(newCfg, process.GetAPIPort()); err != nil {
return false, err
}
process.SetConfig(newCfg)
persistHotConfig(process)
return true, nil
}
fix(xray): restart when a diff strands a client's live session (#6550) * fix(xray): restart when a diff strands a client's live session Disabling or deleting a client took it out of the generated config and the hot path applied that with AlterInbound/RemoveUser, which only drops the credential (vless, vmess, trojan and shadowsocks all keep the established session running) -- so the panel showed a disabled client whose connection kept passing traffic, and the core offers no API to close one session. A diff that removes a user without re-adding the same email under the same tag is that case: honour the operator's restart-on-client-disable setting and let the caller replace the process, which is already how an auto-disabled client loses its session. An edit re-adds the email and keeps the hot path. * chore(i18n): cover manual disable and delete in the restart-setting description The setting now also decides what happens when a client is disabled or deleted by hand, so the description cannot keep naming only the automatic path. All 13 locales updated in the same commit to keep the wording consistent. * fix(xray): reach the guard from the manual switch and from every protocol Round-1 findings on this PR. The guard sat in tryHotApply, but a manual disable or delete applies through runtime.Runtime and finishes with needRestart false, so none of the three RestartXray schedulers fired and the predicate was never reached: the session in #6533 kept flowing. The apply layer now asks for the restart the setting promises when the client actually leaves the config, on the single-client update and delete paths and on bulk disable, and only for local inbounds so a node row cannot make the master restart its own core. The predicate itself could not fire for shadowsocks or hysteria either, because RemovedUsers is only produced for the protocols diffInboundUsers will diff. The diff now also compares settings.clients of an inbound present in both configs, which is the one shape every account list shares, so those protocols reach the guard through the inbound instead of through nothing. TestManualClientDisableHonoursRestartSetting fails without the apply-layer fix ("needRestart = false, want true" with the setting on) and TestHotDiffDropsUsersOnProtocolsItCannotDiff fails without the diff fix -- both watched red. The two three-line comments this PR added are back inside the cap. * docs(i18n): stop scoping restartXrayOnClientDisable to auto-disable The setting now covers a client disabled or deleted by hand as well, so its title no longer says "Auto" in all 13 locales, and the docs callouts in en, ru, zh and fa describe the same behaviour instead of the auto-only one.
2026-09-15 05:39:40 -07:00
// The core's RemoveUser drops the credential only, so a disabled or deleted
// client needs the restart this setting asks for.
if s.restartToDropClients(diff) {
logger.Info("hot apply: clients left the config, restarting to drop their live sessions")
return false, nil
fix(xray): restart when a diff strands a client's live session (#6550) * fix(xray): restart when a diff strands a client's live session Disabling or deleting a client took it out of the generated config and the hot path applied that with AlterInbound/RemoveUser, which only drops the credential (vless, vmess, trojan and shadowsocks all keep the established session running) -- so the panel showed a disabled client whose connection kept passing traffic, and the core offers no API to close one session. A diff that removes a user without re-adding the same email under the same tag is that case: honour the operator's restart-on-client-disable setting and let the caller replace the process, which is already how an auto-disabled client loses its session. An edit re-adds the email and keeps the hot path. * chore(i18n): cover manual disable and delete in the restart-setting description The setting now also decides what happens when a client is disabled or deleted by hand, so the description cannot keep naming only the automatic path. All 13 locales updated in the same commit to keep the wording consistent. * fix(xray): reach the guard from the manual switch and from every protocol Round-1 findings on this PR. The guard sat in tryHotApply, but a manual disable or delete applies through runtime.Runtime and finishes with needRestart false, so none of the three RestartXray schedulers fired and the predicate was never reached: the session in #6533 kept flowing. The apply layer now asks for the restart the setting promises when the client actually leaves the config, on the single-client update and delete paths and on bulk disable, and only for local inbounds so a node row cannot make the master restart its own core. The predicate itself could not fire for shadowsocks or hysteria either, because RemovedUsers is only produced for the protocols diffInboundUsers will diff. The diff now also compares settings.clients of an inbound present in both configs, which is the one shape every account list shares, so those protocols reach the guard through the inbound instead of through nothing. TestManualClientDisableHonoursRestartSetting fails without the apply-layer fix ("needRestart = false, want true" with the setting on) and TestHotDiffDropsUsersOnProtocolsItCannotDiff fails without the diff fix -- both watched red. The two three-line comments this PR added are back inside the cap. * docs(i18n): stop scoping restartXrayOnClientDisable to auto-disable The setting now covers a client disabled or deleted by hand as well, so its title no longer says "Auto" in all 13 locales, and the docs callouts in en, ru, zh and fa describe the same behaviour instead of the auto-only one.
2026-09-15 05:39:40 -07:00
}
if tuicRequiresOrdinaryRemoval(oldCfg, diff.TuicRelayChanges) {
return false, nil
}
apiPort := process.GetAPIPort()
if apiPort <= 0 {
if len(diff.TuicRelayChanges) > 0 {
return false, errors.New("TUIC relay update requires the Xray HandlerService API")
}
return false, nil
}
// A dedicated client: s.xrayAPI may be in use by traffic polling on other
// service instances and is reset around restarts.
hotAPI := xray.XrayAPI{}
if err := hotAPI.Init(apiPort); err != nil {
logger.Debug("hot apply: failed to init xray api:", err)
if len(diff.TuicRelayChanges) > 0 {
return false, err
}
return false, nil
}
defer hotAPI.Close()
staged, relayErr := reconcileTuicHotConfig(&hotAPI, oldCfg, newCfg, diff.TuicRelayChanges)
defer staged.finish()
effective := staged.effective
intermediate := *oldCfg
intermediate.InboundConfigs = nil
for _, inbound := range oldCfg.InboundConfigs {
if !inbound.TuicRelay {
intermediate.InboundConfigs = append(intermediate.InboundConfigs, inbound)
}
}
for _, inbound := range effective.InboundConfigs {
if inbound.TuicRelay {
intermediate.InboundConfigs = append(intermediate.InboundConfigs, inbound)
}
}
ordinary, hot := xray.ComputeHotDiff(&intermediate, effective)
if !hot || len(ordinary.TuicRelayChanges) > 0 {
relayErr = errors.Join(relayErr, staged.rollback(&hotAPI))
intermediate.InboundConfigs = oldCfg.InboundConfigs
process.SetConfig(&intermediate)
persistHotConfig(process)
return false, errors.Join(relayErr, errors.New("independent Xray changes could not be planned after TUIC reconciliation"))
}
if !applyXrayHotDiff(&hotAPI, ordinary) {
relayErr = errors.Join(relayErr, staged.rollback(&hotAPI))
intermediate.InboundConfigs = oldCfg.InboundConfigs
process.SetConfig(&intermediate)
persistHotConfig(process)
if relayErr != nil {
return false, relayErr
}
return false, nil
}
balancerOverrides.reapply(process, &hotAPI, effective)
process.SetConfig(effective)
persistHotConfig(process)
if relayErr != nil {
return false, relayErr
}
return true, nil
}
func applyXrayHotDiff(hotAPI *xray.XrayAPI, diff *xray.HotDiff) bool {
// Removals first so changed handlers and port swaps never collide with
// the additions that follow.
for _, u := range diff.RemovedUsers {
if err := hotAPI.RemoveUser(u.Tag, u.Email); err != nil && !xray.IsMissingHandlerErr(err) {
logger.Info("hot apply: remove user [", u.Email, "] from [", u.Tag, "] failed:", err)
return false
}
}
for _, tag := range diff.RemovedInboundTags {
if err := hotAPI.DelInbound(tag); err != nil && !xray.IsMissingHandlerErr(err) {
logger.Info("hot apply: remove inbound [", tag, "] failed:", err)
return false
}
}
for _, tag := range diff.RemovedOutboundTags {
if err := hotAPI.DelOutbound(tag); err != nil && !xray.IsMissingHandlerErr(err) {
logger.Info("hot apply: remove outbound [", tag, "] failed:", err)
return false
}
}
for _, ob := range diff.AddedOutbounds {
if err := addOutboundReconciling(hotAPI, ob); err != nil {
logger.Info("hot apply: add outbound failed:", err)
return false
}
}
for _, ib := range diff.AddedInbounds {
if err := addInboundReconciling(hotAPI, ib); err != nil {
logger.Info("hot apply: add inbound failed:", err)
return false
}
}
for _, u := range diff.AddedUsers {
if err := addUserReconciling(hotAPI, u); err != nil {
logger.Info("hot apply: add user [", u.Email, "] to [", u.Tag, "] failed:", err)
return false
}
}
if diff.RoutingConfig != nil {
if err := hotAPI.ApplyRoutingConfig(diff.RoutingConfig); err != nil {
logger.Info("hot apply: apply routing config failed:", err)
return false
}
}
return true
}
// persistHotConfig refreshes config.json after a hot apply; a write failure is
// logged only, since the running core already has the change.
func persistHotConfig(process *xray.Process) {
if err := process.PersistConfig(); err != nil {
logger.Warning("hot apply: failed to update config.json:", err)
}
}
// addUserReconciling adds a user, and on an email conflict (the user was
// already applied through the runtime API) replaces the existing user instead.
func addUserReconciling(api *xray.XrayAPI, u xray.UserOp) error {
err := api.AddUser(u.Protocol, u.Tag, u.User)
if err == nil || !xray.IsUserExistsErr(err) {
return err
}
if delErr := api.RemoveUser(u.Tag, u.Email); delErr != nil && !xray.IsMissingHandlerErr(delErr) {
return delErr
}
return api.AddUser(u.Protocol, u.Tag, u.User)
}
// addInboundReconciling adds an inbound, and on a tag conflict (the handler
// was already created through the runtime API while the stored snapshot was
// stale) replaces the existing handler instead.
func addInboundReconciling(api *xray.XrayAPI, inbound []byte) error {
err := api.AddInbound(inbound)
if err == nil || !xray.IsExistingTagErr(err) {
return err
}
var meta struct {
Tag string `json:"tag"`
}
if jsonErr := json.Unmarshal(inbound, &meta); jsonErr != nil || meta.Tag == "" {
return err
}
if delErr := api.DelInbound(meta.Tag); delErr != nil && !xray.IsMissingHandlerErr(delErr) {
return delErr
}
return api.AddInbound(inbound)
}
// addOutboundReconciling mirrors addInboundReconciling for outbounds.
func addOutboundReconciling(api *xray.XrayAPI, outbound []byte) error {
err := api.AddOutbound(outbound)
if err == nil || !xray.IsExistingTagErr(err) {
return err
}
var meta struct {
Tag string `json:"tag"`
}
if jsonErr := json.Unmarshal(outbound, &meta); jsonErr != nil || meta.Tag == "" {
return err
}
if delErr := api.DelOutbound(meta.Tag); delErr != nil && !xray.IsMissingHandlerErr(delErr) {
return delErr
}
return api.AddOutbound(outbound)
}
2025-09-20 00:35:50 -07:00
// StopXray stops the running Xray process.
2023-02-09 11:18:06 -08:00
func (s *XrayService) StopXray() error {
lock.Lock()
defer lock.Unlock()
isManuallyStopped.Store(true)
2024-07-08 14:08:00 -07:00
logger.Debug("Attempting to stop Xray...")
process := currentXrayProcess()
if process != nil && process.IsRunning() {
return process.Stop()
2023-02-09 11:18:06 -08:00
}
return ErrXrayNotRunning
2023-02-09 11:18:06 -08:00
}
2025-09-20 00:35:50 -07:00
// SetToNeedRestart marks that Xray needs to be restarted.
2023-02-09 11:18:06 -08:00
func (s *XrayService) SetToNeedRestart() {
isNeedXrayRestart.Store(true)
}
// SetToNeedReconcile schedules config reconciliation without forcing an unchanged core to restart.
func (s *XrayService) SetToNeedReconcile() { isNeedXrayReconcile.Store(true) }
Vue3 migration (#4198) * docs(migration): Phase 1 inventory — Vue 2 / AD-Vue 1 surface area Captures the breakage surface for the Vue 3 + Ant Design Vue 4 + Vite migration: 17,650 lines across 69 templates, 3,145 a-* component instances across 63 files, with per-pattern counts and file lists. Key findings: - No Vue filters anywhere — dodges a major Vue 3 breaking change - 358 v-model uses; AD-Vue 4 absorbs most, custom components don't - 233 <template slot="X"> usages must become <template #X> - 49 scopedSlots: { ... } column defs need new slots: { ... } shape - a-icon is removed in AD-Vue 4 — every icon must be imported Establishes the 8-phase order; Phase 2 (Vite toolchain) is next. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): Phase 2 — scaffold Vite + Vue 3 + AD-Vue 4 Adds a frontend/ directory that lives alongside the legacy web/html/ Vue 2 templates during the migration. Vite builds into ../web/dist/ so the Go binary will be able to embed the result via embed.FS once Phase 4 starts moving real pages over. - package.json pins Vue 3.5, Ant Design Vue 4.2, Vite 6, vue-i18n 10 - vite.config.js: dev server on :5173 with API proxy to the Go panel on :2053; build output to ../web/dist/ - src/App.vue is currently a smoke-test placeholder — delete once the first real page (login) lands in Phase 4 - node_modules and dist are already ignored at repo root To verify locally: cd frontend && npm install && npm run dev Pages will be migrated one at a time on the vue3-migration branch. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): Phase 3 — port utils, models, axios, websocket as ES modules Ports the framework-agnostic JS from web/assets/js/ into frontend/src/ so Vue 3 pages can import what they need without relying on script-tag globals. - web/assets/js/util/index.js (927 lines, 21 classes) → frontend/src/utils/legacy.js + a barrel at utils/index.js. All classes are now named exports. - Vue.prototype.$message in HttpUtil → direct import of `message` from ant-design-vue (Vue 3 has no Vue.prototype). - RandomUtil.randomShadowsocksPassword previously defaulted to SSMethods.BLAKE3_AES_256_GCM from inbound.js, creating a circular import. Replaced with the literal string default. - MediaQueryMixin (Vue 2 mixin) removed. Replaced by composables/useMediaQuery.js — Vue 3 composable returning reactive `isMobile`. - axios-init.js wrapped as setupAxios(); Qs global → npm `qs`. - websocket.js exported as WebSocketClient class; the implicit window.wsClient global is gone — pages instantiate it themselves. - model/{inbound,outbound,dbinbound,setting,reality_targets}.js copied with `export` added on every top-level declaration. Imports between models and utils are wired up explicitly. - subscription.js deferred to Phase 5 (it's a Vue 2 mount, not a util). - App.vue smoke test exercises SizeFormatter / RandomUtil / Wireguard / useMediaQuery so the user can verify Phase 3 with `npm run dev`. Run `cd frontend && npm install && npm run dev` — qs was added so a fresh install is required. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 4 — port login.html to Vue 3 + AD-Vue 4 + Vite 8 First real page in the new toolchain. Multi-page Vite: each migrated page is its own entry. login.html now lives at frontend/login.html with a thin entrypoint at frontend/src/login.js mounting LoginPage.vue. Vite 6 → Vite 8.0.11 (per user request). Requires Node 20.19+ or 22.12+. @vitejs/plugin-vue bumped to ^6.0.6 (peers vite ^8). Ant Design Vue stays on 4.2.6 — there is no AD-Vue 6. Vue 2 → Vue 3 / AD-Vue 1 → AD-Vue 4 syntax changes hit on this page: - new Vue({ el, delimiters, data, methods }) → createApp + <script setup> - mounted() → onMounted() - <template slot="X"> → <template #X> - <a-icon slot="prefix" type="user"> → <template #prefix><UserOutlined /> </template> with explicit @ant-design/icons-vue imports - v-model.trim → v-model:value (AD-Vue 4 uses named v-model on inputs) Three legacy features deferred so Phase 4 stays small: - i18n (Phase 7 wires up vue-i18n) - theme switcher (custom component pending Phase 5) - headline word-cycle animation (purely aesthetic) Run `cd frontend && npm install && npm run dev`, open http://localhost:5173/login.html. With Go panel running on :2053 the form submits real credentials via the configured proxy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5a — theme system + Vite 8 + vue-i18n 11 Bumps Vite to 8.0.11 (npm install picked up 6.4.2 from the stale lockfile; clean install resolves the new constraint). Bumps vue-i18n to 11.1.4 since v10 was just EOL'd. Migrates aThemeSwitch.html — the two-flavor theme picker + global themeSwitcher object — into: - composables/useTheme.js: single reactive `theme` state with toggleTheme / toggleUltra. Boot side-effect applies the stored theme to <body>/<html> before Vue renders; watchEffect persists changes back to localStorage. - components/ThemeSwitch.vue: full menu version for the main panel. - components/ThemeSwitchLogin.vue: login-popover version. AD-Vue 1 → 4 changes hit on this component: - <a-icon type="bulb" :theme="filled|outlined"> dropped — replaced by explicit BulbFilled / BulbOutlined imports from @ant-design/icons-vue, swapped via <component :is="BulbIcon"> - Vue.component('a-theme-switch', { ... }) global registration → SFC + per-page import - this.$message.config(...) (Vue 2 instance method) → message.config(...) imported from ant-design-vue, called once in login.js at boot Login page now surfaces a settings button → popover → theme picker. Known gap: web/assets/css/custom.min.css isn't yet imported into the new bundle, so toggling dark mode currently only re-themes AD-Vue's own components, not the panel chrome. The body class is still toggled so behavior is correct; visual fidelity returns when custom.css is ported or directly imported. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5b — port four shared components to Vue 3 CustomStatistic.vue and SettingListItem.vue are mechanical Vue.component → SFC ports. AppSidebar.vue: AD-Vue 4 dropped <a-icon :type="dynamic">, so the five sidebar icons (dashboard/user/setting/tool/logout) live in a name→component map and render via <component :is>. The legacy <a-drawer slot="handle"> hack is replaced with a sibling fixed- position toggle button. Tab paths take basePath/requestUri as props instead of pulling them from Go template scope. TableSortable.vue: the biggest Vue 3 rewrite of this phase. - $listeners is gone — replaced by inheritAttrs: false + explicit attrs forwarding - scopedSlots: this.$scopedSlots collapsed into Vue 3's unified slots object — just iterate Object.keys(this.slots) and forward - Vue 2 h(tag, { props, on, scopedSlots }, children) → Vue 3 h(tag, { ...props, ...on }, slotsObject) - 'a-table' string → resolveComponent('a-table') so app.use(Antd) registration is honored - inject: ['sortable'] (Options API) → inject('sortable', null) (Composition API) inside the trigger child - beforeDestroy → beforeUnmount - customRow's return shape flattened (no nested props/on/class) Two intentional skips, documented in the migration doc: - aClientTable.html — slot fragments, not a component. Migrates inline with inbounds.html (new Phase 5f). - aPersianDatepicker.html — wraps a Persian-only third-party lib; defer until settings.html lands. Build verified with vite 8.0.11. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): anchor Vite dev proxy so /login.html isn't forwarded The /login proxy entry was matching any path starting with /login — including /login.html, which Vite is supposed to serve itself. Without the Go backend running, this caused ECONNREFUSED noise on every page load. Switched to regex patterns anchored with ^...$ so only the bare backend paths (/login, /logout, /getTwoFactorEnable) and explicit sub-routes (/panel/*, /server/*) get proxied. Static .html files Vite serves directly are no longer matched. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): real dark mode + silence dev proxy ECONNREFUSED noise Two issues from running login.html against no Go backend: 1. Dark mode toggled the body class but didn't actually re-theme any AD-Vue components. The legacy panel relied on custom.min.css which we haven't ported. AD-Vue 4 ships its own dark algorithm — wrap LoginPage in <a-config-provider :theme="{ algorithm }"> driven by our useTheme state, and AD-Vue restyles every component for free. Page chrome (background, card, title) gets explicit .is-dark CSS since the algorithm only covers AD-Vue components. 2. Vite logged every failed proxy attempt loudly. When the Go panel isn't running locally that's pure noise. Added a configure() callback that swallows ECONNREFUSED specifically; real errors (timeouts, 5xx, anything else) still surface. Both fixes are dev-experience only — production build is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): use legacy panel palette for login page dark mode Earlier dark mode used invented colors (#141a26 page bg, #1f2937 card) that didn't match the rest of the panel. Replaced with the actual values from web/assets/css/custom.min.css: light dark ultra-dark bg #c7ebe2 bg #222d42 bg #0f2d32 card #fff card #151f31 card #0c0e12 title #008771 title #fff/.92 title #fff/.92 Drove everything off CSS custom properties on .login-app so the .is-dark / .is-ultra class swap is a few var overrides instead of duplicating selectors. Also restored the legacy card metrics (2rem radius, 4rem 3rem padding, 2rem title) so the new page matches the old panel's geometry, not just its colors. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave layout + recolor for dark mode The wave SVG had inline fill="#c7ebe2" (mint) on the bottom wave, so in dark/ultra-dark mode it rendered as a pale-white blob against the dark page. Stripped the inline fills, drove them off CSS variables that swap with .is-dark / .is-ultra: light: green tints + #c7ebe2 (mint) on the bottom wave dark: #222d42 across all four waves ultra-dark: #0f2d32 The wave was also positioned wrong — anchored to the top 200px of the viewport with absolute positioning. Restored the legacy layout: - .waves-header is fixed to the top of the viewport with z-index -1 so the form floats over it - .waves-inner-header pushes the wave SVG down to ~50vh with a 50vh-tall solid block of the page color - .waves SVG itself is 15vh tall, sitting at the bottom of that block Net effect: top half is solid-colored, then a wavy edge transitions into the rest of the page, with the form centered on top — matching the legacy panel exactly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): bring wave-header to front so the wave actually shows Two layering bugs were hiding the wave entirely: 1. .ant-layout-content had background: var(--bg-page) which painted an opaque rectangle covering the full content area — including the fixed wave-header behind it. Made the layout/content transparent and moved the bg paint up to .login-app (the outer ant-layout). 2. .waves-header had z-index: -1 which on its own was fine, but with .ant-layout-content opaque on top it was doubly buried. Promoted the wave-header to z-index: 0 and gave the form .login-row z-index: 1, so the form sits above the wave and the wave sits above the page-bg. Also set --bg-page to the legacy mint (#c7ebe2) for light mode so the bottom half of the page below the wave matches the legacy panel (was white). Dark mode stays at the surface-100/login-wave palette. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave animation timings + dark page bg Two reasons the bottom wave looked static in dark/ultra-dark: 1. Animation durations were 7s/10s/13s/20s. Legacy uses 4s/7s/10s/13s. The 20s on the bottom wave was so slow that against the low dark- mode contrast it read as motionless. Restored the legacy timings. 2. --bg-page in dark mode was #151f31 (card color / surface-100), but the legacy .under uses surface-200 (#222d42) — that's the color of the bottom half of the page, the same as the wave fill, so the wave appears to flow into the page rather than meeting a hard edge. Now it does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): restore Hello/Welcome headline cycle on login Earlier I deferred the legacy headline word-cycling animation as "purely aesthetic". Restored it: the title now alternates between 'Hello' and 'Welcome' every 2 seconds, matching the legacy panel. The legacy implementation toggled .is-visible / .is-hidden classes on two <b> elements via setTimeout chains and DOM querying. Replaced with a reactive ref + Vue 3 <Transition mode="out-in"> so the fade between words is declarative — no manual DOM manipulation, and the interval is properly cleaned up in onBeforeUnmount. The earlier "Welcome to 3x-ui" string was wrong on two counts: it should be just "Welcome", and it should be one of two cycling words with "Hello" preceding it. Ultra-dark palette already matched legacy after the prior wave timing fix; no additional changes needed there beyond the animation speeds that now also apply to ultra-dark via the shared CSS rules. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): correct dark login bg + give ultra-dark wave real contrast Two related fixes: 1. Default-dark wave-header bg was wrong. I had #0a2227, but that's the *ultra-dark* override; default dark uses --dark-color-background = #0a1222. Now the dark-mode top half is the legacy purple-blue instead of teal. 2. Ultra-dark wave fill is intentionally near-identical to its bg in the legacy palette (#0f2d32 vs #0a2227, ~5/11/11 RGB delta), which makes the wave look static even though the animation is running. Bumped --wave-fill / --wave-fill-bottom to #1f4d52 in ultra-dark only — far enough above the bg that the motion reads, while staying within the same teal hue family. Also corrected ultra-dark --bg-page back to #0f2d32 (was briefly #0c0e12, which is the card color, not the page color). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): drop ultra-dark bottom-wave seam line Last fix made the wave fill #1f4d52 in ultra-dark for both top-three waves and the bottom wave, which gave visible motion but exposed a hard horizontal line where the bottom wave's flat lower edge met the page bg (#0f2d32). The user noticed it as "the wave at the bottom not moving its like a line" — they were seeing the SVG's clipped bottom edge, not the wave itself. Solution: only the top three waves get the brighter fill (those carry the visible motion). The bottom wave reverts to #0f2d32 = --bg-page, so its flat bottom edge merges seamlessly into the page below. Net effect: motion is still visible (from waves 2 and 3), and there's no seam line at the bottom of the SVG. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-i — index.html dashboard shell Replaces the smoke-test App.vue with a real IndexPage shell so the /index.html route now boots the actual dashboard layout in Vue 3: - a-config-provider drives AD-Vue 4's dark algorithm from useTheme (same pattern as LoginPage) - AppSidebar (Phase 5b component) is wired in with basePath + requestUri props - a-spin loading state with placeholder card while we build out the rest of the page - Page palette mirrors the legacy: light #f0f2f5, dark #0a1222 (--dark-color-background), ultra-dark #21242a The 1,805-line legacy index.html is too big for one commit. Split into five sub-phases on the todo list: ii) status cards + /server/status polling, iii) xray status card, iv) logs/backup/panel-update modals, v) custom-geo section. frontend/src/App.vue and frontend/src/main.js (smoke-test scaffold) are removed — both purposes now served by IndexPage and index.js. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-ii — live status cards on the dashboard Adds the CPU / memory / swap / disk dashboard cards to IndexPage, backed by a useStatus() composable that polls /panel/api/server/status every 2 s and a Status / CurTotal model ported from the legacy inline classes in index.html. - models/status.js — Status & CurTotal classes (CurTotal exposes reactive .percent and .color computed-style getters; Status maps the API payload + xray state to color/message strings) - composables/useStatus.js — 2s polling with shallowRef so each fetch swaps the whole Status object atomically. WebSocket integration intentionally deferred — the legacy panel falls back to this same 2s polling when its websocket drops, so we ship the proven path first and add WS on top in a later sub-phase. - pages/index/StatusCard.vue — four a-progress dashboard widgets in a 2x2 grid (mobile collapses to a 1x4). CPU widget exposes a history button; the modal it opens is part of 5c-iv. - IndexPage now consumes both, plus useMediaQuery so the layout responds to viewport changes. AD-Vue 4 changes: <a-icon type="area-chart"|"history"> dropped in favor of explicit AreaChartOutlined / HistoryOutlined imports. <a-tooltip slot="title"> → <template #title>. i18n strings still hardcoded English (Phase 7 wires up vue-i18n). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iii — xray status card + stop/restart controls XrayStatusCard.vue renders the right-hand card on the dashboard: - Title with mobile-only version tag (matches the legacy collapse) - Animated badge for the running/stop/error states. The pulsing dot comes from xray-pulse keyframes (renamed from runningAnimation in legacy custom.min.css). Color rings on the badge use the legacy's per-state border-color overrides on .ant-badge-status-processing. - Error state replaces the badge with a popover that surfaces the multi-line errorMsg + a logs shortcut. - Action row at the bottom: optional logs (when ipLimitEnable), stop, restart, and version switch. IndexPage now wires: - POST /panel/api/server/stopXrayService and /restartXrayService, followed by a refresh() so the status card reflects the new state without waiting for the next poll tick - POST /panel/setting/defaultSettings to read ipLimitEnable - Stub handlers for the panel-logs / xray-logs / version-switch / cpu-history modals — those land in 5c-iv AD-Vue 4 changes hit on this card: - <a-icon type="bars|poweroff|reload|tool"> → explicit BarsOutlined / PoweroffOutlined / ReloadOutlined / ToolOutlined - <span slot="title|content"> → <template #title|#content> - The .xray-*-animation classes ship as global <style> (not scoped) so they pierce AD-Vue's internal .ant-badge-status-* DOM. i18n still hardcoded English; Phase 7 wires vue-i18n. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (a) — panel update / logs / backup modals Adds three of the six dashboard modals plus a Quick Actions card that surfaces them. The remaining three (xray logs, version picker, CPU history sparkline) ship in 5c-iv-b. - PanelUpdateModal.vue — current vs latest version, "update now" button. Confirm dialog → POST /panel/api/server/updatePanel, then poll /server/status for up to 90s until the new panel answers, then reload. - LogModal.vue — panel logs viewer. Filters: rows (10-500), level (debug/info/notice/warning/error), syslog toggle. Auto-fetches on open and on every filter change. Color-coded timestamps and levels via inline span styles. Download button writes the raw log to x-ui.log via FileManager.downloadTextFile. - BackupModal.vue — db export (window.location to /getDb) and import (FormData upload to /importDB, then panel restart + reload). - Quick Actions card surfaces Logs / Backup / Update buttons and shows an orange update badge (extra slot) when an update is available. Modal-busy pattern: long-running operations (update, import) emit a `busy` event with a tip; IndexPage flips its a-spin overlay so the user sees a loading message while the panel is restarting. AD-Vue 4 changes: - v-model on <a-modal> renamed to v-model:open - v-model on <a-input>/<a-select>/<a-checkbox> uses the named v-model:value / v-model:checked pattern - <a-icon type="..."> dropped — explicit Ant icon imports (BarsOutlined, CloudServerOutlined, CloudDownloadOutlined, DownloadOutlined, UploadOutlined, SyncOutlined) - Modal.confirm() replaces this.$confirm() since setup() has no `this` Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (b) — cpu-history / xray-logs / xray-version modals Wires up the three remaining dashboard buttons that were stubbed in 5c-iv (a): the CPU history button on StatusCard, the xray-logs button in XrayStatusCard's error popover and ipLimitEnable action, and the "Switch xray" button in XrayStatusCard's action footer. - Sparkline.vue: shared SVG line chart (composition-API port of the inline Vue 2 component). Per-instance gradient id avoids defs collisions between sparklines on the same page. - CpuHistoryModal.vue: bucket dropdown (2m/30m/1h/2h/3h/5h) drives GET /panel/api/server/cpuHistory/{bucket}; renders via Sparkline. - XrayLogModal.vue: rows + filter + direct/blocked/proxy checkboxes; POST /panel/api/server/xraylogs/{rows} returns access-log entries rendered as a colored HTML table; download button serializes to text. - VersionModal.vue: collapse with Xray panel (radio list of versions from getXrayVersion, install via installXray/{version}) and Geofiles panel (per-file reload + Update all). CustomGeo collapse panel is Phase 5c-v. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-v — custom-geo section in VersionModal Adds the third collapse panel ("Custom geo") that lets users register external geosite/geoip files referenced by routing rules via ext:<filename>:tag. Backend endpoints are unchanged. - CustomGeoSection.vue: bordered table over /panel/api/custom-geo/list with per-row edit, download (refetch), and delete actions, plus an Add button and Update-all. Lazy-loads the list when the parent collapse opens this panel — closed panels don't fetch. - CustomGeoFormModal.vue: shared add/edit form with the same alias regex (^[a-z0-9_-]+$) and URL validation as legacy. Type and alias are immutable when editing — backend rejects changes anyway. - ext:<filename>:tag value is click-to-copy via ClipboardManager. - Relative time is computed inline (no moment dep); tooltip shows the absolute timestamp. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-i — settings page shell + dirty tracking Adds the settings entry as a new Vite multi-page input. Lays down the shared page chrome (sidebar, save bar, restart, security alert) and the AllSetting fetch/dirty-poll lifecycle so 5d-ii through 5d-vi can drop in tab partials without re-implementing it. - settings.html + src/settings.js: third Vite entry; mounts SettingsPage. - SettingsPage.vue: page chrome with the legacy two-button save/restart bar, conf-alerts banner, and 5 a-tabs (4 always-visible + the formats tab gated on subJsonEnable || subClashEnable). Each tab body is an a-empty placeholder until 5d-ii…vi fill them in. - useAllSetting.js composable: POST /panel/setting/all on mount, mirrors the legacy 1s busy-loop dirty check via setInterval, and exposes fetchAll/saveAll. saveDisabled flips off as soon as the user diverges from the server snapshot. - restartPanel rebuilds the URL (host/port/scheme/base path) from the saved settings so users land on the new endpoint after a port or cert change. - models/setting.js: adopts the @/utils alias and a leading file-level doc — semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-ii — settings General tab Ports the panel/general partial (the largest single tab) — six collapse panels: General, Notifications, Certificates, External traffic webhook, Date and time, LDAP. - GeneralTab.vue receives the reactive AllSetting via props and binds fields directly with v-model:value; SettingsPage stays the sole fetch/save owner. - remarkModel/remarkSeparator surfaced as computed v-models that read+write the underlying single-string field (legacy stores them packed as <separator><orderedKeys>, e.g. "-ieo"). - LDAP inbound-tags select binds to a CSV ↔ array computed; inbound options come from /panel/api/inbounds/list on mount. - Language select stays cookie-based via LanguageManager and reloads on change — same UX as legacy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iii — settings Security tab + 2FA modal Ports the panel/security partial: change-credentials form and 2FA toggle. The 2FA modal is a new shared component since enabling 2FA, disabling 2FA, and changing credentials all funnel through it with slightly different copy. - TwoFactorModal.vue: 'set' flow renders a QR code + manual key + a 6-digit verifier; 'confirm' flow renders just the verifier. The parent passes a confirm(success) callback that fires only when the entered code matches the live TOTP value (otpauth lib). - SecurityTab.vue: holds the local user form (oldUsername/oldPassword/ new*), POSTs /panel/setting/updateUser, and on success force-redirects to logout. When 2FA is on, the credentials change goes through the confirm-modal first. - toggleTwoFactor leaves the switch read-only (the v-bound :checked matches AllSetting) and only flips after the modal succeeds, so cancelling out leaves state unchanged. - Adds otpauth ^9.5.1 dep (qrious was already present). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iv — settings Telegram tab Ports the panel/telegram partial: bot enable/token/chatId/lang in the General panel, schedule/backup/login/CPU-threshold in Notifications, and proxy/API-server overrides in the third panel. All bindings live on the shared AllSetting reactive — no fetch/save logic in this tab. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-v — settings Subscription general tab Ports the subscription/general partial — four collapse panels covering the master enable switches, presentation/template fields, certs, and update interval. - Sub path goes through a strip-on-input + normalize-on-blur computed: legacy stripped `:` and `*` and ensured the value starts and ends with a single `/` — same here. - Both `subEnableRouting` and the announce/profile/title/support URLs are bound directly on AllSetting. - The "Subscription URI override" placeholder mirrors the legacy pattern for the manual full-URL form. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-vi — settings Subscription formats tab Ports the subscription/json partial — paths/URIs for the JSON and Clash formats plus the four packed-JSON sub-fields: fragment, noises, mux, and direct routing rules. - subJsonFragment / subJsonMux / subJsonNoises / subJsonRules are each a JSON string on the wire; the tab exposes their fields as computed v-models that read+write the underlying JSON. Toggling a top-level switch off resets the field to "" (matches legacy semantics). - Direct routing rules surface the IP and domain entries of the seed rule array as multi-select tag inputs; setting/removing tags edits the rules array in place rather than rebuilding it from scratch, so manually-added rules are preserved. - Tab is gated on subJsonEnable || subClashEnable in the parent (only rendered when the user actually opted into one of those formats). This closes Phase 5d — full settings page parity with the legacy panel across all five tabs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): route /panel/<route> to migrated pages in dev The sidebar links to production-style URLs like /panel/settings, but in dev that gets proxied to the legacy Go template — which fails because we haven't loaded the legacy asset chain. Add a proxy bypass so /panel and /panel/settings are served from index.html / settings.html on the Vite dev server itself. Unmigrated routes (inbounds, xray) still proxy to Go. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(csrf): expose token endpoint for SPA pages and fetch it from axios The legacy panel pages got their CSRF token from a <meta name="csrf-token"> tag rendered by Go. SPA pages built by Vite don't have that, so every unsafe (POST/PUT/DELETE) request from them was hitting CSRFMiddleware with no token and getting 403 — visible as the settings page being stuck on "Loading…" because POST /panel/setting/all failed. - web/controller/xui.go: GET /panel/csrf-token returns the session token. Lives under the xui group so checkLogin still gates it; the CSRFMiddleware on the same group is a no-op for GET. - frontend/src/api/axios-init.js: cache the token at module scope and lazy-fetch it when a non-safe request needs one. Seed from the meta tag first when present (legacy compat). On a 403 response, drop the cache and retry once — handles the case where a server restart rotated the token after the SPA loaded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): keep sidebar links absolute when basePath is empty The dashboard sidebar built tab keys as basePath + 'panel/...'. In dev the window-injected basePath is '' so the resulting key was a relative path like 'panel/settings'. When the browser resolved that against the current /panel/settings URL it produced /panel/panel/settings — visible as broken navigation between Dashboard and Settings. Force a leading slash so the keys are always absolute regardless of whether the host injected a basePath. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-i — inbounds page shell + list fetch Adds the inbounds entry as a fourth Vite multi-page input and wires /panel/inbounds through the dev proxy bypass. Lays down the page chrome (sidebar, summary statistics card, refresh button) and the fetch lifecycle composable so 5f-ii onward can drop in the table columns and the modals without re-implementing it. - inbounds.html + src/inbounds.js: fourth Vite entry; mounts InboundsPage. - InboundsPage.vue: sidebar + summary card (totals over up/down, all-time, inbound count, client tags) + a basic table with enable/ remark/port/protocol/traffic/expiry columns. Row actions, popovers, search/filter, auto-refresh, and the WebSocket delta path are all deferred to subsequent 5f subphases. - useInbounds.js composable: GET /panel/api/inbounds/list + POST /panel/api/inbounds/onlines + POST /panel/api/inbounds/lastOnline + POST /panel/setting/defaultSettings, then computes the per-inbound clientCount roll-ups (active/deactive/depleted/expiring/ online/comments) the table popovers consume. - models/dbinbound.js + models/inbound.js: switched the legacy-utils import to the @/utils alias for consistency with the rest of the app. Semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-ii — inbound list table + search/filter + auto-refresh Fleshes out the inbound list with the full column set, search & filter toolbar, row enable toggle wired to /panel/api/inbounds/setEnable/:id, and a per-row action dropdown that emits events the parent will route to modals as those land in 5f-iii through 5f-vii. - InboundList.vue (new): toolbar (Add inbound + General actions dropdown + Refresh + auto-refresh popover), search-or-filter switch with the legacy radio buttons (Active/Disabled/Depleted/Depleting/ Online), and a a-table with desktop and mobile column variants. Cells use AD-Vue 4's #bodyCell slot — protocol/clients/traffic/ allTime/expiry/info cells render the same popovers and tags as legacy. Row enable switch is optimistic with rollback on POST failure. - visibleInbounds computed mirrors the legacy search and filter projection: deep search through dbInbound + clients, or filter reduces inbound.settings.clients to the selected bucket so the table only shows matching client rows. - Auto-refresh interval is read/written to localStorage with the same keys (`isRefreshEnabled`, `refreshInterval`) as the legacy panel. WebSocket delta updates are still deferred. - Action menu emits event payloads {key, dbInbound}; the parent currently shows a "coming in later 5f subphase" toast for each. Modals (edit/qr/clone/delete/reset/info/clients) land in 5f-iii through 5f-vii. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): wrap popover-table rows in <tbody> Vue's template compiler warned that <tr> can't be a direct child of <table> per the HTML spec; the browser silently inserts a <tbody> wrapper but Vue's SSR/hydration path doesn't, which can cause hydration mismatches. Add explicit <tbody> in both popover tables (traffic + mobile-info). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii — inbound add/edit modal + delete/clone/reset Wires up the inbound CRUD flows. The protocol-specific and transport- specific forms are still ahead in 5f-iii-b — for now the modal exposes those as JSON textareas so users can both edit existing inbounds without losing settings and create new ones from default templates. - InboundFormModal.vue: tabbed modal with a full Basics tab (enable, remark, protocol, listen, port, total GB, traffic reset, expiry date) and three JSON-edit tabs (Settings, Stream, Sniffing). Add mode stamps a fresh template per protocol via Inbound.Settings.getSettings(protocol); changing the protocol in add mode restamps the JSON. Edit mode pretty-prints the existing JSON so the user sees the same fields they save back. - POST /panel/api/inbounds/add or /panel/api/inbounds/update/:id on submit; on success the parent refreshes the list and the modal closes. Malformed JSON in any of the three textareas surfaces a message.error and aborts the save without losing user input. - InboundsPage.vue: wires the row action menu to real handlers — edit (opens the modal in edit mode), delete, reset-traffic, clone, reset-clients, del-depleted-clients all go through Modal.confirm and refresh on success. General actions menu wires reset-inbounds / reset-clients / del-depleted-clients the same way. Remaining actions (qrcode/info/import/export/copyClients) still toast as "coming soon" — those land in 5f-iv and 5f-v. - Adds dayjs ^1.11.20 dep for the a-date-picker v-model interop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iv — client add/edit + bulk-add modals Wires per-inbound client management. Both flows go through the same addClient/updateClient endpoints as legacy; the modals just funnel the form state into the right shape (`{id, settings: '{"clients": [...]}'}`). - ClientFormModal.vue: protocol-aware single-client editor — email/ password/id/auth/security/flow/subId/tgId/comment/ipLimit/totalGB/ expiry/renewal fields are shown/hidden per protocol like legacy. Edit mode displays the per-client traffic stats with a reset button; IP-limit log is read on click and clearable. Random helpers (sync icon next to each label) regenerate UUID/email/ password/sub-id values. - ClientBulkModal.vue: 1–500 clients in one POST, with the legacy five email-generation modes (Random / +Prefix / +Num / +Postfix / Pure-Prefix-Num-Postfix). Builds clients via the protocol-aware factory and concatenates their toString() output into a single settings.clients JSON array. - InboundsPage.vue: opens both modals from the row action menu (`addClient` / `addBulkClient`). They both refresh the inbound list on success. - Outstanding row actions still toast as "coming soon": qrcode, showInfo, copyClients, clipboard. Those land in 5f-v / 5f-vi. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-v — inbound info + QR-code modals Wires the row "info" and "qrcode" actions and ports the legacy inbound_info_modal end-to-end. The info modal handles every protocol the legacy panel did: • multi-user (VMess/VLess/Trojan/SS-multi/Hysteria) — per-client table + share links + per-link QR; • SS single-user — share link + QR; • WireGuard — full peer table with downloadable peer-N.conf and a wg:// share link per peer; • Mixed/HTTP/Tunnel — connection-detail tables. - QrPanel.vue: shared link card (header tag, copy button, optional download button, optional QR canvas, monospace footer with the raw value). Per-instance QRious instances are repainted on value/size change. - InboundInfoModal.vue: full info modal. Subscription URL block keys off subSettings.subURI/subJsonURI; IP-log lazy-loads on open and surfaces refresh + clear; tg-id, last-online, depleted/enabled tags all match legacy. - QrCodeModal.vue: lighter modal used for the row "qrcode" action on SS-single and WireGuard inbounds (just the QRs, no info table). - InboundsPage.vue: wires both flows. checkFallback() reproduces the legacy logic — when an inbound listens on a unix-socket fallback (`@<name>`), the link generator is pointed at the root inbound that owns the listen address so QRs/links carry the public host:port + the right TLS state. Multi-client navigation (focusing a specific client's links) is deferred to 5f-vi where the per-inbound expand- row table will pass the email through. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vi — per-inbound client expand-row table Each multi-user inbound row in the list now expands to show its client roster, mirroring the legacy aClientTable component. - ClientRowTable.vue: inner a-table with full desktop column set (action icons / enable / online / client-with-status-dot / traffic with progress bar / all-time / expiry with reset cycle) and a collapsed mobile variant (single dropdown menu + popover info). Self-contained: stats are looked up via a per-inbound email->stats Map; per-client confirms (reset/delete) live on the row. - The component emits typed events (edit/qrcode/info/reset-traffic/ delete/toggle-enable) — InboundsPage routes them back to the existing client and info modals (with `findClientIndex` so the modal opens focused on the right client). - InboundList.vue: hooks ClientRowTable into the a-table's expandedRowRender slot; row-class-name `hide-expand-icon` and a scoped CSS rule hide the chevron for non-multi-user inbounds (HTTP/Mixed/Tunnel/WireGuard/SS-single) so they keep looking flat. - toggle-enable-client routes through updateClient with the same `{id, settings: '{"clients": [...]}'}` shape as the other modals, so backend parsing stays single-pathed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii-b — replace inbound modal JSON textareas with structured forms Rewrites InboundFormModal to look like the legacy panel: structured forms for the common case, with a compact "Advanced (JSON)" fallback for the rare bits we don't yet have UI for. Tabs: • Basics — enable/remark/protocol/listen/port/total/trafficReset/expiry • Protocol — protocol-aware: VMess/VLess/Trojan/SS-multi/Hysteria in add mode embed an inline first-client form (email + ID/password/auth, security, flow, subId, comment, total GB, expiry); edit mode shows a clients-count summary table; VLess: decryption/encryption inputs; SS: method dropdown that re-randomizes password and propagates method change to the multi-user array (matches legacy SSMethodChange); HTTP/Mixed: accounts table with add/remove rows + Mixed auth/udp/ip toggles; Tunnel: address/port/network/followRedirect; WireGuard: secretKey/pubKey (regen via Wireguard.generateKeypair) + per-peer fields with PSK regen + allowedIPs add/remove + keepAlive. • Stream — only when canEnableStream(); transport selector with structured forms for TCP (proxy-protocol, http camouflage), WS (host/path/heartbeat/headers), gRPC (serviceName, multiMode), HTTPUpgrade (host/path). KCP/XHTTP fall back to the Advanced tab with an alert banner. Security selector with TLS (sni/alpn/ fingerprint) and Reality (target/serverNames/keypair-gen via /panel/api/server/getNewX25519Cert / shortIds / fingerprint). • Sniffing — enabled/destOverride/metadataOnly/routeOnly/ ipsExcluded/domainsExcluded as structured fields. • Advanced (JSON) — raw streamSettings + sniffing JSON for users reaching KCP/XHTTP/sockopt/finalmask/full TLS cert arrays. The stream JSON is auto-synced from the live model whenever the structured fields change. State source of truth is a deeply-reactive Inbound + DBInbound pair cloned on open; submit serializes via inbound.settings.toString() + inbound.stream.toString() so the wire shape matches the legacy panel byte-for-byte. streamNetworkChange semantics (clear flow when TLS/Reality unavailable, reset finalmask.udp when not KCP) are preserved. Vision Seed for VLess + finer-grained TCP HTTP camouflage + the full TLS cert/ECH editor will land in 5f-iii-c. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vii — shared text/prompt modals + remaining export/import wiring Wires up the last batch of inbound row + general actions that were toasting "coming soon": export-inbound-links, export-subs (per-inbound and global), export-all-links, import-inbound, and the clipboard JSON peek. Two small shared components back them — both can be reused by the xray page later. - TextModal.vue (shared): read-only multi-line viewer with a copy button and an optional download button when fileName is set. Replaces the legacy txtModal which the inbounds page used for every link export. - PromptModal.vue (shared): generic title + input/textarea + confirm callback, with the legacy keybindings (Enter submits in single-line mode; Ctrl+S submits in textarea mode). Used here for import-inbound but also by xray-config edits in Phase 6. - InboundsPage.vue: drops the toast stubs for `import`/`export`/`subs` on the general-actions menu and `export`/`subs`/`clipboard` on the per-row menu, routing each through openText / openPrompt + the appropriate model helper (genInboundLinks, etc.). The copyClients cross-inbound modal stays toast-stubbed — that's its own dedicated legacy modal worth its own commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-i — xray page scaffold + Advanced JSON tab The fifth and last legacy page comes online. Tabs are scaffolded with a-empty placeholders for the structured editors (Basics / Routing / Outbounds / Balancers / DNS) so navigation is stable; the Advanced (JSON) tab is fully functional and lets power users edit the raw xraySetting tree exactly like the legacy CodeMirror pane. - xray.html + src/xray.js: fifth Vite multi-page entry, mounted as XrayPage; vite.config.js routes /panel/xray and /panel/xray/ to it through the dev proxy bypass alongside the other pages. - XrayPage.vue: page chrome with the Save / Restart-xray bar, restart- output popover (surfaces /panel/xray/getXrayResult content when startup fails), 6 a-tabs, and a textarea-backed Advanced JSON editor. CodeMirror is intentionally not pulled in — the textarea works for every modern browser and keeps the bundle slim while structured editors land in 6-ii through 6-v. - useXraySetting.js composable: POST /panel/xray/ on mount, mirrors the settings-page 1s busy-loop dirty check for both xraySetting and outboundTestUrl, and exposes saveAll + restartXray. The dirty flag relies on string equality of the pretty-printed JSON, so reformat-only edits don't enable Save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-ii — xray Basics tab structured editor Replaces the placeholder on the Basics tab with a structured form for the most-touched fields of the xray template — outbound + routing strategy, log levels, traffic stat counters, and the "basic routing" shortcuts (block torrent / IPs / domains, direct IPs / domains, IPv4 forced, WARP / NordVPN routing). - useXraySetting.js: hoists a parsed `templateSettings` reactive alongside the JSON string, with two cooperating watches that keep them in sync. Editing structured fields stringifies into xraySetting for the dirty-poll + Advanced JSON tab; editing the JSON re-parses into templateSettings only when valid, so structured tabs stay readable mid-edit. - BasicsTab.vue: collapse panels mirror the legacy partial — General, Statistics, Logs, Basic routing. Every input is a computed v-model reading/writing into templateSettings; the routing-rule shortcuts funnel through ruleGetter/ruleSetter which match the legacy templateRuleGetter/templateRuleSetter behavior (replace-first, drop-duplicates, pop-the-rule-when-empty). Direct/IPv4 setters also call syncOutbound() to provision/prune the matching outbound. - XrayPage.vue: imports BasicsTab + derives `warpExist`/`nordExist` from the parsed templateSettings. WARP/NordVPN provisioning modals are still placeholders that toast — those land in 6-v with the routing/outbound editors. Default tab flips back to Basics so users land on the structured editor. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iii — xray Routing tab + rule modal Replaces the Routing tab placeholder with a full editor for templateSettings.routing.rules: - RoutingTab.vue: a-table over the parsed rules with the legacy six- column layout (action / source / network / destination / inbound / outbound) and the same "lead value + N more" pill renderer for multi-value criteria. Mobile drops source/network/destination for readability. Per-row dropdown handles edit / move-up / move-down / delete; the array-mutation reordering replaces the legacy jQuery Sortable drag handle without pulling in a sortable lib. - RuleFormModal.vue: full form mirroring xray_rule_modal.html — CSV inputs for sourceIP/sourcePort/vlessRoute/ip/domain/user/port, Network select, Protocol multi-select, Attrs key/value pairs, inbound-tag multi-select sourced from templateSettings.inbounds + parent inboundTags + dnsTag, outbound-tag single-select sourced from templateSettings.outbounds + clientReverseTags, and balancerTag from templateSettings.routing.balancers. Submit serializes via the same shape the legacy `getResult` produces (CSV → array, drop empty fields). - XrayPage.vue: imports RoutingTab and exposes inboundTags + clientReverseTags from useXraySetting so the modal can populate its tag pools. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iv — xray Outbounds tab + outbound modal Replaces the Outbounds tab placeholder with a full table + add/edit flow. The 1.3k-line legacy outbound modal is condensed to a tabbed modal with structured Basics fields (tag/protocol/sendThrough/domain strategy) and JSON tabs for the protocol-specific settings + stream trees — same approach the Inbound modal uses, and a power user can still edit the same trees via the page-level Advanced (JSON) tab. - useXraySetting.js: adds fetchOutboundsTraffic + resetOutboundsTraffic + testOutbound. Test states are tracked per outbound index so the row's Test button can show loading + the Test-result column can render the response delay / status / error. - OutboundsTab.vue: full table (action / identity / address / traffic / test result / test) plus a card-list mobile variant with the same row dropdown (set-first / edit / move up/down / reset traffic / delete). outboundAddresses() reproduces the legacy findOutboundAddress logic so each protocol's host:port list is rendered consistently. Add/edit go through OutboundFormModal, delete goes through Modal.confirm, reset traffic posts to /panel/xray/resetOutboundsTraffic with the row's tag (or "-alltags-" from the toolbar). - OutboundFormModal.vue: tag/protocol/sendThrough/domainStrategy on the Basics tab; settings + streamSettings as raw JSON on their respective tabs. Tag-collision check happens client-side before emitting; malformed JSON aborts the save with a message.error. - XrayPage.vue: imports OutboundsTab and wires the test action to the composable's testOutbound helper. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-v — xray Balancers tab + DNS placeholder Brings Balancers to full parity with the legacy panel and adds a DNS tab placeholder that exposes the full dns/fakedns trees as JSON so users can edit them without falling through to Advanced. - BalancerFormModal.vue: tag (with duplicate-tag warning across other balancers), strategy (random/roundRobin/leastLoad/leastPing), selector tag-mode multi-select sourced from existing outbound tags + free-form additions, fallback. Disable-on-invalid is driven by the duplicateTag + emptySelector computed flags. - BalancersTab.vue: empty state with a single "Add balancer" CTA; populated state shows the legacy 4-column table (action / tag / strategy / selector / fallback) with per-row edit + delete in a dropdown. On submit the wire shape preserves the `strategy: { type }` nesting only when the strategy is non-default, matching the legacy emit. Tag renames also chase across routing.rules.balancerTag references so existing rules don't dangle. - DnsTab.vue: master enable switch + raw JSON for `dns` and `fakedns`. Legacy had a dedicated server-by-server editor + a fakedns row editor; both are big enough to deserve their own commits, and the JSON path supports every field today. WARP / NordVPN provisioning modals still toast as "coming soon" — those are third-party API integrations worth their own commits. The xray page now has structured editors for Basics / Routing / Outbounds / Balancers and JSON editors for DNS / Advanced — every xray tab the legacy panel offered is functional. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(server): Phase 8 — cut HTML routes over to web/dist/ Production cutover. Every user-facing HTML route now serves the Vue-3-built bundle from web/dist/ instead of rendering the legacy Go template; the long-hashed Vite assets are served at /assets/ from the same embedded filesystem. The legacy templates in web/html/ and the legacy static tree in web/assets/ are kept on disk for now in case a quick revert is needed, but nothing the binary serves references them. What changed: - web.go: a new //go:embed dist/* feeds the controller package via a SetDistFS hand-off before controller construction. The static /assets/ route is rebound: in dev to web/dist/assets/ on disk so Vite's incremental rebuilds show up live; in prod to the embedded dist via wrapDistFS (rooted one level deeper than wrapAssetsFS). - controller/dist.go: serveDistPage helper used by every HTML handler. Reads dist/<name> from the embedded FS and applies two transforms before sending: 1. injects <script>window.__X_UI_BASE_PATH__="..."</script> just before </head> so AppSidebar links resolve under the panel's basePath. 2. when basePath != "/", rewrites Vite's absolute /assets/ URLs to <basePath>assets/ so installs running under a custom URL prefix load the bundle where the static handler lives. HTML responses go out with no-cache so panel upgrades reach users on the next refresh; hashed JS/CSS stays cacheable. - controller/index.go: IndexController.index now serves dist/login.html for logged-out callers (the redirect for logged-in users is unchanged). - controller/xui.go: XUIController.{index,inbounds,settings,xraySettings} each become a one-line wrapper around serveDistPage. Smoke checklist for the maintainer: - run `cd frontend && npm run build` to refresh web/dist/ before building the Go binary (the embed snapshot is taken at compile time); - visit /panel/, /panel/inbounds, /panel/settings, /panel/xray and confirm each loads its Vue page; - log out and log back in to verify the login flow; - confirm the sidebar links navigate correctly under your install's basePath; - POST flows (e.g. saving settings) still need the CSRF token — that endpoint (/panel/csrf-token, added earlier) is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-vi — WARP + NordVPN provisioning modals Replaces the toast stubs on the Basics tab and Outbounds toolbar with the legacy WARP + NordVPN provisioning flows. Both modals now stage their wireguard outbounds back into templateSettings.outbounds through the same event channels OutboundsTab uses, so the existing add / reset / delete / refresh-traffic surface keeps working. - WarpModal.vue: empty state shows a single Create button that generates a wireguard keypair locally (Wireguard.generateKeypair) and posts it to /panel/xray/warp/reg; populated state surfaces the access_token / device_id / license_key / private_key, lets the user upgrade to WARP+ via /panel/xray/warp/license, refreshes the account info from /panel/xray/warp/config (plan / quota / usage in human-readable bytes), and stages a wireguard outbound with the WARP-specific reserved-byte encoding pulled from client_id. Add / Reset / Delete go through events the parent routes back to templateSettings.outbounds. - NordModal.vue: dual-tab login (NordVPN access token → /panel/xray/nord/reg, or paste a NordLynx private key → /panel/xray/nord/setKey). Once authenticated, country / city / server selectors fetch from /panel/xray/nord/{countries,servers}, servers sort by load ascending, the lowest-load server in the current city auto-selects. Reset emits oldTag/newTag so the parent renames matching routing rules in place; logout emits a remove-routing-rules event with prefix `nord-` to purge any dangling references. - XrayPage.vue: holds warpOpen / nordOpen flags, ensures the outbounds array exists before mutating it, and wires the modal events (add-outbound / reset-outbound / remove-outbound / remove-routing-rules) to in-place edits of templateSettings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 7 — vue-i18n wired up + login page translated Sets up vue-i18n on top of the panel's existing TOML translation files. The Go side stays the source of truth — translators continue to edit web/translation/*.toml; a sync script snapshots those files into per-locale JSON the Vue bundle imports. The login page is translated end-to-end as a worked example; remaining pages can be converted incrementally without infrastructure churn. What's in the box: - scripts/sync-locales.mjs: small TOML→JSON converter that walks web/translation/*.toml and writes frontend/src/locales/<code>.json. Handles the narrow subset of TOML the panel uses (flat key/value pairs + dotted [section.subsection] heads). Wired as a `prebuild` + `predev` script so production builds always include the latest strings without a manual step. - src/i18n/index.js: createI18n() in composition mode with all 13 locales emitted as their own Vite chunks. The active locale (read from the same `lang` cookie LanguageManager has always managed) plus the en-US fallback are eagerly loaded; the rest are dynamically importable via a loadLocale(code) helper. This keeps the per-page bundle the user actually downloads small — only ~30 KB of strings end up in the initial payload, vs ~220 KB if all 13 were eager. - All five page entries (index/login/settings/inbounds/xray) wire the i18n plugin into createApp via .use(i18n). - LoginPage.vue: t(...) replaces hardcoded English on the username / password / 2FA placeholders, the submit button label, and the Settings popover title. The Hello/Welcome headline cycle stays hardcoded — those are stylistic, not labels. The 'Hello'/'Welcome' cycle stays in English deliberately; the rest of the migration's components still ship hardcoded English and will be converted page by page in follow-up commits. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate page chrome — sidebar, save bars, tabs, summary cards Replaces hardcoded English with t() calls in the components every user sees on every page load. The translations themselves come from the existing TOML files via the sync script — no new strings, no new locale keys. Per component: - AppSidebar.vue: 5 menu titles (dashboard / inbounds / settings / xray / logout). Computed so the sidebar re-renders when the cookie-driven locale flips on reload. - IndexPage.vue: Quick actions card title + Logs / Backup / Up-to- date / Update buttons. - StatusCard.vue: CPU / Memory / Swap / Storage labels + logical-processors / frequency tooltips. - XrayStatusCard.vue: card title + error popover header + Stop / Restart / Switch xray action labels (kept the v-prefix version string as-is — it's content, not a label). - SettingsPage.vue: 5 tab titles + Save / Restart-panel buttons + unsaved-changes warning. - XrayPage.vue: 6 tab titles + Save / Restart-xray buttons + unsaved-changes warning. - InboundsPage.vue: 5 summary-stat card titles. - InboundList.vue: 10 column titles (computed for live locale), Add inbound / General actions buttons + every dropdown menu item, search placeholder, filter radio labels, popover titles (disabled / depleted / depleting / online), traffic + info popover row labels. Total: ~75 strings localised across 8 files. The remaining English labels live in the per-tab settings forms, the form modals (Inbound / Client / Outbound / Rule / Balancer / WARP / Nord), and the per-row table cell helpers — all incremental work that doesn't touch infrastructure. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate every remaining English string on the index page Closes the index page's i18n coverage. Combined with the page-chrome commit, every label users see on the dashboard is now sourced from the TOML translation files. Per file: - IndexPage.vue: loading-spinner tip (initial + dynamic). - BackupModal.vue: modal title, both list-item titles + descriptions ("Back up" / "Restore"), in-flight busy tips ("Importing database…" / "Restarting panel…"). - PanelUpdateModal.vue: modal title, update-available alert, current/latest version row labels, "Up to date" tag + label, primary action button. Modal.confirm now uses the translated panelUpdateDialog / panelUpdateDialogDesc with #version# substitution; success toast uses panelUpdateStartedPopover. - LogModal.vue: title slot ("Logs"). The Debug/Info/Notice/Warning/ Error log-level options stay literal — they're xray's wire values, not user-facing labels (matches the existing settings-page choice). - XrayLogModal.vue: title + Filter label. Direct/Blocked/Proxy stay literal for the same reason. - VersionModal.vue: modal title + xray-switch alert + per-file tooltip + "Update all" button + custom-geo collapse header. The Modal.confirm flows for switchXrayVersion + updateGeofile use translated dialog/desc with #version# / #filename# substitution. - CpuHistoryModal.vue: title slot. - CustomGeoSection.vue: routing-hint alert, Add / Update-all buttons, every column title (computed for live locale), copy/edit/download/ delete tooltips, copy toast, delete-confirm modal, empty-state text. - CustomGeoFormModal.vue: add/edit titles, OK/cancel labels, Type/ Alias/URL field labels, alias placeholder, all three validation toasts. Total: ~50 strings localised across 8 index-page files. The Hello / Welcome login headline cycle and a handful of literal xray wire values (Direct/Blocked/Proxy/log levels) are intentionally kept hardcoded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): Phase 7-c — translate settings, inbounds modals, xray tabs Continues the page-by-page translation pass started in cb37dd55 — runs every user-visible string on settings (General/Security/Telegram/Sub), inbounds (Client/QR/Info modals), and xray (Routing/Balancer/Rule/Warp/ Nord/Basics/Outbounds tabs) through useI18n. Updates the TOML→JSON sync script to escape `@` (vue-i18n parses it as a linked-format prefix) and refreshes all 13 locale files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): Phase 9 — restore index dashboard, fix login/CSRF, port legacy styles - Index dashboard regains the 8 cards that were lost in the SPA port (3X-UI panel info, Operation Hours, System Load, Usage, Overall Speed, Total Data, IP Addresses, Connection Stats), plus a Config button that shows the live xray config.json. Version display falls back through panelUpdateInfo → window.__X_UI_CUR_VER__ → '?' so dev mode isn't blank. - Xray config no longer hangs on load: useXraySetting surfaces failures instead of leaving a perpetual spinner, and the Vite dev proxy stops hijacking POST requests to migrated routes (only GETs get bypassed). - Inbound page no longer throws __asyncLoader/emitsOptions errors — inbound.js was missing imports (NumberFormatter, SizeFormatter, Wireguard) and InboundList kept emitting after unmount. - Login round-trip works after logout: a public /csrf-token endpoint bootstraps the SPA before authentication, axios caches the token module-level, and the dev 401 handler navigates to /login.html instead of reloading the dashboard into a redirect loop. - legacy.css mirrors the legacy panel's surface/text variables so dark and ultra-dark themes match main; every SPA entry imports it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray DNS section to match main branch DnsTab now exposes every field the legacy panel did — top-level toggles (tag, hosts, queryStrategy, disableCache/queryConcurrency, fallback strategy, client subnet), the servers table with per-row strategy and domain/expectIP/unexpectedIP overrides, and the Fake DNS pool. The new DnsServerModal covers the full add/edit flow and collapses to a bare string when the user only sets an address — matching the wire shape the legacy form emits for plain DNS entries like "8.8.8.8". Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray outbound modal with structured per-protocol forms Replaces the JSON textareas with the same shape the legacy panel uses: all 11 outbound protocols (vmess/vless/trojan/shadowsocks/socks/http/ mixed/wireguard/tun/dns/loopback/blackhole/freedom) get dedicated fields, every transport (TCP/KCP/WS/gRPC/HTTPUpgrade/XHTTP) gets its own panel, and TLS/Reality/sockopt/Mux are configured through the same controls as the inbound side. Brings the SPA outbound editor to parity with main so users no longer have to drop into raw JSON. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): bring inbound modal to full parity with main branch Switches the default protocol on add to VLESS, fixes a crash when adding a Mixed account (the constructor is SocksAccount, not MixedAccount), and fills in the fields the SPA was previously delegating to the Advanced JSON tab: - TLS: cipher suites, min/max version, reject SNI / disable system root / session resumption switches, the certificate array with per-row Path-or-Content toggle (Set Default pulls from /panel/setting/ defaultSettings), One Time Loading, Usage / Build Chain, plus ECH key/config with a Get New ECH Cert button. - Reality: xver, target/SNI sync icons (uses getRandomRealityTarget), max time diff, min/max client version, short IDs randomizer, SpiderX, mldsa65 seed/verify with Get New Seed. - Stream: full structured forms for every transport — TCP HTTP camouflage gets its request/response editor, mKCP gets MTU/TTI/uplink/ downlink/CWND/maxSendingWindow, WebSocket / gRPC (now with Authority) / HTTPUpgrade get headers + proxy-protocol toggles, XHTTP gets the full SplitHTTPConfig surface (mode-aware fields, padding obfs, session/sequence placement, uplink data, no-SSE). - New External Proxy section and a structured Sockopt block (mark, TCP keepalive/timeout/clamp, fast open, MPTCP, penetrate, V6Only, domain strategy, congestion, TProxy, dialer/interface, trusted XFF). - VLESS gets the legacy X25519 / ML-KEM-768 buttons that fetch fresh decryption/encryption blocks via /panel/api/server/getNewVlessEnc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): add FinalMask UI (TCP/UDP masks + QUIC params) to inbound and outbound Mirrors web/html/form/stream/stream_finalmask.html as a shared FinalMaskForm component used by both modals — they share the same StreamSettings shape (addTcpMask/addUdpMask/finalmask/enableQuicParams) so a single template handles both. Surfaces: - TCP masks for raw/tcp/httpupgrade/ws/grpc/xhttp networks: fragment, sudoku, and header-custom (with the 2D clients/servers groups, each row supporting array/str/hex/base64 packets and a randomize button for base64). - UDP masks for hysteria protocol or kcp network: hysteria gets just salamander; kcp gets the full type list (mkcp variants, header-*, xdns/xicmp, header-custom with flat client/server lists, and noise). Switching to xdns shrinks the kcp MTU to 900 to match the legacy panel's behavior. - QUIC Params for hysteria or xhttp: congestion (incl. brutal up/down fields), debug, UDP hop ports/interval, idle/keepalive timeouts, path-MTU discovery toggle, and the four receive-window tunables. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): remove duplicate Outbound test URL from xray Advanced tab The Basics tab already exposes this field through BasicsTab — duplicating it on the Advanced tab let two inputs race the same ref and only added clutter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): unify theming on vanilla AD-Vue light/dark/ultra-dark The legacy panel CSS (custom.min.css ported as legacy.css) tinted every non-primary button teal-green via .dark .ant-btn:not(.ant-btn-primary) overrides while AD-Vue 4's darkAlgorithm kept primary buttons blue — producing the mixed blue/green button look on dark mode. Drop legacy.css entirely and let AD-Vue 4's algorithms own the palette. Centralize antdThemeConfig in useTheme.js so every page resolves to the same source of truth (light = defaultAlgorithm, dark = darkAlgorithm, ultra-dark = darkAlgorithm + deeper colorBgBase/Layout/Container/ Elevated tokens). Each page's <a-config-provider> now imports the shared computed instead of defining its own copy. Drops the 67 KB legacy CSS chunk; per-page CSS bundles fall to ≤5.9 KB. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): restore computed import in Settings + Xray pages When 5f1aba28 dropped the local antdThemeConfig computed (now shared from useTheme), it also stripped `computed` from the import list — but both pages still call computed() elsewhere (confAlerts, advanced-tab helpers). Re-adds it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): retheme dashboard gauges to AD-Vue blue and shrink them - StatusCard's CPU/RAM/Swap/Storage dashboards rendered at AD-Vue's default 120px width which made the percent text balloon to ~36px. Drop to 90px (70px on mobile) so the gauge fits the rest of the card. - The CurTotal.color thresholds still hardcoded the legacy teal/orange palette (#008771 / #f37b24 / #cf3c3c). Switch to AD-Vue's primary / warning / danger tokens (#1677ff / #faad14 / #ff4d4f) so the gauges match the rest of the panel under both light and dark themes. - XrayStatusCard's running-animation badge ring also still pointed at the deleted --color-primary-100 var; hardcode the new primary blue. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: shorten backupTitle to "Backup & Restore" across all 13 locales The backup modal header was the second-longest title in the dashboard on every locale ("Database Backup & Restore" / "Резервне копіювання та відновлення бази даних" / etc). Drop the "Database / Veritabanı / 数据库" qualifier — the modal already lives under the "Database" column, so the shorter form reads cleaner on narrow viewports. Updated both the .toml source-of-truth files and the synced .json locales (re-running scripts/sync-locales.mjs). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: collapse two translation databases into a single web/translation/<lang>.json set The Vue SPA had been reading from frontend/src/locales/*.json while the Go binary still loaded web/translation/translate.*.toml — and a sync-locales.mjs pre-build step kept the two in lockstep, with TOML as the source of truth. Now that go-i18n v2.6.1 already flattens nested JSON via recGetMessages/addChildMessages, both runtimes can share one file per locale. - Move the 13 nested-JSON locale files to web/translation/<lang>.json so they live alongside the Go //go:embed translation/* directive. - Switch web/locale/locale.go from toml.Unmarshal to json.Unmarshal (and drop the pelletier/go-toml import — it's now indirect-only). Confirmed via a smoke test that pages.index.cpu, subscription.title, tgbot.commands.help, and menu.settings all resolve in en-US, fa-IR, ru-RU, and zh-CN. - Repoint Vue's i18n loader at the new path (../../../web/translation/ *.json glob) and drop the moved-here pathDelimiter comment that no longer applies. - Delete the 13 legacy translate.*.toml files and the sync-locales.mjs script + its npm pre-script hooks (predev/prebuild/i18n:sync). The Telegram bot and subscription page still get their messages because they were reading the same MessageIDs the JSON files now produce. - Update copilot-instructions.md so the next contributor knows where the canonical translation files live. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): redesign expand-row + retheme client visuals When you expanded an inbound row, the nested <a-table> inside ClientRowTable burst out of the parent's scroll-x box — its .ant-spin-container ended up wider than the parent's narrow .ant-table-cell, so the child looked oversized while the parent looked squeezed. Replace the nested table with a CSS-grid layout that owns its sizing, sits flush inside the expanded cell, and collapses to a 3-column layout on mobile (action menu, client identity, info popover). While in there, fix three other client-row visuals: - The Unicode infinity glyph (U+221E) renders as an "m"-shaped character in some system fonts (Windows Segoe UI in particular). Add a shared <InfinityIcon /> SVG component (legacy panel's path) and use it in ClientRowTable, InboundList, and InboundInfoModal — desktop and mobile cells. - The "unlimited quota" traffic bar passed :percent="100" with no stroke-color, so AD-Vue auto-coloured it success-green. Pin it to the AD-Vue purple token (#722ed1) so it reads as the no-limit sentinel rather than another usage state. - ColorUtils + the in-row statsExpColor still hardcoded the legacy teal/orange/red/purple palette (#008771 / #f37b24 / #cf3c3c / #7a316f). Map them onto AD-Vue 4's success/warning/danger/purple tokens (#52c41a / #faad14 / #ff4d4f / #722ed1) so badges, tags, and progress bars all match the rest of the panel. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): darken light-theme page bg so cards stand out The light-theme --bg-page was #f0f2f5 — close enough to AD-Vue's #fff card background that the cards faded into the page. Bump it to #e6e8ec (a more visibly distinct gray) so cards lift cleanly off the surface. Dark and ultra-dark stay where they were. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): shrink dashboard percent text and surface the unfinished arc Two follow-up tweaks to the dashboard gauges: - AD-Vue scales the percent text from the SVG, not from :width, so the 90px gauges still rendered the number at ~27px. Pin .ant-progress-text to 14px via :deep() and trim the gauge to 70px (60px on mobile) so the whole card stays compact. - The default trail (rgba(0,0,0,0.06) / rgba(255,255,255,0.08)) was invisible on the light-theme card. Pass an explicit rgba(128,128,128,0.25) trail-color so the unfinished portion is visible under both light and dark themes. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): migrate subpage.html to Vue 3 SPA The subscription info page was the last page still rendered by Go templates. Move it to the Vite multi-page setup so the whole panel loads through one toolchain. Frontend: SubPage.vue mounts at /sub/<id>?html=1 and reads window.__SUB_PAGE_DATA__ for the parsed view-model (traffic / quota / expiry + rendered share links). Fix descriptions borders against the light-theme card by painting the row divider on each cell's bottom edge — AD-Vue's <tr> border doesn't render reliably under border-collapse:collapse. Backend: serveSubPage reads dist/subpage.html, injects window.__X_UI_BASE_PATH__ + window.__SUB_PAGE_DATA__ before </head>, and rewrites Vite's absolute /assets/ URLs when the panel runs under a URL prefix. Drop the legacy template-FuncMap wiring and switch the sub server's static mount from web/assets to web/dist/assets. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): inbound modal QR + tabs + restored TLS fallbacks Per-client QR action: the qr icon on the expand-row table opened the big info modal instead of the QR modal. Route it to QrCodeModal and extend that modal with a `client` prop so genAllLinks() produces the per-client share URLs (and per-peer remarks for WireGuard). Inbound's Data redesign: split the dense single-page view into three tabs — Inbound, Client, Subscription. Drop every QR rendering from this modal (QrCodeModal is the QR home now). Each row in the Inbound tab is one label/value pair instead of the legacy 2-column grid, and long values like the VLESS encryption blob render as a wrapping code block with a copy button so they can't blow out the dialog. The Subscription tab renders sub URL + JSON URL as clickable anchors that open in a new tab. Restored TLS fallbacks UI: the model already exposed VLESSSettings.Fallback / TrojanSettings.Fallback with addFallback / delFallback / fallbackToJson, but the form modal never surfaced them during the Vue 3 migration. Re-add the legacy form (SNI, ALPN, Path, Destination, PROXY) on the protocol tab, gated on TCP transport plus (for VLESS) encryption=none — same conditions as main. Column widths: Protocol 70→130 and All-time Traffic 60→95 in the inbound list; All-time Traffic 90→130 in the client expand-row, so the header text fits and tags don't get squeezed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): navy dark theme + rounded inbound/client corners Dark theme picks up a refined navy palette (page #0a1426, cards #142340, sider #0d1d33) so the sidebar blends with the rest of the surface; ultra-dark stays neutral black. Resolves the previous mismatch where AD-Vue 4 hardcoded #001529 / #002140 for the sider, trigger and dark Menu items via Layout.colorBgHeader / colorBgTrigger and Menu's colorItemBg — overrides go through the component-token map now. Round the inbound table's outer corners (header start/end + last row end) and wrap the client expand-row grid in a 1px / 8px-radius border so the list reads as a contained block instead of a flush rectangle. Linter-driven whitespace cleanup across inbounds/*.vue rolled into the same commit since it can't be split out cleanly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): xray tab fixes — modal close, tag validation, full XHTTP, reset to default Modal close: BalancersTab / OutboundsTab / RoutingTab confirmDelete used arrow expressions that returned splice's removed-items array. AD-Vue 4 treats truthy non-thenables from onOk as "still pending" and never closes the dialog (see ActionButton.js:103-106), so the confirm modal stayed open. Wrap the body so onOk returns undefined and AD-Vue auto-closes. Tag validation: outbound + balancer modals only flipped between warning/success on duplicate, leaving the empty case as a green ✓. Split into a 3-state computed — error (empty) / warning (duplicate) / success — and wire a help message so the input clearly explains why the OK button is disabled. Reset to default: re-add the legacy "Reset to Default" panel at the bottom of BasicsTab. Calls /panel/setting/getDefaultJsonConfig and overwrites templateSettings; the existing watch re-stringifies so the JSON tab + dirty-poll see the new state. Restored Basics option lists from main: IPs (4→10, +Vietnam/Spain/ Indonesia/Ukraine/Türkiye/Brazil), DomainsOptions (4→10, +regex entries), BlockDomainsOptions (5→17, +Malware/Phishing/Adult/regex), ServicesOptions (Reddit/Speedtest in, off-template Microsoft out). Outbound form parity with main: • Reverse Sniffing UI for VLESS — toggle + destOverride checkboxes (HTTP/TLS/QUIC/FAKEDNS) + Metadata/Route Only + IPs/Domains excluded multi-selects, gated on reverseTag being set. • Full XHTTP transport — request headers list, Max Upload Size / Min Upload Interval (packet-up), Padding Obfs Mode + sub-fields, Uplink HTTP Method, Session/Sequence/UplinkData placement + keys, No gRPC Header (stream-up/stream-one), expanded XMUX with Max Concurrency/Connections/Reuse/Request/Reusable/Keep-alive. Strip a-divider from the outbound form per request — replaced with plain section/item heading divs so the labels and per-row delete icons stay but the horizontal rule is gone. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): xray Advanced tab parity + finalmask gating Advanced tab was a single textarea bound to the full xraySetting blob. Restore the legacy 4-way view: a radio group toggles between All / Inbounds / Outbounds / Routing Rules, and the textarea reads/writes the matching slice through templateSettings. Added the legacy header ("Advanced Xray Configuration Template" + description) so the page introduces itself like main. Outbound finalmask leaked into protocols that don't have a stream (Freedom / Blackhole / DNS / Socks / HTTP / Wireguard) because the v-if only checked outbound.stream. Gate the whole FinalMaskForm on outbound.canEnableStream() to match main. Drop the leading divider inside FinalMaskForm — its parent already provides separation, so the rule above "TCP Masks" was redundant. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound Advanced tab live mirror + QR exact-fit sizing Advanced tab in the inbound modal showed stale state. The watch only refreshed advancedJson.stream, so toggling the Sniffing switch in the Sniffing tab left the Advanced JSON showing the prior value. And encryption — stored on inbound.settings.encryption, not on stream — never appeared at all because Advanced only exposed stream + sniffing. Split the watch into three (stream / sniffing / settings) and add a settings textarea so encryption / clients / fallbacks live alongside the existing two views. The submit() path now reads settings from the JSON tab too (falling back to inbound.settings.toString()) so power-user edits in Advanced override the structured form on save. QR canvas: when a longer share-URL bumps the QR matrix size, QRious falls back to floor(canvasSize / matrixWidth) and centers the pattern, leaving a white margin (e.g. matrix=41, size=180 → 8px gap). Pre-pick the QR version from the URL byte length and set canvas size to a multiple of matrixWidth × pixelSize so the pattern always fills it edge-to-edge — no white margin even after toggling encryption on. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound stream tidy-up + QR sizing + dev proxy Stream tab clean-up: drop the seven a-divider rules in the inbound form's Stream tab — replace the labelled ones (Request / Response / Security) with a section-heading div that matches the outbound modal, delete the empty rules above TLS sub-blocks / External Proxy / Sockopt. Empty header-list form-items also leaked margin space below each "Add header" button across TCP / WS / HTTPUpgrade / XHTTP — gate each on headers.length > 0 so they vanish until the user adds one. QR panel: drop the link text under the canvas (the user already has a copy button on the header). Pin the canvas display size to a fixed 240px square via :style + image-rendering: pixelated/crisp-edges so a dense WireGuard config QR and its sparser link share the same on-screen footprint without blurring. Dev proxy: Node's AggregateError wraps connection failures whenever DNS returns more than one address (::1 + 127.0.0.1) and the code lands on the inner errors, not the outer. The existing handler only checked err.code so the ECONNREFUSED stack still spammed the log when the Go backend was down. Walk err.errors too, print one friendly line ("backend not reachable — start the Go server"), then stay quiet for the rest of the session. Vendor splitting + chunk-size warning: split node_modules into stable vendor-* chunks so each page only ships the deps it uses and the browser caches them across versions. ant-design-vue stays as a single chunk because its components share internals; raise the chunk-size warning to 1500kB so the build stays quiet (its 1.4MB minified gzips to ~410kB on the wire). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): info-modal cleanup + 2FA QR + outbound link import - 2FA QR: matrix-snap canvas + opaque background to drop white margin - Inbound info modal: stack Mixed/HTTP/Tunnel as info-rows, hide tab strip when only the Inbound tab applies - Add inline VLESS Reverse tag input on first-client form - Hide Protocol tab for TUN (no form yet) - Outbound link converter: route through Outbound.fromLink so vless/trojan/ss/hysteria(2) imports work alongside vmess; fix stray implicit global in fromLink Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): jalali calendar + drop legacy moment-jalali - Wire Calendar Type setting to a real Jalali datepicker via vue3-persian-datetime-picker, gated by useDatepicker composable - DateTimePicker wrapper swaps between AD-Vue and Persian picker; keeps dayjs v-model contract so existing forms/setters work unchanged - Theme picker popup explicitly per body.dark / data-theme=ultra-dark (AD-Vue 4 doesn't expose CSS vars, so var() fallbacks defaulted to white); fix invisible disabled days, SVG arrow fills, popup clipping via append-to="body" - Replace stray moment() calls in dbinbound/inbound models with dayjs; the legacy global was undefined under ESM and broke the inbounds list whenever any inbound had expiryTime > 0 - Remove legacy moment-jalali / persian-datepicker / aPersianDatepicker assets — replaced by the Vue 3 picker Note: dark/ultra background of the date popup still renders white in some cases — pending follow-up. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): jalali popup theming + full-month layout - Re-prefix popup selectors with .vpd-wrapper (popup root that travels with appendTo='body'), not .vpd-main (which stays at the input); paints the popup's dark/ultra background again - Drop the 1px border on .vpd-content — with box-sizing: border-box it ate 2px from the day-row width, wrapping the 7th cell of every row and hiding days 18-31 of months that needed a 5th week Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat: render dates in Jalali when Calendar Type is jalalian - IntlUtil.formatDate accepts an optional calendar arg; appends the BCP-47 -u-ca-persian extension so Intl renders Jalali across all UI languages, not just fa-IR - Plumb the panel's datepicker setting into the SubPage via the Go injection (window.__SUB_PAGE_DATA__.datepicker) - Panel pages (inbound list/info, client row, xray log) read the same setting through the useDatepicker composable so the whole panel stays consistent Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): ultra-dark page tint + mobile-friendly inbound view - Drop --bg-page from #21242a (lighter than the cards) to #050505 in ultra-dark across index/sub/settings/inbounds/xray, so cards consistently elevate over the page - Hide the inline sider's children + collapse-trigger and zero its width below 768px; the floating drawer-handle remains the menu trigger - Inbounds page mobile pass: tighten content-area + card padding; flex-wrap the filter bar instead of stacking; shrink table cell padding so all 4 mobile columns fit; bump expand / action / info icon hit targets - Per-client expand row on mobile: soft-tinted rounded cards instead of hairline borders, larger action / info touch targets, more legible email typography, bigger status badge dot Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove legacy template + asset trees and dead Go template engine - Delete web/html/ entirely (page templates, form/, modals/, component/, common/, settings/) — every route is served from web/dist/ now via serveDistPage; nothing in the binary referenced these - Delete web/assets/ entirely (jQuery-era ant-design-vue, axios, moment, codemirror, qrcode/qs/uri/vue/otpauth, custom CSS, Vazirmatn font); Vite bundles all of this into web/dist/assets - Drop the Gin HTML template wiring: remove //go:embed assets + //go:embed html/*, the assetsFS/htmlFS vars, the wrapAssetsFS adapter, EmbeddedHTML / EmbeddedAssets exports, getHtmlFiles / getHtmlTemplate, the i18nWebFunc/funcMap and SetFuncMap call, and the dev/prod template-engine branch — only StaticFS for /assets/ is needed now - Remove dead html()/getContext() helpers and unused imports from web/controller/util.go (no c.HTML(...) callers remain) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound expand chevron position + cpu history layout - Push the inbound table's expand chevron away from the left edge with margin-inline + cell padding so it isn't flush against the corner - Move "Timeframe: …" caption above the chart (was below); restore the line that the previous edit removed - Fix x-axis time labels being clipped at the bottom of the cpu chart — the offset (paddingTop+drawHeight+22 = 222) exceeded the SVG viewBox height (220); dropped to +14 so labels sit at y=214 with room for descenders - Move the SVG axis text colors out of <style scoped> into a global block — Vue's scoped CSS doesn't always hash-attribute SVG <text> descendants, so the dark-mode overrides via :global() weren't matching; bumped opacity 0.55 → 0.85 for legibility on navy/black Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(login): language picker in settings popover + fluid card sizing - Add language select alongside the theme switch (mirrors SubPage) - Bind headline to pages.login.hello / pages.login.title so the "Hello / Welcome" cycle re-translates with the active locale - Replace AD-Vue 5-breakpoint grid with clamp() sizing so the card scales smoothly instead of jumping ~33% at each breakpoint - Pin horizontal padding so input width stays stable on large viewports Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): organize entry HTML + bootstrap JS into folders - Move entry HTML files: frontend/*.html -> frontend/html/*.html - Move per-page bootstrap modules: src/{index,login,settings,inbounds,xray,subpage}.js -> src/entries/ - Update vite.config rollup inputs and dev-mode MIGRATED_ROUTES to /html/<page>.html - Build output now lands at web/dist/html/<page>.html - serveDistPage and subController updated to read from dist/html/ Cleans up the flat frontend/ root which previously interleaved 6 HTML files with package.json, README, src/, etc. The src/ root similarly gets rid of 6 entry .js files mixed in alongside api/, components/, models/, etc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove obsolete vue3 phase1 inventory doc The migration is well past phase 1 — the inventory doc has rotted and the live state lives in the codebase plus the plan files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): merge utils/legacy.js into utils/index.js The barrel was a placeholder for an eventual split that hasn't happened. Collapsing the two files removes one layer of indirection and the misleading "legacy" name (the contents are still actively used by the migrated SPA). - Move all 930 lines from utils/legacy.js into utils/index.js - Delete utils/legacy.js - Update direct import in models/outbound.js to '@/utils' - Drop a stale legacy.js reference in InboundFormModal comment Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(frontend): keep entry HTML files at frontend/ root The earlier move to frontend/html/ made dev-mode URLs ugly (http://localhost:5173/html/index.html instead of plain /). The folder didn't add real value — it just hid 6 files behind a non-conventional layout. Reverting that piece while keeping src/entries/ (which is a genuine separation between page bootstrap and the rest of src/). - HTML files back at frontend/<page>.html - Vite rollupOptions.input + MIGRATED_ROUTES restored to flat paths - Build output is web/dist/<page>.html again - web/controller/dist.go and sub/subController.go read from dist/<name> Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): bump eslint to 10 + add flat config + clean lint warnings - Upgrade eslint 9.39 -> 10.3 and eslint-plugin-vue 9.33 -> 10.9 - Add eslint.config.js (flat config required by ESLint 10) with vue3-recommended rules, sensible defaults, and exemptions for the project's existing formatting style - Drop --ext from the lint script (removed in ESLint 10) - vue/no-mutating-props is left off because the form-modal pattern ports straight from Vue 2 (parent passes a reactive object, child mutates it); a real fix is an architectural rewire, separate task Lint warning cleanup: - utils/index.js: var -> let/const in the X25519 routines, replace obj.hasOwnProperty(...) with Object.prototype.hasOwnProperty.call(...) - Remove unused imports (reactive, ref, Inbound) in ClientFormModal, InboundInfoModal, QrCodeModal, DnsServerModal, OutboundFormModal, SubPage; remove unused locals (isClientOnline, ONLINE_GRACE_MS, fetchAll, isSocks, isHTTP, _antdAlgorithm) - XrayStatusCard: declare 'open-logs' on defineEmits (was emitted but not declared) - RuleFormModal: rename v-for var t -> tag (shadowed useI18n's t) - Drop stale eslint-disable directives (no-new, no-unused-vars) - OutboundsTab/InboundList: drop redundant initial null assigns - InboundInfoModal/OutboundFormModal: explicit eslint-disable for the intentional local-ref-shadows-prop pattern in modal drafts `npm run lint` now passes with 0 errors and 0 warnings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): one client identity across multiple inbounds via subId Lets the operator add the same email under the same subId to several inbounds. Xray reports traffic per email, so a single client_traffics row acts as the shared accumulator — no aggregation overhead, quota and expiry stay consistent. - Email validation allows duplicates only when subId matches - AddClientStat upserts via OnConflict DoNothing (idempotent on rerun) - Stat/IP rows survive client deletion when a sibling inbound still references the email - enrichClientStats tops up GORM-preloaded stats with rows whose inbound_id points at a sibling, so every panel view sees usage - disableInvalidClients cascades enable=false and syncs the row's total/expiry into every sibling JSON when the shared identity expires - DelDepletedClients removes the depleted client from all referencing inbounds, batched - Subscription services dedupe traffic by email so shared quota is counted once Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * docs(frontend): rewrite README for multi-page Vue 3 layout Reflects the current state — embedded build, per-route HTML entries, ESLint 10 flat config, src/ layout, and the steps to add a new page. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): drop deprecated rimraf/glob/inflight transitive deps vue3-persian-datetime-picker pinned moment-jalaali to ^0.9.4, which pulled rimraf@3 → glob@7 → inflight@1. inflight in particular leaks memory and is unmaintained. Override moment-jalaali to ^0.10.4 (same runtime API, dropped the legacy build deps) so npm install no longer warns and the dep tree is 12 packages lighter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): multi-node panel orchestration (CRUD, deployment, traffic sync, sub per-node) - Node model + service + controller (/panel/api/nodes/*) with bearer-token apiToken auth - Heartbeat job @every 10s; status/latency/xrayVersion surfaced in Nodes UI - Runtime abstraction (Local + Remote) so inbound/client mutations target the inbound's owning node instead of always hitting the local xray - Inbounds gain optional NodeID; tag-based correlation with remote panel (no RemoteInboundID column needed) - NodeTrafficSyncJob @every 10s pulls absolute counters + online/lastOnline from each enabled+online node and writes them into central DB; 30s reset grace window prevents post-reset overwrite - Reset propagation to nodes (best-effort) on client/inbound/all reset paths - Subscription server uses node.Address for inbounds with NodeID, falling back to existing host resolution for local inbounds - Frontend: Nodes page, "Deploy to" select in inbound form, Node column on inbound list, hostOverride threaded through genAllLinks/QR/Info modals Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(stats): system history modal + per-node CPU/Mem trends across all locales Backend - web/service/metric_history.go: generic in-memory ring buffer with two singletons — system-wide (cpu/mem/netUp/netDown/online/load1/5/15) and per-node (cpu/mem) keyed by node id - ServerService.AppendStatusSample writes all 8 metrics every 2s on the same tick; AppendCpuSample/AggregateCpuHistory kept for back-compat - NodeService.UpdateHeartbeat appends cpu/mem only on online ticks so offline gaps render as missing data, not phantom dips - New routes: GET /panel/api/server/history/:metric/:bucket and GET /panel/api/nodes/history/:id/:metric/:bucket, both whitelisted Frontend - Sparkline component generalized: arbitrary value range (auto-scale when valueMax=null), pluggable yFormatter/tooltipFormatter for B/s, client counts, load averages - SystemHistoryModal replaces CpuHistoryModal with tabs for every metric; opened from a tag on the 3X-UI card next to Documentation - NodeHistoryPanel: expandable row on the Nodes table showing per-node CPU and Mem trends, refreshed every 15s Localization - Backfill systemHistoryTitle / trendLast2Min / pages.inbounds.{node, deployTo, localPanel} and the entire pages.nodes block (51 keys including statusValues + toasts) into all 11 non-en/fa locales: ar-EG, es-ES, id-ID, ja-JP, pt-BR, ru-RU, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(embed): include underscore-prefixed Vite chunks in dist FS go:embed silently excludes files whose names start with `_` or `.`, so the `_plugin-vue_export-helper-<hash>.js` chunk that Vite/rolldown emits for @vitejs/plugin-vue was missing from the production binary. First import at runtime hit a 404 and the SPA failed to mount — blank page on every page load, no error in the server logs because the asset 404 was just a static-handler miss. Switched the directive to `//go:embed all:dist` which keeps the same root layout but disables the underscore/dot exclusion rule. Dev mode was unaffected (it serves dist/assets/ from disk, not the embedded FS). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * ci: build frontend bundle before Go compile in release.yml + Dockerfile Phase 8 cut all panel HTML routes over to web/dist/ and embedded the Vite bundle into the Go binary via //go:embed all:dist. web/dist/ is .gitignored, so on a fresh CI checkout it doesn't exist — every Go build since Phase 8 has been failing with "pattern dist: no matching files found" or producing a binary that 404s on first asset request. release.yml: add a setup-node@v4 + npm ci + npm run build trio before the existing go build step in both the Linux matrix job (7 arches) and the Windows job. npm cache is keyed on frontend/package-lock.json. Dockerfile: add a node:22-alpine frontend stage that runs npm ci + npm run build and emits to /src/web/dist (via vite.config.js's outDir). The golang builder stage then COPY --from=frontend /src/web/dist into ./web/dist before the go build, so embed.FS sees the bundle. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(ws): live updates on inbounds/xray/nodes pages, drop polling + manual refresh Replaces the legacy polling + manual-refresh model with WebSocket pushes across the three live-data pages. The hub already broadcast traffic / client_stats / outbounds; this wires the frontend to consume them and adds a new `nodes` channel for the heartbeat job's snapshot. Frontend - new useWebSocket composable: page-scoped singleton WebSocketClient, lifecycle-managed on/off, leaves disconnect to page-unload - inbounds: useInbounds gains applyTrafficEvent / applyClientStatsEvent / applyInvalidate that merge counters and online/lastOnline in place; InboundsPage subscribes; InboundList drops the auto-refresh popover, the refresh button, and the now-unused refreshing prop - xray outbounds: useXraySetting gains applyOutboundsEvent; XrayPage subscribes; OutboundsTab drops the refresh button + emit - nodes: useNodes gains applyNodesEvent and stops the 5s setInterval/visibilitychange polling; NodesPage subscribes; NodeList drops the refresh button and ReloadOutlined import Backend - web/websocket: new MessageTypeNodes + BroadcastNodes notifier - node_heartbeat_job: after wg.Wait(), reload the table once and BroadcastNodes(updated). Gated on websocket.HasClients() so a panel with no open browser doesn't spend the DB read Bug fixes spotted in this pass - websocket.js #buildUrl defaulted basePath to '' when the global was missing (dev mode), producing `ws://host:portws` and a SyntaxError on the WebSocket constructor. Fall back to '/' and ensure leading slash. - vite.config.js: forward /ws to ws://localhost:2053 with ws:true so dev (5173) reaches the Go backend's WebSocket - NodeFormModal: a-input-password's visibilityToggle is Boolean in AntD Vue 4; the v3-era object form (`{ visible, 'onUpdate:visible' }`) triggered a Vue prop-type warning. Drop the override (default true shows the eye icon and toggles internally) and remove the orphaned tokenVisible ref Translations - pages.inbounds.autoRefresh / autoRefreshInterval: removed from all 13 locales (UI gone) - pages.nodes.refresh: removed from all 13 locales (UI gone) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): hide Node column when no nodes are defined Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-09 08:38:48 -07:00
// GetXrayAPIPort returns the port the local xray process is listening on
// for its gRPC HandlerService, or 0 when xray isn't currently running.
// Exposed for the runtime package's LocalRuntime adapter without a
// service-package import cycle.
Vue3 migration (#4198) * docs(migration): Phase 1 inventory — Vue 2 / AD-Vue 1 surface area Captures the breakage surface for the Vue 3 + Ant Design Vue 4 + Vite migration: 17,650 lines across 69 templates, 3,145 a-* component instances across 63 files, with per-pattern counts and file lists. Key findings: - No Vue filters anywhere — dodges a major Vue 3 breaking change - 358 v-model uses; AD-Vue 4 absorbs most, custom components don't - 233 <template slot="X"> usages must become <template #X> - 49 scopedSlots: { ... } column defs need new slots: { ... } shape - a-icon is removed in AD-Vue 4 — every icon must be imported Establishes the 8-phase order; Phase 2 (Vite toolchain) is next. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): Phase 2 — scaffold Vite + Vue 3 + AD-Vue 4 Adds a frontend/ directory that lives alongside the legacy web/html/ Vue 2 templates during the migration. Vite builds into ../web/dist/ so the Go binary will be able to embed the result via embed.FS once Phase 4 starts moving real pages over. - package.json pins Vue 3.5, Ant Design Vue 4.2, Vite 6, vue-i18n 10 - vite.config.js: dev server on :5173 with API proxy to the Go panel on :2053; build output to ../web/dist/ - src/App.vue is currently a smoke-test placeholder — delete once the first real page (login) lands in Phase 4 - node_modules and dist are already ignored at repo root To verify locally: cd frontend && npm install && npm run dev Pages will be migrated one at a time on the vue3-migration branch. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): Phase 3 — port utils, models, axios, websocket as ES modules Ports the framework-agnostic JS from web/assets/js/ into frontend/src/ so Vue 3 pages can import what they need without relying on script-tag globals. - web/assets/js/util/index.js (927 lines, 21 classes) → frontend/src/utils/legacy.js + a barrel at utils/index.js. All classes are now named exports. - Vue.prototype.$message in HttpUtil → direct import of `message` from ant-design-vue (Vue 3 has no Vue.prototype). - RandomUtil.randomShadowsocksPassword previously defaulted to SSMethods.BLAKE3_AES_256_GCM from inbound.js, creating a circular import. Replaced with the literal string default. - MediaQueryMixin (Vue 2 mixin) removed. Replaced by composables/useMediaQuery.js — Vue 3 composable returning reactive `isMobile`. - axios-init.js wrapped as setupAxios(); Qs global → npm `qs`. - websocket.js exported as WebSocketClient class; the implicit window.wsClient global is gone — pages instantiate it themselves. - model/{inbound,outbound,dbinbound,setting,reality_targets}.js copied with `export` added on every top-level declaration. Imports between models and utils are wired up explicitly. - subscription.js deferred to Phase 5 (it's a Vue 2 mount, not a util). - App.vue smoke test exercises SizeFormatter / RandomUtil / Wireguard / useMediaQuery so the user can verify Phase 3 with `npm run dev`. Run `cd frontend && npm install && npm run dev` — qs was added so a fresh install is required. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 4 — port login.html to Vue 3 + AD-Vue 4 + Vite 8 First real page in the new toolchain. Multi-page Vite: each migrated page is its own entry. login.html now lives at frontend/login.html with a thin entrypoint at frontend/src/login.js mounting LoginPage.vue. Vite 6 → Vite 8.0.11 (per user request). Requires Node 20.19+ or 22.12+. @vitejs/plugin-vue bumped to ^6.0.6 (peers vite ^8). Ant Design Vue stays on 4.2.6 — there is no AD-Vue 6. Vue 2 → Vue 3 / AD-Vue 1 → AD-Vue 4 syntax changes hit on this page: - new Vue({ el, delimiters, data, methods }) → createApp + <script setup> - mounted() → onMounted() - <template slot="X"> → <template #X> - <a-icon slot="prefix" type="user"> → <template #prefix><UserOutlined /> </template> with explicit @ant-design/icons-vue imports - v-model.trim → v-model:value (AD-Vue 4 uses named v-model on inputs) Three legacy features deferred so Phase 4 stays small: - i18n (Phase 7 wires up vue-i18n) - theme switcher (custom component pending Phase 5) - headline word-cycle animation (purely aesthetic) Run `cd frontend && npm install && npm run dev`, open http://localhost:5173/login.html. With Go panel running on :2053 the form submits real credentials via the configured proxy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5a — theme system + Vite 8 + vue-i18n 11 Bumps Vite to 8.0.11 (npm install picked up 6.4.2 from the stale lockfile; clean install resolves the new constraint). Bumps vue-i18n to 11.1.4 since v10 was just EOL'd. Migrates aThemeSwitch.html — the two-flavor theme picker + global themeSwitcher object — into: - composables/useTheme.js: single reactive `theme` state with toggleTheme / toggleUltra. Boot side-effect applies the stored theme to <body>/<html> before Vue renders; watchEffect persists changes back to localStorage. - components/ThemeSwitch.vue: full menu version for the main panel. - components/ThemeSwitchLogin.vue: login-popover version. AD-Vue 1 → 4 changes hit on this component: - <a-icon type="bulb" :theme="filled|outlined"> dropped — replaced by explicit BulbFilled / BulbOutlined imports from @ant-design/icons-vue, swapped via <component :is="BulbIcon"> - Vue.component('a-theme-switch', { ... }) global registration → SFC + per-page import - this.$message.config(...) (Vue 2 instance method) → message.config(...) imported from ant-design-vue, called once in login.js at boot Login page now surfaces a settings button → popover → theme picker. Known gap: web/assets/css/custom.min.css isn't yet imported into the new bundle, so toggling dark mode currently only re-themes AD-Vue's own components, not the panel chrome. The body class is still toggled so behavior is correct; visual fidelity returns when custom.css is ported or directly imported. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5b — port four shared components to Vue 3 CustomStatistic.vue and SettingListItem.vue are mechanical Vue.component → SFC ports. AppSidebar.vue: AD-Vue 4 dropped <a-icon :type="dynamic">, so the five sidebar icons (dashboard/user/setting/tool/logout) live in a name→component map and render via <component :is>. The legacy <a-drawer slot="handle"> hack is replaced with a sibling fixed- position toggle button. Tab paths take basePath/requestUri as props instead of pulling them from Go template scope. TableSortable.vue: the biggest Vue 3 rewrite of this phase. - $listeners is gone — replaced by inheritAttrs: false + explicit attrs forwarding - scopedSlots: this.$scopedSlots collapsed into Vue 3's unified slots object — just iterate Object.keys(this.slots) and forward - Vue 2 h(tag, { props, on, scopedSlots }, children) → Vue 3 h(tag, { ...props, ...on }, slotsObject) - 'a-table' string → resolveComponent('a-table') so app.use(Antd) registration is honored - inject: ['sortable'] (Options API) → inject('sortable', null) (Composition API) inside the trigger child - beforeDestroy → beforeUnmount - customRow's return shape flattened (no nested props/on/class) Two intentional skips, documented in the migration doc: - aClientTable.html — slot fragments, not a component. Migrates inline with inbounds.html (new Phase 5f). - aPersianDatepicker.html — wraps a Persian-only third-party lib; defer until settings.html lands. Build verified with vite 8.0.11. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): anchor Vite dev proxy so /login.html isn't forwarded The /login proxy entry was matching any path starting with /login — including /login.html, which Vite is supposed to serve itself. Without the Go backend running, this caused ECONNREFUSED noise on every page load. Switched to regex patterns anchored with ^...$ so only the bare backend paths (/login, /logout, /getTwoFactorEnable) and explicit sub-routes (/panel/*, /server/*) get proxied. Static .html files Vite serves directly are no longer matched. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): real dark mode + silence dev proxy ECONNREFUSED noise Two issues from running login.html against no Go backend: 1. Dark mode toggled the body class but didn't actually re-theme any AD-Vue components. The legacy panel relied on custom.min.css which we haven't ported. AD-Vue 4 ships its own dark algorithm — wrap LoginPage in <a-config-provider :theme="{ algorithm }"> driven by our useTheme state, and AD-Vue restyles every component for free. Page chrome (background, card, title) gets explicit .is-dark CSS since the algorithm only covers AD-Vue components. 2. Vite logged every failed proxy attempt loudly. When the Go panel isn't running locally that's pure noise. Added a configure() callback that swallows ECONNREFUSED specifically; real errors (timeouts, 5xx, anything else) still surface. Both fixes are dev-experience only — production build is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): use legacy panel palette for login page dark mode Earlier dark mode used invented colors (#141a26 page bg, #1f2937 card) that didn't match the rest of the panel. Replaced with the actual values from web/assets/css/custom.min.css: light dark ultra-dark bg #c7ebe2 bg #222d42 bg #0f2d32 card #fff card #151f31 card #0c0e12 title #008771 title #fff/.92 title #fff/.92 Drove everything off CSS custom properties on .login-app so the .is-dark / .is-ultra class swap is a few var overrides instead of duplicating selectors. Also restored the legacy card metrics (2rem radius, 4rem 3rem padding, 2rem title) so the new page matches the old panel's geometry, not just its colors. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave layout + recolor for dark mode The wave SVG had inline fill="#c7ebe2" (mint) on the bottom wave, so in dark/ultra-dark mode it rendered as a pale-white blob against the dark page. Stripped the inline fills, drove them off CSS variables that swap with .is-dark / .is-ultra: light: green tints + #c7ebe2 (mint) on the bottom wave dark: #222d42 across all four waves ultra-dark: #0f2d32 The wave was also positioned wrong — anchored to the top 200px of the viewport with absolute positioning. Restored the legacy layout: - .waves-header is fixed to the top of the viewport with z-index -1 so the form floats over it - .waves-inner-header pushes the wave SVG down to ~50vh with a 50vh-tall solid block of the page color - .waves SVG itself is 15vh tall, sitting at the bottom of that block Net effect: top half is solid-colored, then a wavy edge transitions into the rest of the page, with the form centered on top — matching the legacy panel exactly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): bring wave-header to front so the wave actually shows Two layering bugs were hiding the wave entirely: 1. .ant-layout-content had background: var(--bg-page) which painted an opaque rectangle covering the full content area — including the fixed wave-header behind it. Made the layout/content transparent and moved the bg paint up to .login-app (the outer ant-layout). 2. .waves-header had z-index: -1 which on its own was fine, but with .ant-layout-content opaque on top it was doubly buried. Promoted the wave-header to z-index: 0 and gave the form .login-row z-index: 1, so the form sits above the wave and the wave sits above the page-bg. Also set --bg-page to the legacy mint (#c7ebe2) for light mode so the bottom half of the page below the wave matches the legacy panel (was white). Dark mode stays at the surface-100/login-wave palette. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave animation timings + dark page bg Two reasons the bottom wave looked static in dark/ultra-dark: 1. Animation durations were 7s/10s/13s/20s. Legacy uses 4s/7s/10s/13s. The 20s on the bottom wave was so slow that against the low dark- mode contrast it read as motionless. Restored the legacy timings. 2. --bg-page in dark mode was #151f31 (card color / surface-100), but the legacy .under uses surface-200 (#222d42) — that's the color of the bottom half of the page, the same as the wave fill, so the wave appears to flow into the page rather than meeting a hard edge. Now it does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): restore Hello/Welcome headline cycle on login Earlier I deferred the legacy headline word-cycling animation as "purely aesthetic". Restored it: the title now alternates between 'Hello' and 'Welcome' every 2 seconds, matching the legacy panel. The legacy implementation toggled .is-visible / .is-hidden classes on two <b> elements via setTimeout chains and DOM querying. Replaced with a reactive ref + Vue 3 <Transition mode="out-in"> so the fade between words is declarative — no manual DOM manipulation, and the interval is properly cleaned up in onBeforeUnmount. The earlier "Welcome to 3x-ui" string was wrong on two counts: it should be just "Welcome", and it should be one of two cycling words with "Hello" preceding it. Ultra-dark palette already matched legacy after the prior wave timing fix; no additional changes needed there beyond the animation speeds that now also apply to ultra-dark via the shared CSS rules. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): correct dark login bg + give ultra-dark wave real contrast Two related fixes: 1. Default-dark wave-header bg was wrong. I had #0a2227, but that's the *ultra-dark* override; default dark uses --dark-color-background = #0a1222. Now the dark-mode top half is the legacy purple-blue instead of teal. 2. Ultra-dark wave fill is intentionally near-identical to its bg in the legacy palette (#0f2d32 vs #0a2227, ~5/11/11 RGB delta), which makes the wave look static even though the animation is running. Bumped --wave-fill / --wave-fill-bottom to #1f4d52 in ultra-dark only — far enough above the bg that the motion reads, while staying within the same teal hue family. Also corrected ultra-dark --bg-page back to #0f2d32 (was briefly #0c0e12, which is the card color, not the page color). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): drop ultra-dark bottom-wave seam line Last fix made the wave fill #1f4d52 in ultra-dark for both top-three waves and the bottom wave, which gave visible motion but exposed a hard horizontal line where the bottom wave's flat lower edge met the page bg (#0f2d32). The user noticed it as "the wave at the bottom not moving its like a line" — they were seeing the SVG's clipped bottom edge, not the wave itself. Solution: only the top three waves get the brighter fill (those carry the visible motion). The bottom wave reverts to #0f2d32 = --bg-page, so its flat bottom edge merges seamlessly into the page below. Net effect: motion is still visible (from waves 2 and 3), and there's no seam line at the bottom of the SVG. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-i — index.html dashboard shell Replaces the smoke-test App.vue with a real IndexPage shell so the /index.html route now boots the actual dashboard layout in Vue 3: - a-config-provider drives AD-Vue 4's dark algorithm from useTheme (same pattern as LoginPage) - AppSidebar (Phase 5b component) is wired in with basePath + requestUri props - a-spin loading state with placeholder card while we build out the rest of the page - Page palette mirrors the legacy: light #f0f2f5, dark #0a1222 (--dark-color-background), ultra-dark #21242a The 1,805-line legacy index.html is too big for one commit. Split into five sub-phases on the todo list: ii) status cards + /server/status polling, iii) xray status card, iv) logs/backup/panel-update modals, v) custom-geo section. frontend/src/App.vue and frontend/src/main.js (smoke-test scaffold) are removed — both purposes now served by IndexPage and index.js. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-ii — live status cards on the dashboard Adds the CPU / memory / swap / disk dashboard cards to IndexPage, backed by a useStatus() composable that polls /panel/api/server/status every 2 s and a Status / CurTotal model ported from the legacy inline classes in index.html. - models/status.js — Status & CurTotal classes (CurTotal exposes reactive .percent and .color computed-style getters; Status maps the API payload + xray state to color/message strings) - composables/useStatus.js — 2s polling with shallowRef so each fetch swaps the whole Status object atomically. WebSocket integration intentionally deferred — the legacy panel falls back to this same 2s polling when its websocket drops, so we ship the proven path first and add WS on top in a later sub-phase. - pages/index/StatusCard.vue — four a-progress dashboard widgets in a 2x2 grid (mobile collapses to a 1x4). CPU widget exposes a history button; the modal it opens is part of 5c-iv. - IndexPage now consumes both, plus useMediaQuery so the layout responds to viewport changes. AD-Vue 4 changes: <a-icon type="area-chart"|"history"> dropped in favor of explicit AreaChartOutlined / HistoryOutlined imports. <a-tooltip slot="title"> → <template #title>. i18n strings still hardcoded English (Phase 7 wires up vue-i18n). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iii — xray status card + stop/restart controls XrayStatusCard.vue renders the right-hand card on the dashboard: - Title with mobile-only version tag (matches the legacy collapse) - Animated badge for the running/stop/error states. The pulsing dot comes from xray-pulse keyframes (renamed from runningAnimation in legacy custom.min.css). Color rings on the badge use the legacy's per-state border-color overrides on .ant-badge-status-processing. - Error state replaces the badge with a popover that surfaces the multi-line errorMsg + a logs shortcut. - Action row at the bottom: optional logs (when ipLimitEnable), stop, restart, and version switch. IndexPage now wires: - POST /panel/api/server/stopXrayService and /restartXrayService, followed by a refresh() so the status card reflects the new state without waiting for the next poll tick - POST /panel/setting/defaultSettings to read ipLimitEnable - Stub handlers for the panel-logs / xray-logs / version-switch / cpu-history modals — those land in 5c-iv AD-Vue 4 changes hit on this card: - <a-icon type="bars|poweroff|reload|tool"> → explicit BarsOutlined / PoweroffOutlined / ReloadOutlined / ToolOutlined - <span slot="title|content"> → <template #title|#content> - The .xray-*-animation classes ship as global <style> (not scoped) so they pierce AD-Vue's internal .ant-badge-status-* DOM. i18n still hardcoded English; Phase 7 wires vue-i18n. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (a) — panel update / logs / backup modals Adds three of the six dashboard modals plus a Quick Actions card that surfaces them. The remaining three (xray logs, version picker, CPU history sparkline) ship in 5c-iv-b. - PanelUpdateModal.vue — current vs latest version, "update now" button. Confirm dialog → POST /panel/api/server/updatePanel, then poll /server/status for up to 90s until the new panel answers, then reload. - LogModal.vue — panel logs viewer. Filters: rows (10-500), level (debug/info/notice/warning/error), syslog toggle. Auto-fetches on open and on every filter change. Color-coded timestamps and levels via inline span styles. Download button writes the raw log to x-ui.log via FileManager.downloadTextFile. - BackupModal.vue — db export (window.location to /getDb) and import (FormData upload to /importDB, then panel restart + reload). - Quick Actions card surfaces Logs / Backup / Update buttons and shows an orange update badge (extra slot) when an update is available. Modal-busy pattern: long-running operations (update, import) emit a `busy` event with a tip; IndexPage flips its a-spin overlay so the user sees a loading message while the panel is restarting. AD-Vue 4 changes: - v-model on <a-modal> renamed to v-model:open - v-model on <a-input>/<a-select>/<a-checkbox> uses the named v-model:value / v-model:checked pattern - <a-icon type="..."> dropped — explicit Ant icon imports (BarsOutlined, CloudServerOutlined, CloudDownloadOutlined, DownloadOutlined, UploadOutlined, SyncOutlined) - Modal.confirm() replaces this.$confirm() since setup() has no `this` Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (b) — cpu-history / xray-logs / xray-version modals Wires up the three remaining dashboard buttons that were stubbed in 5c-iv (a): the CPU history button on StatusCard, the xray-logs button in XrayStatusCard's error popover and ipLimitEnable action, and the "Switch xray" button in XrayStatusCard's action footer. - Sparkline.vue: shared SVG line chart (composition-API port of the inline Vue 2 component). Per-instance gradient id avoids defs collisions between sparklines on the same page. - CpuHistoryModal.vue: bucket dropdown (2m/30m/1h/2h/3h/5h) drives GET /panel/api/server/cpuHistory/{bucket}; renders via Sparkline. - XrayLogModal.vue: rows + filter + direct/blocked/proxy checkboxes; POST /panel/api/server/xraylogs/{rows} returns access-log entries rendered as a colored HTML table; download button serializes to text. - VersionModal.vue: collapse with Xray panel (radio list of versions from getXrayVersion, install via installXray/{version}) and Geofiles panel (per-file reload + Update all). CustomGeo collapse panel is Phase 5c-v. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-v — custom-geo section in VersionModal Adds the third collapse panel ("Custom geo") that lets users register external geosite/geoip files referenced by routing rules via ext:<filename>:tag. Backend endpoints are unchanged. - CustomGeoSection.vue: bordered table over /panel/api/custom-geo/list with per-row edit, download (refetch), and delete actions, plus an Add button and Update-all. Lazy-loads the list when the parent collapse opens this panel — closed panels don't fetch. - CustomGeoFormModal.vue: shared add/edit form with the same alias regex (^[a-z0-9_-]+$) and URL validation as legacy. Type and alias are immutable when editing — backend rejects changes anyway. - ext:<filename>:tag value is click-to-copy via ClipboardManager. - Relative time is computed inline (no moment dep); tooltip shows the absolute timestamp. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-i — settings page shell + dirty tracking Adds the settings entry as a new Vite multi-page input. Lays down the shared page chrome (sidebar, save bar, restart, security alert) and the AllSetting fetch/dirty-poll lifecycle so 5d-ii through 5d-vi can drop in tab partials without re-implementing it. - settings.html + src/settings.js: third Vite entry; mounts SettingsPage. - SettingsPage.vue: page chrome with the legacy two-button save/restart bar, conf-alerts banner, and 5 a-tabs (4 always-visible + the formats tab gated on subJsonEnable || subClashEnable). Each tab body is an a-empty placeholder until 5d-ii…vi fill them in. - useAllSetting.js composable: POST /panel/setting/all on mount, mirrors the legacy 1s busy-loop dirty check via setInterval, and exposes fetchAll/saveAll. saveDisabled flips off as soon as the user diverges from the server snapshot. - restartPanel rebuilds the URL (host/port/scheme/base path) from the saved settings so users land on the new endpoint after a port or cert change. - models/setting.js: adopts the @/utils alias and a leading file-level doc — semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-ii — settings General tab Ports the panel/general partial (the largest single tab) — six collapse panels: General, Notifications, Certificates, External traffic webhook, Date and time, LDAP. - GeneralTab.vue receives the reactive AllSetting via props and binds fields directly with v-model:value; SettingsPage stays the sole fetch/save owner. - remarkModel/remarkSeparator surfaced as computed v-models that read+write the underlying single-string field (legacy stores them packed as <separator><orderedKeys>, e.g. "-ieo"). - LDAP inbound-tags select binds to a CSV ↔ array computed; inbound options come from /panel/api/inbounds/list on mount. - Language select stays cookie-based via LanguageManager and reloads on change — same UX as legacy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iii — settings Security tab + 2FA modal Ports the panel/security partial: change-credentials form and 2FA toggle. The 2FA modal is a new shared component since enabling 2FA, disabling 2FA, and changing credentials all funnel through it with slightly different copy. - TwoFactorModal.vue: 'set' flow renders a QR code + manual key + a 6-digit verifier; 'confirm' flow renders just the verifier. The parent passes a confirm(success) callback that fires only when the entered code matches the live TOTP value (otpauth lib). - SecurityTab.vue: holds the local user form (oldUsername/oldPassword/ new*), POSTs /panel/setting/updateUser, and on success force-redirects to logout. When 2FA is on, the credentials change goes through the confirm-modal first. - toggleTwoFactor leaves the switch read-only (the v-bound :checked matches AllSetting) and only flips after the modal succeeds, so cancelling out leaves state unchanged. - Adds otpauth ^9.5.1 dep (qrious was already present). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iv — settings Telegram tab Ports the panel/telegram partial: bot enable/token/chatId/lang in the General panel, schedule/backup/login/CPU-threshold in Notifications, and proxy/API-server overrides in the third panel. All bindings live on the shared AllSetting reactive — no fetch/save logic in this tab. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-v — settings Subscription general tab Ports the subscription/general partial — four collapse panels covering the master enable switches, presentation/template fields, certs, and update interval. - Sub path goes through a strip-on-input + normalize-on-blur computed: legacy stripped `:` and `*` and ensured the value starts and ends with a single `/` — same here. - Both `subEnableRouting` and the announce/profile/title/support URLs are bound directly on AllSetting. - The "Subscription URI override" placeholder mirrors the legacy pattern for the manual full-URL form. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-vi — settings Subscription formats tab Ports the subscription/json partial — paths/URIs for the JSON and Clash formats plus the four packed-JSON sub-fields: fragment, noises, mux, and direct routing rules. - subJsonFragment / subJsonMux / subJsonNoises / subJsonRules are each a JSON string on the wire; the tab exposes their fields as computed v-models that read+write the underlying JSON. Toggling a top-level switch off resets the field to "" (matches legacy semantics). - Direct routing rules surface the IP and domain entries of the seed rule array as multi-select tag inputs; setting/removing tags edits the rules array in place rather than rebuilding it from scratch, so manually-added rules are preserved. - Tab is gated on subJsonEnable || subClashEnable in the parent (only rendered when the user actually opted into one of those formats). This closes Phase 5d — full settings page parity with the legacy panel across all five tabs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): route /panel/<route> to migrated pages in dev The sidebar links to production-style URLs like /panel/settings, but in dev that gets proxied to the legacy Go template — which fails because we haven't loaded the legacy asset chain. Add a proxy bypass so /panel and /panel/settings are served from index.html / settings.html on the Vite dev server itself. Unmigrated routes (inbounds, xray) still proxy to Go. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(csrf): expose token endpoint for SPA pages and fetch it from axios The legacy panel pages got their CSRF token from a <meta name="csrf-token"> tag rendered by Go. SPA pages built by Vite don't have that, so every unsafe (POST/PUT/DELETE) request from them was hitting CSRFMiddleware with no token and getting 403 — visible as the settings page being stuck on "Loading…" because POST /panel/setting/all failed. - web/controller/xui.go: GET /panel/csrf-token returns the session token. Lives under the xui group so checkLogin still gates it; the CSRFMiddleware on the same group is a no-op for GET. - frontend/src/api/axios-init.js: cache the token at module scope and lazy-fetch it when a non-safe request needs one. Seed from the meta tag first when present (legacy compat). On a 403 response, drop the cache and retry once — handles the case where a server restart rotated the token after the SPA loaded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): keep sidebar links absolute when basePath is empty The dashboard sidebar built tab keys as basePath + 'panel/...'. In dev the window-injected basePath is '' so the resulting key was a relative path like 'panel/settings'. When the browser resolved that against the current /panel/settings URL it produced /panel/panel/settings — visible as broken navigation between Dashboard and Settings. Force a leading slash so the keys are always absolute regardless of whether the host injected a basePath. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-i — inbounds page shell + list fetch Adds the inbounds entry as a fourth Vite multi-page input and wires /panel/inbounds through the dev proxy bypass. Lays down the page chrome (sidebar, summary statistics card, refresh button) and the fetch lifecycle composable so 5f-ii onward can drop in the table columns and the modals without re-implementing it. - inbounds.html + src/inbounds.js: fourth Vite entry; mounts InboundsPage. - InboundsPage.vue: sidebar + summary card (totals over up/down, all-time, inbound count, client tags) + a basic table with enable/ remark/port/protocol/traffic/expiry columns. Row actions, popovers, search/filter, auto-refresh, and the WebSocket delta path are all deferred to subsequent 5f subphases. - useInbounds.js composable: GET /panel/api/inbounds/list + POST /panel/api/inbounds/onlines + POST /panel/api/inbounds/lastOnline + POST /panel/setting/defaultSettings, then computes the per-inbound clientCount roll-ups (active/deactive/depleted/expiring/ online/comments) the table popovers consume. - models/dbinbound.js + models/inbound.js: switched the legacy-utils import to the @/utils alias for consistency with the rest of the app. Semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-ii — inbound list table + search/filter + auto-refresh Fleshes out the inbound list with the full column set, search & filter toolbar, row enable toggle wired to /panel/api/inbounds/setEnable/:id, and a per-row action dropdown that emits events the parent will route to modals as those land in 5f-iii through 5f-vii. - InboundList.vue (new): toolbar (Add inbound + General actions dropdown + Refresh + auto-refresh popover), search-or-filter switch with the legacy radio buttons (Active/Disabled/Depleted/Depleting/ Online), and a a-table with desktop and mobile column variants. Cells use AD-Vue 4's #bodyCell slot — protocol/clients/traffic/ allTime/expiry/info cells render the same popovers and tags as legacy. Row enable switch is optimistic with rollback on POST failure. - visibleInbounds computed mirrors the legacy search and filter projection: deep search through dbInbound + clients, or filter reduces inbound.settings.clients to the selected bucket so the table only shows matching client rows. - Auto-refresh interval is read/written to localStorage with the same keys (`isRefreshEnabled`, `refreshInterval`) as the legacy panel. WebSocket delta updates are still deferred. - Action menu emits event payloads {key, dbInbound}; the parent currently shows a "coming in later 5f subphase" toast for each. Modals (edit/qr/clone/delete/reset/info/clients) land in 5f-iii through 5f-vii. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): wrap popover-table rows in <tbody> Vue's template compiler warned that <tr> can't be a direct child of <table> per the HTML spec; the browser silently inserts a <tbody> wrapper but Vue's SSR/hydration path doesn't, which can cause hydration mismatches. Add explicit <tbody> in both popover tables (traffic + mobile-info). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii — inbound add/edit modal + delete/clone/reset Wires up the inbound CRUD flows. The protocol-specific and transport- specific forms are still ahead in 5f-iii-b — for now the modal exposes those as JSON textareas so users can both edit existing inbounds without losing settings and create new ones from default templates. - InboundFormModal.vue: tabbed modal with a full Basics tab (enable, remark, protocol, listen, port, total GB, traffic reset, expiry date) and three JSON-edit tabs (Settings, Stream, Sniffing). Add mode stamps a fresh template per protocol via Inbound.Settings.getSettings(protocol); changing the protocol in add mode restamps the JSON. Edit mode pretty-prints the existing JSON so the user sees the same fields they save back. - POST /panel/api/inbounds/add or /panel/api/inbounds/update/:id on submit; on success the parent refreshes the list and the modal closes. Malformed JSON in any of the three textareas surfaces a message.error and aborts the save without losing user input. - InboundsPage.vue: wires the row action menu to real handlers — edit (opens the modal in edit mode), delete, reset-traffic, clone, reset-clients, del-depleted-clients all go through Modal.confirm and refresh on success. General actions menu wires reset-inbounds / reset-clients / del-depleted-clients the same way. Remaining actions (qrcode/info/import/export/copyClients) still toast as "coming soon" — those land in 5f-iv and 5f-v. - Adds dayjs ^1.11.20 dep for the a-date-picker v-model interop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iv — client add/edit + bulk-add modals Wires per-inbound client management. Both flows go through the same addClient/updateClient endpoints as legacy; the modals just funnel the form state into the right shape (`{id, settings: '{"clients": [...]}'}`). - ClientFormModal.vue: protocol-aware single-client editor — email/ password/id/auth/security/flow/subId/tgId/comment/ipLimit/totalGB/ expiry/renewal fields are shown/hidden per protocol like legacy. Edit mode displays the per-client traffic stats with a reset button; IP-limit log is read on click and clearable. Random helpers (sync icon next to each label) regenerate UUID/email/ password/sub-id values. - ClientBulkModal.vue: 1–500 clients in one POST, with the legacy five email-generation modes (Random / +Prefix / +Num / +Postfix / Pure-Prefix-Num-Postfix). Builds clients via the protocol-aware factory and concatenates their toString() output into a single settings.clients JSON array. - InboundsPage.vue: opens both modals from the row action menu (`addClient` / `addBulkClient`). They both refresh the inbound list on success. - Outstanding row actions still toast as "coming soon": qrcode, showInfo, copyClients, clipboard. Those land in 5f-v / 5f-vi. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-v — inbound info + QR-code modals Wires the row "info" and "qrcode" actions and ports the legacy inbound_info_modal end-to-end. The info modal handles every protocol the legacy panel did: • multi-user (VMess/VLess/Trojan/SS-multi/Hysteria) — per-client table + share links + per-link QR; • SS single-user — share link + QR; • WireGuard — full peer table with downloadable peer-N.conf and a wg:// share link per peer; • Mixed/HTTP/Tunnel — connection-detail tables. - QrPanel.vue: shared link card (header tag, copy button, optional download button, optional QR canvas, monospace footer with the raw value). Per-instance QRious instances are repainted on value/size change. - InboundInfoModal.vue: full info modal. Subscription URL block keys off subSettings.subURI/subJsonURI; IP-log lazy-loads on open and surfaces refresh + clear; tg-id, last-online, depleted/enabled tags all match legacy. - QrCodeModal.vue: lighter modal used for the row "qrcode" action on SS-single and WireGuard inbounds (just the QRs, no info table). - InboundsPage.vue: wires both flows. checkFallback() reproduces the legacy logic — when an inbound listens on a unix-socket fallback (`@<name>`), the link generator is pointed at the root inbound that owns the listen address so QRs/links carry the public host:port + the right TLS state. Multi-client navigation (focusing a specific client's links) is deferred to 5f-vi where the per-inbound expand- row table will pass the email through. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vi — per-inbound client expand-row table Each multi-user inbound row in the list now expands to show its client roster, mirroring the legacy aClientTable component. - ClientRowTable.vue: inner a-table with full desktop column set (action icons / enable / online / client-with-status-dot / traffic with progress bar / all-time / expiry with reset cycle) and a collapsed mobile variant (single dropdown menu + popover info). Self-contained: stats are looked up via a per-inbound email->stats Map; per-client confirms (reset/delete) live on the row. - The component emits typed events (edit/qrcode/info/reset-traffic/ delete/toggle-enable) — InboundsPage routes them back to the existing client and info modals (with `findClientIndex` so the modal opens focused on the right client). - InboundList.vue: hooks ClientRowTable into the a-table's expandedRowRender slot; row-class-name `hide-expand-icon` and a scoped CSS rule hide the chevron for non-multi-user inbounds (HTTP/Mixed/Tunnel/WireGuard/SS-single) so they keep looking flat. - toggle-enable-client routes through updateClient with the same `{id, settings: '{"clients": [...]}'}` shape as the other modals, so backend parsing stays single-pathed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii-b — replace inbound modal JSON textareas with structured forms Rewrites InboundFormModal to look like the legacy panel: structured forms for the common case, with a compact "Advanced (JSON)" fallback for the rare bits we don't yet have UI for. Tabs: • Basics — enable/remark/protocol/listen/port/total/trafficReset/expiry • Protocol — protocol-aware: VMess/VLess/Trojan/SS-multi/Hysteria in add mode embed an inline first-client form (email + ID/password/auth, security, flow, subId, comment, total GB, expiry); edit mode shows a clients-count summary table; VLess: decryption/encryption inputs; SS: method dropdown that re-randomizes password and propagates method change to the multi-user array (matches legacy SSMethodChange); HTTP/Mixed: accounts table with add/remove rows + Mixed auth/udp/ip toggles; Tunnel: address/port/network/followRedirect; WireGuard: secretKey/pubKey (regen via Wireguard.generateKeypair) + per-peer fields with PSK regen + allowedIPs add/remove + keepAlive. • Stream — only when canEnableStream(); transport selector with structured forms for TCP (proxy-protocol, http camouflage), WS (host/path/heartbeat/headers), gRPC (serviceName, multiMode), HTTPUpgrade (host/path). KCP/XHTTP fall back to the Advanced tab with an alert banner. Security selector with TLS (sni/alpn/ fingerprint) and Reality (target/serverNames/keypair-gen via /panel/api/server/getNewX25519Cert / shortIds / fingerprint). • Sniffing — enabled/destOverride/metadataOnly/routeOnly/ ipsExcluded/domainsExcluded as structured fields. • Advanced (JSON) — raw streamSettings + sniffing JSON for users reaching KCP/XHTTP/sockopt/finalmask/full TLS cert arrays. The stream JSON is auto-synced from the live model whenever the structured fields change. State source of truth is a deeply-reactive Inbound + DBInbound pair cloned on open; submit serializes via inbound.settings.toString() + inbound.stream.toString() so the wire shape matches the legacy panel byte-for-byte. streamNetworkChange semantics (clear flow when TLS/Reality unavailable, reset finalmask.udp when not KCP) are preserved. Vision Seed for VLess + finer-grained TCP HTTP camouflage + the full TLS cert/ECH editor will land in 5f-iii-c. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vii — shared text/prompt modals + remaining export/import wiring Wires up the last batch of inbound row + general actions that were toasting "coming soon": export-inbound-links, export-subs (per-inbound and global), export-all-links, import-inbound, and the clipboard JSON peek. Two small shared components back them — both can be reused by the xray page later. - TextModal.vue (shared): read-only multi-line viewer with a copy button and an optional download button when fileName is set. Replaces the legacy txtModal which the inbounds page used for every link export. - PromptModal.vue (shared): generic title + input/textarea + confirm callback, with the legacy keybindings (Enter submits in single-line mode; Ctrl+S submits in textarea mode). Used here for import-inbound but also by xray-config edits in Phase 6. - InboundsPage.vue: drops the toast stubs for `import`/`export`/`subs` on the general-actions menu and `export`/`subs`/`clipboard` on the per-row menu, routing each through openText / openPrompt + the appropriate model helper (genInboundLinks, etc.). The copyClients cross-inbound modal stays toast-stubbed — that's its own dedicated legacy modal worth its own commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-i — xray page scaffold + Advanced JSON tab The fifth and last legacy page comes online. Tabs are scaffolded with a-empty placeholders for the structured editors (Basics / Routing / Outbounds / Balancers / DNS) so navigation is stable; the Advanced (JSON) tab is fully functional and lets power users edit the raw xraySetting tree exactly like the legacy CodeMirror pane. - xray.html + src/xray.js: fifth Vite multi-page entry, mounted as XrayPage; vite.config.js routes /panel/xray and /panel/xray/ to it through the dev proxy bypass alongside the other pages. - XrayPage.vue: page chrome with the Save / Restart-xray bar, restart- output popover (surfaces /panel/xray/getXrayResult content when startup fails), 6 a-tabs, and a textarea-backed Advanced JSON editor. CodeMirror is intentionally not pulled in — the textarea works for every modern browser and keeps the bundle slim while structured editors land in 6-ii through 6-v. - useXraySetting.js composable: POST /panel/xray/ on mount, mirrors the settings-page 1s busy-loop dirty check for both xraySetting and outboundTestUrl, and exposes saveAll + restartXray. The dirty flag relies on string equality of the pretty-printed JSON, so reformat-only edits don't enable Save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-ii — xray Basics tab structured editor Replaces the placeholder on the Basics tab with a structured form for the most-touched fields of the xray template — outbound + routing strategy, log levels, traffic stat counters, and the "basic routing" shortcuts (block torrent / IPs / domains, direct IPs / domains, IPv4 forced, WARP / NordVPN routing). - useXraySetting.js: hoists a parsed `templateSettings` reactive alongside the JSON string, with two cooperating watches that keep them in sync. Editing structured fields stringifies into xraySetting for the dirty-poll + Advanced JSON tab; editing the JSON re-parses into templateSettings only when valid, so structured tabs stay readable mid-edit. - BasicsTab.vue: collapse panels mirror the legacy partial — General, Statistics, Logs, Basic routing. Every input is a computed v-model reading/writing into templateSettings; the routing-rule shortcuts funnel through ruleGetter/ruleSetter which match the legacy templateRuleGetter/templateRuleSetter behavior (replace-first, drop-duplicates, pop-the-rule-when-empty). Direct/IPv4 setters also call syncOutbound() to provision/prune the matching outbound. - XrayPage.vue: imports BasicsTab + derives `warpExist`/`nordExist` from the parsed templateSettings. WARP/NordVPN provisioning modals are still placeholders that toast — those land in 6-v with the routing/outbound editors. Default tab flips back to Basics so users land on the structured editor. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iii — xray Routing tab + rule modal Replaces the Routing tab placeholder with a full editor for templateSettings.routing.rules: - RoutingTab.vue: a-table over the parsed rules with the legacy six- column layout (action / source / network / destination / inbound / outbound) and the same "lead value + N more" pill renderer for multi-value criteria. Mobile drops source/network/destination for readability. Per-row dropdown handles edit / move-up / move-down / delete; the array-mutation reordering replaces the legacy jQuery Sortable drag handle without pulling in a sortable lib. - RuleFormModal.vue: full form mirroring xray_rule_modal.html — CSV inputs for sourceIP/sourcePort/vlessRoute/ip/domain/user/port, Network select, Protocol multi-select, Attrs key/value pairs, inbound-tag multi-select sourced from templateSettings.inbounds + parent inboundTags + dnsTag, outbound-tag single-select sourced from templateSettings.outbounds + clientReverseTags, and balancerTag from templateSettings.routing.balancers. Submit serializes via the same shape the legacy `getResult` produces (CSV → array, drop empty fields). - XrayPage.vue: imports RoutingTab and exposes inboundTags + clientReverseTags from useXraySetting so the modal can populate its tag pools. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iv — xray Outbounds tab + outbound modal Replaces the Outbounds tab placeholder with a full table + add/edit flow. The 1.3k-line legacy outbound modal is condensed to a tabbed modal with structured Basics fields (tag/protocol/sendThrough/domain strategy) and JSON tabs for the protocol-specific settings + stream trees — same approach the Inbound modal uses, and a power user can still edit the same trees via the page-level Advanced (JSON) tab. - useXraySetting.js: adds fetchOutboundsTraffic + resetOutboundsTraffic + testOutbound. Test states are tracked per outbound index so the row's Test button can show loading + the Test-result column can render the response delay / status / error. - OutboundsTab.vue: full table (action / identity / address / traffic / test result / test) plus a card-list mobile variant with the same row dropdown (set-first / edit / move up/down / reset traffic / delete). outboundAddresses() reproduces the legacy findOutboundAddress logic so each protocol's host:port list is rendered consistently. Add/edit go through OutboundFormModal, delete goes through Modal.confirm, reset traffic posts to /panel/xray/resetOutboundsTraffic with the row's tag (or "-alltags-" from the toolbar). - OutboundFormModal.vue: tag/protocol/sendThrough/domainStrategy on the Basics tab; settings + streamSettings as raw JSON on their respective tabs. Tag-collision check happens client-side before emitting; malformed JSON aborts the save with a message.error. - XrayPage.vue: imports OutboundsTab and wires the test action to the composable's testOutbound helper. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-v — xray Balancers tab + DNS placeholder Brings Balancers to full parity with the legacy panel and adds a DNS tab placeholder that exposes the full dns/fakedns trees as JSON so users can edit them without falling through to Advanced. - BalancerFormModal.vue: tag (with duplicate-tag warning across other balancers), strategy (random/roundRobin/leastLoad/leastPing), selector tag-mode multi-select sourced from existing outbound tags + free-form additions, fallback. Disable-on-invalid is driven by the duplicateTag + emptySelector computed flags. - BalancersTab.vue: empty state with a single "Add balancer" CTA; populated state shows the legacy 4-column table (action / tag / strategy / selector / fallback) with per-row edit + delete in a dropdown. On submit the wire shape preserves the `strategy: { type }` nesting only when the strategy is non-default, matching the legacy emit. Tag renames also chase across routing.rules.balancerTag references so existing rules don't dangle. - DnsTab.vue: master enable switch + raw JSON for `dns` and `fakedns`. Legacy had a dedicated server-by-server editor + a fakedns row editor; both are big enough to deserve their own commits, and the JSON path supports every field today. WARP / NordVPN provisioning modals still toast as "coming soon" — those are third-party API integrations worth their own commits. The xray page now has structured editors for Basics / Routing / Outbounds / Balancers and JSON editors for DNS / Advanced — every xray tab the legacy panel offered is functional. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(server): Phase 8 — cut HTML routes over to web/dist/ Production cutover. Every user-facing HTML route now serves the Vue-3-built bundle from web/dist/ instead of rendering the legacy Go template; the long-hashed Vite assets are served at /assets/ from the same embedded filesystem. The legacy templates in web/html/ and the legacy static tree in web/assets/ are kept on disk for now in case a quick revert is needed, but nothing the binary serves references them. What changed: - web.go: a new //go:embed dist/* feeds the controller package via a SetDistFS hand-off before controller construction. The static /assets/ route is rebound: in dev to web/dist/assets/ on disk so Vite's incremental rebuilds show up live; in prod to the embedded dist via wrapDistFS (rooted one level deeper than wrapAssetsFS). - controller/dist.go: serveDistPage helper used by every HTML handler. Reads dist/<name> from the embedded FS and applies two transforms before sending: 1. injects <script>window.__X_UI_BASE_PATH__="..."</script> just before </head> so AppSidebar links resolve under the panel's basePath. 2. when basePath != "/", rewrites Vite's absolute /assets/ URLs to <basePath>assets/ so installs running under a custom URL prefix load the bundle where the static handler lives. HTML responses go out with no-cache so panel upgrades reach users on the next refresh; hashed JS/CSS stays cacheable. - controller/index.go: IndexController.index now serves dist/login.html for logged-out callers (the redirect for logged-in users is unchanged). - controller/xui.go: XUIController.{index,inbounds,settings,xraySettings} each become a one-line wrapper around serveDistPage. Smoke checklist for the maintainer: - run `cd frontend && npm run build` to refresh web/dist/ before building the Go binary (the embed snapshot is taken at compile time); - visit /panel/, /panel/inbounds, /panel/settings, /panel/xray and confirm each loads its Vue page; - log out and log back in to verify the login flow; - confirm the sidebar links navigate correctly under your install's basePath; - POST flows (e.g. saving settings) still need the CSRF token — that endpoint (/panel/csrf-token, added earlier) is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-vi — WARP + NordVPN provisioning modals Replaces the toast stubs on the Basics tab and Outbounds toolbar with the legacy WARP + NordVPN provisioning flows. Both modals now stage their wireguard outbounds back into templateSettings.outbounds through the same event channels OutboundsTab uses, so the existing add / reset / delete / refresh-traffic surface keeps working. - WarpModal.vue: empty state shows a single Create button that generates a wireguard keypair locally (Wireguard.generateKeypair) and posts it to /panel/xray/warp/reg; populated state surfaces the access_token / device_id / license_key / private_key, lets the user upgrade to WARP+ via /panel/xray/warp/license, refreshes the account info from /panel/xray/warp/config (plan / quota / usage in human-readable bytes), and stages a wireguard outbound with the WARP-specific reserved-byte encoding pulled from client_id. Add / Reset / Delete go through events the parent routes back to templateSettings.outbounds. - NordModal.vue: dual-tab login (NordVPN access token → /panel/xray/nord/reg, or paste a NordLynx private key → /panel/xray/nord/setKey). Once authenticated, country / city / server selectors fetch from /panel/xray/nord/{countries,servers}, servers sort by load ascending, the lowest-load server in the current city auto-selects. Reset emits oldTag/newTag so the parent renames matching routing rules in place; logout emits a remove-routing-rules event with prefix `nord-` to purge any dangling references. - XrayPage.vue: holds warpOpen / nordOpen flags, ensures the outbounds array exists before mutating it, and wires the modal events (add-outbound / reset-outbound / remove-outbound / remove-routing-rules) to in-place edits of templateSettings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 7 — vue-i18n wired up + login page translated Sets up vue-i18n on top of the panel's existing TOML translation files. The Go side stays the source of truth — translators continue to edit web/translation/*.toml; a sync script snapshots those files into per-locale JSON the Vue bundle imports. The login page is translated end-to-end as a worked example; remaining pages can be converted incrementally without infrastructure churn. What's in the box: - scripts/sync-locales.mjs: small TOML→JSON converter that walks web/translation/*.toml and writes frontend/src/locales/<code>.json. Handles the narrow subset of TOML the panel uses (flat key/value pairs + dotted [section.subsection] heads). Wired as a `prebuild` + `predev` script so production builds always include the latest strings without a manual step. - src/i18n/index.js: createI18n() in composition mode with all 13 locales emitted as their own Vite chunks. The active locale (read from the same `lang` cookie LanguageManager has always managed) plus the en-US fallback are eagerly loaded; the rest are dynamically importable via a loadLocale(code) helper. This keeps the per-page bundle the user actually downloads small — only ~30 KB of strings end up in the initial payload, vs ~220 KB if all 13 were eager. - All five page entries (index/login/settings/inbounds/xray) wire the i18n plugin into createApp via .use(i18n). - LoginPage.vue: t(...) replaces hardcoded English on the username / password / 2FA placeholders, the submit button label, and the Settings popover title. The Hello/Welcome headline cycle stays hardcoded — those are stylistic, not labels. The 'Hello'/'Welcome' cycle stays in English deliberately; the rest of the migration's components still ship hardcoded English and will be converted page by page in follow-up commits. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate page chrome — sidebar, save bars, tabs, summary cards Replaces hardcoded English with t() calls in the components every user sees on every page load. The translations themselves come from the existing TOML files via the sync script — no new strings, no new locale keys. Per component: - AppSidebar.vue: 5 menu titles (dashboard / inbounds / settings / xray / logout). Computed so the sidebar re-renders when the cookie-driven locale flips on reload. - IndexPage.vue: Quick actions card title + Logs / Backup / Up-to- date / Update buttons. - StatusCard.vue: CPU / Memory / Swap / Storage labels + logical-processors / frequency tooltips. - XrayStatusCard.vue: card title + error popover header + Stop / Restart / Switch xray action labels (kept the v-prefix version string as-is — it's content, not a label). - SettingsPage.vue: 5 tab titles + Save / Restart-panel buttons + unsaved-changes warning. - XrayPage.vue: 6 tab titles + Save / Restart-xray buttons + unsaved-changes warning. - InboundsPage.vue: 5 summary-stat card titles. - InboundList.vue: 10 column titles (computed for live locale), Add inbound / General actions buttons + every dropdown menu item, search placeholder, filter radio labels, popover titles (disabled / depleted / depleting / online), traffic + info popover row labels. Total: ~75 strings localised across 8 files. The remaining English labels live in the per-tab settings forms, the form modals (Inbound / Client / Outbound / Rule / Balancer / WARP / Nord), and the per-row table cell helpers — all incremental work that doesn't touch infrastructure. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate every remaining English string on the index page Closes the index page's i18n coverage. Combined with the page-chrome commit, every label users see on the dashboard is now sourced from the TOML translation files. Per file: - IndexPage.vue: loading-spinner tip (initial + dynamic). - BackupModal.vue: modal title, both list-item titles + descriptions ("Back up" / "Restore"), in-flight busy tips ("Importing database…" / "Restarting panel…"). - PanelUpdateModal.vue: modal title, update-available alert, current/latest version row labels, "Up to date" tag + label, primary action button. Modal.confirm now uses the translated panelUpdateDialog / panelUpdateDialogDesc with #version# substitution; success toast uses panelUpdateStartedPopover. - LogModal.vue: title slot ("Logs"). The Debug/Info/Notice/Warning/ Error log-level options stay literal — they're xray's wire values, not user-facing labels (matches the existing settings-page choice). - XrayLogModal.vue: title + Filter label. Direct/Blocked/Proxy stay literal for the same reason. - VersionModal.vue: modal title + xray-switch alert + per-file tooltip + "Update all" button + custom-geo collapse header. The Modal.confirm flows for switchXrayVersion + updateGeofile use translated dialog/desc with #version# / #filename# substitution. - CpuHistoryModal.vue: title slot. - CustomGeoSection.vue: routing-hint alert, Add / Update-all buttons, every column title (computed for live locale), copy/edit/download/ delete tooltips, copy toast, delete-confirm modal, empty-state text. - CustomGeoFormModal.vue: add/edit titles, OK/cancel labels, Type/ Alias/URL field labels, alias placeholder, all three validation toasts. Total: ~50 strings localised across 8 index-page files. The Hello / Welcome login headline cycle and a handful of literal xray wire values (Direct/Blocked/Proxy/log levels) are intentionally kept hardcoded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): Phase 7-c — translate settings, inbounds modals, xray tabs Continues the page-by-page translation pass started in cb37dd55 — runs every user-visible string on settings (General/Security/Telegram/Sub), inbounds (Client/QR/Info modals), and xray (Routing/Balancer/Rule/Warp/ Nord/Basics/Outbounds tabs) through useI18n. Updates the TOML→JSON sync script to escape `@` (vue-i18n parses it as a linked-format prefix) and refreshes all 13 locale files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): Phase 9 — restore index dashboard, fix login/CSRF, port legacy styles - Index dashboard regains the 8 cards that were lost in the SPA port (3X-UI panel info, Operation Hours, System Load, Usage, Overall Speed, Total Data, IP Addresses, Connection Stats), plus a Config button that shows the live xray config.json. Version display falls back through panelUpdateInfo → window.__X_UI_CUR_VER__ → '?' so dev mode isn't blank. - Xray config no longer hangs on load: useXraySetting surfaces failures instead of leaving a perpetual spinner, and the Vite dev proxy stops hijacking POST requests to migrated routes (only GETs get bypassed). - Inbound page no longer throws __asyncLoader/emitsOptions errors — inbound.js was missing imports (NumberFormatter, SizeFormatter, Wireguard) and InboundList kept emitting after unmount. - Login round-trip works after logout: a public /csrf-token endpoint bootstraps the SPA before authentication, axios caches the token module-level, and the dev 401 handler navigates to /login.html instead of reloading the dashboard into a redirect loop. - legacy.css mirrors the legacy panel's surface/text variables so dark and ultra-dark themes match main; every SPA entry imports it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray DNS section to match main branch DnsTab now exposes every field the legacy panel did — top-level toggles (tag, hosts, queryStrategy, disableCache/queryConcurrency, fallback strategy, client subnet), the servers table with per-row strategy and domain/expectIP/unexpectedIP overrides, and the Fake DNS pool. The new DnsServerModal covers the full add/edit flow and collapses to a bare string when the user only sets an address — matching the wire shape the legacy form emits for plain DNS entries like "8.8.8.8". Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray outbound modal with structured per-protocol forms Replaces the JSON textareas with the same shape the legacy panel uses: all 11 outbound protocols (vmess/vless/trojan/shadowsocks/socks/http/ mixed/wireguard/tun/dns/loopback/blackhole/freedom) get dedicated fields, every transport (TCP/KCP/WS/gRPC/HTTPUpgrade/XHTTP) gets its own panel, and TLS/Reality/sockopt/Mux are configured through the same controls as the inbound side. Brings the SPA outbound editor to parity with main so users no longer have to drop into raw JSON. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): bring inbound modal to full parity with main branch Switches the default protocol on add to VLESS, fixes a crash when adding a Mixed account (the constructor is SocksAccount, not MixedAccount), and fills in the fields the SPA was previously delegating to the Advanced JSON tab: - TLS: cipher suites, min/max version, reject SNI / disable system root / session resumption switches, the certificate array with per-row Path-or-Content toggle (Set Default pulls from /panel/setting/ defaultSettings), One Time Loading, Usage / Build Chain, plus ECH key/config with a Get New ECH Cert button. - Reality: xver, target/SNI sync icons (uses getRandomRealityTarget), max time diff, min/max client version, short IDs randomizer, SpiderX, mldsa65 seed/verify with Get New Seed. - Stream: full structured forms for every transport — TCP HTTP camouflage gets its request/response editor, mKCP gets MTU/TTI/uplink/ downlink/CWND/maxSendingWindow, WebSocket / gRPC (now with Authority) / HTTPUpgrade get headers + proxy-protocol toggles, XHTTP gets the full SplitHTTPConfig surface (mode-aware fields, padding obfs, session/sequence placement, uplink data, no-SSE). - New External Proxy section and a structured Sockopt block (mark, TCP keepalive/timeout/clamp, fast open, MPTCP, penetrate, V6Only, domain strategy, congestion, TProxy, dialer/interface, trusted XFF). - VLESS gets the legacy X25519 / ML-KEM-768 buttons that fetch fresh decryption/encryption blocks via /panel/api/server/getNewVlessEnc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): add FinalMask UI (TCP/UDP masks + QUIC params) to inbound and outbound Mirrors web/html/form/stream/stream_finalmask.html as a shared FinalMaskForm component used by both modals — they share the same StreamSettings shape (addTcpMask/addUdpMask/finalmask/enableQuicParams) so a single template handles both. Surfaces: - TCP masks for raw/tcp/httpupgrade/ws/grpc/xhttp networks: fragment, sudoku, and header-custom (with the 2D clients/servers groups, each row supporting array/str/hex/base64 packets and a randomize button for base64). - UDP masks for hysteria protocol or kcp network: hysteria gets just salamander; kcp gets the full type list (mkcp variants, header-*, xdns/xicmp, header-custom with flat client/server lists, and noise). Switching to xdns shrinks the kcp MTU to 900 to match the legacy panel's behavior. - QUIC Params for hysteria or xhttp: congestion (incl. brutal up/down fields), debug, UDP hop ports/interval, idle/keepalive timeouts, path-MTU discovery toggle, and the four receive-window tunables. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): remove duplicate Outbound test URL from xray Advanced tab The Basics tab already exposes this field through BasicsTab — duplicating it on the Advanced tab let two inputs race the same ref and only added clutter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): unify theming on vanilla AD-Vue light/dark/ultra-dark The legacy panel CSS (custom.min.css ported as legacy.css) tinted every non-primary button teal-green via .dark .ant-btn:not(.ant-btn-primary) overrides while AD-Vue 4's darkAlgorithm kept primary buttons blue — producing the mixed blue/green button look on dark mode. Drop legacy.css entirely and let AD-Vue 4's algorithms own the palette. Centralize antdThemeConfig in useTheme.js so every page resolves to the same source of truth (light = defaultAlgorithm, dark = darkAlgorithm, ultra-dark = darkAlgorithm + deeper colorBgBase/Layout/Container/ Elevated tokens). Each page's <a-config-provider> now imports the shared computed instead of defining its own copy. Drops the 67 KB legacy CSS chunk; per-page CSS bundles fall to ≤5.9 KB. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): restore computed import in Settings + Xray pages When 5f1aba28 dropped the local antdThemeConfig computed (now shared from useTheme), it also stripped `computed` from the import list — but both pages still call computed() elsewhere (confAlerts, advanced-tab helpers). Re-adds it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): retheme dashboard gauges to AD-Vue blue and shrink them - StatusCard's CPU/RAM/Swap/Storage dashboards rendered at AD-Vue's default 120px width which made the percent text balloon to ~36px. Drop to 90px (70px on mobile) so the gauge fits the rest of the card. - The CurTotal.color thresholds still hardcoded the legacy teal/orange palette (#008771 / #f37b24 / #cf3c3c). Switch to AD-Vue's primary / warning / danger tokens (#1677ff / #faad14 / #ff4d4f) so the gauges match the rest of the panel under both light and dark themes. - XrayStatusCard's running-animation badge ring also still pointed at the deleted --color-primary-100 var; hardcode the new primary blue. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: shorten backupTitle to "Backup & Restore" across all 13 locales The backup modal header was the second-longest title in the dashboard on every locale ("Database Backup & Restore" / "Резервне копіювання та відновлення бази даних" / etc). Drop the "Database / Veritabanı / 数据库" qualifier — the modal already lives under the "Database" column, so the shorter form reads cleaner on narrow viewports. Updated both the .toml source-of-truth files and the synced .json locales (re-running scripts/sync-locales.mjs). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: collapse two translation databases into a single web/translation/<lang>.json set The Vue SPA had been reading from frontend/src/locales/*.json while the Go binary still loaded web/translation/translate.*.toml — and a sync-locales.mjs pre-build step kept the two in lockstep, with TOML as the source of truth. Now that go-i18n v2.6.1 already flattens nested JSON via recGetMessages/addChildMessages, both runtimes can share one file per locale. - Move the 13 nested-JSON locale files to web/translation/<lang>.json so they live alongside the Go //go:embed translation/* directive. - Switch web/locale/locale.go from toml.Unmarshal to json.Unmarshal (and drop the pelletier/go-toml import — it's now indirect-only). Confirmed via a smoke test that pages.index.cpu, subscription.title, tgbot.commands.help, and menu.settings all resolve in en-US, fa-IR, ru-RU, and zh-CN. - Repoint Vue's i18n loader at the new path (../../../web/translation/ *.json glob) and drop the moved-here pathDelimiter comment that no longer applies. - Delete the 13 legacy translate.*.toml files and the sync-locales.mjs script + its npm pre-script hooks (predev/prebuild/i18n:sync). The Telegram bot and subscription page still get their messages because they were reading the same MessageIDs the JSON files now produce. - Update copilot-instructions.md so the next contributor knows where the canonical translation files live. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): redesign expand-row + retheme client visuals When you expanded an inbound row, the nested <a-table> inside ClientRowTable burst out of the parent's scroll-x box — its .ant-spin-container ended up wider than the parent's narrow .ant-table-cell, so the child looked oversized while the parent looked squeezed. Replace the nested table with a CSS-grid layout that owns its sizing, sits flush inside the expanded cell, and collapses to a 3-column layout on mobile (action menu, client identity, info popover). While in there, fix three other client-row visuals: - The Unicode infinity glyph (U+221E) renders as an "m"-shaped character in some system fonts (Windows Segoe UI in particular). Add a shared <InfinityIcon /> SVG component (legacy panel's path) and use it in ClientRowTable, InboundList, and InboundInfoModal — desktop and mobile cells. - The "unlimited quota" traffic bar passed :percent="100" with no stroke-color, so AD-Vue auto-coloured it success-green. Pin it to the AD-Vue purple token (#722ed1) so it reads as the no-limit sentinel rather than another usage state. - ColorUtils + the in-row statsExpColor still hardcoded the legacy teal/orange/red/purple palette (#008771 / #f37b24 / #cf3c3c / #7a316f). Map them onto AD-Vue 4's success/warning/danger/purple tokens (#52c41a / #faad14 / #ff4d4f / #722ed1) so badges, tags, and progress bars all match the rest of the panel. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): darken light-theme page bg so cards stand out The light-theme --bg-page was #f0f2f5 — close enough to AD-Vue's #fff card background that the cards faded into the page. Bump it to #e6e8ec (a more visibly distinct gray) so cards lift cleanly off the surface. Dark and ultra-dark stay where they were. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): shrink dashboard percent text and surface the unfinished arc Two follow-up tweaks to the dashboard gauges: - AD-Vue scales the percent text from the SVG, not from :width, so the 90px gauges still rendered the number at ~27px. Pin .ant-progress-text to 14px via :deep() and trim the gauge to 70px (60px on mobile) so the whole card stays compact. - The default trail (rgba(0,0,0,0.06) / rgba(255,255,255,0.08)) was invisible on the light-theme card. Pass an explicit rgba(128,128,128,0.25) trail-color so the unfinished portion is visible under both light and dark themes. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): migrate subpage.html to Vue 3 SPA The subscription info page was the last page still rendered by Go templates. Move it to the Vite multi-page setup so the whole panel loads through one toolchain. Frontend: SubPage.vue mounts at /sub/<id>?html=1 and reads window.__SUB_PAGE_DATA__ for the parsed view-model (traffic / quota / expiry + rendered share links). Fix descriptions borders against the light-theme card by painting the row divider on each cell's bottom edge — AD-Vue's <tr> border doesn't render reliably under border-collapse:collapse. Backend: serveSubPage reads dist/subpage.html, injects window.__X_UI_BASE_PATH__ + window.__SUB_PAGE_DATA__ before </head>, and rewrites Vite's absolute /assets/ URLs when the panel runs under a URL prefix. Drop the legacy template-FuncMap wiring and switch the sub server's static mount from web/assets to web/dist/assets. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): inbound modal QR + tabs + restored TLS fallbacks Per-client QR action: the qr icon on the expand-row table opened the big info modal instead of the QR modal. Route it to QrCodeModal and extend that modal with a `client` prop so genAllLinks() produces the per-client share URLs (and per-peer remarks for WireGuard). Inbound's Data redesign: split the dense single-page view into three tabs — Inbound, Client, Subscription. Drop every QR rendering from this modal (QrCodeModal is the QR home now). Each row in the Inbound tab is one label/value pair instead of the legacy 2-column grid, and long values like the VLESS encryption blob render as a wrapping code block with a copy button so they can't blow out the dialog. The Subscription tab renders sub URL + JSON URL as clickable anchors that open in a new tab. Restored TLS fallbacks UI: the model already exposed VLESSSettings.Fallback / TrojanSettings.Fallback with addFallback / delFallback / fallbackToJson, but the form modal never surfaced them during the Vue 3 migration. Re-add the legacy form (SNI, ALPN, Path, Destination, PROXY) on the protocol tab, gated on TCP transport plus (for VLESS) encryption=none — same conditions as main. Column widths: Protocol 70→130 and All-time Traffic 60→95 in the inbound list; All-time Traffic 90→130 in the client expand-row, so the header text fits and tags don't get squeezed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): navy dark theme + rounded inbound/client corners Dark theme picks up a refined navy palette (page #0a1426, cards #142340, sider #0d1d33) so the sidebar blends with the rest of the surface; ultra-dark stays neutral black. Resolves the previous mismatch where AD-Vue 4 hardcoded #001529 / #002140 for the sider, trigger and dark Menu items via Layout.colorBgHeader / colorBgTrigger and Menu's colorItemBg — overrides go through the component-token map now. Round the inbound table's outer corners (header start/end + last row end) and wrap the client expand-row grid in a 1px / 8px-radius border so the list reads as a contained block instead of a flush rectangle. Linter-driven whitespace cleanup across inbounds/*.vue rolled into the same commit since it can't be split out cleanly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): xray tab fixes — modal close, tag validation, full XHTTP, reset to default Modal close: BalancersTab / OutboundsTab / RoutingTab confirmDelete used arrow expressions that returned splice's removed-items array. AD-Vue 4 treats truthy non-thenables from onOk as "still pending" and never closes the dialog (see ActionButton.js:103-106), so the confirm modal stayed open. Wrap the body so onOk returns undefined and AD-Vue auto-closes. Tag validation: outbound + balancer modals only flipped between warning/success on duplicate, leaving the empty case as a green ✓. Split into a 3-state computed — error (empty) / warning (duplicate) / success — and wire a help message so the input clearly explains why the OK button is disabled. Reset to default: re-add the legacy "Reset to Default" panel at the bottom of BasicsTab. Calls /panel/setting/getDefaultJsonConfig and overwrites templateSettings; the existing watch re-stringifies so the JSON tab + dirty-poll see the new state. Restored Basics option lists from main: IPs (4→10, +Vietnam/Spain/ Indonesia/Ukraine/Türkiye/Brazil), DomainsOptions (4→10, +regex entries), BlockDomainsOptions (5→17, +Malware/Phishing/Adult/regex), ServicesOptions (Reddit/Speedtest in, off-template Microsoft out). Outbound form parity with main: • Reverse Sniffing UI for VLESS — toggle + destOverride checkboxes (HTTP/TLS/QUIC/FAKEDNS) + Metadata/Route Only + IPs/Domains excluded multi-selects, gated on reverseTag being set. • Full XHTTP transport — request headers list, Max Upload Size / Min Upload Interval (packet-up), Padding Obfs Mode + sub-fields, Uplink HTTP Method, Session/Sequence/UplinkData placement + keys, No gRPC Header (stream-up/stream-one), expanded XMUX with Max Concurrency/Connections/Reuse/Request/Reusable/Keep-alive. Strip a-divider from the outbound form per request — replaced with plain section/item heading divs so the labels and per-row delete icons stay but the horizontal rule is gone. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): xray Advanced tab parity + finalmask gating Advanced tab was a single textarea bound to the full xraySetting blob. Restore the legacy 4-way view: a radio group toggles between All / Inbounds / Outbounds / Routing Rules, and the textarea reads/writes the matching slice through templateSettings. Added the legacy header ("Advanced Xray Configuration Template" + description) so the page introduces itself like main. Outbound finalmask leaked into protocols that don't have a stream (Freedom / Blackhole / DNS / Socks / HTTP / Wireguard) because the v-if only checked outbound.stream. Gate the whole FinalMaskForm on outbound.canEnableStream() to match main. Drop the leading divider inside FinalMaskForm — its parent already provides separation, so the rule above "TCP Masks" was redundant. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound Advanced tab live mirror + QR exact-fit sizing Advanced tab in the inbound modal showed stale state. The watch only refreshed advancedJson.stream, so toggling the Sniffing switch in the Sniffing tab left the Advanced JSON showing the prior value. And encryption — stored on inbound.settings.encryption, not on stream — never appeared at all because Advanced only exposed stream + sniffing. Split the watch into three (stream / sniffing / settings) and add a settings textarea so encryption / clients / fallbacks live alongside the existing two views. The submit() path now reads settings from the JSON tab too (falling back to inbound.settings.toString()) so power-user edits in Advanced override the structured form on save. QR canvas: when a longer share-URL bumps the QR matrix size, QRious falls back to floor(canvasSize / matrixWidth) and centers the pattern, leaving a white margin (e.g. matrix=41, size=180 → 8px gap). Pre-pick the QR version from the URL byte length and set canvas size to a multiple of matrixWidth × pixelSize so the pattern always fills it edge-to-edge — no white margin even after toggling encryption on. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound stream tidy-up + QR sizing + dev proxy Stream tab clean-up: drop the seven a-divider rules in the inbound form's Stream tab — replace the labelled ones (Request / Response / Security) with a section-heading div that matches the outbound modal, delete the empty rules above TLS sub-blocks / External Proxy / Sockopt. Empty header-list form-items also leaked margin space below each "Add header" button across TCP / WS / HTTPUpgrade / XHTTP — gate each on headers.length > 0 so they vanish until the user adds one. QR panel: drop the link text under the canvas (the user already has a copy button on the header). Pin the canvas display size to a fixed 240px square via :style + image-rendering: pixelated/crisp-edges so a dense WireGuard config QR and its sparser link share the same on-screen footprint without blurring. Dev proxy: Node's AggregateError wraps connection failures whenever DNS returns more than one address (::1 + 127.0.0.1) and the code lands on the inner errors, not the outer. The existing handler only checked err.code so the ECONNREFUSED stack still spammed the log when the Go backend was down. Walk err.errors too, print one friendly line ("backend not reachable — start the Go server"), then stay quiet for the rest of the session. Vendor splitting + chunk-size warning: split node_modules into stable vendor-* chunks so each page only ships the deps it uses and the browser caches them across versions. ant-design-vue stays as a single chunk because its components share internals; raise the chunk-size warning to 1500kB so the build stays quiet (its 1.4MB minified gzips to ~410kB on the wire). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): info-modal cleanup + 2FA QR + outbound link import - 2FA QR: matrix-snap canvas + opaque background to drop white margin - Inbound info modal: stack Mixed/HTTP/Tunnel as info-rows, hide tab strip when only the Inbound tab applies - Add inline VLESS Reverse tag input on first-client form - Hide Protocol tab for TUN (no form yet) - Outbound link converter: route through Outbound.fromLink so vless/trojan/ss/hysteria(2) imports work alongside vmess; fix stray implicit global in fromLink Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): jalali calendar + drop legacy moment-jalali - Wire Calendar Type setting to a real Jalali datepicker via vue3-persian-datetime-picker, gated by useDatepicker composable - DateTimePicker wrapper swaps between AD-Vue and Persian picker; keeps dayjs v-model contract so existing forms/setters work unchanged - Theme picker popup explicitly per body.dark / data-theme=ultra-dark (AD-Vue 4 doesn't expose CSS vars, so var() fallbacks defaulted to white); fix invisible disabled days, SVG arrow fills, popup clipping via append-to="body" - Replace stray moment() calls in dbinbound/inbound models with dayjs; the legacy global was undefined under ESM and broke the inbounds list whenever any inbound had expiryTime > 0 - Remove legacy moment-jalali / persian-datepicker / aPersianDatepicker assets — replaced by the Vue 3 picker Note: dark/ultra background of the date popup still renders white in some cases — pending follow-up. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): jalali popup theming + full-month layout - Re-prefix popup selectors with .vpd-wrapper (popup root that travels with appendTo='body'), not .vpd-main (which stays at the input); paints the popup's dark/ultra background again - Drop the 1px border on .vpd-content — with box-sizing: border-box it ate 2px from the day-row width, wrapping the 7th cell of every row and hiding days 18-31 of months that needed a 5th week Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat: render dates in Jalali when Calendar Type is jalalian - IntlUtil.formatDate accepts an optional calendar arg; appends the BCP-47 -u-ca-persian extension so Intl renders Jalali across all UI languages, not just fa-IR - Plumb the panel's datepicker setting into the SubPage via the Go injection (window.__SUB_PAGE_DATA__.datepicker) - Panel pages (inbound list/info, client row, xray log) read the same setting through the useDatepicker composable so the whole panel stays consistent Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): ultra-dark page tint + mobile-friendly inbound view - Drop --bg-page from #21242a (lighter than the cards) to #050505 in ultra-dark across index/sub/settings/inbounds/xray, so cards consistently elevate over the page - Hide the inline sider's children + collapse-trigger and zero its width below 768px; the floating drawer-handle remains the menu trigger - Inbounds page mobile pass: tighten content-area + card padding; flex-wrap the filter bar instead of stacking; shrink table cell padding so all 4 mobile columns fit; bump expand / action / info icon hit targets - Per-client expand row on mobile: soft-tinted rounded cards instead of hairline borders, larger action / info touch targets, more legible email typography, bigger status badge dot Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove legacy template + asset trees and dead Go template engine - Delete web/html/ entirely (page templates, form/, modals/, component/, common/, settings/) — every route is served from web/dist/ now via serveDistPage; nothing in the binary referenced these - Delete web/assets/ entirely (jQuery-era ant-design-vue, axios, moment, codemirror, qrcode/qs/uri/vue/otpauth, custom CSS, Vazirmatn font); Vite bundles all of this into web/dist/assets - Drop the Gin HTML template wiring: remove //go:embed assets + //go:embed html/*, the assetsFS/htmlFS vars, the wrapAssetsFS adapter, EmbeddedHTML / EmbeddedAssets exports, getHtmlFiles / getHtmlTemplate, the i18nWebFunc/funcMap and SetFuncMap call, and the dev/prod template-engine branch — only StaticFS for /assets/ is needed now - Remove dead html()/getContext() helpers and unused imports from web/controller/util.go (no c.HTML(...) callers remain) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound expand chevron position + cpu history layout - Push the inbound table's expand chevron away from the left edge with margin-inline + cell padding so it isn't flush against the corner - Move "Timeframe: …" caption above the chart (was below); restore the line that the previous edit removed - Fix x-axis time labels being clipped at the bottom of the cpu chart — the offset (paddingTop+drawHeight+22 = 222) exceeded the SVG viewBox height (220); dropped to +14 so labels sit at y=214 with room for descenders - Move the SVG axis text colors out of <style scoped> into a global block — Vue's scoped CSS doesn't always hash-attribute SVG <text> descendants, so the dark-mode overrides via :global() weren't matching; bumped opacity 0.55 → 0.85 for legibility on navy/black Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(login): language picker in settings popover + fluid card sizing - Add language select alongside the theme switch (mirrors SubPage) - Bind headline to pages.login.hello / pages.login.title so the "Hello / Welcome" cycle re-translates with the active locale - Replace AD-Vue 5-breakpoint grid with clamp() sizing so the card scales smoothly instead of jumping ~33% at each breakpoint - Pin horizontal padding so input width stays stable on large viewports Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): organize entry HTML + bootstrap JS into folders - Move entry HTML files: frontend/*.html -> frontend/html/*.html - Move per-page bootstrap modules: src/{index,login,settings,inbounds,xray,subpage}.js -> src/entries/ - Update vite.config rollup inputs and dev-mode MIGRATED_ROUTES to /html/<page>.html - Build output now lands at web/dist/html/<page>.html - serveDistPage and subController updated to read from dist/html/ Cleans up the flat frontend/ root which previously interleaved 6 HTML files with package.json, README, src/, etc. The src/ root similarly gets rid of 6 entry .js files mixed in alongside api/, components/, models/, etc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove obsolete vue3 phase1 inventory doc The migration is well past phase 1 — the inventory doc has rotted and the live state lives in the codebase plus the plan files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): merge utils/legacy.js into utils/index.js The barrel was a placeholder for an eventual split that hasn't happened. Collapsing the two files removes one layer of indirection and the misleading "legacy" name (the contents are still actively used by the migrated SPA). - Move all 930 lines from utils/legacy.js into utils/index.js - Delete utils/legacy.js - Update direct import in models/outbound.js to '@/utils' - Drop a stale legacy.js reference in InboundFormModal comment Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(frontend): keep entry HTML files at frontend/ root The earlier move to frontend/html/ made dev-mode URLs ugly (http://localhost:5173/html/index.html instead of plain /). The folder didn't add real value — it just hid 6 files behind a non-conventional layout. Reverting that piece while keeping src/entries/ (which is a genuine separation between page bootstrap and the rest of src/). - HTML files back at frontend/<page>.html - Vite rollupOptions.input + MIGRATED_ROUTES restored to flat paths - Build output is web/dist/<page>.html again - web/controller/dist.go and sub/subController.go read from dist/<name> Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): bump eslint to 10 + add flat config + clean lint warnings - Upgrade eslint 9.39 -> 10.3 and eslint-plugin-vue 9.33 -> 10.9 - Add eslint.config.js (flat config required by ESLint 10) with vue3-recommended rules, sensible defaults, and exemptions for the project's existing formatting style - Drop --ext from the lint script (removed in ESLint 10) - vue/no-mutating-props is left off because the form-modal pattern ports straight from Vue 2 (parent passes a reactive object, child mutates it); a real fix is an architectural rewire, separate task Lint warning cleanup: - utils/index.js: var -> let/const in the X25519 routines, replace obj.hasOwnProperty(...) with Object.prototype.hasOwnProperty.call(...) - Remove unused imports (reactive, ref, Inbound) in ClientFormModal, InboundInfoModal, QrCodeModal, DnsServerModal, OutboundFormModal, SubPage; remove unused locals (isClientOnline, ONLINE_GRACE_MS, fetchAll, isSocks, isHTTP, _antdAlgorithm) - XrayStatusCard: declare 'open-logs' on defineEmits (was emitted but not declared) - RuleFormModal: rename v-for var t -> tag (shadowed useI18n's t) - Drop stale eslint-disable directives (no-new, no-unused-vars) - OutboundsTab/InboundList: drop redundant initial null assigns - InboundInfoModal/OutboundFormModal: explicit eslint-disable for the intentional local-ref-shadows-prop pattern in modal drafts `npm run lint` now passes with 0 errors and 0 warnings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): one client identity across multiple inbounds via subId Lets the operator add the same email under the same subId to several inbounds. Xray reports traffic per email, so a single client_traffics row acts as the shared accumulator — no aggregation overhead, quota and expiry stay consistent. - Email validation allows duplicates only when subId matches - AddClientStat upserts via OnConflict DoNothing (idempotent on rerun) - Stat/IP rows survive client deletion when a sibling inbound still references the email - enrichClientStats tops up GORM-preloaded stats with rows whose inbound_id points at a sibling, so every panel view sees usage - disableInvalidClients cascades enable=false and syncs the row's total/expiry into every sibling JSON when the shared identity expires - DelDepletedClients removes the depleted client from all referencing inbounds, batched - Subscription services dedupe traffic by email so shared quota is counted once Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * docs(frontend): rewrite README for multi-page Vue 3 layout Reflects the current state — embedded build, per-route HTML entries, ESLint 10 flat config, src/ layout, and the steps to add a new page. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): drop deprecated rimraf/glob/inflight transitive deps vue3-persian-datetime-picker pinned moment-jalaali to ^0.9.4, which pulled rimraf@3 → glob@7 → inflight@1. inflight in particular leaks memory and is unmaintained. Override moment-jalaali to ^0.10.4 (same runtime API, dropped the legacy build deps) so npm install no longer warns and the dep tree is 12 packages lighter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): multi-node panel orchestration (CRUD, deployment, traffic sync, sub per-node) - Node model + service + controller (/panel/api/nodes/*) with bearer-token apiToken auth - Heartbeat job @every 10s; status/latency/xrayVersion surfaced in Nodes UI - Runtime abstraction (Local + Remote) so inbound/client mutations target the inbound's owning node instead of always hitting the local xray - Inbounds gain optional NodeID; tag-based correlation with remote panel (no RemoteInboundID column needed) - NodeTrafficSyncJob @every 10s pulls absolute counters + online/lastOnline from each enabled+online node and writes them into central DB; 30s reset grace window prevents post-reset overwrite - Reset propagation to nodes (best-effort) on client/inbound/all reset paths - Subscription server uses node.Address for inbounds with NodeID, falling back to existing host resolution for local inbounds - Frontend: Nodes page, "Deploy to" select in inbound form, Node column on inbound list, hostOverride threaded through genAllLinks/QR/Info modals Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(stats): system history modal + per-node CPU/Mem trends across all locales Backend - web/service/metric_history.go: generic in-memory ring buffer with two singletons — system-wide (cpu/mem/netUp/netDown/online/load1/5/15) and per-node (cpu/mem) keyed by node id - ServerService.AppendStatusSample writes all 8 metrics every 2s on the same tick; AppendCpuSample/AggregateCpuHistory kept for back-compat - NodeService.UpdateHeartbeat appends cpu/mem only on online ticks so offline gaps render as missing data, not phantom dips - New routes: GET /panel/api/server/history/:metric/:bucket and GET /panel/api/nodes/history/:id/:metric/:bucket, both whitelisted Frontend - Sparkline component generalized: arbitrary value range (auto-scale when valueMax=null), pluggable yFormatter/tooltipFormatter for B/s, client counts, load averages - SystemHistoryModal replaces CpuHistoryModal with tabs for every metric; opened from a tag on the 3X-UI card next to Documentation - NodeHistoryPanel: expandable row on the Nodes table showing per-node CPU and Mem trends, refreshed every 15s Localization - Backfill systemHistoryTitle / trendLast2Min / pages.inbounds.{node, deployTo, localPanel} and the entire pages.nodes block (51 keys including statusValues + toasts) into all 11 non-en/fa locales: ar-EG, es-ES, id-ID, ja-JP, pt-BR, ru-RU, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(embed): include underscore-prefixed Vite chunks in dist FS go:embed silently excludes files whose names start with `_` or `.`, so the `_plugin-vue_export-helper-<hash>.js` chunk that Vite/rolldown emits for @vitejs/plugin-vue was missing from the production binary. First import at runtime hit a 404 and the SPA failed to mount — blank page on every page load, no error in the server logs because the asset 404 was just a static-handler miss. Switched the directive to `//go:embed all:dist` which keeps the same root layout but disables the underscore/dot exclusion rule. Dev mode was unaffected (it serves dist/assets/ from disk, not the embedded FS). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * ci: build frontend bundle before Go compile in release.yml + Dockerfile Phase 8 cut all panel HTML routes over to web/dist/ and embedded the Vite bundle into the Go binary via //go:embed all:dist. web/dist/ is .gitignored, so on a fresh CI checkout it doesn't exist — every Go build since Phase 8 has been failing with "pattern dist: no matching files found" or producing a binary that 404s on first asset request. release.yml: add a setup-node@v4 + npm ci + npm run build trio before the existing go build step in both the Linux matrix job (7 arches) and the Windows job. npm cache is keyed on frontend/package-lock.json. Dockerfile: add a node:22-alpine frontend stage that runs npm ci + npm run build and emits to /src/web/dist (via vite.config.js's outDir). The golang builder stage then COPY --from=frontend /src/web/dist into ./web/dist before the go build, so embed.FS sees the bundle. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(ws): live updates on inbounds/xray/nodes pages, drop polling + manual refresh Replaces the legacy polling + manual-refresh model with WebSocket pushes across the three live-data pages. The hub already broadcast traffic / client_stats / outbounds; this wires the frontend to consume them and adds a new `nodes` channel for the heartbeat job's snapshot. Frontend - new useWebSocket composable: page-scoped singleton WebSocketClient, lifecycle-managed on/off, leaves disconnect to page-unload - inbounds: useInbounds gains applyTrafficEvent / applyClientStatsEvent / applyInvalidate that merge counters and online/lastOnline in place; InboundsPage subscribes; InboundList drops the auto-refresh popover, the refresh button, and the now-unused refreshing prop - xray outbounds: useXraySetting gains applyOutboundsEvent; XrayPage subscribes; OutboundsTab drops the refresh button + emit - nodes: useNodes gains applyNodesEvent and stops the 5s setInterval/visibilitychange polling; NodesPage subscribes; NodeList drops the refresh button and ReloadOutlined import Backend - web/websocket: new MessageTypeNodes + BroadcastNodes notifier - node_heartbeat_job: after wg.Wait(), reload the table once and BroadcastNodes(updated). Gated on websocket.HasClients() so a panel with no open browser doesn't spend the DB read Bug fixes spotted in this pass - websocket.js #buildUrl defaulted basePath to '' when the global was missing (dev mode), producing `ws://host:portws` and a SyntaxError on the WebSocket constructor. Fall back to '/' and ensure leading slash. - vite.config.js: forward /ws to ws://localhost:2053 with ws:true so dev (5173) reaches the Go backend's WebSocket - NodeFormModal: a-input-password's visibilityToggle is Boolean in AntD Vue 4; the v3-era object form (`{ visible, 'onUpdate:visible' }`) triggered a Vue prop-type warning. Drop the override (default true shows the eye icon and toggles internally) and remove the orphaned tokenVisible ref Translations - pages.inbounds.autoRefresh / autoRefreshInterval: removed from all 13 locales (UI gone) - pages.nodes.refresh: removed from all 13 locales (UI gone) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): hide Node column when no nodes are defined Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-09 08:38:48 -07:00
func (s *XrayService) GetXrayAPIPort() int {
process := currentXrayProcess()
if process == nil || !process.IsRunning() {
Vue3 migration (#4198) * docs(migration): Phase 1 inventory — Vue 2 / AD-Vue 1 surface area Captures the breakage surface for the Vue 3 + Ant Design Vue 4 + Vite migration: 17,650 lines across 69 templates, 3,145 a-* component instances across 63 files, with per-pattern counts and file lists. Key findings: - No Vue filters anywhere — dodges a major Vue 3 breaking change - 358 v-model uses; AD-Vue 4 absorbs most, custom components don't - 233 <template slot="X"> usages must become <template #X> - 49 scopedSlots: { ... } column defs need new slots: { ... } shape - a-icon is removed in AD-Vue 4 — every icon must be imported Establishes the 8-phase order; Phase 2 (Vite toolchain) is next. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): Phase 2 — scaffold Vite + Vue 3 + AD-Vue 4 Adds a frontend/ directory that lives alongside the legacy web/html/ Vue 2 templates during the migration. Vite builds into ../web/dist/ so the Go binary will be able to embed the result via embed.FS once Phase 4 starts moving real pages over. - package.json pins Vue 3.5, Ant Design Vue 4.2, Vite 6, vue-i18n 10 - vite.config.js: dev server on :5173 with API proxy to the Go panel on :2053; build output to ../web/dist/ - src/App.vue is currently a smoke-test placeholder — delete once the first real page (login) lands in Phase 4 - node_modules and dist are already ignored at repo root To verify locally: cd frontend && npm install && npm run dev Pages will be migrated one at a time on the vue3-migration branch. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): Phase 3 — port utils, models, axios, websocket as ES modules Ports the framework-agnostic JS from web/assets/js/ into frontend/src/ so Vue 3 pages can import what they need without relying on script-tag globals. - web/assets/js/util/index.js (927 lines, 21 classes) → frontend/src/utils/legacy.js + a barrel at utils/index.js. All classes are now named exports. - Vue.prototype.$message in HttpUtil → direct import of `message` from ant-design-vue (Vue 3 has no Vue.prototype). - RandomUtil.randomShadowsocksPassword previously defaulted to SSMethods.BLAKE3_AES_256_GCM from inbound.js, creating a circular import. Replaced with the literal string default. - MediaQueryMixin (Vue 2 mixin) removed. Replaced by composables/useMediaQuery.js — Vue 3 composable returning reactive `isMobile`. - axios-init.js wrapped as setupAxios(); Qs global → npm `qs`. - websocket.js exported as WebSocketClient class; the implicit window.wsClient global is gone — pages instantiate it themselves. - model/{inbound,outbound,dbinbound,setting,reality_targets}.js copied with `export` added on every top-level declaration. Imports between models and utils are wired up explicitly. - subscription.js deferred to Phase 5 (it's a Vue 2 mount, not a util). - App.vue smoke test exercises SizeFormatter / RandomUtil / Wireguard / useMediaQuery so the user can verify Phase 3 with `npm run dev`. Run `cd frontend && npm install && npm run dev` — qs was added so a fresh install is required. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 4 — port login.html to Vue 3 + AD-Vue 4 + Vite 8 First real page in the new toolchain. Multi-page Vite: each migrated page is its own entry. login.html now lives at frontend/login.html with a thin entrypoint at frontend/src/login.js mounting LoginPage.vue. Vite 6 → Vite 8.0.11 (per user request). Requires Node 20.19+ or 22.12+. @vitejs/plugin-vue bumped to ^6.0.6 (peers vite ^8). Ant Design Vue stays on 4.2.6 — there is no AD-Vue 6. Vue 2 → Vue 3 / AD-Vue 1 → AD-Vue 4 syntax changes hit on this page: - new Vue({ el, delimiters, data, methods }) → createApp + <script setup> - mounted() → onMounted() - <template slot="X"> → <template #X> - <a-icon slot="prefix" type="user"> → <template #prefix><UserOutlined /> </template> with explicit @ant-design/icons-vue imports - v-model.trim → v-model:value (AD-Vue 4 uses named v-model on inputs) Three legacy features deferred so Phase 4 stays small: - i18n (Phase 7 wires up vue-i18n) - theme switcher (custom component pending Phase 5) - headline word-cycle animation (purely aesthetic) Run `cd frontend && npm install && npm run dev`, open http://localhost:5173/login.html. With Go panel running on :2053 the form submits real credentials via the configured proxy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5a — theme system + Vite 8 + vue-i18n 11 Bumps Vite to 8.0.11 (npm install picked up 6.4.2 from the stale lockfile; clean install resolves the new constraint). Bumps vue-i18n to 11.1.4 since v10 was just EOL'd. Migrates aThemeSwitch.html — the two-flavor theme picker + global themeSwitcher object — into: - composables/useTheme.js: single reactive `theme` state with toggleTheme / toggleUltra. Boot side-effect applies the stored theme to <body>/<html> before Vue renders; watchEffect persists changes back to localStorage. - components/ThemeSwitch.vue: full menu version for the main panel. - components/ThemeSwitchLogin.vue: login-popover version. AD-Vue 1 → 4 changes hit on this component: - <a-icon type="bulb" :theme="filled|outlined"> dropped — replaced by explicit BulbFilled / BulbOutlined imports from @ant-design/icons-vue, swapped via <component :is="BulbIcon"> - Vue.component('a-theme-switch', { ... }) global registration → SFC + per-page import - this.$message.config(...) (Vue 2 instance method) → message.config(...) imported from ant-design-vue, called once in login.js at boot Login page now surfaces a settings button → popover → theme picker. Known gap: web/assets/css/custom.min.css isn't yet imported into the new bundle, so toggling dark mode currently only re-themes AD-Vue's own components, not the panel chrome. The body class is still toggled so behavior is correct; visual fidelity returns when custom.css is ported or directly imported. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5b — port four shared components to Vue 3 CustomStatistic.vue and SettingListItem.vue are mechanical Vue.component → SFC ports. AppSidebar.vue: AD-Vue 4 dropped <a-icon :type="dynamic">, so the five sidebar icons (dashboard/user/setting/tool/logout) live in a name→component map and render via <component :is>. The legacy <a-drawer slot="handle"> hack is replaced with a sibling fixed- position toggle button. Tab paths take basePath/requestUri as props instead of pulling them from Go template scope. TableSortable.vue: the biggest Vue 3 rewrite of this phase. - $listeners is gone — replaced by inheritAttrs: false + explicit attrs forwarding - scopedSlots: this.$scopedSlots collapsed into Vue 3's unified slots object — just iterate Object.keys(this.slots) and forward - Vue 2 h(tag, { props, on, scopedSlots }, children) → Vue 3 h(tag, { ...props, ...on }, slotsObject) - 'a-table' string → resolveComponent('a-table') so app.use(Antd) registration is honored - inject: ['sortable'] (Options API) → inject('sortable', null) (Composition API) inside the trigger child - beforeDestroy → beforeUnmount - customRow's return shape flattened (no nested props/on/class) Two intentional skips, documented in the migration doc: - aClientTable.html — slot fragments, not a component. Migrates inline with inbounds.html (new Phase 5f). - aPersianDatepicker.html — wraps a Persian-only third-party lib; defer until settings.html lands. Build verified with vite 8.0.11. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): anchor Vite dev proxy so /login.html isn't forwarded The /login proxy entry was matching any path starting with /login — including /login.html, which Vite is supposed to serve itself. Without the Go backend running, this caused ECONNREFUSED noise on every page load. Switched to regex patterns anchored with ^...$ so only the bare backend paths (/login, /logout, /getTwoFactorEnable) and explicit sub-routes (/panel/*, /server/*) get proxied. Static .html files Vite serves directly are no longer matched. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): real dark mode + silence dev proxy ECONNREFUSED noise Two issues from running login.html against no Go backend: 1. Dark mode toggled the body class but didn't actually re-theme any AD-Vue components. The legacy panel relied on custom.min.css which we haven't ported. AD-Vue 4 ships its own dark algorithm — wrap LoginPage in <a-config-provider :theme="{ algorithm }"> driven by our useTheme state, and AD-Vue restyles every component for free. Page chrome (background, card, title) gets explicit .is-dark CSS since the algorithm only covers AD-Vue components. 2. Vite logged every failed proxy attempt loudly. When the Go panel isn't running locally that's pure noise. Added a configure() callback that swallows ECONNREFUSED specifically; real errors (timeouts, 5xx, anything else) still surface. Both fixes are dev-experience only — production build is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): use legacy panel palette for login page dark mode Earlier dark mode used invented colors (#141a26 page bg, #1f2937 card) that didn't match the rest of the panel. Replaced with the actual values from web/assets/css/custom.min.css: light dark ultra-dark bg #c7ebe2 bg #222d42 bg #0f2d32 card #fff card #151f31 card #0c0e12 title #008771 title #fff/.92 title #fff/.92 Drove everything off CSS custom properties on .login-app so the .is-dark / .is-ultra class swap is a few var overrides instead of duplicating selectors. Also restored the legacy card metrics (2rem radius, 4rem 3rem padding, 2rem title) so the new page matches the old panel's geometry, not just its colors. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave layout + recolor for dark mode The wave SVG had inline fill="#c7ebe2" (mint) on the bottom wave, so in dark/ultra-dark mode it rendered as a pale-white blob against the dark page. Stripped the inline fills, drove them off CSS variables that swap with .is-dark / .is-ultra: light: green tints + #c7ebe2 (mint) on the bottom wave dark: #222d42 across all four waves ultra-dark: #0f2d32 The wave was also positioned wrong — anchored to the top 200px of the viewport with absolute positioning. Restored the legacy layout: - .waves-header is fixed to the top of the viewport with z-index -1 so the form floats over it - .waves-inner-header pushes the wave SVG down to ~50vh with a 50vh-tall solid block of the page color - .waves SVG itself is 15vh tall, sitting at the bottom of that block Net effect: top half is solid-colored, then a wavy edge transitions into the rest of the page, with the form centered on top — matching the legacy panel exactly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): bring wave-header to front so the wave actually shows Two layering bugs were hiding the wave entirely: 1. .ant-layout-content had background: var(--bg-page) which painted an opaque rectangle covering the full content area — including the fixed wave-header behind it. Made the layout/content transparent and moved the bg paint up to .login-app (the outer ant-layout). 2. .waves-header had z-index: -1 which on its own was fine, but with .ant-layout-content opaque on top it was doubly buried. Promoted the wave-header to z-index: 0 and gave the form .login-row z-index: 1, so the form sits above the wave and the wave sits above the page-bg. Also set --bg-page to the legacy mint (#c7ebe2) for light mode so the bottom half of the page below the wave matches the legacy panel (was white). Dark mode stays at the surface-100/login-wave palette. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave animation timings + dark page bg Two reasons the bottom wave looked static in dark/ultra-dark: 1. Animation durations were 7s/10s/13s/20s. Legacy uses 4s/7s/10s/13s. The 20s on the bottom wave was so slow that against the low dark- mode contrast it read as motionless. Restored the legacy timings. 2. --bg-page in dark mode was #151f31 (card color / surface-100), but the legacy .under uses surface-200 (#222d42) — that's the color of the bottom half of the page, the same as the wave fill, so the wave appears to flow into the page rather than meeting a hard edge. Now it does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): restore Hello/Welcome headline cycle on login Earlier I deferred the legacy headline word-cycling animation as "purely aesthetic". Restored it: the title now alternates between 'Hello' and 'Welcome' every 2 seconds, matching the legacy panel. The legacy implementation toggled .is-visible / .is-hidden classes on two <b> elements via setTimeout chains and DOM querying. Replaced with a reactive ref + Vue 3 <Transition mode="out-in"> so the fade between words is declarative — no manual DOM manipulation, and the interval is properly cleaned up in onBeforeUnmount. The earlier "Welcome to 3x-ui" string was wrong on two counts: it should be just "Welcome", and it should be one of two cycling words with "Hello" preceding it. Ultra-dark palette already matched legacy after the prior wave timing fix; no additional changes needed there beyond the animation speeds that now also apply to ultra-dark via the shared CSS rules. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): correct dark login bg + give ultra-dark wave real contrast Two related fixes: 1. Default-dark wave-header bg was wrong. I had #0a2227, but that's the *ultra-dark* override; default dark uses --dark-color-background = #0a1222. Now the dark-mode top half is the legacy purple-blue instead of teal. 2. Ultra-dark wave fill is intentionally near-identical to its bg in the legacy palette (#0f2d32 vs #0a2227, ~5/11/11 RGB delta), which makes the wave look static even though the animation is running. Bumped --wave-fill / --wave-fill-bottom to #1f4d52 in ultra-dark only — far enough above the bg that the motion reads, while staying within the same teal hue family. Also corrected ultra-dark --bg-page back to #0f2d32 (was briefly #0c0e12, which is the card color, not the page color). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): drop ultra-dark bottom-wave seam line Last fix made the wave fill #1f4d52 in ultra-dark for both top-three waves and the bottom wave, which gave visible motion but exposed a hard horizontal line where the bottom wave's flat lower edge met the page bg (#0f2d32). The user noticed it as "the wave at the bottom not moving its like a line" — they were seeing the SVG's clipped bottom edge, not the wave itself. Solution: only the top three waves get the brighter fill (those carry the visible motion). The bottom wave reverts to #0f2d32 = --bg-page, so its flat bottom edge merges seamlessly into the page below. Net effect: motion is still visible (from waves 2 and 3), and there's no seam line at the bottom of the SVG. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-i — index.html dashboard shell Replaces the smoke-test App.vue with a real IndexPage shell so the /index.html route now boots the actual dashboard layout in Vue 3: - a-config-provider drives AD-Vue 4's dark algorithm from useTheme (same pattern as LoginPage) - AppSidebar (Phase 5b component) is wired in with basePath + requestUri props - a-spin loading state with placeholder card while we build out the rest of the page - Page palette mirrors the legacy: light #f0f2f5, dark #0a1222 (--dark-color-background), ultra-dark #21242a The 1,805-line legacy index.html is too big for one commit. Split into five sub-phases on the todo list: ii) status cards + /server/status polling, iii) xray status card, iv) logs/backup/panel-update modals, v) custom-geo section. frontend/src/App.vue and frontend/src/main.js (smoke-test scaffold) are removed — both purposes now served by IndexPage and index.js. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-ii — live status cards on the dashboard Adds the CPU / memory / swap / disk dashboard cards to IndexPage, backed by a useStatus() composable that polls /panel/api/server/status every 2 s and a Status / CurTotal model ported from the legacy inline classes in index.html. - models/status.js — Status & CurTotal classes (CurTotal exposes reactive .percent and .color computed-style getters; Status maps the API payload + xray state to color/message strings) - composables/useStatus.js — 2s polling with shallowRef so each fetch swaps the whole Status object atomically. WebSocket integration intentionally deferred — the legacy panel falls back to this same 2s polling when its websocket drops, so we ship the proven path first and add WS on top in a later sub-phase. - pages/index/StatusCard.vue — four a-progress dashboard widgets in a 2x2 grid (mobile collapses to a 1x4). CPU widget exposes a history button; the modal it opens is part of 5c-iv. - IndexPage now consumes both, plus useMediaQuery so the layout responds to viewport changes. AD-Vue 4 changes: <a-icon type="area-chart"|"history"> dropped in favor of explicit AreaChartOutlined / HistoryOutlined imports. <a-tooltip slot="title"> → <template #title>. i18n strings still hardcoded English (Phase 7 wires up vue-i18n). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iii — xray status card + stop/restart controls XrayStatusCard.vue renders the right-hand card on the dashboard: - Title with mobile-only version tag (matches the legacy collapse) - Animated badge for the running/stop/error states. The pulsing dot comes from xray-pulse keyframes (renamed from runningAnimation in legacy custom.min.css). Color rings on the badge use the legacy's per-state border-color overrides on .ant-badge-status-processing. - Error state replaces the badge with a popover that surfaces the multi-line errorMsg + a logs shortcut. - Action row at the bottom: optional logs (when ipLimitEnable), stop, restart, and version switch. IndexPage now wires: - POST /panel/api/server/stopXrayService and /restartXrayService, followed by a refresh() so the status card reflects the new state without waiting for the next poll tick - POST /panel/setting/defaultSettings to read ipLimitEnable - Stub handlers for the panel-logs / xray-logs / version-switch / cpu-history modals — those land in 5c-iv AD-Vue 4 changes hit on this card: - <a-icon type="bars|poweroff|reload|tool"> → explicit BarsOutlined / PoweroffOutlined / ReloadOutlined / ToolOutlined - <span slot="title|content"> → <template #title|#content> - The .xray-*-animation classes ship as global <style> (not scoped) so they pierce AD-Vue's internal .ant-badge-status-* DOM. i18n still hardcoded English; Phase 7 wires vue-i18n. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (a) — panel update / logs / backup modals Adds three of the six dashboard modals plus a Quick Actions card that surfaces them. The remaining three (xray logs, version picker, CPU history sparkline) ship in 5c-iv-b. - PanelUpdateModal.vue — current vs latest version, "update now" button. Confirm dialog → POST /panel/api/server/updatePanel, then poll /server/status for up to 90s until the new panel answers, then reload. - LogModal.vue — panel logs viewer. Filters: rows (10-500), level (debug/info/notice/warning/error), syslog toggle. Auto-fetches on open and on every filter change. Color-coded timestamps and levels via inline span styles. Download button writes the raw log to x-ui.log via FileManager.downloadTextFile. - BackupModal.vue — db export (window.location to /getDb) and import (FormData upload to /importDB, then panel restart + reload). - Quick Actions card surfaces Logs / Backup / Update buttons and shows an orange update badge (extra slot) when an update is available. Modal-busy pattern: long-running operations (update, import) emit a `busy` event with a tip; IndexPage flips its a-spin overlay so the user sees a loading message while the panel is restarting. AD-Vue 4 changes: - v-model on <a-modal> renamed to v-model:open - v-model on <a-input>/<a-select>/<a-checkbox> uses the named v-model:value / v-model:checked pattern - <a-icon type="..."> dropped — explicit Ant icon imports (BarsOutlined, CloudServerOutlined, CloudDownloadOutlined, DownloadOutlined, UploadOutlined, SyncOutlined) - Modal.confirm() replaces this.$confirm() since setup() has no `this` Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (b) — cpu-history / xray-logs / xray-version modals Wires up the three remaining dashboard buttons that were stubbed in 5c-iv (a): the CPU history button on StatusCard, the xray-logs button in XrayStatusCard's error popover and ipLimitEnable action, and the "Switch xray" button in XrayStatusCard's action footer. - Sparkline.vue: shared SVG line chart (composition-API port of the inline Vue 2 component). Per-instance gradient id avoids defs collisions between sparklines on the same page. - CpuHistoryModal.vue: bucket dropdown (2m/30m/1h/2h/3h/5h) drives GET /panel/api/server/cpuHistory/{bucket}; renders via Sparkline. - XrayLogModal.vue: rows + filter + direct/blocked/proxy checkboxes; POST /panel/api/server/xraylogs/{rows} returns access-log entries rendered as a colored HTML table; download button serializes to text. - VersionModal.vue: collapse with Xray panel (radio list of versions from getXrayVersion, install via installXray/{version}) and Geofiles panel (per-file reload + Update all). CustomGeo collapse panel is Phase 5c-v. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-v — custom-geo section in VersionModal Adds the third collapse panel ("Custom geo") that lets users register external geosite/geoip files referenced by routing rules via ext:<filename>:tag. Backend endpoints are unchanged. - CustomGeoSection.vue: bordered table over /panel/api/custom-geo/list with per-row edit, download (refetch), and delete actions, plus an Add button and Update-all. Lazy-loads the list when the parent collapse opens this panel — closed panels don't fetch. - CustomGeoFormModal.vue: shared add/edit form with the same alias regex (^[a-z0-9_-]+$) and URL validation as legacy. Type and alias are immutable when editing — backend rejects changes anyway. - ext:<filename>:tag value is click-to-copy via ClipboardManager. - Relative time is computed inline (no moment dep); tooltip shows the absolute timestamp. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-i — settings page shell + dirty tracking Adds the settings entry as a new Vite multi-page input. Lays down the shared page chrome (sidebar, save bar, restart, security alert) and the AllSetting fetch/dirty-poll lifecycle so 5d-ii through 5d-vi can drop in tab partials without re-implementing it. - settings.html + src/settings.js: third Vite entry; mounts SettingsPage. - SettingsPage.vue: page chrome with the legacy two-button save/restart bar, conf-alerts banner, and 5 a-tabs (4 always-visible + the formats tab gated on subJsonEnable || subClashEnable). Each tab body is an a-empty placeholder until 5d-ii…vi fill them in. - useAllSetting.js composable: POST /panel/setting/all on mount, mirrors the legacy 1s busy-loop dirty check via setInterval, and exposes fetchAll/saveAll. saveDisabled flips off as soon as the user diverges from the server snapshot. - restartPanel rebuilds the URL (host/port/scheme/base path) from the saved settings so users land on the new endpoint after a port or cert change. - models/setting.js: adopts the @/utils alias and a leading file-level doc — semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-ii — settings General tab Ports the panel/general partial (the largest single tab) — six collapse panels: General, Notifications, Certificates, External traffic webhook, Date and time, LDAP. - GeneralTab.vue receives the reactive AllSetting via props and binds fields directly with v-model:value; SettingsPage stays the sole fetch/save owner. - remarkModel/remarkSeparator surfaced as computed v-models that read+write the underlying single-string field (legacy stores them packed as <separator><orderedKeys>, e.g. "-ieo"). - LDAP inbound-tags select binds to a CSV ↔ array computed; inbound options come from /panel/api/inbounds/list on mount. - Language select stays cookie-based via LanguageManager and reloads on change — same UX as legacy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iii — settings Security tab + 2FA modal Ports the panel/security partial: change-credentials form and 2FA toggle. The 2FA modal is a new shared component since enabling 2FA, disabling 2FA, and changing credentials all funnel through it with slightly different copy. - TwoFactorModal.vue: 'set' flow renders a QR code + manual key + a 6-digit verifier; 'confirm' flow renders just the verifier. The parent passes a confirm(success) callback that fires only when the entered code matches the live TOTP value (otpauth lib). - SecurityTab.vue: holds the local user form (oldUsername/oldPassword/ new*), POSTs /panel/setting/updateUser, and on success force-redirects to logout. When 2FA is on, the credentials change goes through the confirm-modal first. - toggleTwoFactor leaves the switch read-only (the v-bound :checked matches AllSetting) and only flips after the modal succeeds, so cancelling out leaves state unchanged. - Adds otpauth ^9.5.1 dep (qrious was already present). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iv — settings Telegram tab Ports the panel/telegram partial: bot enable/token/chatId/lang in the General panel, schedule/backup/login/CPU-threshold in Notifications, and proxy/API-server overrides in the third panel. All bindings live on the shared AllSetting reactive — no fetch/save logic in this tab. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-v — settings Subscription general tab Ports the subscription/general partial — four collapse panels covering the master enable switches, presentation/template fields, certs, and update interval. - Sub path goes through a strip-on-input + normalize-on-blur computed: legacy stripped `:` and `*` and ensured the value starts and ends with a single `/` — same here. - Both `subEnableRouting` and the announce/profile/title/support URLs are bound directly on AllSetting. - The "Subscription URI override" placeholder mirrors the legacy pattern for the manual full-URL form. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-vi — settings Subscription formats tab Ports the subscription/json partial — paths/URIs for the JSON and Clash formats plus the four packed-JSON sub-fields: fragment, noises, mux, and direct routing rules. - subJsonFragment / subJsonMux / subJsonNoises / subJsonRules are each a JSON string on the wire; the tab exposes their fields as computed v-models that read+write the underlying JSON. Toggling a top-level switch off resets the field to "" (matches legacy semantics). - Direct routing rules surface the IP and domain entries of the seed rule array as multi-select tag inputs; setting/removing tags edits the rules array in place rather than rebuilding it from scratch, so manually-added rules are preserved. - Tab is gated on subJsonEnable || subClashEnable in the parent (only rendered when the user actually opted into one of those formats). This closes Phase 5d — full settings page parity with the legacy panel across all five tabs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): route /panel/<route> to migrated pages in dev The sidebar links to production-style URLs like /panel/settings, but in dev that gets proxied to the legacy Go template — which fails because we haven't loaded the legacy asset chain. Add a proxy bypass so /panel and /panel/settings are served from index.html / settings.html on the Vite dev server itself. Unmigrated routes (inbounds, xray) still proxy to Go. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(csrf): expose token endpoint for SPA pages and fetch it from axios The legacy panel pages got their CSRF token from a <meta name="csrf-token"> tag rendered by Go. SPA pages built by Vite don't have that, so every unsafe (POST/PUT/DELETE) request from them was hitting CSRFMiddleware with no token and getting 403 — visible as the settings page being stuck on "Loading…" because POST /panel/setting/all failed. - web/controller/xui.go: GET /panel/csrf-token returns the session token. Lives under the xui group so checkLogin still gates it; the CSRFMiddleware on the same group is a no-op for GET. - frontend/src/api/axios-init.js: cache the token at module scope and lazy-fetch it when a non-safe request needs one. Seed from the meta tag first when present (legacy compat). On a 403 response, drop the cache and retry once — handles the case where a server restart rotated the token after the SPA loaded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): keep sidebar links absolute when basePath is empty The dashboard sidebar built tab keys as basePath + 'panel/...'. In dev the window-injected basePath is '' so the resulting key was a relative path like 'panel/settings'. When the browser resolved that against the current /panel/settings URL it produced /panel/panel/settings — visible as broken navigation between Dashboard and Settings. Force a leading slash so the keys are always absolute regardless of whether the host injected a basePath. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-i — inbounds page shell + list fetch Adds the inbounds entry as a fourth Vite multi-page input and wires /panel/inbounds through the dev proxy bypass. Lays down the page chrome (sidebar, summary statistics card, refresh button) and the fetch lifecycle composable so 5f-ii onward can drop in the table columns and the modals without re-implementing it. - inbounds.html + src/inbounds.js: fourth Vite entry; mounts InboundsPage. - InboundsPage.vue: sidebar + summary card (totals over up/down, all-time, inbound count, client tags) + a basic table with enable/ remark/port/protocol/traffic/expiry columns. Row actions, popovers, search/filter, auto-refresh, and the WebSocket delta path are all deferred to subsequent 5f subphases. - useInbounds.js composable: GET /panel/api/inbounds/list + POST /panel/api/inbounds/onlines + POST /panel/api/inbounds/lastOnline + POST /panel/setting/defaultSettings, then computes the per-inbound clientCount roll-ups (active/deactive/depleted/expiring/ online/comments) the table popovers consume. - models/dbinbound.js + models/inbound.js: switched the legacy-utils import to the @/utils alias for consistency with the rest of the app. Semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-ii — inbound list table + search/filter + auto-refresh Fleshes out the inbound list with the full column set, search & filter toolbar, row enable toggle wired to /panel/api/inbounds/setEnable/:id, and a per-row action dropdown that emits events the parent will route to modals as those land in 5f-iii through 5f-vii. - InboundList.vue (new): toolbar (Add inbound + General actions dropdown + Refresh + auto-refresh popover), search-or-filter switch with the legacy radio buttons (Active/Disabled/Depleted/Depleting/ Online), and a a-table with desktop and mobile column variants. Cells use AD-Vue 4's #bodyCell slot — protocol/clients/traffic/ allTime/expiry/info cells render the same popovers and tags as legacy. Row enable switch is optimistic with rollback on POST failure. - visibleInbounds computed mirrors the legacy search and filter projection: deep search through dbInbound + clients, or filter reduces inbound.settings.clients to the selected bucket so the table only shows matching client rows. - Auto-refresh interval is read/written to localStorage with the same keys (`isRefreshEnabled`, `refreshInterval`) as the legacy panel. WebSocket delta updates are still deferred. - Action menu emits event payloads {key, dbInbound}; the parent currently shows a "coming in later 5f subphase" toast for each. Modals (edit/qr/clone/delete/reset/info/clients) land in 5f-iii through 5f-vii. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): wrap popover-table rows in <tbody> Vue's template compiler warned that <tr> can't be a direct child of <table> per the HTML spec; the browser silently inserts a <tbody> wrapper but Vue's SSR/hydration path doesn't, which can cause hydration mismatches. Add explicit <tbody> in both popover tables (traffic + mobile-info). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii — inbound add/edit modal + delete/clone/reset Wires up the inbound CRUD flows. The protocol-specific and transport- specific forms are still ahead in 5f-iii-b — for now the modal exposes those as JSON textareas so users can both edit existing inbounds without losing settings and create new ones from default templates. - InboundFormModal.vue: tabbed modal with a full Basics tab (enable, remark, protocol, listen, port, total GB, traffic reset, expiry date) and three JSON-edit tabs (Settings, Stream, Sniffing). Add mode stamps a fresh template per protocol via Inbound.Settings.getSettings(protocol); changing the protocol in add mode restamps the JSON. Edit mode pretty-prints the existing JSON so the user sees the same fields they save back. - POST /panel/api/inbounds/add or /panel/api/inbounds/update/:id on submit; on success the parent refreshes the list and the modal closes. Malformed JSON in any of the three textareas surfaces a message.error and aborts the save without losing user input. - InboundsPage.vue: wires the row action menu to real handlers — edit (opens the modal in edit mode), delete, reset-traffic, clone, reset-clients, del-depleted-clients all go through Modal.confirm and refresh on success. General actions menu wires reset-inbounds / reset-clients / del-depleted-clients the same way. Remaining actions (qrcode/info/import/export/copyClients) still toast as "coming soon" — those land in 5f-iv and 5f-v. - Adds dayjs ^1.11.20 dep for the a-date-picker v-model interop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iv — client add/edit + bulk-add modals Wires per-inbound client management. Both flows go through the same addClient/updateClient endpoints as legacy; the modals just funnel the form state into the right shape (`{id, settings: '{"clients": [...]}'}`). - ClientFormModal.vue: protocol-aware single-client editor — email/ password/id/auth/security/flow/subId/tgId/comment/ipLimit/totalGB/ expiry/renewal fields are shown/hidden per protocol like legacy. Edit mode displays the per-client traffic stats with a reset button; IP-limit log is read on click and clearable. Random helpers (sync icon next to each label) regenerate UUID/email/ password/sub-id values. - ClientBulkModal.vue: 1–500 clients in one POST, with the legacy five email-generation modes (Random / +Prefix / +Num / +Postfix / Pure-Prefix-Num-Postfix). Builds clients via the protocol-aware factory and concatenates their toString() output into a single settings.clients JSON array. - InboundsPage.vue: opens both modals from the row action menu (`addClient` / `addBulkClient`). They both refresh the inbound list on success. - Outstanding row actions still toast as "coming soon": qrcode, showInfo, copyClients, clipboard. Those land in 5f-v / 5f-vi. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-v — inbound info + QR-code modals Wires the row "info" and "qrcode" actions and ports the legacy inbound_info_modal end-to-end. The info modal handles every protocol the legacy panel did: • multi-user (VMess/VLess/Trojan/SS-multi/Hysteria) — per-client table + share links + per-link QR; • SS single-user — share link + QR; • WireGuard — full peer table with downloadable peer-N.conf and a wg:// share link per peer; • Mixed/HTTP/Tunnel — connection-detail tables. - QrPanel.vue: shared link card (header tag, copy button, optional download button, optional QR canvas, monospace footer with the raw value). Per-instance QRious instances are repainted on value/size change. - InboundInfoModal.vue: full info modal. Subscription URL block keys off subSettings.subURI/subJsonURI; IP-log lazy-loads on open and surfaces refresh + clear; tg-id, last-online, depleted/enabled tags all match legacy. - QrCodeModal.vue: lighter modal used for the row "qrcode" action on SS-single and WireGuard inbounds (just the QRs, no info table). - InboundsPage.vue: wires both flows. checkFallback() reproduces the legacy logic — when an inbound listens on a unix-socket fallback (`@<name>`), the link generator is pointed at the root inbound that owns the listen address so QRs/links carry the public host:port + the right TLS state. Multi-client navigation (focusing a specific client's links) is deferred to 5f-vi where the per-inbound expand- row table will pass the email through. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vi — per-inbound client expand-row table Each multi-user inbound row in the list now expands to show its client roster, mirroring the legacy aClientTable component. - ClientRowTable.vue: inner a-table with full desktop column set (action icons / enable / online / client-with-status-dot / traffic with progress bar / all-time / expiry with reset cycle) and a collapsed mobile variant (single dropdown menu + popover info). Self-contained: stats are looked up via a per-inbound email->stats Map; per-client confirms (reset/delete) live on the row. - The component emits typed events (edit/qrcode/info/reset-traffic/ delete/toggle-enable) — InboundsPage routes them back to the existing client and info modals (with `findClientIndex` so the modal opens focused on the right client). - InboundList.vue: hooks ClientRowTable into the a-table's expandedRowRender slot; row-class-name `hide-expand-icon` and a scoped CSS rule hide the chevron for non-multi-user inbounds (HTTP/Mixed/Tunnel/WireGuard/SS-single) so they keep looking flat. - toggle-enable-client routes through updateClient with the same `{id, settings: '{"clients": [...]}'}` shape as the other modals, so backend parsing stays single-pathed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii-b — replace inbound modal JSON textareas with structured forms Rewrites InboundFormModal to look like the legacy panel: structured forms for the common case, with a compact "Advanced (JSON)" fallback for the rare bits we don't yet have UI for. Tabs: • Basics — enable/remark/protocol/listen/port/total/trafficReset/expiry • Protocol — protocol-aware: VMess/VLess/Trojan/SS-multi/Hysteria in add mode embed an inline first-client form (email + ID/password/auth, security, flow, subId, comment, total GB, expiry); edit mode shows a clients-count summary table; VLess: decryption/encryption inputs; SS: method dropdown that re-randomizes password and propagates method change to the multi-user array (matches legacy SSMethodChange); HTTP/Mixed: accounts table with add/remove rows + Mixed auth/udp/ip toggles; Tunnel: address/port/network/followRedirect; WireGuard: secretKey/pubKey (regen via Wireguard.generateKeypair) + per-peer fields with PSK regen + allowedIPs add/remove + keepAlive. • Stream — only when canEnableStream(); transport selector with structured forms for TCP (proxy-protocol, http camouflage), WS (host/path/heartbeat/headers), gRPC (serviceName, multiMode), HTTPUpgrade (host/path). KCP/XHTTP fall back to the Advanced tab with an alert banner. Security selector with TLS (sni/alpn/ fingerprint) and Reality (target/serverNames/keypair-gen via /panel/api/server/getNewX25519Cert / shortIds / fingerprint). • Sniffing — enabled/destOverride/metadataOnly/routeOnly/ ipsExcluded/domainsExcluded as structured fields. • Advanced (JSON) — raw streamSettings + sniffing JSON for users reaching KCP/XHTTP/sockopt/finalmask/full TLS cert arrays. The stream JSON is auto-synced from the live model whenever the structured fields change. State source of truth is a deeply-reactive Inbound + DBInbound pair cloned on open; submit serializes via inbound.settings.toString() + inbound.stream.toString() so the wire shape matches the legacy panel byte-for-byte. streamNetworkChange semantics (clear flow when TLS/Reality unavailable, reset finalmask.udp when not KCP) are preserved. Vision Seed for VLess + finer-grained TCP HTTP camouflage + the full TLS cert/ECH editor will land in 5f-iii-c. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vii — shared text/prompt modals + remaining export/import wiring Wires up the last batch of inbound row + general actions that were toasting "coming soon": export-inbound-links, export-subs (per-inbound and global), export-all-links, import-inbound, and the clipboard JSON peek. Two small shared components back them — both can be reused by the xray page later. - TextModal.vue (shared): read-only multi-line viewer with a copy button and an optional download button when fileName is set. Replaces the legacy txtModal which the inbounds page used for every link export. - PromptModal.vue (shared): generic title + input/textarea + confirm callback, with the legacy keybindings (Enter submits in single-line mode; Ctrl+S submits in textarea mode). Used here for import-inbound but also by xray-config edits in Phase 6. - InboundsPage.vue: drops the toast stubs for `import`/`export`/`subs` on the general-actions menu and `export`/`subs`/`clipboard` on the per-row menu, routing each through openText / openPrompt + the appropriate model helper (genInboundLinks, etc.). The copyClients cross-inbound modal stays toast-stubbed — that's its own dedicated legacy modal worth its own commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-i — xray page scaffold + Advanced JSON tab The fifth and last legacy page comes online. Tabs are scaffolded with a-empty placeholders for the structured editors (Basics / Routing / Outbounds / Balancers / DNS) so navigation is stable; the Advanced (JSON) tab is fully functional and lets power users edit the raw xraySetting tree exactly like the legacy CodeMirror pane. - xray.html + src/xray.js: fifth Vite multi-page entry, mounted as XrayPage; vite.config.js routes /panel/xray and /panel/xray/ to it through the dev proxy bypass alongside the other pages. - XrayPage.vue: page chrome with the Save / Restart-xray bar, restart- output popover (surfaces /panel/xray/getXrayResult content when startup fails), 6 a-tabs, and a textarea-backed Advanced JSON editor. CodeMirror is intentionally not pulled in — the textarea works for every modern browser and keeps the bundle slim while structured editors land in 6-ii through 6-v. - useXraySetting.js composable: POST /panel/xray/ on mount, mirrors the settings-page 1s busy-loop dirty check for both xraySetting and outboundTestUrl, and exposes saveAll + restartXray. The dirty flag relies on string equality of the pretty-printed JSON, so reformat-only edits don't enable Save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-ii — xray Basics tab structured editor Replaces the placeholder on the Basics tab with a structured form for the most-touched fields of the xray template — outbound + routing strategy, log levels, traffic stat counters, and the "basic routing" shortcuts (block torrent / IPs / domains, direct IPs / domains, IPv4 forced, WARP / NordVPN routing). - useXraySetting.js: hoists a parsed `templateSettings` reactive alongside the JSON string, with two cooperating watches that keep them in sync. Editing structured fields stringifies into xraySetting for the dirty-poll + Advanced JSON tab; editing the JSON re-parses into templateSettings only when valid, so structured tabs stay readable mid-edit. - BasicsTab.vue: collapse panels mirror the legacy partial — General, Statistics, Logs, Basic routing. Every input is a computed v-model reading/writing into templateSettings; the routing-rule shortcuts funnel through ruleGetter/ruleSetter which match the legacy templateRuleGetter/templateRuleSetter behavior (replace-first, drop-duplicates, pop-the-rule-when-empty). Direct/IPv4 setters also call syncOutbound() to provision/prune the matching outbound. - XrayPage.vue: imports BasicsTab + derives `warpExist`/`nordExist` from the parsed templateSettings. WARP/NordVPN provisioning modals are still placeholders that toast — those land in 6-v with the routing/outbound editors. Default tab flips back to Basics so users land on the structured editor. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iii — xray Routing tab + rule modal Replaces the Routing tab placeholder with a full editor for templateSettings.routing.rules: - RoutingTab.vue: a-table over the parsed rules with the legacy six- column layout (action / source / network / destination / inbound / outbound) and the same "lead value + N more" pill renderer for multi-value criteria. Mobile drops source/network/destination for readability. Per-row dropdown handles edit / move-up / move-down / delete; the array-mutation reordering replaces the legacy jQuery Sortable drag handle without pulling in a sortable lib. - RuleFormModal.vue: full form mirroring xray_rule_modal.html — CSV inputs for sourceIP/sourcePort/vlessRoute/ip/domain/user/port, Network select, Protocol multi-select, Attrs key/value pairs, inbound-tag multi-select sourced from templateSettings.inbounds + parent inboundTags + dnsTag, outbound-tag single-select sourced from templateSettings.outbounds + clientReverseTags, and balancerTag from templateSettings.routing.balancers. Submit serializes via the same shape the legacy `getResult` produces (CSV → array, drop empty fields). - XrayPage.vue: imports RoutingTab and exposes inboundTags + clientReverseTags from useXraySetting so the modal can populate its tag pools. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iv — xray Outbounds tab + outbound modal Replaces the Outbounds tab placeholder with a full table + add/edit flow. The 1.3k-line legacy outbound modal is condensed to a tabbed modal with structured Basics fields (tag/protocol/sendThrough/domain strategy) and JSON tabs for the protocol-specific settings + stream trees — same approach the Inbound modal uses, and a power user can still edit the same trees via the page-level Advanced (JSON) tab. - useXraySetting.js: adds fetchOutboundsTraffic + resetOutboundsTraffic + testOutbound. Test states are tracked per outbound index so the row's Test button can show loading + the Test-result column can render the response delay / status / error. - OutboundsTab.vue: full table (action / identity / address / traffic / test result / test) plus a card-list mobile variant with the same row dropdown (set-first / edit / move up/down / reset traffic / delete). outboundAddresses() reproduces the legacy findOutboundAddress logic so each protocol's host:port list is rendered consistently. Add/edit go through OutboundFormModal, delete goes through Modal.confirm, reset traffic posts to /panel/xray/resetOutboundsTraffic with the row's tag (or "-alltags-" from the toolbar). - OutboundFormModal.vue: tag/protocol/sendThrough/domainStrategy on the Basics tab; settings + streamSettings as raw JSON on their respective tabs. Tag-collision check happens client-side before emitting; malformed JSON aborts the save with a message.error. - XrayPage.vue: imports OutboundsTab and wires the test action to the composable's testOutbound helper. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-v — xray Balancers tab + DNS placeholder Brings Balancers to full parity with the legacy panel and adds a DNS tab placeholder that exposes the full dns/fakedns trees as JSON so users can edit them without falling through to Advanced. - BalancerFormModal.vue: tag (with duplicate-tag warning across other balancers), strategy (random/roundRobin/leastLoad/leastPing), selector tag-mode multi-select sourced from existing outbound tags + free-form additions, fallback. Disable-on-invalid is driven by the duplicateTag + emptySelector computed flags. - BalancersTab.vue: empty state with a single "Add balancer" CTA; populated state shows the legacy 4-column table (action / tag / strategy / selector / fallback) with per-row edit + delete in a dropdown. On submit the wire shape preserves the `strategy: { type }` nesting only when the strategy is non-default, matching the legacy emit. Tag renames also chase across routing.rules.balancerTag references so existing rules don't dangle. - DnsTab.vue: master enable switch + raw JSON for `dns` and `fakedns`. Legacy had a dedicated server-by-server editor + a fakedns row editor; both are big enough to deserve their own commits, and the JSON path supports every field today. WARP / NordVPN provisioning modals still toast as "coming soon" — those are third-party API integrations worth their own commits. The xray page now has structured editors for Basics / Routing / Outbounds / Balancers and JSON editors for DNS / Advanced — every xray tab the legacy panel offered is functional. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(server): Phase 8 — cut HTML routes over to web/dist/ Production cutover. Every user-facing HTML route now serves the Vue-3-built bundle from web/dist/ instead of rendering the legacy Go template; the long-hashed Vite assets are served at /assets/ from the same embedded filesystem. The legacy templates in web/html/ and the legacy static tree in web/assets/ are kept on disk for now in case a quick revert is needed, but nothing the binary serves references them. What changed: - web.go: a new //go:embed dist/* feeds the controller package via a SetDistFS hand-off before controller construction. The static /assets/ route is rebound: in dev to web/dist/assets/ on disk so Vite's incremental rebuilds show up live; in prod to the embedded dist via wrapDistFS (rooted one level deeper than wrapAssetsFS). - controller/dist.go: serveDistPage helper used by every HTML handler. Reads dist/<name> from the embedded FS and applies two transforms before sending: 1. injects <script>window.__X_UI_BASE_PATH__="..."</script> just before </head> so AppSidebar links resolve under the panel's basePath. 2. when basePath != "/", rewrites Vite's absolute /assets/ URLs to <basePath>assets/ so installs running under a custom URL prefix load the bundle where the static handler lives. HTML responses go out with no-cache so panel upgrades reach users on the next refresh; hashed JS/CSS stays cacheable. - controller/index.go: IndexController.index now serves dist/login.html for logged-out callers (the redirect for logged-in users is unchanged). - controller/xui.go: XUIController.{index,inbounds,settings,xraySettings} each become a one-line wrapper around serveDistPage. Smoke checklist for the maintainer: - run `cd frontend && npm run build` to refresh web/dist/ before building the Go binary (the embed snapshot is taken at compile time); - visit /panel/, /panel/inbounds, /panel/settings, /panel/xray and confirm each loads its Vue page; - log out and log back in to verify the login flow; - confirm the sidebar links navigate correctly under your install's basePath; - POST flows (e.g. saving settings) still need the CSRF token — that endpoint (/panel/csrf-token, added earlier) is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-vi — WARP + NordVPN provisioning modals Replaces the toast stubs on the Basics tab and Outbounds toolbar with the legacy WARP + NordVPN provisioning flows. Both modals now stage their wireguard outbounds back into templateSettings.outbounds through the same event channels OutboundsTab uses, so the existing add / reset / delete / refresh-traffic surface keeps working. - WarpModal.vue: empty state shows a single Create button that generates a wireguard keypair locally (Wireguard.generateKeypair) and posts it to /panel/xray/warp/reg; populated state surfaces the access_token / device_id / license_key / private_key, lets the user upgrade to WARP+ via /panel/xray/warp/license, refreshes the account info from /panel/xray/warp/config (plan / quota / usage in human-readable bytes), and stages a wireguard outbound with the WARP-specific reserved-byte encoding pulled from client_id. Add / Reset / Delete go through events the parent routes back to templateSettings.outbounds. - NordModal.vue: dual-tab login (NordVPN access token → /panel/xray/nord/reg, or paste a NordLynx private key → /panel/xray/nord/setKey). Once authenticated, country / city / server selectors fetch from /panel/xray/nord/{countries,servers}, servers sort by load ascending, the lowest-load server in the current city auto-selects. Reset emits oldTag/newTag so the parent renames matching routing rules in place; logout emits a remove-routing-rules event with prefix `nord-` to purge any dangling references. - XrayPage.vue: holds warpOpen / nordOpen flags, ensures the outbounds array exists before mutating it, and wires the modal events (add-outbound / reset-outbound / remove-outbound / remove-routing-rules) to in-place edits of templateSettings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 7 — vue-i18n wired up + login page translated Sets up vue-i18n on top of the panel's existing TOML translation files. The Go side stays the source of truth — translators continue to edit web/translation/*.toml; a sync script snapshots those files into per-locale JSON the Vue bundle imports. The login page is translated end-to-end as a worked example; remaining pages can be converted incrementally without infrastructure churn. What's in the box: - scripts/sync-locales.mjs: small TOML→JSON converter that walks web/translation/*.toml and writes frontend/src/locales/<code>.json. Handles the narrow subset of TOML the panel uses (flat key/value pairs + dotted [section.subsection] heads). Wired as a `prebuild` + `predev` script so production builds always include the latest strings without a manual step. - src/i18n/index.js: createI18n() in composition mode with all 13 locales emitted as their own Vite chunks. The active locale (read from the same `lang` cookie LanguageManager has always managed) plus the en-US fallback are eagerly loaded; the rest are dynamically importable via a loadLocale(code) helper. This keeps the per-page bundle the user actually downloads small — only ~30 KB of strings end up in the initial payload, vs ~220 KB if all 13 were eager. - All five page entries (index/login/settings/inbounds/xray) wire the i18n plugin into createApp via .use(i18n). - LoginPage.vue: t(...) replaces hardcoded English on the username / password / 2FA placeholders, the submit button label, and the Settings popover title. The Hello/Welcome headline cycle stays hardcoded — those are stylistic, not labels. The 'Hello'/'Welcome' cycle stays in English deliberately; the rest of the migration's components still ship hardcoded English and will be converted page by page in follow-up commits. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate page chrome — sidebar, save bars, tabs, summary cards Replaces hardcoded English with t() calls in the components every user sees on every page load. The translations themselves come from the existing TOML files via the sync script — no new strings, no new locale keys. Per component: - AppSidebar.vue: 5 menu titles (dashboard / inbounds / settings / xray / logout). Computed so the sidebar re-renders when the cookie-driven locale flips on reload. - IndexPage.vue: Quick actions card title + Logs / Backup / Up-to- date / Update buttons. - StatusCard.vue: CPU / Memory / Swap / Storage labels + logical-processors / frequency tooltips. - XrayStatusCard.vue: card title + error popover header + Stop / Restart / Switch xray action labels (kept the v-prefix version string as-is — it's content, not a label). - SettingsPage.vue: 5 tab titles + Save / Restart-panel buttons + unsaved-changes warning. - XrayPage.vue: 6 tab titles + Save / Restart-xray buttons + unsaved-changes warning. - InboundsPage.vue: 5 summary-stat card titles. - InboundList.vue: 10 column titles (computed for live locale), Add inbound / General actions buttons + every dropdown menu item, search placeholder, filter radio labels, popover titles (disabled / depleted / depleting / online), traffic + info popover row labels. Total: ~75 strings localised across 8 files. The remaining English labels live in the per-tab settings forms, the form modals (Inbound / Client / Outbound / Rule / Balancer / WARP / Nord), and the per-row table cell helpers — all incremental work that doesn't touch infrastructure. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate every remaining English string on the index page Closes the index page's i18n coverage. Combined with the page-chrome commit, every label users see on the dashboard is now sourced from the TOML translation files. Per file: - IndexPage.vue: loading-spinner tip (initial + dynamic). - BackupModal.vue: modal title, both list-item titles + descriptions ("Back up" / "Restore"), in-flight busy tips ("Importing database…" / "Restarting panel…"). - PanelUpdateModal.vue: modal title, update-available alert, current/latest version row labels, "Up to date" tag + label, primary action button. Modal.confirm now uses the translated panelUpdateDialog / panelUpdateDialogDesc with #version# substitution; success toast uses panelUpdateStartedPopover. - LogModal.vue: title slot ("Logs"). The Debug/Info/Notice/Warning/ Error log-level options stay literal — they're xray's wire values, not user-facing labels (matches the existing settings-page choice). - XrayLogModal.vue: title + Filter label. Direct/Blocked/Proxy stay literal for the same reason. - VersionModal.vue: modal title + xray-switch alert + per-file tooltip + "Update all" button + custom-geo collapse header. The Modal.confirm flows for switchXrayVersion + updateGeofile use translated dialog/desc with #version# / #filename# substitution. - CpuHistoryModal.vue: title slot. - CustomGeoSection.vue: routing-hint alert, Add / Update-all buttons, every column title (computed for live locale), copy/edit/download/ delete tooltips, copy toast, delete-confirm modal, empty-state text. - CustomGeoFormModal.vue: add/edit titles, OK/cancel labels, Type/ Alias/URL field labels, alias placeholder, all three validation toasts. Total: ~50 strings localised across 8 index-page files. The Hello / Welcome login headline cycle and a handful of literal xray wire values (Direct/Blocked/Proxy/log levels) are intentionally kept hardcoded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): Phase 7-c — translate settings, inbounds modals, xray tabs Continues the page-by-page translation pass started in cb37dd55 — runs every user-visible string on settings (General/Security/Telegram/Sub), inbounds (Client/QR/Info modals), and xray (Routing/Balancer/Rule/Warp/ Nord/Basics/Outbounds tabs) through useI18n. Updates the TOML→JSON sync script to escape `@` (vue-i18n parses it as a linked-format prefix) and refreshes all 13 locale files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): Phase 9 — restore index dashboard, fix login/CSRF, port legacy styles - Index dashboard regains the 8 cards that were lost in the SPA port (3X-UI panel info, Operation Hours, System Load, Usage, Overall Speed, Total Data, IP Addresses, Connection Stats), plus a Config button that shows the live xray config.json. Version display falls back through panelUpdateInfo → window.__X_UI_CUR_VER__ → '?' so dev mode isn't blank. - Xray config no longer hangs on load: useXraySetting surfaces failures instead of leaving a perpetual spinner, and the Vite dev proxy stops hijacking POST requests to migrated routes (only GETs get bypassed). - Inbound page no longer throws __asyncLoader/emitsOptions errors — inbound.js was missing imports (NumberFormatter, SizeFormatter, Wireguard) and InboundList kept emitting after unmount. - Login round-trip works after logout: a public /csrf-token endpoint bootstraps the SPA before authentication, axios caches the token module-level, and the dev 401 handler navigates to /login.html instead of reloading the dashboard into a redirect loop. - legacy.css mirrors the legacy panel's surface/text variables so dark and ultra-dark themes match main; every SPA entry imports it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray DNS section to match main branch DnsTab now exposes every field the legacy panel did — top-level toggles (tag, hosts, queryStrategy, disableCache/queryConcurrency, fallback strategy, client subnet), the servers table with per-row strategy and domain/expectIP/unexpectedIP overrides, and the Fake DNS pool. The new DnsServerModal covers the full add/edit flow and collapses to a bare string when the user only sets an address — matching the wire shape the legacy form emits for plain DNS entries like "8.8.8.8". Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray outbound modal with structured per-protocol forms Replaces the JSON textareas with the same shape the legacy panel uses: all 11 outbound protocols (vmess/vless/trojan/shadowsocks/socks/http/ mixed/wireguard/tun/dns/loopback/blackhole/freedom) get dedicated fields, every transport (TCP/KCP/WS/gRPC/HTTPUpgrade/XHTTP) gets its own panel, and TLS/Reality/sockopt/Mux are configured through the same controls as the inbound side. Brings the SPA outbound editor to parity with main so users no longer have to drop into raw JSON. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): bring inbound modal to full parity with main branch Switches the default protocol on add to VLESS, fixes a crash when adding a Mixed account (the constructor is SocksAccount, not MixedAccount), and fills in the fields the SPA was previously delegating to the Advanced JSON tab: - TLS: cipher suites, min/max version, reject SNI / disable system root / session resumption switches, the certificate array with per-row Path-or-Content toggle (Set Default pulls from /panel/setting/ defaultSettings), One Time Loading, Usage / Build Chain, plus ECH key/config with a Get New ECH Cert button. - Reality: xver, target/SNI sync icons (uses getRandomRealityTarget), max time diff, min/max client version, short IDs randomizer, SpiderX, mldsa65 seed/verify with Get New Seed. - Stream: full structured forms for every transport — TCP HTTP camouflage gets its request/response editor, mKCP gets MTU/TTI/uplink/ downlink/CWND/maxSendingWindow, WebSocket / gRPC (now with Authority) / HTTPUpgrade get headers + proxy-protocol toggles, XHTTP gets the full SplitHTTPConfig surface (mode-aware fields, padding obfs, session/sequence placement, uplink data, no-SSE). - New External Proxy section and a structured Sockopt block (mark, TCP keepalive/timeout/clamp, fast open, MPTCP, penetrate, V6Only, domain strategy, congestion, TProxy, dialer/interface, trusted XFF). - VLESS gets the legacy X25519 / ML-KEM-768 buttons that fetch fresh decryption/encryption blocks via /panel/api/server/getNewVlessEnc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): add FinalMask UI (TCP/UDP masks + QUIC params) to inbound and outbound Mirrors web/html/form/stream/stream_finalmask.html as a shared FinalMaskForm component used by both modals — they share the same StreamSettings shape (addTcpMask/addUdpMask/finalmask/enableQuicParams) so a single template handles both. Surfaces: - TCP masks for raw/tcp/httpupgrade/ws/grpc/xhttp networks: fragment, sudoku, and header-custom (with the 2D clients/servers groups, each row supporting array/str/hex/base64 packets and a randomize button for base64). - UDP masks for hysteria protocol or kcp network: hysteria gets just salamander; kcp gets the full type list (mkcp variants, header-*, xdns/xicmp, header-custom with flat client/server lists, and noise). Switching to xdns shrinks the kcp MTU to 900 to match the legacy panel's behavior. - QUIC Params for hysteria or xhttp: congestion (incl. brutal up/down fields), debug, UDP hop ports/interval, idle/keepalive timeouts, path-MTU discovery toggle, and the four receive-window tunables. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): remove duplicate Outbound test URL from xray Advanced tab The Basics tab already exposes this field through BasicsTab — duplicating it on the Advanced tab let two inputs race the same ref and only added clutter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): unify theming on vanilla AD-Vue light/dark/ultra-dark The legacy panel CSS (custom.min.css ported as legacy.css) tinted every non-primary button teal-green via .dark .ant-btn:not(.ant-btn-primary) overrides while AD-Vue 4's darkAlgorithm kept primary buttons blue — producing the mixed blue/green button look on dark mode. Drop legacy.css entirely and let AD-Vue 4's algorithms own the palette. Centralize antdThemeConfig in useTheme.js so every page resolves to the same source of truth (light = defaultAlgorithm, dark = darkAlgorithm, ultra-dark = darkAlgorithm + deeper colorBgBase/Layout/Container/ Elevated tokens). Each page's <a-config-provider> now imports the shared computed instead of defining its own copy. Drops the 67 KB legacy CSS chunk; per-page CSS bundles fall to ≤5.9 KB. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): restore computed import in Settings + Xray pages When 5f1aba28 dropped the local antdThemeConfig computed (now shared from useTheme), it also stripped `computed` from the import list — but both pages still call computed() elsewhere (confAlerts, advanced-tab helpers). Re-adds it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): retheme dashboard gauges to AD-Vue blue and shrink them - StatusCard's CPU/RAM/Swap/Storage dashboards rendered at AD-Vue's default 120px width which made the percent text balloon to ~36px. Drop to 90px (70px on mobile) so the gauge fits the rest of the card. - The CurTotal.color thresholds still hardcoded the legacy teal/orange palette (#008771 / #f37b24 / #cf3c3c). Switch to AD-Vue's primary / warning / danger tokens (#1677ff / #faad14 / #ff4d4f) so the gauges match the rest of the panel under both light and dark themes. - XrayStatusCard's running-animation badge ring also still pointed at the deleted --color-primary-100 var; hardcode the new primary blue. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: shorten backupTitle to "Backup & Restore" across all 13 locales The backup modal header was the second-longest title in the dashboard on every locale ("Database Backup & Restore" / "Резервне копіювання та відновлення бази даних" / etc). Drop the "Database / Veritabanı / 数据库" qualifier — the modal already lives under the "Database" column, so the shorter form reads cleaner on narrow viewports. Updated both the .toml source-of-truth files and the synced .json locales (re-running scripts/sync-locales.mjs). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: collapse two translation databases into a single web/translation/<lang>.json set The Vue SPA had been reading from frontend/src/locales/*.json while the Go binary still loaded web/translation/translate.*.toml — and a sync-locales.mjs pre-build step kept the two in lockstep, with TOML as the source of truth. Now that go-i18n v2.6.1 already flattens nested JSON via recGetMessages/addChildMessages, both runtimes can share one file per locale. - Move the 13 nested-JSON locale files to web/translation/<lang>.json so they live alongside the Go //go:embed translation/* directive. - Switch web/locale/locale.go from toml.Unmarshal to json.Unmarshal (and drop the pelletier/go-toml import — it's now indirect-only). Confirmed via a smoke test that pages.index.cpu, subscription.title, tgbot.commands.help, and menu.settings all resolve in en-US, fa-IR, ru-RU, and zh-CN. - Repoint Vue's i18n loader at the new path (../../../web/translation/ *.json glob) and drop the moved-here pathDelimiter comment that no longer applies. - Delete the 13 legacy translate.*.toml files and the sync-locales.mjs script + its npm pre-script hooks (predev/prebuild/i18n:sync). The Telegram bot and subscription page still get their messages because they were reading the same MessageIDs the JSON files now produce. - Update copilot-instructions.md so the next contributor knows where the canonical translation files live. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): redesign expand-row + retheme client visuals When you expanded an inbound row, the nested <a-table> inside ClientRowTable burst out of the parent's scroll-x box — its .ant-spin-container ended up wider than the parent's narrow .ant-table-cell, so the child looked oversized while the parent looked squeezed. Replace the nested table with a CSS-grid layout that owns its sizing, sits flush inside the expanded cell, and collapses to a 3-column layout on mobile (action menu, client identity, info popover). While in there, fix three other client-row visuals: - The Unicode infinity glyph (U+221E) renders as an "m"-shaped character in some system fonts (Windows Segoe UI in particular). Add a shared <InfinityIcon /> SVG component (legacy panel's path) and use it in ClientRowTable, InboundList, and InboundInfoModal — desktop and mobile cells. - The "unlimited quota" traffic bar passed :percent="100" with no stroke-color, so AD-Vue auto-coloured it success-green. Pin it to the AD-Vue purple token (#722ed1) so it reads as the no-limit sentinel rather than another usage state. - ColorUtils + the in-row statsExpColor still hardcoded the legacy teal/orange/red/purple palette (#008771 / #f37b24 / #cf3c3c / #7a316f). Map them onto AD-Vue 4's success/warning/danger/purple tokens (#52c41a / #faad14 / #ff4d4f / #722ed1) so badges, tags, and progress bars all match the rest of the panel. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): darken light-theme page bg so cards stand out The light-theme --bg-page was #f0f2f5 — close enough to AD-Vue's #fff card background that the cards faded into the page. Bump it to #e6e8ec (a more visibly distinct gray) so cards lift cleanly off the surface. Dark and ultra-dark stay where they were. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): shrink dashboard percent text and surface the unfinished arc Two follow-up tweaks to the dashboard gauges: - AD-Vue scales the percent text from the SVG, not from :width, so the 90px gauges still rendered the number at ~27px. Pin .ant-progress-text to 14px via :deep() and trim the gauge to 70px (60px on mobile) so the whole card stays compact. - The default trail (rgba(0,0,0,0.06) / rgba(255,255,255,0.08)) was invisible on the light-theme card. Pass an explicit rgba(128,128,128,0.25) trail-color so the unfinished portion is visible under both light and dark themes. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): migrate subpage.html to Vue 3 SPA The subscription info page was the last page still rendered by Go templates. Move it to the Vite multi-page setup so the whole panel loads through one toolchain. Frontend: SubPage.vue mounts at /sub/<id>?html=1 and reads window.__SUB_PAGE_DATA__ for the parsed view-model (traffic / quota / expiry + rendered share links). Fix descriptions borders against the light-theme card by painting the row divider on each cell's bottom edge — AD-Vue's <tr> border doesn't render reliably under border-collapse:collapse. Backend: serveSubPage reads dist/subpage.html, injects window.__X_UI_BASE_PATH__ + window.__SUB_PAGE_DATA__ before </head>, and rewrites Vite's absolute /assets/ URLs when the panel runs under a URL prefix. Drop the legacy template-FuncMap wiring and switch the sub server's static mount from web/assets to web/dist/assets. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): inbound modal QR + tabs + restored TLS fallbacks Per-client QR action: the qr icon on the expand-row table opened the big info modal instead of the QR modal. Route it to QrCodeModal and extend that modal with a `client` prop so genAllLinks() produces the per-client share URLs (and per-peer remarks for WireGuard). Inbound's Data redesign: split the dense single-page view into three tabs — Inbound, Client, Subscription. Drop every QR rendering from this modal (QrCodeModal is the QR home now). Each row in the Inbound tab is one label/value pair instead of the legacy 2-column grid, and long values like the VLESS encryption blob render as a wrapping code block with a copy button so they can't blow out the dialog. The Subscription tab renders sub URL + JSON URL as clickable anchors that open in a new tab. Restored TLS fallbacks UI: the model already exposed VLESSSettings.Fallback / TrojanSettings.Fallback with addFallback / delFallback / fallbackToJson, but the form modal never surfaced them during the Vue 3 migration. Re-add the legacy form (SNI, ALPN, Path, Destination, PROXY) on the protocol tab, gated on TCP transport plus (for VLESS) encryption=none — same conditions as main. Column widths: Protocol 70→130 and All-time Traffic 60→95 in the inbound list; All-time Traffic 90→130 in the client expand-row, so the header text fits and tags don't get squeezed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): navy dark theme + rounded inbound/client corners Dark theme picks up a refined navy palette (page #0a1426, cards #142340, sider #0d1d33) so the sidebar blends with the rest of the surface; ultra-dark stays neutral black. Resolves the previous mismatch where AD-Vue 4 hardcoded #001529 / #002140 for the sider, trigger and dark Menu items via Layout.colorBgHeader / colorBgTrigger and Menu's colorItemBg — overrides go through the component-token map now. Round the inbound table's outer corners (header start/end + last row end) and wrap the client expand-row grid in a 1px / 8px-radius border so the list reads as a contained block instead of a flush rectangle. Linter-driven whitespace cleanup across inbounds/*.vue rolled into the same commit since it can't be split out cleanly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): xray tab fixes — modal close, tag validation, full XHTTP, reset to default Modal close: BalancersTab / OutboundsTab / RoutingTab confirmDelete used arrow expressions that returned splice's removed-items array. AD-Vue 4 treats truthy non-thenables from onOk as "still pending" and never closes the dialog (see ActionButton.js:103-106), so the confirm modal stayed open. Wrap the body so onOk returns undefined and AD-Vue auto-closes. Tag validation: outbound + balancer modals only flipped between warning/success on duplicate, leaving the empty case as a green ✓. Split into a 3-state computed — error (empty) / warning (duplicate) / success — and wire a help message so the input clearly explains why the OK button is disabled. Reset to default: re-add the legacy "Reset to Default" panel at the bottom of BasicsTab. Calls /panel/setting/getDefaultJsonConfig and overwrites templateSettings; the existing watch re-stringifies so the JSON tab + dirty-poll see the new state. Restored Basics option lists from main: IPs (4→10, +Vietnam/Spain/ Indonesia/Ukraine/Türkiye/Brazil), DomainsOptions (4→10, +regex entries), BlockDomainsOptions (5→17, +Malware/Phishing/Adult/regex), ServicesOptions (Reddit/Speedtest in, off-template Microsoft out). Outbound form parity with main: • Reverse Sniffing UI for VLESS — toggle + destOverride checkboxes (HTTP/TLS/QUIC/FAKEDNS) + Metadata/Route Only + IPs/Domains excluded multi-selects, gated on reverseTag being set. • Full XHTTP transport — request headers list, Max Upload Size / Min Upload Interval (packet-up), Padding Obfs Mode + sub-fields, Uplink HTTP Method, Session/Sequence/UplinkData placement + keys, No gRPC Header (stream-up/stream-one), expanded XMUX with Max Concurrency/Connections/Reuse/Request/Reusable/Keep-alive. Strip a-divider from the outbound form per request — replaced with plain section/item heading divs so the labels and per-row delete icons stay but the horizontal rule is gone. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): xray Advanced tab parity + finalmask gating Advanced tab was a single textarea bound to the full xraySetting blob. Restore the legacy 4-way view: a radio group toggles between All / Inbounds / Outbounds / Routing Rules, and the textarea reads/writes the matching slice through templateSettings. Added the legacy header ("Advanced Xray Configuration Template" + description) so the page introduces itself like main. Outbound finalmask leaked into protocols that don't have a stream (Freedom / Blackhole / DNS / Socks / HTTP / Wireguard) because the v-if only checked outbound.stream. Gate the whole FinalMaskForm on outbound.canEnableStream() to match main. Drop the leading divider inside FinalMaskForm — its parent already provides separation, so the rule above "TCP Masks" was redundant. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound Advanced tab live mirror + QR exact-fit sizing Advanced tab in the inbound modal showed stale state. The watch only refreshed advancedJson.stream, so toggling the Sniffing switch in the Sniffing tab left the Advanced JSON showing the prior value. And encryption — stored on inbound.settings.encryption, not on stream — never appeared at all because Advanced only exposed stream + sniffing. Split the watch into three (stream / sniffing / settings) and add a settings textarea so encryption / clients / fallbacks live alongside the existing two views. The submit() path now reads settings from the JSON tab too (falling back to inbound.settings.toString()) so power-user edits in Advanced override the structured form on save. QR canvas: when a longer share-URL bumps the QR matrix size, QRious falls back to floor(canvasSize / matrixWidth) and centers the pattern, leaving a white margin (e.g. matrix=41, size=180 → 8px gap). Pre-pick the QR version from the URL byte length and set canvas size to a multiple of matrixWidth × pixelSize so the pattern always fills it edge-to-edge — no white margin even after toggling encryption on. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound stream tidy-up + QR sizing + dev proxy Stream tab clean-up: drop the seven a-divider rules in the inbound form's Stream tab — replace the labelled ones (Request / Response / Security) with a section-heading div that matches the outbound modal, delete the empty rules above TLS sub-blocks / External Proxy / Sockopt. Empty header-list form-items also leaked margin space below each "Add header" button across TCP / WS / HTTPUpgrade / XHTTP — gate each on headers.length > 0 so they vanish until the user adds one. QR panel: drop the link text under the canvas (the user already has a copy button on the header). Pin the canvas display size to a fixed 240px square via :style + image-rendering: pixelated/crisp-edges so a dense WireGuard config QR and its sparser link share the same on-screen footprint without blurring. Dev proxy: Node's AggregateError wraps connection failures whenever DNS returns more than one address (::1 + 127.0.0.1) and the code lands on the inner errors, not the outer. The existing handler only checked err.code so the ECONNREFUSED stack still spammed the log when the Go backend was down. Walk err.errors too, print one friendly line ("backend not reachable — start the Go server"), then stay quiet for the rest of the session. Vendor splitting + chunk-size warning: split node_modules into stable vendor-* chunks so each page only ships the deps it uses and the browser caches them across versions. ant-design-vue stays as a single chunk because its components share internals; raise the chunk-size warning to 1500kB so the build stays quiet (its 1.4MB minified gzips to ~410kB on the wire). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): info-modal cleanup + 2FA QR + outbound link import - 2FA QR: matrix-snap canvas + opaque background to drop white margin - Inbound info modal: stack Mixed/HTTP/Tunnel as info-rows, hide tab strip when only the Inbound tab applies - Add inline VLESS Reverse tag input on first-client form - Hide Protocol tab for TUN (no form yet) - Outbound link converter: route through Outbound.fromLink so vless/trojan/ss/hysteria(2) imports work alongside vmess; fix stray implicit global in fromLink Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): jalali calendar + drop legacy moment-jalali - Wire Calendar Type setting to a real Jalali datepicker via vue3-persian-datetime-picker, gated by useDatepicker composable - DateTimePicker wrapper swaps between AD-Vue and Persian picker; keeps dayjs v-model contract so existing forms/setters work unchanged - Theme picker popup explicitly per body.dark / data-theme=ultra-dark (AD-Vue 4 doesn't expose CSS vars, so var() fallbacks defaulted to white); fix invisible disabled days, SVG arrow fills, popup clipping via append-to="body" - Replace stray moment() calls in dbinbound/inbound models with dayjs; the legacy global was undefined under ESM and broke the inbounds list whenever any inbound had expiryTime > 0 - Remove legacy moment-jalali / persian-datepicker / aPersianDatepicker assets — replaced by the Vue 3 picker Note: dark/ultra background of the date popup still renders white in some cases — pending follow-up. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): jalali popup theming + full-month layout - Re-prefix popup selectors with .vpd-wrapper (popup root that travels with appendTo='body'), not .vpd-main (which stays at the input); paints the popup's dark/ultra background again - Drop the 1px border on .vpd-content — with box-sizing: border-box it ate 2px from the day-row width, wrapping the 7th cell of every row and hiding days 18-31 of months that needed a 5th week Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat: render dates in Jalali when Calendar Type is jalalian - IntlUtil.formatDate accepts an optional calendar arg; appends the BCP-47 -u-ca-persian extension so Intl renders Jalali across all UI languages, not just fa-IR - Plumb the panel's datepicker setting into the SubPage via the Go injection (window.__SUB_PAGE_DATA__.datepicker) - Panel pages (inbound list/info, client row, xray log) read the same setting through the useDatepicker composable so the whole panel stays consistent Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): ultra-dark page tint + mobile-friendly inbound view - Drop --bg-page from #21242a (lighter than the cards) to #050505 in ultra-dark across index/sub/settings/inbounds/xray, so cards consistently elevate over the page - Hide the inline sider's children + collapse-trigger and zero its width below 768px; the floating drawer-handle remains the menu trigger - Inbounds page mobile pass: tighten content-area + card padding; flex-wrap the filter bar instead of stacking; shrink table cell padding so all 4 mobile columns fit; bump expand / action / info icon hit targets - Per-client expand row on mobile: soft-tinted rounded cards instead of hairline borders, larger action / info touch targets, more legible email typography, bigger status badge dot Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove legacy template + asset trees and dead Go template engine - Delete web/html/ entirely (page templates, form/, modals/, component/, common/, settings/) — every route is served from web/dist/ now via serveDistPage; nothing in the binary referenced these - Delete web/assets/ entirely (jQuery-era ant-design-vue, axios, moment, codemirror, qrcode/qs/uri/vue/otpauth, custom CSS, Vazirmatn font); Vite bundles all of this into web/dist/assets - Drop the Gin HTML template wiring: remove //go:embed assets + //go:embed html/*, the assetsFS/htmlFS vars, the wrapAssetsFS adapter, EmbeddedHTML / EmbeddedAssets exports, getHtmlFiles / getHtmlTemplate, the i18nWebFunc/funcMap and SetFuncMap call, and the dev/prod template-engine branch — only StaticFS for /assets/ is needed now - Remove dead html()/getContext() helpers and unused imports from web/controller/util.go (no c.HTML(...) callers remain) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound expand chevron position + cpu history layout - Push the inbound table's expand chevron away from the left edge with margin-inline + cell padding so it isn't flush against the corner - Move "Timeframe: …" caption above the chart (was below); restore the line that the previous edit removed - Fix x-axis time labels being clipped at the bottom of the cpu chart — the offset (paddingTop+drawHeight+22 = 222) exceeded the SVG viewBox height (220); dropped to +14 so labels sit at y=214 with room for descenders - Move the SVG axis text colors out of <style scoped> into a global block — Vue's scoped CSS doesn't always hash-attribute SVG <text> descendants, so the dark-mode overrides via :global() weren't matching; bumped opacity 0.55 → 0.85 for legibility on navy/black Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(login): language picker in settings popover + fluid card sizing - Add language select alongside the theme switch (mirrors SubPage) - Bind headline to pages.login.hello / pages.login.title so the "Hello / Welcome" cycle re-translates with the active locale - Replace AD-Vue 5-breakpoint grid with clamp() sizing so the card scales smoothly instead of jumping ~33% at each breakpoint - Pin horizontal padding so input width stays stable on large viewports Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): organize entry HTML + bootstrap JS into folders - Move entry HTML files: frontend/*.html -> frontend/html/*.html - Move per-page bootstrap modules: src/{index,login,settings,inbounds,xray,subpage}.js -> src/entries/ - Update vite.config rollup inputs and dev-mode MIGRATED_ROUTES to /html/<page>.html - Build output now lands at web/dist/html/<page>.html - serveDistPage and subController updated to read from dist/html/ Cleans up the flat frontend/ root which previously interleaved 6 HTML files with package.json, README, src/, etc. The src/ root similarly gets rid of 6 entry .js files mixed in alongside api/, components/, models/, etc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove obsolete vue3 phase1 inventory doc The migration is well past phase 1 — the inventory doc has rotted and the live state lives in the codebase plus the plan files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): merge utils/legacy.js into utils/index.js The barrel was a placeholder for an eventual split that hasn't happened. Collapsing the two files removes one layer of indirection and the misleading "legacy" name (the contents are still actively used by the migrated SPA). - Move all 930 lines from utils/legacy.js into utils/index.js - Delete utils/legacy.js - Update direct import in models/outbound.js to '@/utils' - Drop a stale legacy.js reference in InboundFormModal comment Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(frontend): keep entry HTML files at frontend/ root The earlier move to frontend/html/ made dev-mode URLs ugly (http://localhost:5173/html/index.html instead of plain /). The folder didn't add real value — it just hid 6 files behind a non-conventional layout. Reverting that piece while keeping src/entries/ (which is a genuine separation between page bootstrap and the rest of src/). - HTML files back at frontend/<page>.html - Vite rollupOptions.input + MIGRATED_ROUTES restored to flat paths - Build output is web/dist/<page>.html again - web/controller/dist.go and sub/subController.go read from dist/<name> Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): bump eslint to 10 + add flat config + clean lint warnings - Upgrade eslint 9.39 -> 10.3 and eslint-plugin-vue 9.33 -> 10.9 - Add eslint.config.js (flat config required by ESLint 10) with vue3-recommended rules, sensible defaults, and exemptions for the project's existing formatting style - Drop --ext from the lint script (removed in ESLint 10) - vue/no-mutating-props is left off because the form-modal pattern ports straight from Vue 2 (parent passes a reactive object, child mutates it); a real fix is an architectural rewire, separate task Lint warning cleanup: - utils/index.js: var -> let/const in the X25519 routines, replace obj.hasOwnProperty(...) with Object.prototype.hasOwnProperty.call(...) - Remove unused imports (reactive, ref, Inbound) in ClientFormModal, InboundInfoModal, QrCodeModal, DnsServerModal, OutboundFormModal, SubPage; remove unused locals (isClientOnline, ONLINE_GRACE_MS, fetchAll, isSocks, isHTTP, _antdAlgorithm) - XrayStatusCard: declare 'open-logs' on defineEmits (was emitted but not declared) - RuleFormModal: rename v-for var t -> tag (shadowed useI18n's t) - Drop stale eslint-disable directives (no-new, no-unused-vars) - OutboundsTab/InboundList: drop redundant initial null assigns - InboundInfoModal/OutboundFormModal: explicit eslint-disable for the intentional local-ref-shadows-prop pattern in modal drafts `npm run lint` now passes with 0 errors and 0 warnings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): one client identity across multiple inbounds via subId Lets the operator add the same email under the same subId to several inbounds. Xray reports traffic per email, so a single client_traffics row acts as the shared accumulator — no aggregation overhead, quota and expiry stay consistent. - Email validation allows duplicates only when subId matches - AddClientStat upserts via OnConflict DoNothing (idempotent on rerun) - Stat/IP rows survive client deletion when a sibling inbound still references the email - enrichClientStats tops up GORM-preloaded stats with rows whose inbound_id points at a sibling, so every panel view sees usage - disableInvalidClients cascades enable=false and syncs the row's total/expiry into every sibling JSON when the shared identity expires - DelDepletedClients removes the depleted client from all referencing inbounds, batched - Subscription services dedupe traffic by email so shared quota is counted once Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * docs(frontend): rewrite README for multi-page Vue 3 layout Reflects the current state — embedded build, per-route HTML entries, ESLint 10 flat config, src/ layout, and the steps to add a new page. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): drop deprecated rimraf/glob/inflight transitive deps vue3-persian-datetime-picker pinned moment-jalaali to ^0.9.4, which pulled rimraf@3 → glob@7 → inflight@1. inflight in particular leaks memory and is unmaintained. Override moment-jalaali to ^0.10.4 (same runtime API, dropped the legacy build deps) so npm install no longer warns and the dep tree is 12 packages lighter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): multi-node panel orchestration (CRUD, deployment, traffic sync, sub per-node) - Node model + service + controller (/panel/api/nodes/*) with bearer-token apiToken auth - Heartbeat job @every 10s; status/latency/xrayVersion surfaced in Nodes UI - Runtime abstraction (Local + Remote) so inbound/client mutations target the inbound's owning node instead of always hitting the local xray - Inbounds gain optional NodeID; tag-based correlation with remote panel (no RemoteInboundID column needed) - NodeTrafficSyncJob @every 10s pulls absolute counters + online/lastOnline from each enabled+online node and writes them into central DB; 30s reset grace window prevents post-reset overwrite - Reset propagation to nodes (best-effort) on client/inbound/all reset paths - Subscription server uses node.Address for inbounds with NodeID, falling back to existing host resolution for local inbounds - Frontend: Nodes page, "Deploy to" select in inbound form, Node column on inbound list, hostOverride threaded through genAllLinks/QR/Info modals Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(stats): system history modal + per-node CPU/Mem trends across all locales Backend - web/service/metric_history.go: generic in-memory ring buffer with two singletons — system-wide (cpu/mem/netUp/netDown/online/load1/5/15) and per-node (cpu/mem) keyed by node id - ServerService.AppendStatusSample writes all 8 metrics every 2s on the same tick; AppendCpuSample/AggregateCpuHistory kept for back-compat - NodeService.UpdateHeartbeat appends cpu/mem only on online ticks so offline gaps render as missing data, not phantom dips - New routes: GET /panel/api/server/history/:metric/:bucket and GET /panel/api/nodes/history/:id/:metric/:bucket, both whitelisted Frontend - Sparkline component generalized: arbitrary value range (auto-scale when valueMax=null), pluggable yFormatter/tooltipFormatter for B/s, client counts, load averages - SystemHistoryModal replaces CpuHistoryModal with tabs for every metric; opened from a tag on the 3X-UI card next to Documentation - NodeHistoryPanel: expandable row on the Nodes table showing per-node CPU and Mem trends, refreshed every 15s Localization - Backfill systemHistoryTitle / trendLast2Min / pages.inbounds.{node, deployTo, localPanel} and the entire pages.nodes block (51 keys including statusValues + toasts) into all 11 non-en/fa locales: ar-EG, es-ES, id-ID, ja-JP, pt-BR, ru-RU, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(embed): include underscore-prefixed Vite chunks in dist FS go:embed silently excludes files whose names start with `_` or `.`, so the `_plugin-vue_export-helper-<hash>.js` chunk that Vite/rolldown emits for @vitejs/plugin-vue was missing from the production binary. First import at runtime hit a 404 and the SPA failed to mount — blank page on every page load, no error in the server logs because the asset 404 was just a static-handler miss. Switched the directive to `//go:embed all:dist` which keeps the same root layout but disables the underscore/dot exclusion rule. Dev mode was unaffected (it serves dist/assets/ from disk, not the embedded FS). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * ci: build frontend bundle before Go compile in release.yml + Dockerfile Phase 8 cut all panel HTML routes over to web/dist/ and embedded the Vite bundle into the Go binary via //go:embed all:dist. web/dist/ is .gitignored, so on a fresh CI checkout it doesn't exist — every Go build since Phase 8 has been failing with "pattern dist: no matching files found" or producing a binary that 404s on first asset request. release.yml: add a setup-node@v4 + npm ci + npm run build trio before the existing go build step in both the Linux matrix job (7 arches) and the Windows job. npm cache is keyed on frontend/package-lock.json. Dockerfile: add a node:22-alpine frontend stage that runs npm ci + npm run build and emits to /src/web/dist (via vite.config.js's outDir). The golang builder stage then COPY --from=frontend /src/web/dist into ./web/dist before the go build, so embed.FS sees the bundle. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(ws): live updates on inbounds/xray/nodes pages, drop polling + manual refresh Replaces the legacy polling + manual-refresh model with WebSocket pushes across the three live-data pages. The hub already broadcast traffic / client_stats / outbounds; this wires the frontend to consume them and adds a new `nodes` channel for the heartbeat job's snapshot. Frontend - new useWebSocket composable: page-scoped singleton WebSocketClient, lifecycle-managed on/off, leaves disconnect to page-unload - inbounds: useInbounds gains applyTrafficEvent / applyClientStatsEvent / applyInvalidate that merge counters and online/lastOnline in place; InboundsPage subscribes; InboundList drops the auto-refresh popover, the refresh button, and the now-unused refreshing prop - xray outbounds: useXraySetting gains applyOutboundsEvent; XrayPage subscribes; OutboundsTab drops the refresh button + emit - nodes: useNodes gains applyNodesEvent and stops the 5s setInterval/visibilitychange polling; NodesPage subscribes; NodeList drops the refresh button and ReloadOutlined import Backend - web/websocket: new MessageTypeNodes + BroadcastNodes notifier - node_heartbeat_job: after wg.Wait(), reload the table once and BroadcastNodes(updated). Gated on websocket.HasClients() so a panel with no open browser doesn't spend the DB read Bug fixes spotted in this pass - websocket.js #buildUrl defaulted basePath to '' when the global was missing (dev mode), producing `ws://host:portws` and a SyntaxError on the WebSocket constructor. Fall back to '/' and ensure leading slash. - vite.config.js: forward /ws to ws://localhost:2053 with ws:true so dev (5173) reaches the Go backend's WebSocket - NodeFormModal: a-input-password's visibilityToggle is Boolean in AntD Vue 4; the v3-era object form (`{ visible, 'onUpdate:visible' }`) triggered a Vue prop-type warning. Drop the override (default true shows the eye icon and toggles internally) and remove the orphaned tokenVisible ref Translations - pages.inbounds.autoRefresh / autoRefreshInterval: removed from all 13 locales (UI gone) - pages.nodes.refresh: removed from all 13 locales (UI gone) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): hide Node column when no nodes are defined Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-09 08:38:48 -07:00
return 0
}
return process.GetAPIPort()
Vue3 migration (#4198) * docs(migration): Phase 1 inventory — Vue 2 / AD-Vue 1 surface area Captures the breakage surface for the Vue 3 + Ant Design Vue 4 + Vite migration: 17,650 lines across 69 templates, 3,145 a-* component instances across 63 files, with per-pattern counts and file lists. Key findings: - No Vue filters anywhere — dodges a major Vue 3 breaking change - 358 v-model uses; AD-Vue 4 absorbs most, custom components don't - 233 <template slot="X"> usages must become <template #X> - 49 scopedSlots: { ... } column defs need new slots: { ... } shape - a-icon is removed in AD-Vue 4 — every icon must be imported Establishes the 8-phase order; Phase 2 (Vite toolchain) is next. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): Phase 2 — scaffold Vite + Vue 3 + AD-Vue 4 Adds a frontend/ directory that lives alongside the legacy web/html/ Vue 2 templates during the migration. Vite builds into ../web/dist/ so the Go binary will be able to embed the result via embed.FS once Phase 4 starts moving real pages over. - package.json pins Vue 3.5, Ant Design Vue 4.2, Vite 6, vue-i18n 10 - vite.config.js: dev server on :5173 with API proxy to the Go panel on :2053; build output to ../web/dist/ - src/App.vue is currently a smoke-test placeholder — delete once the first real page (login) lands in Phase 4 - node_modules and dist are already ignored at repo root To verify locally: cd frontend && npm install && npm run dev Pages will be migrated one at a time on the vue3-migration branch. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): Phase 3 — port utils, models, axios, websocket as ES modules Ports the framework-agnostic JS from web/assets/js/ into frontend/src/ so Vue 3 pages can import what they need without relying on script-tag globals. - web/assets/js/util/index.js (927 lines, 21 classes) → frontend/src/utils/legacy.js + a barrel at utils/index.js. All classes are now named exports. - Vue.prototype.$message in HttpUtil → direct import of `message` from ant-design-vue (Vue 3 has no Vue.prototype). - RandomUtil.randomShadowsocksPassword previously defaulted to SSMethods.BLAKE3_AES_256_GCM from inbound.js, creating a circular import. Replaced with the literal string default. - MediaQueryMixin (Vue 2 mixin) removed. Replaced by composables/useMediaQuery.js — Vue 3 composable returning reactive `isMobile`. - axios-init.js wrapped as setupAxios(); Qs global → npm `qs`. - websocket.js exported as WebSocketClient class; the implicit window.wsClient global is gone — pages instantiate it themselves. - model/{inbound,outbound,dbinbound,setting,reality_targets}.js copied with `export` added on every top-level declaration. Imports between models and utils are wired up explicitly. - subscription.js deferred to Phase 5 (it's a Vue 2 mount, not a util). - App.vue smoke test exercises SizeFormatter / RandomUtil / Wireguard / useMediaQuery so the user can verify Phase 3 with `npm run dev`. Run `cd frontend && npm install && npm run dev` — qs was added so a fresh install is required. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 4 — port login.html to Vue 3 + AD-Vue 4 + Vite 8 First real page in the new toolchain. Multi-page Vite: each migrated page is its own entry. login.html now lives at frontend/login.html with a thin entrypoint at frontend/src/login.js mounting LoginPage.vue. Vite 6 → Vite 8.0.11 (per user request). Requires Node 20.19+ or 22.12+. @vitejs/plugin-vue bumped to ^6.0.6 (peers vite ^8). Ant Design Vue stays on 4.2.6 — there is no AD-Vue 6. Vue 2 → Vue 3 / AD-Vue 1 → AD-Vue 4 syntax changes hit on this page: - new Vue({ el, delimiters, data, methods }) → createApp + <script setup> - mounted() → onMounted() - <template slot="X"> → <template #X> - <a-icon slot="prefix" type="user"> → <template #prefix><UserOutlined /> </template> with explicit @ant-design/icons-vue imports - v-model.trim → v-model:value (AD-Vue 4 uses named v-model on inputs) Three legacy features deferred so Phase 4 stays small: - i18n (Phase 7 wires up vue-i18n) - theme switcher (custom component pending Phase 5) - headline word-cycle animation (purely aesthetic) Run `cd frontend && npm install && npm run dev`, open http://localhost:5173/login.html. With Go panel running on :2053 the form submits real credentials via the configured proxy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5a — theme system + Vite 8 + vue-i18n 11 Bumps Vite to 8.0.11 (npm install picked up 6.4.2 from the stale lockfile; clean install resolves the new constraint). Bumps vue-i18n to 11.1.4 since v10 was just EOL'd. Migrates aThemeSwitch.html — the two-flavor theme picker + global themeSwitcher object — into: - composables/useTheme.js: single reactive `theme` state with toggleTheme / toggleUltra. Boot side-effect applies the stored theme to <body>/<html> before Vue renders; watchEffect persists changes back to localStorage. - components/ThemeSwitch.vue: full menu version for the main panel. - components/ThemeSwitchLogin.vue: login-popover version. AD-Vue 1 → 4 changes hit on this component: - <a-icon type="bulb" :theme="filled|outlined"> dropped — replaced by explicit BulbFilled / BulbOutlined imports from @ant-design/icons-vue, swapped via <component :is="BulbIcon"> - Vue.component('a-theme-switch', { ... }) global registration → SFC + per-page import - this.$message.config(...) (Vue 2 instance method) → message.config(...) imported from ant-design-vue, called once in login.js at boot Login page now surfaces a settings button → popover → theme picker. Known gap: web/assets/css/custom.min.css isn't yet imported into the new bundle, so toggling dark mode currently only re-themes AD-Vue's own components, not the panel chrome. The body class is still toggled so behavior is correct; visual fidelity returns when custom.css is ported or directly imported. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5b — port four shared components to Vue 3 CustomStatistic.vue and SettingListItem.vue are mechanical Vue.component → SFC ports. AppSidebar.vue: AD-Vue 4 dropped <a-icon :type="dynamic">, so the five sidebar icons (dashboard/user/setting/tool/logout) live in a name→component map and render via <component :is>. The legacy <a-drawer slot="handle"> hack is replaced with a sibling fixed- position toggle button. Tab paths take basePath/requestUri as props instead of pulling them from Go template scope. TableSortable.vue: the biggest Vue 3 rewrite of this phase. - $listeners is gone — replaced by inheritAttrs: false + explicit attrs forwarding - scopedSlots: this.$scopedSlots collapsed into Vue 3's unified slots object — just iterate Object.keys(this.slots) and forward - Vue 2 h(tag, { props, on, scopedSlots }, children) → Vue 3 h(tag, { ...props, ...on }, slotsObject) - 'a-table' string → resolveComponent('a-table') so app.use(Antd) registration is honored - inject: ['sortable'] (Options API) → inject('sortable', null) (Composition API) inside the trigger child - beforeDestroy → beforeUnmount - customRow's return shape flattened (no nested props/on/class) Two intentional skips, documented in the migration doc: - aClientTable.html — slot fragments, not a component. Migrates inline with inbounds.html (new Phase 5f). - aPersianDatepicker.html — wraps a Persian-only third-party lib; defer until settings.html lands. Build verified with vite 8.0.11. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): anchor Vite dev proxy so /login.html isn't forwarded The /login proxy entry was matching any path starting with /login — including /login.html, which Vite is supposed to serve itself. Without the Go backend running, this caused ECONNREFUSED noise on every page load. Switched to regex patterns anchored with ^...$ so only the bare backend paths (/login, /logout, /getTwoFactorEnable) and explicit sub-routes (/panel/*, /server/*) get proxied. Static .html files Vite serves directly are no longer matched. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): real dark mode + silence dev proxy ECONNREFUSED noise Two issues from running login.html against no Go backend: 1. Dark mode toggled the body class but didn't actually re-theme any AD-Vue components. The legacy panel relied on custom.min.css which we haven't ported. AD-Vue 4 ships its own dark algorithm — wrap LoginPage in <a-config-provider :theme="{ algorithm }"> driven by our useTheme state, and AD-Vue restyles every component for free. Page chrome (background, card, title) gets explicit .is-dark CSS since the algorithm only covers AD-Vue components. 2. Vite logged every failed proxy attempt loudly. When the Go panel isn't running locally that's pure noise. Added a configure() callback that swallows ECONNREFUSED specifically; real errors (timeouts, 5xx, anything else) still surface. Both fixes are dev-experience only — production build is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): use legacy panel palette for login page dark mode Earlier dark mode used invented colors (#141a26 page bg, #1f2937 card) that didn't match the rest of the panel. Replaced with the actual values from web/assets/css/custom.min.css: light dark ultra-dark bg #c7ebe2 bg #222d42 bg #0f2d32 card #fff card #151f31 card #0c0e12 title #008771 title #fff/.92 title #fff/.92 Drove everything off CSS custom properties on .login-app so the .is-dark / .is-ultra class swap is a few var overrides instead of duplicating selectors. Also restored the legacy card metrics (2rem radius, 4rem 3rem padding, 2rem title) so the new page matches the old panel's geometry, not just its colors. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave layout + recolor for dark mode The wave SVG had inline fill="#c7ebe2" (mint) on the bottom wave, so in dark/ultra-dark mode it rendered as a pale-white blob against the dark page. Stripped the inline fills, drove them off CSS variables that swap with .is-dark / .is-ultra: light: green tints + #c7ebe2 (mint) on the bottom wave dark: #222d42 across all four waves ultra-dark: #0f2d32 The wave was also positioned wrong — anchored to the top 200px of the viewport with absolute positioning. Restored the legacy layout: - .waves-header is fixed to the top of the viewport with z-index -1 so the form floats over it - .waves-inner-header pushes the wave SVG down to ~50vh with a 50vh-tall solid block of the page color - .waves SVG itself is 15vh tall, sitting at the bottom of that block Net effect: top half is solid-colored, then a wavy edge transitions into the rest of the page, with the form centered on top — matching the legacy panel exactly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): bring wave-header to front so the wave actually shows Two layering bugs were hiding the wave entirely: 1. .ant-layout-content had background: var(--bg-page) which painted an opaque rectangle covering the full content area — including the fixed wave-header behind it. Made the layout/content transparent and moved the bg paint up to .login-app (the outer ant-layout). 2. .waves-header had z-index: -1 which on its own was fine, but with .ant-layout-content opaque on top it was doubly buried. Promoted the wave-header to z-index: 0 and gave the form .login-row z-index: 1, so the form sits above the wave and the wave sits above the page-bg. Also set --bg-page to the legacy mint (#c7ebe2) for light mode so the bottom half of the page below the wave matches the legacy panel (was white). Dark mode stays at the surface-100/login-wave palette. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): match legacy wave animation timings + dark page bg Two reasons the bottom wave looked static in dark/ultra-dark: 1. Animation durations were 7s/10s/13s/20s. Legacy uses 4s/7s/10s/13s. The 20s on the bottom wave was so slow that against the low dark- mode contrast it read as motionless. Restored the legacy timings. 2. --bg-page in dark mode was #151f31 (card color / surface-100), but the legacy .under uses surface-200 (#222d42) — that's the color of the bottom half of the page, the same as the wave fill, so the wave appears to flow into the page rather than meeting a hard edge. Now it does. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): restore Hello/Welcome headline cycle on login Earlier I deferred the legacy headline word-cycling animation as "purely aesthetic". Restored it: the title now alternates between 'Hello' and 'Welcome' every 2 seconds, matching the legacy panel. The legacy implementation toggled .is-visible / .is-hidden classes on two <b> elements via setTimeout chains and DOM querying. Replaced with a reactive ref + Vue 3 <Transition mode="out-in"> so the fade between words is declarative — no manual DOM manipulation, and the interval is properly cleaned up in onBeforeUnmount. The earlier "Welcome to 3x-ui" string was wrong on two counts: it should be just "Welcome", and it should be one of two cycling words with "Hello" preceding it. Ultra-dark palette already matched legacy after the prior wave timing fix; no additional changes needed there beyond the animation speeds that now also apply to ultra-dark via the shared CSS rules. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): correct dark login bg + give ultra-dark wave real contrast Two related fixes: 1. Default-dark wave-header bg was wrong. I had #0a2227, but that's the *ultra-dark* override; default dark uses --dark-color-background = #0a1222. Now the dark-mode top half is the legacy purple-blue instead of teal. 2. Ultra-dark wave fill is intentionally near-identical to its bg in the legacy palette (#0f2d32 vs #0a2227, ~5/11/11 RGB delta), which makes the wave look static even though the animation is running. Bumped --wave-fill / --wave-fill-bottom to #1f4d52 in ultra-dark only — far enough above the bg that the motion reads, while staying within the same teal hue family. Also corrected ultra-dark --bg-page back to #0f2d32 (was briefly #0c0e12, which is the card color, not the page color). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): drop ultra-dark bottom-wave seam line Last fix made the wave fill #1f4d52 in ultra-dark for both top-three waves and the bottom wave, which gave visible motion but exposed a hard horizontal line where the bottom wave's flat lower edge met the page bg (#0f2d32). The user noticed it as "the wave at the bottom not moving its like a line" — they were seeing the SVG's clipped bottom edge, not the wave itself. Solution: only the top three waves get the brighter fill (those carry the visible motion). The bottom wave reverts to #0f2d32 = --bg-page, so its flat bottom edge merges seamlessly into the page below. Net effect: motion is still visible (from waves 2 and 3), and there's no seam line at the bottom of the SVG. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-i — index.html dashboard shell Replaces the smoke-test App.vue with a real IndexPage shell so the /index.html route now boots the actual dashboard layout in Vue 3: - a-config-provider drives AD-Vue 4's dark algorithm from useTheme (same pattern as LoginPage) - AppSidebar (Phase 5b component) is wired in with basePath + requestUri props - a-spin loading state with placeholder card while we build out the rest of the page - Page palette mirrors the legacy: light #f0f2f5, dark #0a1222 (--dark-color-background), ultra-dark #21242a The 1,805-line legacy index.html is too big for one commit. Split into five sub-phases on the todo list: ii) status cards + /server/status polling, iii) xray status card, iv) logs/backup/panel-update modals, v) custom-geo section. frontend/src/App.vue and frontend/src/main.js (smoke-test scaffold) are removed — both purposes now served by IndexPage and index.js. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-ii — live status cards on the dashboard Adds the CPU / memory / swap / disk dashboard cards to IndexPage, backed by a useStatus() composable that polls /panel/api/server/status every 2 s and a Status / CurTotal model ported from the legacy inline classes in index.html. - models/status.js — Status & CurTotal classes (CurTotal exposes reactive .percent and .color computed-style getters; Status maps the API payload + xray state to color/message strings) - composables/useStatus.js — 2s polling with shallowRef so each fetch swaps the whole Status object atomically. WebSocket integration intentionally deferred — the legacy panel falls back to this same 2s polling when its websocket drops, so we ship the proven path first and add WS on top in a later sub-phase. - pages/index/StatusCard.vue — four a-progress dashboard widgets in a 2x2 grid (mobile collapses to a 1x4). CPU widget exposes a history button; the modal it opens is part of 5c-iv. - IndexPage now consumes both, plus useMediaQuery so the layout responds to viewport changes. AD-Vue 4 changes: <a-icon type="area-chart"|"history"> dropped in favor of explicit AreaChartOutlined / HistoryOutlined imports. <a-tooltip slot="title"> → <template #title>. i18n strings still hardcoded English (Phase 7 wires up vue-i18n). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iii — xray status card + stop/restart controls XrayStatusCard.vue renders the right-hand card on the dashboard: - Title with mobile-only version tag (matches the legacy collapse) - Animated badge for the running/stop/error states. The pulsing dot comes from xray-pulse keyframes (renamed from runningAnimation in legacy custom.min.css). Color rings on the badge use the legacy's per-state border-color overrides on .ant-badge-status-processing. - Error state replaces the badge with a popover that surfaces the multi-line errorMsg + a logs shortcut. - Action row at the bottom: optional logs (when ipLimitEnable), stop, restart, and version switch. IndexPage now wires: - POST /panel/api/server/stopXrayService and /restartXrayService, followed by a refresh() so the status card reflects the new state without waiting for the next poll tick - POST /panel/setting/defaultSettings to read ipLimitEnable - Stub handlers for the panel-logs / xray-logs / version-switch / cpu-history modals — those land in 5c-iv AD-Vue 4 changes hit on this card: - <a-icon type="bars|poweroff|reload|tool"> → explicit BarsOutlined / PoweroffOutlined / ReloadOutlined / ToolOutlined - <span slot="title|content"> → <template #title|#content> - The .xray-*-animation classes ship as global <style> (not scoped) so they pierce AD-Vue's internal .ant-badge-status-* DOM. i18n still hardcoded English; Phase 7 wires vue-i18n. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (a) — panel update / logs / backup modals Adds three of the six dashboard modals plus a Quick Actions card that surfaces them. The remaining three (xray logs, version picker, CPU history sparkline) ship in 5c-iv-b. - PanelUpdateModal.vue — current vs latest version, "update now" button. Confirm dialog → POST /panel/api/server/updatePanel, then poll /server/status for up to 90s until the new panel answers, then reload. - LogModal.vue — panel logs viewer. Filters: rows (10-500), level (debug/info/notice/warning/error), syslog toggle. Auto-fetches on open and on every filter change. Color-coded timestamps and levels via inline span styles. Download button writes the raw log to x-ui.log via FileManager.downloadTextFile. - BackupModal.vue — db export (window.location to /getDb) and import (FormData upload to /importDB, then panel restart + reload). - Quick Actions card surfaces Logs / Backup / Update buttons and shows an orange update badge (extra slot) when an update is available. Modal-busy pattern: long-running operations (update, import) emit a `busy` event with a tip; IndexPage flips its a-spin overlay so the user sees a loading message while the panel is restarting. AD-Vue 4 changes: - v-model on <a-modal> renamed to v-model:open - v-model on <a-input>/<a-select>/<a-checkbox> uses the named v-model:value / v-model:checked pattern - <a-icon type="..."> dropped — explicit Ant icon imports (BarsOutlined, CloudServerOutlined, CloudDownloadOutlined, DownloadOutlined, UploadOutlined, SyncOutlined) - Modal.confirm() replaces this.$confirm() since setup() has no `this` Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-iv (b) — cpu-history / xray-logs / xray-version modals Wires up the three remaining dashboard buttons that were stubbed in 5c-iv (a): the CPU history button on StatusCard, the xray-logs button in XrayStatusCard's error popover and ipLimitEnable action, and the "Switch xray" button in XrayStatusCard's action footer. - Sparkline.vue: shared SVG line chart (composition-API port of the inline Vue 2 component). Per-instance gradient id avoids defs collisions between sparklines on the same page. - CpuHistoryModal.vue: bucket dropdown (2m/30m/1h/2h/3h/5h) drives GET /panel/api/server/cpuHistory/{bucket}; renders via Sparkline. - XrayLogModal.vue: rows + filter + direct/blocked/proxy checkboxes; POST /panel/api/server/xraylogs/{rows} returns access-log entries rendered as a colored HTML table; download button serializes to text. - VersionModal.vue: collapse with Xray panel (radio list of versions from getXrayVersion, install via installXray/{version}) and Geofiles panel (per-file reload + Update all). CustomGeo collapse panel is Phase 5c-v. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5c-v — custom-geo section in VersionModal Adds the third collapse panel ("Custom geo") that lets users register external geosite/geoip files referenced by routing rules via ext:<filename>:tag. Backend endpoints are unchanged. - CustomGeoSection.vue: bordered table over /panel/api/custom-geo/list with per-row edit, download (refetch), and delete actions, plus an Add button and Update-all. Lazy-loads the list when the parent collapse opens this panel — closed panels don't fetch. - CustomGeoFormModal.vue: shared add/edit form with the same alias regex (^[a-z0-9_-]+$) and URL validation as legacy. Type and alias are immutable when editing — backend rejects changes anyway. - ext:<filename>:tag value is click-to-copy via ClipboardManager. - Relative time is computed inline (no moment dep); tooltip shows the absolute timestamp. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-i — settings page shell + dirty tracking Adds the settings entry as a new Vite multi-page input. Lays down the shared page chrome (sidebar, save bar, restart, security alert) and the AllSetting fetch/dirty-poll lifecycle so 5d-ii through 5d-vi can drop in tab partials without re-implementing it. - settings.html + src/settings.js: third Vite entry; mounts SettingsPage. - SettingsPage.vue: page chrome with the legacy two-button save/restart bar, conf-alerts banner, and 5 a-tabs (4 always-visible + the formats tab gated on subJsonEnable || subClashEnable). Each tab body is an a-empty placeholder until 5d-ii…vi fill them in. - useAllSetting.js composable: POST /panel/setting/all on mount, mirrors the legacy 1s busy-loop dirty check via setInterval, and exposes fetchAll/saveAll. saveDisabled flips off as soon as the user diverges from the server snapshot. - restartPanel rebuilds the URL (host/port/scheme/base path) from the saved settings so users land on the new endpoint after a port or cert change. - models/setting.js: adopts the @/utils alias and a leading file-level doc — semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-ii — settings General tab Ports the panel/general partial (the largest single tab) — six collapse panels: General, Notifications, Certificates, External traffic webhook, Date and time, LDAP. - GeneralTab.vue receives the reactive AllSetting via props and binds fields directly with v-model:value; SettingsPage stays the sole fetch/save owner. - remarkModel/remarkSeparator surfaced as computed v-models that read+write the underlying single-string field (legacy stores them packed as <separator><orderedKeys>, e.g. "-ieo"). - LDAP inbound-tags select binds to a CSV ↔ array computed; inbound options come from /panel/api/inbounds/list on mount. - Language select stays cookie-based via LanguageManager and reloads on change — same UX as legacy. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iii — settings Security tab + 2FA modal Ports the panel/security partial: change-credentials form and 2FA toggle. The 2FA modal is a new shared component since enabling 2FA, disabling 2FA, and changing credentials all funnel through it with slightly different copy. - TwoFactorModal.vue: 'set' flow renders a QR code + manual key + a 6-digit verifier; 'confirm' flow renders just the verifier. The parent passes a confirm(success) callback that fires only when the entered code matches the live TOTP value (otpauth lib). - SecurityTab.vue: holds the local user form (oldUsername/oldPassword/ new*), POSTs /panel/setting/updateUser, and on success force-redirects to logout. When 2FA is on, the credentials change goes through the confirm-modal first. - toggleTwoFactor leaves the switch read-only (the v-bound :checked matches AllSetting) and only flips after the modal succeeds, so cancelling out leaves state unchanged. - Adds otpauth ^9.5.1 dep (qrious was already present). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-iv — settings Telegram tab Ports the panel/telegram partial: bot enable/token/chatId/lang in the General panel, schedule/backup/login/CPU-threshold in Notifications, and proxy/API-server overrides in the third panel. All bindings live on the shared AllSetting reactive — no fetch/save logic in this tab. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-v — settings Subscription general tab Ports the subscription/general partial — four collapse panels covering the master enable switches, presentation/template fields, certs, and update interval. - Sub path goes through a strip-on-input + normalize-on-blur computed: legacy stripped `:` and `*` and ensured the value starts and ends with a single `/` — same here. - Both `subEnableRouting` and the announce/profile/title/support URLs are bound directly on AllSetting. - The "Subscription URI override" placeholder mirrors the legacy pattern for the manual full-URL form. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5d-vi — settings Subscription formats tab Ports the subscription/json partial — paths/URIs for the JSON and Clash formats plus the four packed-JSON sub-fields: fragment, noises, mux, and direct routing rules. - subJsonFragment / subJsonMux / subJsonNoises / subJsonRules are each a JSON string on the wire; the tab exposes their fields as computed v-models that read+write the underlying JSON. Toggling a top-level switch off resets the field to "" (matches legacy semantics). - Direct routing rules surface the IP and domain entries of the seed rule array as multi-select tag inputs; setting/removing tags edits the rules array in place rather than rebuilding it from scratch, so manually-added rules are preserved. - Tab is gated on subJsonEnable || subClashEnable in the parent (only rendered when the user actually opted into one of those formats). This closes Phase 5d — full settings page parity with the legacy panel across all five tabs. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): route /panel/<route> to migrated pages in dev The sidebar links to production-style URLs like /panel/settings, but in dev that gets proxied to the legacy Go template — which fails because we haven't loaded the legacy asset chain. Add a proxy bypass so /panel and /panel/settings are served from index.html / settings.html on the Vite dev server itself. Unmigrated routes (inbounds, xray) still proxy to Go. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(csrf): expose token endpoint for SPA pages and fetch it from axios The legacy panel pages got their CSRF token from a <meta name="csrf-token"> tag rendered by Go. SPA pages built by Vite don't have that, so every unsafe (POST/PUT/DELETE) request from them was hitting CSRFMiddleware with no token and getting 403 — visible as the settings page being stuck on "Loading…" because POST /panel/setting/all failed. - web/controller/xui.go: GET /panel/csrf-token returns the session token. Lives under the xui group so checkLogin still gates it; the CSRFMiddleware on the same group is a no-op for GET. - frontend/src/api/axios-init.js: cache the token at module scope and lazy-fetch it when a non-safe request needs one. Seed from the meta tag first when present (legacy compat). On a 403 response, drop the cache and retry once — handles the case where a server restart rotated the token after the SPA loaded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): keep sidebar links absolute when basePath is empty The dashboard sidebar built tab keys as basePath + 'panel/...'. In dev the window-injected basePath is '' so the resulting key was a relative path like 'panel/settings'. When the browser resolved that against the current /panel/settings URL it produced /panel/panel/settings — visible as broken navigation between Dashboard and Settings. Force a leading slash so the keys are always absolute regardless of whether the host injected a basePath. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-i — inbounds page shell + list fetch Adds the inbounds entry as a fourth Vite multi-page input and wires /panel/inbounds through the dev proxy bypass. Lays down the page chrome (sidebar, summary statistics card, refresh button) and the fetch lifecycle composable so 5f-ii onward can drop in the table columns and the modals without re-implementing it. - inbounds.html + src/inbounds.js: fourth Vite entry; mounts InboundsPage. - InboundsPage.vue: sidebar + summary card (totals over up/down, all-time, inbound count, client tags) + a basic table with enable/ remark/port/protocol/traffic/expiry columns. Row actions, popovers, search/filter, auto-refresh, and the WebSocket delta path are all deferred to subsequent 5f subphases. - useInbounds.js composable: GET /panel/api/inbounds/list + POST /panel/api/inbounds/onlines + POST /panel/api/inbounds/lastOnline + POST /panel/setting/defaultSettings, then computes the per-inbound clientCount roll-ups (active/deactive/depleted/expiring/ online/comments) the table popovers consume. - models/dbinbound.js + models/inbound.js: switched the legacy-utils import to the @/utils alias for consistency with the rest of the app. Semantics unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-ii — inbound list table + search/filter + auto-refresh Fleshes out the inbound list with the full column set, search & filter toolbar, row enable toggle wired to /panel/api/inbounds/setEnable/:id, and a per-row action dropdown that emits events the parent will route to modals as those land in 5f-iii through 5f-vii. - InboundList.vue (new): toolbar (Add inbound + General actions dropdown + Refresh + auto-refresh popover), search-or-filter switch with the legacy radio buttons (Active/Disabled/Depleted/Depleting/ Online), and a a-table with desktop and mobile column variants. Cells use AD-Vue 4's #bodyCell slot — protocol/clients/traffic/ allTime/expiry/info cells render the same popovers and tags as legacy. Row enable switch is optimistic with rollback on POST failure. - visibleInbounds computed mirrors the legacy search and filter projection: deep search through dbInbound + clients, or filter reduces inbound.settings.clients to the selected bucket so the table only shows matching client rows. - Auto-refresh interval is read/written to localStorage with the same keys (`isRefreshEnabled`, `refreshInterval`) as the legacy panel. WebSocket delta updates are still deferred. - Action menu emits event payloads {key, dbInbound}; the parent currently shows a "coming in later 5f subphase" toast for each. Modals (edit/qr/clone/delete/reset/info/clients) land in 5f-iii through 5f-vii. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(inbounds): wrap popover-table rows in <tbody> Vue's template compiler warned that <tr> can't be a direct child of <table> per the HTML spec; the browser silently inserts a <tbody> wrapper but Vue's SSR/hydration path doesn't, which can cause hydration mismatches. Add explicit <tbody> in both popover tables (traffic + mobile-info). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii — inbound add/edit modal + delete/clone/reset Wires up the inbound CRUD flows. The protocol-specific and transport- specific forms are still ahead in 5f-iii-b — for now the modal exposes those as JSON textareas so users can both edit existing inbounds without losing settings and create new ones from default templates. - InboundFormModal.vue: tabbed modal with a full Basics tab (enable, remark, protocol, listen, port, total GB, traffic reset, expiry date) and three JSON-edit tabs (Settings, Stream, Sniffing). Add mode stamps a fresh template per protocol via Inbound.Settings.getSettings(protocol); changing the protocol in add mode restamps the JSON. Edit mode pretty-prints the existing JSON so the user sees the same fields they save back. - POST /panel/api/inbounds/add or /panel/api/inbounds/update/:id on submit; on success the parent refreshes the list and the modal closes. Malformed JSON in any of the three textareas surfaces a message.error and aborts the save without losing user input. - InboundsPage.vue: wires the row action menu to real handlers — edit (opens the modal in edit mode), delete, reset-traffic, clone, reset-clients, del-depleted-clients all go through Modal.confirm and refresh on success. General actions menu wires reset-inbounds / reset-clients / del-depleted-clients the same way. Remaining actions (qrcode/info/import/export/copyClients) still toast as "coming soon" — those land in 5f-iv and 5f-v. - Adds dayjs ^1.11.20 dep for the a-date-picker v-model interop. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iv — client add/edit + bulk-add modals Wires per-inbound client management. Both flows go through the same addClient/updateClient endpoints as legacy; the modals just funnel the form state into the right shape (`{id, settings: '{"clients": [...]}'}`). - ClientFormModal.vue: protocol-aware single-client editor — email/ password/id/auth/security/flow/subId/tgId/comment/ipLimit/totalGB/ expiry/renewal fields are shown/hidden per protocol like legacy. Edit mode displays the per-client traffic stats with a reset button; IP-limit log is read on click and clearable. Random helpers (sync icon next to each label) regenerate UUID/email/ password/sub-id values. - ClientBulkModal.vue: 1–500 clients in one POST, with the legacy five email-generation modes (Random / +Prefix / +Num / +Postfix / Pure-Prefix-Num-Postfix). Builds clients via the protocol-aware factory and concatenates their toString() output into a single settings.clients JSON array. - InboundsPage.vue: opens both modals from the row action menu (`addClient` / `addBulkClient`). They both refresh the inbound list on success. - Outstanding row actions still toast as "coming soon": qrcode, showInfo, copyClients, clipboard. Those land in 5f-v / 5f-vi. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-v — inbound info + QR-code modals Wires the row "info" and "qrcode" actions and ports the legacy inbound_info_modal end-to-end. The info modal handles every protocol the legacy panel did: • multi-user (VMess/VLess/Trojan/SS-multi/Hysteria) — per-client table + share links + per-link QR; • SS single-user — share link + QR; • WireGuard — full peer table with downloadable peer-N.conf and a wg:// share link per peer; • Mixed/HTTP/Tunnel — connection-detail tables. - QrPanel.vue: shared link card (header tag, copy button, optional download button, optional QR canvas, monospace footer with the raw value). Per-instance QRious instances are repainted on value/size change. - InboundInfoModal.vue: full info modal. Subscription URL block keys off subSettings.subURI/subJsonURI; IP-log lazy-loads on open and surfaces refresh + clear; tg-id, last-online, depleted/enabled tags all match legacy. - QrCodeModal.vue: lighter modal used for the row "qrcode" action on SS-single and WireGuard inbounds (just the QRs, no info table). - InboundsPage.vue: wires both flows. checkFallback() reproduces the legacy logic — when an inbound listens on a unix-socket fallback (`@<name>`), the link generator is pointed at the root inbound that owns the listen address so QRs/links carry the public host:port + the right TLS state. Multi-client navigation (focusing a specific client's links) is deferred to 5f-vi where the per-inbound expand- row table will pass the email through. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vi — per-inbound client expand-row table Each multi-user inbound row in the list now expands to show its client roster, mirroring the legacy aClientTable component. - ClientRowTable.vue: inner a-table with full desktop column set (action icons / enable / online / client-with-status-dot / traffic with progress bar / all-time / expiry with reset cycle) and a collapsed mobile variant (single dropdown menu + popover info). Self-contained: stats are looked up via a per-inbound email->stats Map; per-client confirms (reset/delete) live on the row. - The component emits typed events (edit/qrcode/info/reset-traffic/ delete/toggle-enable) — InboundsPage routes them back to the existing client and info modals (with `findClientIndex` so the modal opens focused on the right client). - InboundList.vue: hooks ClientRowTable into the a-table's expandedRowRender slot; row-class-name `hide-expand-icon` and a scoped CSS rule hide the chevron for non-multi-user inbounds (HTTP/Mixed/Tunnel/WireGuard/SS-single) so they keep looking flat. - toggle-enable-client routes through updateClient with the same `{id, settings: '{"clients": [...]}'}` shape as the other modals, so backend parsing stays single-pathed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-iii-b — replace inbound modal JSON textareas with structured forms Rewrites InboundFormModal to look like the legacy panel: structured forms for the common case, with a compact "Advanced (JSON)" fallback for the rare bits we don't yet have UI for. Tabs: • Basics — enable/remark/protocol/listen/port/total/trafficReset/expiry • Protocol — protocol-aware: VMess/VLess/Trojan/SS-multi/Hysteria in add mode embed an inline first-client form (email + ID/password/auth, security, flow, subId, comment, total GB, expiry); edit mode shows a clients-count summary table; VLess: decryption/encryption inputs; SS: method dropdown that re-randomizes password and propagates method change to the multi-user array (matches legacy SSMethodChange); HTTP/Mixed: accounts table with add/remove rows + Mixed auth/udp/ip toggles; Tunnel: address/port/network/followRedirect; WireGuard: secretKey/pubKey (regen via Wireguard.generateKeypair) + per-peer fields with PSK regen + allowedIPs add/remove + keepAlive. • Stream — only when canEnableStream(); transport selector with structured forms for TCP (proxy-protocol, http camouflage), WS (host/path/heartbeat/headers), gRPC (serviceName, multiMode), HTTPUpgrade (host/path). KCP/XHTTP fall back to the Advanced tab with an alert banner. Security selector with TLS (sni/alpn/ fingerprint) and Reality (target/serverNames/keypair-gen via /panel/api/server/getNewX25519Cert / shortIds / fingerprint). • Sniffing — enabled/destOverride/metadataOnly/routeOnly/ ipsExcluded/domainsExcluded as structured fields. • Advanced (JSON) — raw streamSettings + sniffing JSON for users reaching KCP/XHTTP/sockopt/finalmask/full TLS cert arrays. The stream JSON is auto-synced from the live model whenever the structured fields change. State source of truth is a deeply-reactive Inbound + DBInbound pair cloned on open; submit serializes via inbound.settings.toString() + inbound.stream.toString() so the wire shape matches the legacy panel byte-for-byte. streamNetworkChange semantics (clear flow when TLS/Reality unavailable, reset finalmask.udp when not KCP) are preserved. Vision Seed for VLess + finer-grained TCP HTTP camouflage + the full TLS cert/ECH editor will land in 5f-iii-c. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 5f-vii — shared text/prompt modals + remaining export/import wiring Wires up the last batch of inbound row + general actions that were toasting "coming soon": export-inbound-links, export-subs (per-inbound and global), export-all-links, import-inbound, and the clipboard JSON peek. Two small shared components back them — both can be reused by the xray page later. - TextModal.vue (shared): read-only multi-line viewer with a copy button and an optional download button when fileName is set. Replaces the legacy txtModal which the inbounds page used for every link export. - PromptModal.vue (shared): generic title + input/textarea + confirm callback, with the legacy keybindings (Enter submits in single-line mode; Ctrl+S submits in textarea mode). Used here for import-inbound but also by xray-config edits in Phase 6. - InboundsPage.vue: drops the toast stubs for `import`/`export`/`subs` on the general-actions menu and `export`/`subs`/`clipboard` on the per-row menu, routing each through openText / openPrompt + the appropriate model helper (genInboundLinks, etc.). The copyClients cross-inbound modal stays toast-stubbed — that's its own dedicated legacy modal worth its own commit. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-i — xray page scaffold + Advanced JSON tab The fifth and last legacy page comes online. Tabs are scaffolded with a-empty placeholders for the structured editors (Basics / Routing / Outbounds / Balancers / DNS) so navigation is stable; the Advanced (JSON) tab is fully functional and lets power users edit the raw xraySetting tree exactly like the legacy CodeMirror pane. - xray.html + src/xray.js: fifth Vite multi-page entry, mounted as XrayPage; vite.config.js routes /panel/xray and /panel/xray/ to it through the dev proxy bypass alongside the other pages. - XrayPage.vue: page chrome with the Save / Restart-xray bar, restart- output popover (surfaces /panel/xray/getXrayResult content when startup fails), 6 a-tabs, and a textarea-backed Advanced JSON editor. CodeMirror is intentionally not pulled in — the textarea works for every modern browser and keeps the bundle slim while structured editors land in 6-ii through 6-v. - useXraySetting.js composable: POST /panel/xray/ on mount, mirrors the settings-page 1s busy-loop dirty check for both xraySetting and outboundTestUrl, and exposes saveAll + restartXray. The dirty flag relies on string equality of the pretty-printed JSON, so reformat-only edits don't enable Save. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-ii — xray Basics tab structured editor Replaces the placeholder on the Basics tab with a structured form for the most-touched fields of the xray template — outbound + routing strategy, log levels, traffic stat counters, and the "basic routing" shortcuts (block torrent / IPs / domains, direct IPs / domains, IPv4 forced, WARP / NordVPN routing). - useXraySetting.js: hoists a parsed `templateSettings` reactive alongside the JSON string, with two cooperating watches that keep them in sync. Editing structured fields stringifies into xraySetting for the dirty-poll + Advanced JSON tab; editing the JSON re-parses into templateSettings only when valid, so structured tabs stay readable mid-edit. - BasicsTab.vue: collapse panels mirror the legacy partial — General, Statistics, Logs, Basic routing. Every input is a computed v-model reading/writing into templateSettings; the routing-rule shortcuts funnel through ruleGetter/ruleSetter which match the legacy templateRuleGetter/templateRuleSetter behavior (replace-first, drop-duplicates, pop-the-rule-when-empty). Direct/IPv4 setters also call syncOutbound() to provision/prune the matching outbound. - XrayPage.vue: imports BasicsTab + derives `warpExist`/`nordExist` from the parsed templateSettings. WARP/NordVPN provisioning modals are still placeholders that toast — those land in 6-v with the routing/outbound editors. Default tab flips back to Basics so users land on the structured editor. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iii — xray Routing tab + rule modal Replaces the Routing tab placeholder with a full editor for templateSettings.routing.rules: - RoutingTab.vue: a-table over the parsed rules with the legacy six- column layout (action / source / network / destination / inbound / outbound) and the same "lead value + N more" pill renderer for multi-value criteria. Mobile drops source/network/destination for readability. Per-row dropdown handles edit / move-up / move-down / delete; the array-mutation reordering replaces the legacy jQuery Sortable drag handle without pulling in a sortable lib. - RuleFormModal.vue: full form mirroring xray_rule_modal.html — CSV inputs for sourceIP/sourcePort/vlessRoute/ip/domain/user/port, Network select, Protocol multi-select, Attrs key/value pairs, inbound-tag multi-select sourced from templateSettings.inbounds + parent inboundTags + dnsTag, outbound-tag single-select sourced from templateSettings.outbounds + clientReverseTags, and balancerTag from templateSettings.routing.balancers. Submit serializes via the same shape the legacy `getResult` produces (CSV → array, drop empty fields). - XrayPage.vue: imports RoutingTab and exposes inboundTags + clientReverseTags from useXraySetting so the modal can populate its tag pools. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-iv — xray Outbounds tab + outbound modal Replaces the Outbounds tab placeholder with a full table + add/edit flow. The 1.3k-line legacy outbound modal is condensed to a tabbed modal with structured Basics fields (tag/protocol/sendThrough/domain strategy) and JSON tabs for the protocol-specific settings + stream trees — same approach the Inbound modal uses, and a power user can still edit the same trees via the page-level Advanced (JSON) tab. - useXraySetting.js: adds fetchOutboundsTraffic + resetOutboundsTraffic + testOutbound. Test states are tracked per outbound index so the row's Test button can show loading + the Test-result column can render the response delay / status / error. - OutboundsTab.vue: full table (action / identity / address / traffic / test result / test) plus a card-list mobile variant with the same row dropdown (set-first / edit / move up/down / reset traffic / delete). outboundAddresses() reproduces the legacy findOutboundAddress logic so each protocol's host:port list is rendered consistently. Add/edit go through OutboundFormModal, delete goes through Modal.confirm, reset traffic posts to /panel/xray/resetOutboundsTraffic with the row's tag (or "-alltags-" from the toolbar). - OutboundFormModal.vue: tag/protocol/sendThrough/domainStrategy on the Basics tab; settings + streamSettings as raw JSON on their respective tabs. Tag-collision check happens client-side before emitting; malformed JSON aborts the save with a message.error. - XrayPage.vue: imports OutboundsTab and wires the test action to the composable's testOutbound helper. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-v — xray Balancers tab + DNS placeholder Brings Balancers to full parity with the legacy panel and adds a DNS tab placeholder that exposes the full dns/fakedns trees as JSON so users can edit them without falling through to Advanced. - BalancerFormModal.vue: tag (with duplicate-tag warning across other balancers), strategy (random/roundRobin/leastLoad/leastPing), selector tag-mode multi-select sourced from existing outbound tags + free-form additions, fallback. Disable-on-invalid is driven by the duplicateTag + emptySelector computed flags. - BalancersTab.vue: empty state with a single "Add balancer" CTA; populated state shows the legacy 4-column table (action / tag / strategy / selector / fallback) with per-row edit + delete in a dropdown. On submit the wire shape preserves the `strategy: { type }` nesting only when the strategy is non-default, matching the legacy emit. Tag renames also chase across routing.rules.balancerTag references so existing rules don't dangle. - DnsTab.vue: master enable switch + raw JSON for `dns` and `fakedns`. Legacy had a dedicated server-by-server editor + a fakedns row editor; both are big enough to deserve their own commits, and the JSON path supports every field today. WARP / NordVPN provisioning modals still toast as "coming soon" — those are third-party API integrations worth their own commits. The xray page now has structured editors for Basics / Routing / Outbounds / Balancers and JSON editors for DNS / Advanced — every xray tab the legacy panel offered is functional. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(server): Phase 8 — cut HTML routes over to web/dist/ Production cutover. Every user-facing HTML route now serves the Vue-3-built bundle from web/dist/ instead of rendering the legacy Go template; the long-hashed Vite assets are served at /assets/ from the same embedded filesystem. The legacy templates in web/html/ and the legacy static tree in web/assets/ are kept on disk for now in case a quick revert is needed, but nothing the binary serves references them. What changed: - web.go: a new //go:embed dist/* feeds the controller package via a SetDistFS hand-off before controller construction. The static /assets/ route is rebound: in dev to web/dist/assets/ on disk so Vite's incremental rebuilds show up live; in prod to the embedded dist via wrapDistFS (rooted one level deeper than wrapAssetsFS). - controller/dist.go: serveDistPage helper used by every HTML handler. Reads dist/<name> from the embedded FS and applies two transforms before sending: 1. injects <script>window.__X_UI_BASE_PATH__="..."</script> just before </head> so AppSidebar links resolve under the panel's basePath. 2. when basePath != "/", rewrites Vite's absolute /assets/ URLs to <basePath>assets/ so installs running under a custom URL prefix load the bundle where the static handler lives. HTML responses go out with no-cache so panel upgrades reach users on the next refresh; hashed JS/CSS stays cacheable. - controller/index.go: IndexController.index now serves dist/login.html for logged-out callers (the redirect for logged-in users is unchanged). - controller/xui.go: XUIController.{index,inbounds,settings,xraySettings} each become a one-line wrapper around serveDistPage. Smoke checklist for the maintainer: - run `cd frontend && npm run build` to refresh web/dist/ before building the Go binary (the embed snapshot is taken at compile time); - visit /panel/, /panel/inbounds, /panel/settings, /panel/xray and confirm each loads its Vue page; - log out and log back in to verify the login flow; - confirm the sidebar links navigate correctly under your install's basePath; - POST flows (e.g. saving settings) still need the CSRF token — that endpoint (/panel/csrf-token, added earlier) is unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 6-vi — WARP + NordVPN provisioning modals Replaces the toast stubs on the Basics tab and Outbounds toolbar with the legacy WARP + NordVPN provisioning flows. Both modals now stage their wireguard outbounds back into templateSettings.outbounds through the same event channels OutboundsTab uses, so the existing add / reset / delete / refresh-traffic surface keeps working. - WarpModal.vue: empty state shows a single Create button that generates a wireguard keypair locally (Wireguard.generateKeypair) and posts it to /panel/xray/warp/reg; populated state surfaces the access_token / device_id / license_key / private_key, lets the user upgrade to WARP+ via /panel/xray/warp/license, refreshes the account info from /panel/xray/warp/config (plan / quota / usage in human-readable bytes), and stages a wireguard outbound with the WARP-specific reserved-byte encoding pulled from client_id. Add / Reset / Delete go through events the parent routes back to templateSettings.outbounds. - NordModal.vue: dual-tab login (NordVPN access token → /panel/xray/nord/reg, or paste a NordLynx private key → /panel/xray/nord/setKey). Once authenticated, country / city / server selectors fetch from /panel/xray/nord/{countries,servers}, servers sort by load ascending, the lowest-load server in the current city auto-selects. Reset emits oldTag/newTag so the parent renames matching routing rules in place; logout emits a remove-routing-rules event with prefix `nord-` to purge any dangling references. - XrayPage.vue: holds warpOpen / nordOpen flags, ensures the outbounds array exists before mutating it, and wires the modal events (add-outbound / reset-outbound / remove-outbound / remove-routing-rules) to in-place edits of templateSettings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): Phase 7 — vue-i18n wired up + login page translated Sets up vue-i18n on top of the panel's existing TOML translation files. The Go side stays the source of truth — translators continue to edit web/translation/*.toml; a sync script snapshots those files into per-locale JSON the Vue bundle imports. The login page is translated end-to-end as a worked example; remaining pages can be converted incrementally without infrastructure churn. What's in the box: - scripts/sync-locales.mjs: small TOML→JSON converter that walks web/translation/*.toml and writes frontend/src/locales/<code>.json. Handles the narrow subset of TOML the panel uses (flat key/value pairs + dotted [section.subsection] heads). Wired as a `prebuild` + `predev` script so production builds always include the latest strings without a manual step. - src/i18n/index.js: createI18n() in composition mode with all 13 locales emitted as their own Vite chunks. The active locale (read from the same `lang` cookie LanguageManager has always managed) plus the en-US fallback are eagerly loaded; the rest are dynamically importable via a loadLocale(code) helper. This keeps the per-page bundle the user actually downloads small — only ~30 KB of strings end up in the initial payload, vs ~220 KB if all 13 were eager. - All five page entries (index/login/settings/inbounds/xray) wire the i18n plugin into createApp via .use(i18n). - LoginPage.vue: t(...) replaces hardcoded English on the username / password / 2FA placeholders, the submit button label, and the Settings popover title. The Hello/Welcome headline cycle stays hardcoded — those are stylistic, not labels. The 'Hello'/'Welcome' cycle stays in English deliberately; the rest of the migration's components still ship hardcoded English and will be converted page by page in follow-up commits. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate page chrome — sidebar, save bars, tabs, summary cards Replaces hardcoded English with t() calls in the components every user sees on every page load. The translations themselves come from the existing TOML files via the sync script — no new strings, no new locale keys. Per component: - AppSidebar.vue: 5 menu titles (dashboard / inbounds / settings / xray / logout). Computed so the sidebar re-renders when the cookie-driven locale flips on reload. - IndexPage.vue: Quick actions card title + Logs / Backup / Up-to- date / Update buttons. - StatusCard.vue: CPU / Memory / Swap / Storage labels + logical-processors / frequency tooltips. - XrayStatusCard.vue: card title + error popover header + Stop / Restart / Switch xray action labels (kept the v-prefix version string as-is — it's content, not a label). - SettingsPage.vue: 5 tab titles + Save / Restart-panel buttons + unsaved-changes warning. - XrayPage.vue: 6 tab titles + Save / Restart-xray buttons + unsaved-changes warning. - InboundsPage.vue: 5 summary-stat card titles. - InboundList.vue: 10 column titles (computed for live locale), Add inbound / General actions buttons + every dropdown menu item, search placeholder, filter radio labels, popover titles (disabled / depleted / depleting / online), traffic + info popover row labels. Total: ~75 strings localised across 8 files. The remaining English labels live in the per-tab settings forms, the form modals (Inbound / Client / Outbound / Rule / Balancer / WARP / Nord), and the per-row table cell helpers — all incremental work that doesn't touch infrastructure. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): translate every remaining English string on the index page Closes the index page's i18n coverage. Combined with the page-chrome commit, every label users see on the dashboard is now sourced from the TOML translation files. Per file: - IndexPage.vue: loading-spinner tip (initial + dynamic). - BackupModal.vue: modal title, both list-item titles + descriptions ("Back up" / "Restore"), in-flight busy tips ("Importing database…" / "Restarting panel…"). - PanelUpdateModal.vue: modal title, update-available alert, current/latest version row labels, "Up to date" tag + label, primary action button. Modal.confirm now uses the translated panelUpdateDialog / panelUpdateDialogDesc with #version# substitution; success toast uses panelUpdateStartedPopover. - LogModal.vue: title slot ("Logs"). The Debug/Info/Notice/Warning/ Error log-level options stay literal — they're xray's wire values, not user-facing labels (matches the existing settings-page choice). - XrayLogModal.vue: title + Filter label. Direct/Blocked/Proxy stay literal for the same reason. - VersionModal.vue: modal title + xray-switch alert + per-file tooltip + "Update all" button + custom-geo collapse header. The Modal.confirm flows for switchXrayVersion + updateGeofile use translated dialog/desc with #version# / #filename# substitution. - CpuHistoryModal.vue: title slot. - CustomGeoSection.vue: routing-hint alert, Add / Update-all buttons, every column title (computed for live locale), copy/edit/download/ delete tooltips, copy toast, delete-confirm modal, empty-state text. - CustomGeoFormModal.vue: add/edit titles, OK/cancel labels, Type/ Alias/URL field labels, alias placeholder, all three validation toasts. Total: ~50 strings localised across 8 index-page files. The Hello / Welcome login headline cycle and a handful of literal xray wire values (Direct/Blocked/Proxy/log levels) are intentionally kept hardcoded. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n(frontend): Phase 7-c — translate settings, inbounds modals, xray tabs Continues the page-by-page translation pass started in cb37dd55 — runs every user-visible string on settings (General/Security/Telegram/Sub), inbounds (Client/QR/Info modals), and xray (Routing/Balancer/Rule/Warp/ Nord/Basics/Outbounds tabs) through useI18n. Updates the TOML→JSON sync script to escape `@` (vue-i18n parses it as a linked-format prefix) and refreshes all 13 locale files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): Phase 9 — restore index dashboard, fix login/CSRF, port legacy styles - Index dashboard regains the 8 cards that were lost in the SPA port (3X-UI panel info, Operation Hours, System Load, Usage, Overall Speed, Total Data, IP Addresses, Connection Stats), plus a Config button that shows the live xray config.json. Version display falls back through panelUpdateInfo → window.__X_UI_CUR_VER__ → '?' so dev mode isn't blank. - Xray config no longer hangs on load: useXraySetting surfaces failures instead of leaving a perpetual spinner, and the Vite dev proxy stops hijacking POST requests to migrated routes (only GETs get bypassed). - Inbound page no longer throws __asyncLoader/emitsOptions errors — inbound.js was missing imports (NumberFormatter, SizeFormatter, Wireguard) and InboundList kept emitting after unmount. - Login round-trip works after logout: a public /csrf-token endpoint bootstraps the SPA before authentication, axios caches the token module-level, and the dev 401 handler navigates to /login.html instead of reloading the dashboard into a redirect loop. - legacy.css mirrors the legacy panel's surface/text variables so dark and ultra-dark themes match main; every SPA entry imports it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray DNS section to match main branch DnsTab now exposes every field the legacy panel did — top-level toggles (tag, hosts, queryStrategy, disableCache/queryConcurrency, fallback strategy, client subnet), the servers table with per-row strategy and domain/expectIP/unexpectedIP overrides, and the Fake DNS pool. The new DnsServerModal covers the full add/edit flow and collapses to a bare string when the user only sets an address — matching the wire shape the legacy form emits for plain DNS entries like "8.8.8.8". Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): rebuild xray outbound modal with structured per-protocol forms Replaces the JSON textareas with the same shape the legacy panel uses: all 11 outbound protocols (vmess/vless/trojan/shadowsocks/socks/http/ mixed/wireguard/tun/dns/loopback/blackhole/freedom) get dedicated fields, every transport (TCP/KCP/WS/gRPC/HTTPUpgrade/XHTTP) gets its own panel, and TLS/Reality/sockopt/Mux are configured through the same controls as the inbound side. Brings the SPA outbound editor to parity with main so users no longer have to drop into raw JSON. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): bring inbound modal to full parity with main branch Switches the default protocol on add to VLESS, fixes a crash when adding a Mixed account (the constructor is SocksAccount, not MixedAccount), and fills in the fields the SPA was previously delegating to the Advanced JSON tab: - TLS: cipher suites, min/max version, reject SNI / disable system root / session resumption switches, the certificate array with per-row Path-or-Content toggle (Set Default pulls from /panel/setting/ defaultSettings), One Time Loading, Usage / Build Chain, plus ECH key/config with a Get New ECH Cert button. - Reality: xver, target/SNI sync icons (uses getRandomRealityTarget), max time diff, min/max client version, short IDs randomizer, SpiderX, mldsa65 seed/verify with Get New Seed. - Stream: full structured forms for every transport — TCP HTTP camouflage gets its request/response editor, mKCP gets MTU/TTI/uplink/ downlink/CWND/maxSendingWindow, WebSocket / gRPC (now with Authority) / HTTPUpgrade get headers + proxy-protocol toggles, XHTTP gets the full SplitHTTPConfig surface (mode-aware fields, padding obfs, session/sequence placement, uplink data, no-SSE). - New External Proxy section and a structured Sockopt block (mark, TCP keepalive/timeout/clamp, fast open, MPTCP, penetrate, V6Only, domain strategy, congestion, TProxy, dialer/interface, trusted XFF). - VLESS gets the legacy X25519 / ML-KEM-768 buttons that fetch fresh decryption/encryption blocks via /panel/api/server/getNewVlessEnc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): add FinalMask UI (TCP/UDP masks + QUIC params) to inbound and outbound Mirrors web/html/form/stream/stream_finalmask.html as a shared FinalMaskForm component used by both modals — they share the same StreamSettings shape (addTcpMask/addUdpMask/finalmask/enableQuicParams) so a single template handles both. Surfaces: - TCP masks for raw/tcp/httpupgrade/ws/grpc/xhttp networks: fragment, sudoku, and header-custom (with the 2D clients/servers groups, each row supporting array/str/hex/base64 packets and a randomize button for base64). - UDP masks for hysteria protocol or kcp network: hysteria gets just salamander; kcp gets the full type list (mkcp variants, header-*, xdns/xicmp, header-custom with flat client/server lists, and noise). Switching to xdns shrinks the kcp MTU to 900 to match the legacy panel's behavior. - QUIC Params for hysteria or xhttp: congestion (incl. brutal up/down fields), debug, UDP hop ports/interval, idle/keepalive timeouts, path-MTU discovery toggle, and the four receive-window tunables. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): remove duplicate Outbound test URL from xray Advanced tab The Basics tab already exposes this field through BasicsTab — duplicating it on the Advanced tab let two inputs race the same ref and only added clutter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): unify theming on vanilla AD-Vue light/dark/ultra-dark The legacy panel CSS (custom.min.css ported as legacy.css) tinted every non-primary button teal-green via .dark .ant-btn:not(.ant-btn-primary) overrides while AD-Vue 4's darkAlgorithm kept primary buttons blue — producing the mixed blue/green button look on dark mode. Drop legacy.css entirely and let AD-Vue 4's algorithms own the palette. Centralize antdThemeConfig in useTheme.js so every page resolves to the same source of truth (light = defaultAlgorithm, dark = darkAlgorithm, ultra-dark = darkAlgorithm + deeper colorBgBase/Layout/Container/ Elevated tokens). Each page's <a-config-provider> now imports the shared computed instead of defining its own copy. Drops the 67 KB legacy CSS chunk; per-page CSS bundles fall to ≤5.9 KB. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): restore computed import in Settings + Xray pages When 5f1aba28 dropped the local antdThemeConfig computed (now shared from useTheme), it also stripped `computed` from the import list — but both pages still call computed() elsewhere (confAlerts, advanced-tab helpers). Re-adds it. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): retheme dashboard gauges to AD-Vue blue and shrink them - StatusCard's CPU/RAM/Swap/Storage dashboards rendered at AD-Vue's default 120px width which made the percent text balloon to ~36px. Drop to 90px (70px on mobile) so the gauge fits the rest of the card. - The CurTotal.color thresholds still hardcoded the legacy teal/orange palette (#008771 / #f37b24 / #cf3c3c). Switch to AD-Vue's primary / warning / danger tokens (#1677ff / #faad14 / #ff4d4f) so the gauges match the rest of the panel under both light and dark themes. - XrayStatusCard's running-animation badge ring also still pointed at the deleted --color-primary-100 var; hardcode the new primary blue. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: shorten backupTitle to "Backup & Restore" across all 13 locales The backup modal header was the second-longest title in the dashboard on every locale ("Database Backup & Restore" / "Резервне копіювання та відновлення бази даних" / etc). Drop the "Database / Veritabanı / 数据库" qualifier — the modal already lives under the "Database" column, so the shorter form reads cleaner on narrow viewports. Updated both the .toml source-of-truth files and the synced .json locales (re-running scripts/sync-locales.mjs). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * i18n: collapse two translation databases into a single web/translation/<lang>.json set The Vue SPA had been reading from frontend/src/locales/*.json while the Go binary still loaded web/translation/translate.*.toml — and a sync-locales.mjs pre-build step kept the two in lockstep, with TOML as the source of truth. Now that go-i18n v2.6.1 already flattens nested JSON via recGetMessages/addChildMessages, both runtimes can share one file per locale. - Move the 13 nested-JSON locale files to web/translation/<lang>.json so they live alongside the Go //go:embed translation/* directive. - Switch web/locale/locale.go from toml.Unmarshal to json.Unmarshal (and drop the pelletier/go-toml import — it's now indirect-only). Confirmed via a smoke test that pages.index.cpu, subscription.title, tgbot.commands.help, and menu.settings all resolve in en-US, fa-IR, ru-RU, and zh-CN. - Repoint Vue's i18n loader at the new path (../../../web/translation/ *.json glob) and drop the moved-here pathDelimiter comment that no longer applies. - Delete the 13 legacy translate.*.toml files and the sync-locales.mjs script + its npm pre-script hooks (predev/prebuild/i18n:sync). The Telegram bot and subscription page still get their messages because they were reading the same MessageIDs the JSON files now produce. - Update copilot-instructions.md so the next contributor knows where the canonical translation files live. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): redesign expand-row + retheme client visuals When you expanded an inbound row, the nested <a-table> inside ClientRowTable burst out of the parent's scroll-x box — its .ant-spin-container ended up wider than the parent's narrow .ant-table-cell, so the child looked oversized while the parent looked squeezed. Replace the nested table with a CSS-grid layout that owns its sizing, sits flush inside the expanded cell, and collapses to a 3-column layout on mobile (action menu, client identity, info popover). While in there, fix three other client-row visuals: - The Unicode infinity glyph (U+221E) renders as an "m"-shaped character in some system fonts (Windows Segoe UI in particular). Add a shared <InfinityIcon /> SVG component (legacy panel's path) and use it in ClientRowTable, InboundList, and InboundInfoModal — desktop and mobile cells. - The "unlimited quota" traffic bar passed :percent="100" with no stroke-color, so AD-Vue auto-coloured it success-green. Pin it to the AD-Vue purple token (#722ed1) so it reads as the no-limit sentinel rather than another usage state. - ColorUtils + the in-row statsExpColor still hardcoded the legacy teal/orange/red/purple palette (#008771 / #f37b24 / #cf3c3c / #7a316f). Map them onto AD-Vue 4's success/warning/danger/purple tokens (#52c41a / #faad14 / #ff4d4f / #722ed1) so badges, tags, and progress bars all match the rest of the panel. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): darken light-theme page bg so cards stand out The light-theme --bg-page was #f0f2f5 — close enough to AD-Vue's #fff card background that the cards faded into the page. Bump it to #e6e8ec (a more visibly distinct gray) so cards lift cleanly off the surface. Dark and ultra-dark stay where they were. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): shrink dashboard percent text and surface the unfinished arc Two follow-up tweaks to the dashboard gauges: - AD-Vue scales the percent text from the SVG, not from :width, so the 90px gauges still rendered the number at ~27px. Pin .ant-progress-text to 14px via :deep() and trim the gauge to 70px (60px on mobile) so the whole card stays compact. - The default trail (rgba(0,0,0,0.06) / rgba(255,255,255,0.08)) was invisible on the light-theme card. Pass an explicit rgba(128,128,128,0.25) trail-color so the unfinished portion is visible under both light and dark themes. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): migrate subpage.html to Vue 3 SPA The subscription info page was the last page still rendered by Go templates. Move it to the Vite multi-page setup so the whole panel loads through one toolchain. Frontend: SubPage.vue mounts at /sub/<id>?html=1 and reads window.__SUB_PAGE_DATA__ for the parsed view-model (traffic / quota / expiry + rendered share links). Fix descriptions borders against the light-theme card by painting the row divider on each cell's bottom edge — AD-Vue's <tr> border doesn't render reliably under border-collapse:collapse. Backend: serveSubPage reads dist/subpage.html, injects window.__X_UI_BASE_PATH__ + window.__SUB_PAGE_DATA__ before </head>, and rewrites Vite's absolute /assets/ URLs when the panel runs under a URL prefix. Drop the legacy template-FuncMap wiring and switch the sub server's static mount from web/assets to web/dist/assets. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): inbound modal QR + tabs + restored TLS fallbacks Per-client QR action: the qr icon on the expand-row table opened the big info modal instead of the QR modal. Route it to QrCodeModal and extend that modal with a `client` prop so genAllLinks() produces the per-client share URLs (and per-peer remarks for WireGuard). Inbound's Data redesign: split the dense single-page view into three tabs — Inbound, Client, Subscription. Drop every QR rendering from this modal (QrCodeModal is the QR home now). Each row in the Inbound tab is one label/value pair instead of the legacy 2-column grid, and long values like the VLESS encryption blob render as a wrapping code block with a copy button so they can't blow out the dialog. The Subscription tab renders sub URL + JSON URL as clickable anchors that open in a new tab. Restored TLS fallbacks UI: the model already exposed VLESSSettings.Fallback / TrojanSettings.Fallback with addFallback / delFallback / fallbackToJson, but the form modal never surfaced them during the Vue 3 migration. Re-add the legacy form (SNI, ALPN, Path, Destination, PROXY) on the protocol tab, gated on TCP transport plus (for VLESS) encryption=none — same conditions as main. Column widths: Protocol 70→130 and All-time Traffic 60→95 in the inbound list; All-time Traffic 90→130 in the client expand-row, so the header text fits and tags don't get squeezed. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): navy dark theme + rounded inbound/client corners Dark theme picks up a refined navy palette (page #0a1426, cards #142340, sider #0d1d33) so the sidebar blends with the rest of the surface; ultra-dark stays neutral black. Resolves the previous mismatch where AD-Vue 4 hardcoded #001529 / #002140 for the sider, trigger and dark Menu items via Layout.colorBgHeader / colorBgTrigger and Menu's colorItemBg — overrides go through the component-token map now. Round the inbound table's outer corners (header start/end + last row end) and wrap the client expand-row grid in a 1px / 8px-radius border so the list reads as a contained block instead of a flush rectangle. Linter-driven whitespace cleanup across inbounds/*.vue rolled into the same commit since it can't be split out cleanly. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): xray tab fixes — modal close, tag validation, full XHTTP, reset to default Modal close: BalancersTab / OutboundsTab / RoutingTab confirmDelete used arrow expressions that returned splice's removed-items array. AD-Vue 4 treats truthy non-thenables from onOk as "still pending" and never closes the dialog (see ActionButton.js:103-106), so the confirm modal stayed open. Wrap the body so onOk returns undefined and AD-Vue auto-closes. Tag validation: outbound + balancer modals only flipped between warning/success on duplicate, leaving the empty case as a green ✓. Split into a 3-state computed — error (empty) / warning (duplicate) / success — and wire a help message so the input clearly explains why the OK button is disabled. Reset to default: re-add the legacy "Reset to Default" panel at the bottom of BasicsTab. Calls /panel/setting/getDefaultJsonConfig and overwrites templateSettings; the existing watch re-stringifies so the JSON tab + dirty-poll see the new state. Restored Basics option lists from main: IPs (4→10, +Vietnam/Spain/ Indonesia/Ukraine/Türkiye/Brazil), DomainsOptions (4→10, +regex entries), BlockDomainsOptions (5→17, +Malware/Phishing/Adult/regex), ServicesOptions (Reddit/Speedtest in, off-template Microsoft out). Outbound form parity with main: • Reverse Sniffing UI for VLESS — toggle + destOverride checkboxes (HTTP/TLS/QUIC/FAKEDNS) + Metadata/Route Only + IPs/Domains excluded multi-selects, gated on reverseTag being set. • Full XHTTP transport — request headers list, Max Upload Size / Min Upload Interval (packet-up), Padding Obfs Mode + sub-fields, Uplink HTTP Method, Session/Sequence/UplinkData placement + keys, No gRPC Header (stream-up/stream-one), expanded XMUX with Max Concurrency/Connections/Reuse/Request/Reusable/Keep-alive. Strip a-divider from the outbound form per request — replaced with plain section/item heading divs so the labels and per-row delete icons stay but the horizontal rule is gone. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): xray Advanced tab parity + finalmask gating Advanced tab was a single textarea bound to the full xraySetting blob. Restore the legacy 4-way view: a radio group toggles between All / Inbounds / Outbounds / Routing Rules, and the textarea reads/writes the matching slice through templateSettings. Added the legacy header ("Advanced Xray Configuration Template" + description) so the page introduces itself like main. Outbound finalmask leaked into protocols that don't have a stream (Freedom / Blackhole / DNS / Socks / HTTP / Wireguard) because the v-if only checked outbound.stream. Gate the whole FinalMaskForm on outbound.canEnableStream() to match main. Drop the leading divider inside FinalMaskForm — its parent already provides separation, so the rule above "TCP Masks" was redundant. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound Advanced tab live mirror + QR exact-fit sizing Advanced tab in the inbound modal showed stale state. The watch only refreshed advancedJson.stream, so toggling the Sniffing switch in the Sniffing tab left the Advanced JSON showing the prior value. And encryption — stored on inbound.settings.encryption, not on stream — never appeared at all because Advanced only exposed stream + sniffing. Split the watch into three (stream / sniffing / settings) and add a settings textarea so encryption / clients / fallbacks live alongside the existing two views. The submit() path now reads settings from the JSON tab too (falling back to inbound.settings.toString()) so power-user edits in Advanced override the structured form on save. QR canvas: when a longer share-URL bumps the QR matrix size, QRious falls back to floor(canvasSize / matrixWidth) and centers the pattern, leaving a white margin (e.g. matrix=41, size=180 → 8px gap). Pre-pick the QR version from the URL byte length and set canvas size to a multiple of matrixWidth × pixelSize so the pattern always fills it edge-to-edge — no white margin even after toggling encryption on. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound stream tidy-up + QR sizing + dev proxy Stream tab clean-up: drop the seven a-divider rules in the inbound form's Stream tab — replace the labelled ones (Request / Response / Security) with a section-heading div that matches the outbound modal, delete the empty rules above TLS sub-blocks / External Proxy / Sockopt. Empty header-list form-items also leaked margin space below each "Add header" button across TCP / WS / HTTPUpgrade / XHTTP — gate each on headers.length > 0 so they vanish until the user adds one. QR panel: drop the link text under the canvas (the user already has a copy button on the header). Pin the canvas display size to a fixed 240px square via :style + image-rendering: pixelated/crisp-edges so a dense WireGuard config QR and its sparser link share the same on-screen footprint without blurring. Dev proxy: Node's AggregateError wraps connection failures whenever DNS returns more than one address (::1 + 127.0.0.1) and the code lands on the inner errors, not the outer. The existing handler only checked err.code so the ECONNREFUSED stack still spammed the log when the Go backend was down. Walk err.errors too, print one friendly line ("backend not reachable — start the Go server"), then stay quiet for the rest of the session. Vendor splitting + chunk-size warning: split node_modules into stable vendor-* chunks so each page only ships the deps it uses and the browser caches them across versions. ant-design-vue stays as a single chunk because its components share internals; raise the chunk-size warning to 1500kB so the build stays quiet (its 1.4MB minified gzips to ~410kB on the wire). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): info-modal cleanup + 2FA QR + outbound link import - 2FA QR: matrix-snap canvas + opaque background to drop white margin - Inbound info modal: stack Mixed/HTTP/Tunnel as info-rows, hide tab strip when only the Inbound tab applies - Add inline VLESS Reverse tag input on first-client form - Hide Protocol tab for TUN (no form yet) - Outbound link converter: route through Outbound.fromLink so vless/trojan/ss/hysteria(2) imports work alongside vmess; fix stray implicit global in fromLink Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): jalali calendar + drop legacy moment-jalali - Wire Calendar Type setting to a real Jalali datepicker via vue3-persian-datetime-picker, gated by useDatepicker composable - DateTimePicker wrapper swaps between AD-Vue and Persian picker; keeps dayjs v-model contract so existing forms/setters work unchanged - Theme picker popup explicitly per body.dark / data-theme=ultra-dark (AD-Vue 4 doesn't expose CSS vars, so var() fallbacks defaulted to white); fix invisible disabled days, SVG arrow fills, popup clipping via append-to="body" - Replace stray moment() calls in dbinbound/inbound models with dayjs; the legacy global was undefined under ESM and broke the inbounds list whenever any inbound had expiryTime > 0 - Remove legacy moment-jalali / persian-datepicker / aPersianDatepicker assets — replaced by the Vue 3 picker Note: dark/ultra background of the date popup still renders white in some cases — pending follow-up. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): jalali popup theming + full-month layout - Re-prefix popup selectors with .vpd-wrapper (popup root that travels with appendTo='body'), not .vpd-main (which stays at the input); paints the popup's dark/ultra background again - Drop the 1px border on .vpd-content — with box-sizing: border-box it ate 2px from the day-row width, wrapping the 7th cell of every row and hiding days 18-31 of months that needed a 5th week Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat: render dates in Jalali when Calendar Type is jalalian - IntlUtil.formatDate accepts an optional calendar arg; appends the BCP-47 -u-ca-persian extension so Intl renders Jalali across all UI languages, not just fa-IR - Plumb the panel's datepicker setting into the SubPage via the Go injection (window.__SUB_PAGE_DATA__.datepicker) - Panel pages (inbound list/info, client row, xray log) read the same setting through the useDatepicker composable so the whole panel stays consistent Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(frontend): ultra-dark page tint + mobile-friendly inbound view - Drop --bg-page from #21242a (lighter than the cards) to #050505 in ultra-dark across index/sub/settings/inbounds/xray, so cards consistently elevate over the page - Hide the inline sider's children + collapse-trigger and zero its width below 768px; the floating drawer-handle remains the menu trigger - Inbounds page mobile pass: tighten content-area + card padding; flex-wrap the filter bar instead of stacking; shrink table cell padding so all 4 mobile columns fit; bump expand / action / info icon hit targets - Per-client expand row on mobile: soft-tinted rounded cards instead of hairline borders, larger action / info touch targets, more legible email typography, bigger status badge dot Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove legacy template + asset trees and dead Go template engine - Delete web/html/ entirely (page templates, form/, modals/, component/, common/, settings/) — every route is served from web/dist/ now via serveDistPage; nothing in the binary referenced these - Delete web/assets/ entirely (jQuery-era ant-design-vue, axios, moment, codemirror, qrcode/qs/uri/vue/otpauth, custom CSS, Vazirmatn font); Vite bundles all of this into web/dist/assets - Drop the Gin HTML template wiring: remove //go:embed assets + //go:embed html/*, the assetsFS/htmlFS vars, the wrapAssetsFS adapter, EmbeddedHTML / EmbeddedAssets exports, getHtmlFiles / getHtmlTemplate, the i18nWebFunc/funcMap and SetFuncMap call, and the dev/prod template-engine branch — only StaticFS for /assets/ is needed now - Remove dead html()/getContext() helpers and unused imports from web/controller/util.go (no c.HTML(...) callers remain) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(frontend): inbound expand chevron position + cpu history layout - Push the inbound table's expand chevron away from the left edge with margin-inline + cell padding so it isn't flush against the corner - Move "Timeframe: …" caption above the chart (was below); restore the line that the previous edit removed - Fix x-axis time labels being clipped at the bottom of the cpu chart — the offset (paddingTop+drawHeight+22 = 222) exceeded the SVG viewBox height (220); dropped to +14 so labels sit at y=214 with room for descenders - Move the SVG axis text colors out of <style scoped> into a global block — Vue's scoped CSS doesn't always hash-attribute SVG <text> descendants, so the dark-mode overrides via :global() weren't matching; bumped opacity 0.55 → 0.85 for legibility on navy/black Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(login): language picker in settings popover + fluid card sizing - Add language select alongside the theme switch (mirrors SubPage) - Bind headline to pages.login.hello / pages.login.title so the "Hello / Welcome" cycle re-translates with the active locale - Replace AD-Vue 5-breakpoint grid with clamp() sizing so the card scales smoothly instead of jumping ~33% at each breakpoint - Pin horizontal padding so input width stays stable on large viewports Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): organize entry HTML + bootstrap JS into folders - Move entry HTML files: frontend/*.html -> frontend/html/*.html - Move per-page bootstrap modules: src/{index,login,settings,inbounds,xray,subpage}.js -> src/entries/ - Update vite.config rollup inputs and dev-mode MIGRATED_ROUTES to /html/<page>.html - Build output now lands at web/dist/html/<page>.html - serveDistPage and subController updated to read from dist/html/ Cleans up the flat frontend/ root which previously interleaved 6 HTML files with package.json, README, src/, etc. The src/ root similarly gets rid of 6 entry .js files mixed in alongside api/, components/, models/, etc. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * chore: remove obsolete vue3 phase1 inventory doc The migration is well past phase 1 — the inventory doc has rotted and the live state lives in the codebase plus the plan files. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * refactor(frontend): merge utils/legacy.js into utils/index.js The barrel was a placeholder for an eventual split that hasn't happened. Collapsing the two files removes one layer of indirection and the misleading "legacy" name (the contents are still actively used by the migrated SPA). - Move all 930 lines from utils/legacy.js into utils/index.js - Delete utils/legacy.js - Update direct import in models/outbound.js to '@/utils' - Drop a stale legacy.js reference in InboundFormModal comment Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * revert(frontend): keep entry HTML files at frontend/ root The earlier move to frontend/html/ made dev-mode URLs ugly (http://localhost:5173/html/index.html instead of plain /). The folder didn't add real value — it just hid 6 files behind a non-conventional layout. Reverting that piece while keeping src/entries/ (which is a genuine separation between page bootstrap and the rest of src/). - HTML files back at frontend/<page>.html - Vite rollupOptions.input + MIGRATED_ROUTES restored to flat paths - Build output is web/dist/<page>.html again - web/controller/dist.go and sub/subController.go read from dist/<name> Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): bump eslint to 10 + add flat config + clean lint warnings - Upgrade eslint 9.39 -> 10.3 and eslint-plugin-vue 9.33 -> 10.9 - Add eslint.config.js (flat config required by ESLint 10) with vue3-recommended rules, sensible defaults, and exemptions for the project's existing formatting style - Drop --ext from the lint script (removed in ESLint 10) - vue/no-mutating-props is left off because the form-modal pattern ports straight from Vue 2 (parent passes a reactive object, child mutates it); a real fix is an architectural rewire, separate task Lint warning cleanup: - utils/index.js: var -> let/const in the X25519 routines, replace obj.hasOwnProperty(...) with Object.prototype.hasOwnProperty.call(...) - Remove unused imports (reactive, ref, Inbound) in ClientFormModal, InboundInfoModal, QrCodeModal, DnsServerModal, OutboundFormModal, SubPage; remove unused locals (isClientOnline, ONLINE_GRACE_MS, fetchAll, isSocks, isHTTP, _antdAlgorithm) - XrayStatusCard: declare 'open-logs' on defineEmits (was emitted but not declared) - RuleFormModal: rename v-for var t -> tag (shadowed useI18n's t) - Drop stale eslint-disable directives (no-new, no-unused-vars) - OutboundsTab/InboundList: drop redundant initial null assigns - InboundInfoModal/OutboundFormModal: explicit eslint-disable for the intentional local-ref-shadows-prop pattern in modal drafts `npm run lint` now passes with 0 errors and 0 warnings. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): one client identity across multiple inbounds via subId Lets the operator add the same email under the same subId to several inbounds. Xray reports traffic per email, so a single client_traffics row acts as the shared accumulator — no aggregation overhead, quota and expiry stay consistent. - Email validation allows duplicates only when subId matches - AddClientStat upserts via OnConflict DoNothing (idempotent on rerun) - Stat/IP rows survive client deletion when a sibling inbound still references the email - enrichClientStats tops up GORM-preloaded stats with rows whose inbound_id points at a sibling, so every panel view sees usage - disableInvalidClients cascades enable=false and syncs the row's total/expiry into every sibling JSON when the shared identity expires - DelDepletedClients removes the depleted client from all referencing inbounds, batched - Subscription services dedupe traffic by email so shared quota is counted once Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * docs(frontend): rewrite README for multi-page Vue 3 layout Reflects the current state — embedded build, per-route HTML entries, ESLint 10 flat config, src/ layout, and the steps to add a new page. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * build(frontend): drop deprecated rimraf/glob/inflight transitive deps vue3-persian-datetime-picker pinned moment-jalaali to ^0.9.4, which pulled rimraf@3 → glob@7 → inflight@1. inflight in particular leaks memory and is unmaintained. Override moment-jalaali to ^0.10.4 (same runtime API, dropped the legacy build deps) so npm install no longer warns and the dep tree is 12 packages lighter. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(nodes): multi-node panel orchestration (CRUD, deployment, traffic sync, sub per-node) - Node model + service + controller (/panel/api/nodes/*) with bearer-token apiToken auth - Heartbeat job @every 10s; status/latency/xrayVersion surfaced in Nodes UI - Runtime abstraction (Local + Remote) so inbound/client mutations target the inbound's owning node instead of always hitting the local xray - Inbounds gain optional NodeID; tag-based correlation with remote panel (no RemoteInboundID column needed) - NodeTrafficSyncJob @every 10s pulls absolute counters + online/lastOnline from each enabled+online node and writes them into central DB; 30s reset grace window prevents post-reset overwrite - Reset propagation to nodes (best-effort) on client/inbound/all reset paths - Subscription server uses node.Address for inbounds with NodeID, falling back to existing host resolution for local inbounds - Frontend: Nodes page, "Deploy to" select in inbound form, Node column on inbound list, hostOverride threaded through genAllLinks/QR/Info modals Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(stats): system history modal + per-node CPU/Mem trends across all locales Backend - web/service/metric_history.go: generic in-memory ring buffer with two singletons — system-wide (cpu/mem/netUp/netDown/online/load1/5/15) and per-node (cpu/mem) keyed by node id - ServerService.AppendStatusSample writes all 8 metrics every 2s on the same tick; AppendCpuSample/AggregateCpuHistory kept for back-compat - NodeService.UpdateHeartbeat appends cpu/mem only on online ticks so offline gaps render as missing data, not phantom dips - New routes: GET /panel/api/server/history/:metric/:bucket and GET /panel/api/nodes/history/:id/:metric/:bucket, both whitelisted Frontend - Sparkline component generalized: arbitrary value range (auto-scale when valueMax=null), pluggable yFormatter/tooltipFormatter for B/s, client counts, load averages - SystemHistoryModal replaces CpuHistoryModal with tabs for every metric; opened from a tag on the 3X-UI card next to Documentation - NodeHistoryPanel: expandable row on the Nodes table showing per-node CPU and Mem trends, refreshed every 15s Localization - Backfill systemHistoryTitle / trendLast2Min / pages.inbounds.{node, deployTo, localPanel} and the entire pages.nodes block (51 keys including statusValues + toasts) into all 11 non-en/fa locales: ar-EG, es-ES, id-ID, ja-JP, pt-BR, ru-RU, tr-TR, uk-UA, vi-VN, zh-CN, zh-TW Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(embed): include underscore-prefixed Vite chunks in dist FS go:embed silently excludes files whose names start with `_` or `.`, so the `_plugin-vue_export-helper-<hash>.js` chunk that Vite/rolldown emits for @vitejs/plugin-vue was missing from the production binary. First import at runtime hit a 404 and the SPA failed to mount — blank page on every page load, no error in the server logs because the asset 404 was just a static-handler miss. Switched the directive to `//go:embed all:dist` which keeps the same root layout but disables the underscore/dot exclusion rule. Dev mode was unaffected (it serves dist/assets/ from disk, not the embedded FS). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * ci: build frontend bundle before Go compile in release.yml + Dockerfile Phase 8 cut all panel HTML routes over to web/dist/ and embedded the Vite bundle into the Go binary via //go:embed all:dist. web/dist/ is .gitignored, so on a fresh CI checkout it doesn't exist — every Go build since Phase 8 has been failing with "pattern dist: no matching files found" or producing a binary that 404s on first asset request. release.yml: add a setup-node@v4 + npm ci + npm run build trio before the existing go build step in both the Linux matrix job (7 arches) and the Windows job. npm cache is keyed on frontend/package-lock.json. Dockerfile: add a node:22-alpine frontend stage that runs npm ci + npm run build and emits to /src/web/dist (via vite.config.js's outDir). The golang builder stage then COPY --from=frontend /src/web/dist into ./web/dist before the go build, so embed.FS sees the bundle. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(ws): live updates on inbounds/xray/nodes pages, drop polling + manual refresh Replaces the legacy polling + manual-refresh model with WebSocket pushes across the three live-data pages. The hub already broadcast traffic / client_stats / outbounds; this wires the frontend to consume them and adds a new `nodes` channel for the heartbeat job's snapshot. Frontend - new useWebSocket composable: page-scoped singleton WebSocketClient, lifecycle-managed on/off, leaves disconnect to page-unload - inbounds: useInbounds gains applyTrafficEvent / applyClientStatsEvent / applyInvalidate that merge counters and online/lastOnline in place; InboundsPage subscribes; InboundList drops the auto-refresh popover, the refresh button, and the now-unused refreshing prop - xray outbounds: useXraySetting gains applyOutboundsEvent; XrayPage subscribes; OutboundsTab drops the refresh button + emit - nodes: useNodes gains applyNodesEvent and stops the 5s setInterval/visibilitychange polling; NodesPage subscribes; NodeList drops the refresh button and ReloadOutlined import Backend - web/websocket: new MessageTypeNodes + BroadcastNodes notifier - node_heartbeat_job: after wg.Wait(), reload the table once and BroadcastNodes(updated). Gated on websocket.HasClients() so a panel with no open browser doesn't spend the DB read Bug fixes spotted in this pass - websocket.js #buildUrl defaulted basePath to '' when the global was missing (dev mode), producing `ws://host:portws` and a SyntaxError on the WebSocket constructor. Fall back to '/' and ensure leading slash. - vite.config.js: forward /ws to ws://localhost:2053 with ws:true so dev (5173) reaches the Go backend's WebSocket - NodeFormModal: a-input-password's visibilityToggle is Boolean in AntD Vue 4; the v3-era object form (`{ visible, 'onUpdate:visible' }`) triggered a Vue prop-type warning. Drop the override (default true shows the eye icon and toggles internally) and remove the orphaned tokenVisible ref Translations - pages.inbounds.autoRefresh / autoRefreshInterval: removed from all 13 locales (UI gone) - pages.nodes.refresh: removed from all 13 locales (UI gone) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * feat(inbounds): hide Node column when no nodes are defined Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-09 08:38:48 -07:00
}
2025-09-20 00:35:50 -07:00
// IsNeedRestartAndSetFalse checks if restart is needed and resets the flag to false.
2023-02-09 11:18:06 -08:00
func (s *XrayService) IsNeedRestartAndSetFalse() bool {
2023-03-17 09:07:49 -07:00
return isNeedXrayRestart.CompareAndSwap(true, false)
2023-02-09 11:18:06 -08:00
}
// ApplyPendingRestart re-arms the original flags when applying fails, so the next tick retries; a
// reconcile retry never becomes a restart, which would disconnect unrelated clients.
Repo-wide self-correcting audit: 54 verified bug fixes (#5970) * fix(email): resolve a name-addr smtpFrom into bare envelope address and display name The save-time validator accepts any RFC 5322 address form, so a value like '3x-ui Panel <panel(at)example.com>' passes validation, but Send and TestConnection fed that raw string to MAIL FROM, which strict servers reject with 501, and buildMessage mangled it into a quoted local part. Parse the configured sender at the point of use: the envelope gets the bare address and, when no explicit sender name is set, the display name embedded in the setting is used for the From header. * fix(email): report a missing sender address from the SMTP connection test TestConnection skipped the empty-from guard that Send enforces, so with no sender and no username configured the test issued the null reverse-path and could report success against a lenient relay while every real notification send kept failing with the missing-sender error. Guard the test path the same way and surface a dedicated translated message. * fix(sub): fall back to the raw subscription when an auto-detected format has no content With format auto-detection enabled, a client whose User-Agent matched the Clash or JSON regex was routed straight to that format handler. For a subscription whose entries convert to neither format (an MTProto-only subscription, for example) the handler returns an empty document and the request ended as 404, breaking a URL that served the raw list before the toggle. The auto-detect branches now serve the detected format only when it produces content and otherwise continue to the raw response; the explicit format endpoints keep answering 404 for empty documents. * fix(node): match prefixed central tags when filtering a selected-mode node snapshot FilterNodeSnapshot compared a node snapshot's inbound tags against the raw selected-tag list with an exact match, while its two siblings (SnapshotHasUnadoptedInbounds and the reconcile tagToCentral map) expand each selected tag to both its bare node-side form and its n<id>- prefixed central form. A panel-created node inbound is recorded in the selected list under the central prefixed tag but reported by the node under the bare tag, so the exact match dropped it from every snapshot and the orphan sweep then deleted its central row one tick after creation. Expand the allowed set with the same prefix flip the siblings use. * fix(client): refuse a bulk quota reduction that would fall to or below zero BulkAdjust clamped a client's new traffic limit with max(total+addBytes, 0). Because 0 is the unlimited sentinel, reducing a client's quota by more than it had left silently granted that client unlimited traffic. The sibling expiry branch already refuses an over-reduction; mirror it for quota so the adjustment is skipped with a clear reason instead of crossing the sentinel. * fix(client): persist a bulk adjustment's applied field even when the sibling field is skipped In a mixed BulkAdjust (both a days delta and a bytes delta), a per-field planning skip such as "unlimited expiry" or "unlimited traffic" was recorded in the same map that gated the client_traffics write. The applied field was already written to the inbound JSON and the clients table, but the enforcement row was left untouched, so the depletion job cut the client on the old limit while the panel showed the new one. Gate the traffic-row write on an actual inbound-processing failure rather than on any planning-phase skip note. * fix(inbound): always create in AddInbound instead of overwriting a row whose id was posted The add controller binds the inbound model's id form field and never clears it, and AddInbound persisted with GORM Save, which updates in place when the primary key is non-zero. A client that reused an existing id (for instance by duplicating an inbound fetched from /get and changing the port) silently overwrote that stored row instead of creating a new inbound. Zero the id at the top of AddInbound, matching how it already zeroes the client-stat ids. * fix(inbound): accept WireGuard clients when creating an inbound AddInbound's per-client validation switch had cases for every protocol except WireGuard, so a WireGuard client fell through to the default branch that requires a non-empty id. WireGuard clients are keyed by their public key and carry no id, so importing a WireGuard inbound or re-adding one to a reconciling node was rejected with "empty client ID". Add a wireguard case that validates the client key, mirroring addInboundClient. * fix(client): stop holding the inbound-lock registry mutex while waiting on one inbound lockInbound acquired the global registry mutex and then blocked on the per-inbound mutex without releasing the registry first. A slow client operation holding one inbound's mutex (for example a bulk delete pushing to an unreachable node) made the next waiter park on that inbound while still holding the registry mutex, which in turn blocked lockInbound for every other inbound — freezing client mutations panel-wide. Release the registry mutex before taking the per-inbound lock. * fix(client): honor keepTraffic when deleting a client that is attached to inbounds Delete, DeleteByEmail and BulkDelete all pass keepTraffic to their final cleanup transaction, but each called the per-inbound delete helper with a hardcoded false. That helper purges the client's traffic, IP and stat rows before the gated cleanup runs, so keepTraffic=true still destroyed all traffic history for any client actually attached to an inbound (the pinned test only covered a record with no inbound mappings). Thread the caller's keepTraffic through to the per-inbound helper at all three call sites. * fix(inbound): defer a local MTProto inbound edit's sidecar push until after commit UpdateInbound applied a local MTProto inbound change by calling the runtime UpdateInbound (which stops/starts the mtg sidecar or talks to it) from inside runSerializedTx. That runs process and network I/O on the single traffic-writer goroutine while a DB transaction is open, so a slow sidecar stalls traffic accounting and every concurrent client mutation, and a later step failing the transaction leaves the sidecar ahead of the rolled-back row. Move the push into the post-commit hook, matching the xray branch. Adds a SetLocalRuntimeOverride test seam mirroring the existing node override so the deferral is regression tested. * fix(client): delete external-link rows when bulk-deleting clients The single-client Delete path removes a client's client_external_links rows, but BulkDelete (and the DelDepleted reaper that routes through it) deleted the record, mappings and traffic while leaving the external-link rows keyed by the now-dead client id, so they accumulated as orphans. Delete them in the same cleanup transaction, keyed by client id like the single path. * fix(inbound): request an xray restart when toggling a routed MTProto inbound AddInbound, DelInbound and UpdateInbound all flag needRestart when an inbound routes MTProto through xray, so the egress SOCKS bridge is regenerated. Only SetInboundEnable's local path omitted it, so toggling a routed MTProto inbound off then on left the bridge out of the running config while the sidecar dialed its loopback port, blackholing that inbound until an unrelated restart. Flag the restart on the local enable path too. * fix(client): apply enable-by-email to every inbound a client is attached to ToggleClientEnableByEmail (Telegram bot) and SetClientEnableByEmail (LDAP sync) resolved a single inbound via the legacy client_traffics pointer and flipped enable only there. A client attached to several inbounds kept connecting through the siblings' running Xray after being disabled, and the next edit could re-enable it everywhere from a stale sibling. Route both through the applyClientFieldByEmail fan-out (the #5039 fix path) so the whole multi-inbound identity is toggled at once, dropping the circular Set/Toggle dependency. * fix(traffic): commit a traffic tick even when a best-effort maintenance helper fails addTrafficLocked stages the inbound and client deltas, then runs three helpers (auto-renew, disable depleted clients, disable depleted inbounds) that are meant to log and continue. All three reused the function-scope err that the deferred commit/rollback inspects, so the last helper's error decided the whole tick: a failure in disableInvalidInbounds rolled back the already-staged traffic while AddTraffic reported success, and because xray had already advanced its counter baseline that traffic was lost for good. Give each best-effort helper its own error variable so only a genuine staging failure rolls the tick back. * fix(traffic): re-enable clients and serialize the write in Reset All Client Traffic ClientService.ResetAllTraffics zeroed up/down but, unlike every sibling reset path, never restored enable=true, so clients that had been auto-disabled for exceeding their quota stayed cut with zero usage after a reset. It also wrote client_traffics directly on the shared DB handle instead of through the serial traffic writer, reintroducing the cross-transaction lock-order deadlock the writer exists to prevent. Restore enable and run the reset inside submitTrafficWrite within one transaction. * fix(traffic): keep node reset propagation out of the serial traffic writer ResetAllTraffics and ResetInboundTraffic performed their remote-node reset HTTP calls inside submitTrafficWrite. Each call can block up to the remote timeout, and Reset All Traffics loops every node serially, so the single traffic-writer goroutine was held for seconds — long enough that the concurrent 5s traffic poll timed out submitting its own write and dropped the deltas it had already drained from xray. Do the DB reset inside the writer, then propagate to the nodes after it returns, matching how the mtproto quota reset is already sequenced. * fix(sub): stop the subscription from 500ing on valid-but-unusual stream settings The raw share-link generators used unchecked type assertions and unguarded array indexing: an empty Reality shortIds/serverNames array (random.Num(0) panics), a tcp-http header with no request block or an empty request.path, a grpc block missing its keys, empty stream settings, and a non-string Host header all panicked mid-generation. Because getSubs loops every client's link with no recover, one such client 500s the entire subscription for everyone. The sibling JSON, Clash and frontend generators already guard these; make the raw generators match with comma-ok assertions and length checks. * fix(sub): tolerate a hysteria inbound without hysteriaSettings in the JSON subscription genHy asserted stream["hysteriaSettings"].(map[string]any) without the comma-ok form, so a hysteria inbound whose StreamSettings omit the hysteriaSettings key (a valid, representable shape the raw generator renders fine) panicked and 500ed the entire JSON subscription. Use comma-ok; the downstream reads already guard each key, so a nil map degrades gracefully. * fix(sub): emit the pinned peer cert sha256 in Clash subscriptions The Clash stream builder computed tlsSettings["pin-sha256"] from the inbound's pinnedPeerCertSha256, but applySecurity's tls case never copied it onto the proxy, so it was written with no reader and silently dropped. Clash subscribers lost certificate pinning while JSON subscribers kept it. Surface pin-sha256 on the proxy in the tls case, matching the JSON emitter. * fix(link): parse the snake_case and extra-blob xhttp fields when importing a share link The panel's share-link emitters (Go and TS) carry advanced xhttp knobs as a snake_case x_padding_bytes plus an extra=<json> payload, but the Go parser's xhttp branch read only top-level camelCase params, so importing an xhttp link via the outbound-subscription feature dropped xPaddingBytes, scMaxEachPostBytes and the rest, silently reverting them to the stream defaults and producing a non-working outbound. Mirror the TS parser: read the snake_case alias, merge the extra JSON blob, then let explicit camelCase params win. * fix(frontend): decode URL-safe base64 when parsing an imported share link Base64.decode called window.atob directly, which rejects the base64url alphabet (- and _) and unpadded input. But the panel's own share-link emitter uses Base64.encode(x, true) (URL-safe, unpadded), and real SIP002 links do too, so importing a Shadowsocks link whose method:password encodes with a - or _ threw, fell back to the raw undecoded string, and produced a wrong method and garbage password (the vmess parser shared the same limitation). Normalize base64url and re-pad before atob so decode round-trips every emitted link. * fix(link): honor the vmess ws path and hysteria2 vcn params on import Two Go/TS parser parity gaps in the outbound share-link import path: parseVmess only applied a ws link's path when the inner JSON also carried a host key, so a generator that omits host dropped the path back to the default; and parseHysteria2 hardcoded verifyPeerCertByName to empty, ignoring the vcn param the panel emits, so a hysteria2 outbound with a decoy SNI and a distinct cert name failed TLS verification after import. The TS parser handles both; make the Go parser match. * fix(ui): stop the sniffing form island from clobbering unrendered fields antd's Form.useWatch only reports registered fields, so while the sniffing toggle was off the island emitted { enabled: false } upward and replaced the full Sniffing object in form state. Saving a VLESS reverse outbound then crashed in sniffingToWire on the missing ipsExcluded array; the loopback outbound and the inbound sniffing tab shared the same hole. Watch the store with preserve: true so unrendered fields keep their values, and seed a missing value from the schema defaults instead of an empty cast. * fix(sub): drop empty remark segments instead of leaving a stray separator expandSegment dropped a "|" segment only when its tokens rendered the unlimited mark, so a segment whose only token resolved to the empty string (a client with no comment, an unlimited client's expiry date) was kept as bare decoration, leaving a trailing "|" or a dangling emoji on every share link's remark. Drop a token-bearing segment whenever none of its tokens produce a real value, while still keeping pure-literal segments. * fix(xray): keep source- and domains-scoped routing rules when an inbound is deleted removeInboundTagFromRules drops a routing rule whose inboundTag list becomes empty only if the rule has no other matcher, but routingMatcherKeys omitted xray-core's canonical source and domains keys. A rule scoped by source or domains (common in hand-authored or imported configs) therefore lost its whole body — including a security-relevant block — when its single listed inbound was deleted, instead of just having the tag trimmed. Recognize source and domains as live matchers. * fix(xray): guard RemoveUser against an uninitialized handler client Every XrayAPI handler method returns an error when HandlerServiceClient is nil, except RemoveUser, which dereferenced it directly. A depletion sweep runs Init with the port ignored and, during a restart window where the fresh process's api port is still 0, Init fails and leaves the client nil — so RemoveUser panicked (recovered by the traffic writer, but re-thrown every poll) instead of returning an error. Add the same nil guard the siblings have. * fix(xray): do not revive a manually stopped Xray on a background restart RestartXray cleared isManuallyStopped unconditionally at its top, so the @30s pending-config cron (and warp/ldap/outbound reconcile jobs) that call RestartXray(false) resurrected an Xray the admin had deliberately stopped — unlike the crash-detector, which honors the manual-stop flag. Skip a non-forced restart while the stop flag is set; only an explicit forced restart clears it. * fix(xray): retry a failed pending-restart instead of dropping the config change The 30s cron consumed the need-restart flag with IsNeedRestartAndSetFalse before calling RestartXray and only logged a failure. If RestartXray failed early (a transient GetXrayConfig DB error) the old process kept running the old config, the crash detector saw a running process and never retried, and the flag stayed cleared — so an admin's saved change silently never reached the core. Move the consume/restart/retry into ApplyPendingRestart, which re-arms the flag on failure so the next tick retries. * fix(xray): synchronize the process version and apiPort fields Start writes p.version and p.apiPort (via refreshVersion/refreshAPIPort) after flipping the process to running, while GetXrayVersion and GetAPIPort read them lock-free from the status and traffic poll goroutines. The struct mutex deliberately excluded these fields, so a restart racing a poll was a real data race — a torn read of the version string header can crash. Extend the mutex to cover version and apiPort, doing the blocking version probe before taking the lock. * fix(settings): detect a wildcard listen collision between the web and sub ports The web/sub same-port check compared the two listen addresses as raw strings, so binding both on all interfaces with different spellings (webListen 0.0.0.0 vs an empty subListen) slipped past validation and only failed at startup with an opaque bind error. Treat any wildcard listen ('', 0.0.0.0, ::) as overlapping so the clash is reported up front, while still allowing two distinct specific addresses to share a port. * fix(db): mark the IP-limit cleanup seeder done on a fresh install ResetIpLimitNoFail2ban is a one-time migration that, on a host without fail2ban, zeroes every existing client's limitIp because the limit can't be enforced. It was missing from the fresh-install fast-path seeder list, so on a brand-new DB it did not run on the first boot but fired on the second — wiping any IP limits the admin had set in between. Add it to the fast-path so a truly fresh install marks it done up front (there is nothing to clean), leaving later admin-set limits intact. * fix(security): dial outbound subscriptions through the SSRF guard The outbound-subscription fetch validated the URL host once (resolving DNS and rejecting private targets) but then fetched with a plain HTTP client that re-resolves the host at dial time, so a subscription domain the attacker controls could pass validation as a public IP and rebind to 127.0.0.1 / a cloud metadata endpoint / an internal host for the actual dial — a blind SSRF into the panel's network. Route the direct fetch (and its redirects) through netsafe.SSRFGuardedDialContext, which resolves, checks and dials the same IP atomically, carrying the subscription's AllowPrivate flag on the request context; a configured egress proxy still dials its loopback bridge unguarded. * fix(security): bound the login-limiter attempts map The login rate limiter keys its records on the caller-supplied username and only evicted a record when that exact key was revisited or the login succeeded. An unauthenticated attacker replaying one CSRF token while rotating a fresh username per request seeded a record that was never revisited, growing the map without bound until the panel OOMs. Cap the map: before inserting a new record, reclaim records whose block has lapsed and whose failures aged out, and if the map is still at the ceiling under a broad flood, drop one so memory can never grow past the cap. * fix(tgbot): require admin for privileged callbacks, not just the first switch answerCallback wraps only its first callback switch in an isAdmin guard; the second switch (server usage, inbound/online enumeration, database backup export, ban logs, mass traffic reset, client creation) ran for every caller. Telegram delivers a callback with the tapping user's id, so a non-admin who can see an admin's inline keyboard — as when the bot runs in a group — could tap Backup and receive the full database and config, or reset all traffic. Default-deny before the second switch: a non-admin may only run the per-user client_* callbacks that resolve their own data from their Telegram id. * fix(eventbus): dispatch each subscriber in its own goroutine The fan-out loop called every subscriber's handler sequentially on the single dispatch goroutine. The email and Telegram notifiers block on network I/O for tens of seconds (or minutes when the remote is slow), so one slow subscriber stalled the whole loop: the 256-slot channel then filled and Publish silently dropped later events — including high-value xray.crash and node.down notifications unrelated to the slow handler. Hand each delivered event to every handler in its own goroutine so a blocking subscriber can no longer stall delivery to the others. safeCall already recovers panics, so a detached handler cannot take down the bus. * fix(integration): cap WARP API response body size doWarpRequest read the response with an unbounded io.ReadAll, unlike the sibling NordVPN client which already caps every read at maxResponseSize. A hostile panel egress proxy or a MITM on the Cloudflare WARP endpoint could stream an arbitrarily large body and force the panel into an unbounded allocation. Wrap the body in an io.LimitReader(maxResponseSize) to match the NordVPN client. * fix(email): bound every SMTP step with a connection deadline The "starttls"/"none" transport delivered through net/smtp.SendMail, which dials with an untimed net.Dial and never sets a socket deadline. When an SMTP server accepted the TCP connection but then stalled (or was a blackhole), the caller was released by Send's 30s select, but the sender goroutine and its socket stayed blocked until the OS TCP timeout — minutes per notification, leaking a goroutine and a connection each time. sendWithTLS dialed with a timeout but likewise armed no deadline on the protocol phase, and TestConnection (called synchronously from the settings handler, with no select guard) could hang the request indefinitely. Replace SendMail with sendPlain, which dials with smtpConnectTimeout and arms conn.SetDeadline(smtpDeadline) before the greeting read, preserving SendMail's opportunistic STARTTLS upgrade. Arm the same deadline in sendWithTLS and TestConnection so every SMTP step is bounded. * fix(server): guard access-log parser against malformed lines GetXrayLogs split each Xray access-log line on whitespace and then read fixed offsets — parts[1] for the timestamp and parts[i+1] after the "from", "accepted" and "email:" markers — without checking the line had that many fields. A truncated or malformed line (the logged destination is attacker-influenced) indexed past the slice and panicked; the panel handler returned a 500 via Gin's recovery. Extract the per-line field parsing into parseAccessLogFields and length guard every positional lookup so a short line yields a partial entry instead of panicking. * fix(server): guard xray key-generator output parsing GetNewX25519Cert, GetNewmldsa65 and GetNewmlkem768 parsed xray's stdout by reading lines[0], lines[1] and each line's second colon-separated field without any length check — unlike GetNewEchCert, which already guards its line count. If the xray binary printed fewer than two lines or reformatted its labels (a version change, or a silent failure that emitted nothing), the fixed slice index panicked and the handler 500'd. Extract the shared parsing into parseXrayKeyPairOutput, which length guards the line count and each label split and returns an error instead of panicking, then route all three generators through it. * fix(tgbot): stop auto-deleted messages from resetting wizard state SendMsgToTgbotDeleteAfter spawns a goroutine that, after the display delay, deleted the transient message and then unconditionally cleared the chat's conversation state. Every caller that ends a wizard step already clears the state synchronously, so that call was redundant — and harmful: if within the delay the user advanced to the next step (a callback sets a fresh awaiting_* state), the late goroutine wiped it, and the user's next message fell through unrecognized, silently dropping their input. Move the delayed deletion into deleteMessageAfterDelay, which only removes the message and no longer touches the conversation state. Guard deleteMessageTgBot against a nil bot so the deletion path is unit-testable. * fix(frontend): refetch a fresh CSRF token on 403 instead of reusing the stale meta tag On a 403 to an unsafe method the client cleared its cached CSRF token and called ensureCsrfToken to retry. But ensureCsrfToken prefers the <meta name="csrf-token"> tag baked into the page, which the production panel always injects, so the "refresh" re-read the same stale token and the /csrf-token refetch was never reached — the retry re-sent the token that had just been rejected and the save failed with an error toast. The token lives in the session and rotates when the session is regenerated (for example re-login in another tab), leaving the tab's baked-in meta token stale. Fetch the current token straight from /csrf-token in the 403 branch so the retry uses the authoritative server value. The existing tests only passed because they strip the meta tag; the new test keeps a stale tag present. * fix(frontend): surface backend error text from failed requests HttpUtil.get/post read the thrown HttpError body as response.data.message, but the backend error envelope (entity.Msg) serializes its text as msg. On any non-2xx JSON response the real reason was therefore dropped and the operator saw only the generic "Request failed with status N" toast. Read response.data.msg first (keeping message and the native error text as fallbacks). The sibling test had pinned the wrong body shape ({ message }); correct it to the real backend shape ({ success:false, msg }) so it exercises the actual envelope. * fix(frontend): share one WebSocket connection across bridge and hooks websocketBridge.ts and useWebSocket.ts each declared their own module-scoped sharedClient plus an identical getSharedClient, so the "shared" client was not shared between them: whenever a page using useWebSocket (Clients/Inbounds) mounted alongside the always-mounted bridge, the panel opened two sockets to /ws. The server then pushed every traffic/stats/nodes/inbounds snapshot to both, doubling WebSocket bandwidth and running two independent reconnect loops, and the hook's socket was never disconnected on unmount. Hoist a single getSharedWebSocketClient into api/websocket.ts and route both the bridge and the hook through it, so exactly one connection is opened. * fix(frontend): guard the outbounds WebSocket handler against non-array payloads onOutbounds wrote the raw WebSocket payload straight into the outboundsTraffic cache, unlike the sibling onNodes/onInbounds handlers which first check Array.isArray. A malformed non-array push (for example an object) would land in the cache with staleTime Infinity; consumers that call .find()/.map() on the outbounds list would then throw and crash the Outbounds tab. Add the same Array.isArray guard so a bad push is ignored. * fix(frontend): key the node table by the computed row key, not id The desktop node table used rowKey="id", but transitive sub-nodes (the read-only rows surfaced from downstream nodes) all carry id 0, so a topology with two or more transitive rows gave React duplicate keys. antd's rowKey prop overrides the row object's own computed `key` (`t-${guid}` for transitive rows, the numeric id otherwise), so the unique key the code already builds was ignored — causing row-state/DOM mis-association on any re-render (heartbeat refetch, address-eye toggle). The mobile card path already keyed by record.key. Key the table by "key" so transitive rows get their distinct t-${guid} identity; direct nodes keep key === id, so row selection (filtered to numeric keys) is unchanged. * fix(frontend): map routing row actions through the rule's real index The routing table hides balancer-loopback rules (`_bl_*`) but keeps each visible row's original index in `key`, then handed antd's positional row index straight to edit/delete/toggle/move/drag — all of which mutate the full, unfiltered routing.rules array. Once a hidden loopback rule precedes a visible one (e.g. a balancer whose fallback is another balancer, plus any rule added afterwards), the positional index no longer matches the array index, so deleting or editing a rule silently hit the wrong one — including destroying the loopback rule that keeps the balancer alive. Add originalRuleIndex to translate a positional row index back through the row's `key`, and route every mutating handler (openEdit, confirmDelete, toggleRule, moveUp/moveDown, drag) through it. When no loopback rows are hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): map outbound row actions through the outbound's real index The outbounds table hides balancer-loopback outbounds (`_bl_*`) but keeps each visible row's original index in `key`, then passed antd's positional row index to edit/delete/move and to the per-row probe (onTest) and its result lookup — all of which address the full, unfiltered outbounds array. Once a hidden loopback outbound precedes a visible one, the positional index diverges from the array index, so deleting or editing an outbound hit the wrong one (its deletion-impact plan and removal targeting the wrong entry), and the test button probed / showed results against the wrong outbound. Add originalOutboundIndex and route the mutating handlers through it; key the probe trigger and test-result columns by record.key. With no loopback rows hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): tolerate a malformed happyEyeballs value in the Xray Basics tab BasicsTab derived directHappyEyeballs by calling HappyEyeballsSchema.parse during render, guarding only against null/non-object. A wrong-typed field (e.g. happyEyeballs.tryDelayMs as a string) or any other shape mismatch — reachable via the Complete Template JSON editor or an imported config — threw straight out of render, white-screening the default Xray landing tab. Use safeParse and fall back to null so a bad value degrades to "no override" instead of crashing the page. * fix(frontend): preserve routing-rule fields the form does not surface The rule form rebuilt the rule from a fixed literal of only the fields it edits, and RoutingTab replaces the rule wholesale on confirm. Fields the form never exposes — localPort, localIP, process, ruleTag, webhook — are in the rule schema and can arrive via the advanced JSON editor or Import Rules; opening such a rule in the form and saving silently dropped them. Carry over every key of the original rule the form does not manage before applying the form-derived fields, so an edit only touches what it surfaces. * fix(frontend): re-sync the sniffing island when its value changes externally The sniffing config editor froze its seed value at mount and only watched its own inner AntD form, never reflecting a later change to the shared RHF `sniffing` path. Because the inbound form mounts every tab with forceRender, the friendly Sniffing tab and the Advanced JSON editor are live at once: editing sniffing in the JSON editor updated the RHF value but not the frozen island, so the next interaction with the friendly tab emitted the stale value and silently discarded the JSON edit. Add an effect that pushes an external value change into the inner form, guarded by the same lastEmitted marker the emit path uses so the island never re-seeds from its own echo and no update loop forms. * fix(frontend): don't drift a client's byte quota on a no-op save The quota field shows the total in GB rounded to two decimals; editing a client and saving converted that display value straight back to bytes. A byte total not aligned to 0.01 GB — one set via the API or an import — was therefore rewritten to the rounded value on any save that never touched the field, losing a few MB each time. Add resolveTotalBytes: keep the original byte total when the displayed GB still matches it, and only re-derive from GB when the user actually changed the field. * fix(eventbus): deliver events on a bounded per-subscriber worker The previous fix dispatched each event to every subscriber with a bare `go safeCall`. That unblocked the dispatch loop, but removed the bus's backpressure: under a login-attempt flood (which both notifier subscribers process without rate-limiting) with email/Telegram enabled, every attempt spawned handler goroutines that each block on network I/O for up to ~30s, with no bound — a goroutine and outbound-connection storm. It also let a subscriber's handler run concurrently with itself, racing the Telegram notifier's lazily-cached hostname. Give each subscriber its own bounded queue drained by a single worker goroutine. Dispatch does a non-blocking send per subscriber (dropping only that subscriber's event when its queue is full), so a slow subscriber still can't stall the others, concurrency is bounded to one in-flight handler per subscriber, per-subscriber event order is preserved, and Stop again waits for in-flight handlers to finish. * fix(frontend): map outbound mobile-card actions through the real index too The desktop outbounds table was keyed by the outbound's real index, but the mobile card list was left keying the probe trigger and every test-state lookup by the positional row index. With a hidden balancer-loopback outbound present, tapping Check on a mobile card probed the wrong outbound and the Test-All results landed on the wrong card. Key onTest and the testResult/isTesting reads by record.key, matching the desktop columns. * fix(frontend): meet WCAG AA contrast on the config-block link text The Storybook accessibility test flagged the share-link <code> block: with no explicit color it inherited a muted grey that renders as #888888 on the #f8f8f8 tertiary-fill background in CI's Chromium — a 3.33:1 contrast, below the 4.5:1 AA threshold. Set the text to the theme's primary text token so the colour is explicit and high-contrast in both light and dark themes instead of depending on an inherited value that varies by browser. * style(sub): simplify a negated conjunction to satisfy staticcheck QF1001 golangci-lint (staticcheck QF1001) flagged the `!(a && b)` guard in expandSegment. Rewrite it via De Morgan's law to the equivalent `!a || !b` form so the linter passes; behavior is unchanged. * fix: close panics and races the audit's own fixes left nearby Second-pass review of the 54-commit self-correcting audit. Each item below was confirmed by reading the surrounding source (and, where practical, the pre-fix code) before being changed; regression tests are included for every behavioral fix. Concurrency: - eventbus: Bus.Subscribe called wg.Add with no synchronization against a concurrent Bus.Stop's wg.Wait, a real "WaitGroup misuse" panic risk (e.g. a Telegram-bot settings save racing panel shutdown/restart). Stop now flips a mu-guarded `stopped` flag before waiting, and Subscribe checks it under the same lock, so Add and Wait can no longer race. Security: - login_limiter: evictForRoom's fallback eviction picked an arbitrary map key, including ones still under an active cooldown - an attacker flooding /login with fresh usernames could evict their own (or anyone's) blocked record and reset the lockout. The fallback now skips actively-blocked records, only falling back to an unconditional evict if the map is somehow entirely full of active blocks (preserves the hard memory cap). Subscription-endpoint panics (reachable by any client hitting /sub): - internal/sub/service.go: applyPathAndHostParams/Obj (ws/httpupgrade/xhttp with no path settings object) and the TLS alpn readers in three places used unchecked type assertions - exactly the bug class abab7cd0 patched elsewhere in the same switch statements, just not these call sites. - internal/sub/json_service.go, clash_service.go: the externalProxy loops in the JSON and Clash generators used unchecked assertions on a legacy/admin-supplied field (missing "port", non-object entry, etc.). - internal/sub/json_service.go: realityData's shortId/serverName selection could assert a non-string array element. Other correctness: - client_traffic.go: ResetAllTraffics (touched by 3eb214d0) still skipped clearing NodeClientTraffic node-sync baselines, unlike its sibling reset paths in the same file - a node's next sync would re-add pre-reset delta on top of the freshly-zeroed counter. - inbound_traffic.go: the traffic-tick tx's Commit/Rollback errors were silently discarded; now logged so a backend-level commit failure (e.g. an aborted Postgres tx from a best-effort helper) doesn't masquerade as a successful tick. - outbound_subscription.go: the new subscriptionFetchClient doc comment was wedged between fetchAndStore's existing comment and fetchAndStore itself, leaving fetchAndStore undocumented and the comment describing the wrong function. Convention cleanup: - Removed narrative // comments added by the audit that violate this repo's no-inline-comment rule (mostly narrating the specific bug/fix rather than a lasting contract, and mostly on new Test functions, which this repo's existing tests never comment) - calibrated against this exact codebase's own pre-existing comment style so legitimate godoc-style doc comments were left alone. --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-07-16 15:33:06 -07:00
func (s *XrayService) ApplyPendingRestart() {
reconcile := isNeedXrayReconcile.Swap(false)
restart := s.IsNeedRestartAndSetFalse()
if !restart && !reconcile {
Repo-wide self-correcting audit: 54 verified bug fixes (#5970) * fix(email): resolve a name-addr smtpFrom into bare envelope address and display name The save-time validator accepts any RFC 5322 address form, so a value like '3x-ui Panel <panel(at)example.com>' passes validation, but Send and TestConnection fed that raw string to MAIL FROM, which strict servers reject with 501, and buildMessage mangled it into a quoted local part. Parse the configured sender at the point of use: the envelope gets the bare address and, when no explicit sender name is set, the display name embedded in the setting is used for the From header. * fix(email): report a missing sender address from the SMTP connection test TestConnection skipped the empty-from guard that Send enforces, so with no sender and no username configured the test issued the null reverse-path and could report success against a lenient relay while every real notification send kept failing with the missing-sender error. Guard the test path the same way and surface a dedicated translated message. * fix(sub): fall back to the raw subscription when an auto-detected format has no content With format auto-detection enabled, a client whose User-Agent matched the Clash or JSON regex was routed straight to that format handler. For a subscription whose entries convert to neither format (an MTProto-only subscription, for example) the handler returns an empty document and the request ended as 404, breaking a URL that served the raw list before the toggle. The auto-detect branches now serve the detected format only when it produces content and otherwise continue to the raw response; the explicit format endpoints keep answering 404 for empty documents. * fix(node): match prefixed central tags when filtering a selected-mode node snapshot FilterNodeSnapshot compared a node snapshot's inbound tags against the raw selected-tag list with an exact match, while its two siblings (SnapshotHasUnadoptedInbounds and the reconcile tagToCentral map) expand each selected tag to both its bare node-side form and its n<id>- prefixed central form. A panel-created node inbound is recorded in the selected list under the central prefixed tag but reported by the node under the bare tag, so the exact match dropped it from every snapshot and the orphan sweep then deleted its central row one tick after creation. Expand the allowed set with the same prefix flip the siblings use. * fix(client): refuse a bulk quota reduction that would fall to or below zero BulkAdjust clamped a client's new traffic limit with max(total+addBytes, 0). Because 0 is the unlimited sentinel, reducing a client's quota by more than it had left silently granted that client unlimited traffic. The sibling expiry branch already refuses an over-reduction; mirror it for quota so the adjustment is skipped with a clear reason instead of crossing the sentinel. * fix(client): persist a bulk adjustment's applied field even when the sibling field is skipped In a mixed BulkAdjust (both a days delta and a bytes delta), a per-field planning skip such as "unlimited expiry" or "unlimited traffic" was recorded in the same map that gated the client_traffics write. The applied field was already written to the inbound JSON and the clients table, but the enforcement row was left untouched, so the depletion job cut the client on the old limit while the panel showed the new one. Gate the traffic-row write on an actual inbound-processing failure rather than on any planning-phase skip note. * fix(inbound): always create in AddInbound instead of overwriting a row whose id was posted The add controller binds the inbound model's id form field and never clears it, and AddInbound persisted with GORM Save, which updates in place when the primary key is non-zero. A client that reused an existing id (for instance by duplicating an inbound fetched from /get and changing the port) silently overwrote that stored row instead of creating a new inbound. Zero the id at the top of AddInbound, matching how it already zeroes the client-stat ids. * fix(inbound): accept WireGuard clients when creating an inbound AddInbound's per-client validation switch had cases for every protocol except WireGuard, so a WireGuard client fell through to the default branch that requires a non-empty id. WireGuard clients are keyed by their public key and carry no id, so importing a WireGuard inbound or re-adding one to a reconciling node was rejected with "empty client ID". Add a wireguard case that validates the client key, mirroring addInboundClient. * fix(client): stop holding the inbound-lock registry mutex while waiting on one inbound lockInbound acquired the global registry mutex and then blocked on the per-inbound mutex without releasing the registry first. A slow client operation holding one inbound's mutex (for example a bulk delete pushing to an unreachable node) made the next waiter park on that inbound while still holding the registry mutex, which in turn blocked lockInbound for every other inbound — freezing client mutations panel-wide. Release the registry mutex before taking the per-inbound lock. * fix(client): honor keepTraffic when deleting a client that is attached to inbounds Delete, DeleteByEmail and BulkDelete all pass keepTraffic to their final cleanup transaction, but each called the per-inbound delete helper with a hardcoded false. That helper purges the client's traffic, IP and stat rows before the gated cleanup runs, so keepTraffic=true still destroyed all traffic history for any client actually attached to an inbound (the pinned test only covered a record with no inbound mappings). Thread the caller's keepTraffic through to the per-inbound helper at all three call sites. * fix(inbound): defer a local MTProto inbound edit's sidecar push until after commit UpdateInbound applied a local MTProto inbound change by calling the runtime UpdateInbound (which stops/starts the mtg sidecar or talks to it) from inside runSerializedTx. That runs process and network I/O on the single traffic-writer goroutine while a DB transaction is open, so a slow sidecar stalls traffic accounting and every concurrent client mutation, and a later step failing the transaction leaves the sidecar ahead of the rolled-back row. Move the push into the post-commit hook, matching the xray branch. Adds a SetLocalRuntimeOverride test seam mirroring the existing node override so the deferral is regression tested. * fix(client): delete external-link rows when bulk-deleting clients The single-client Delete path removes a client's client_external_links rows, but BulkDelete (and the DelDepleted reaper that routes through it) deleted the record, mappings and traffic while leaving the external-link rows keyed by the now-dead client id, so they accumulated as orphans. Delete them in the same cleanup transaction, keyed by client id like the single path. * fix(inbound): request an xray restart when toggling a routed MTProto inbound AddInbound, DelInbound and UpdateInbound all flag needRestart when an inbound routes MTProto through xray, so the egress SOCKS bridge is regenerated. Only SetInboundEnable's local path omitted it, so toggling a routed MTProto inbound off then on left the bridge out of the running config while the sidecar dialed its loopback port, blackholing that inbound until an unrelated restart. Flag the restart on the local enable path too. * fix(client): apply enable-by-email to every inbound a client is attached to ToggleClientEnableByEmail (Telegram bot) and SetClientEnableByEmail (LDAP sync) resolved a single inbound via the legacy client_traffics pointer and flipped enable only there. A client attached to several inbounds kept connecting through the siblings' running Xray after being disabled, and the next edit could re-enable it everywhere from a stale sibling. Route both through the applyClientFieldByEmail fan-out (the #5039 fix path) so the whole multi-inbound identity is toggled at once, dropping the circular Set/Toggle dependency. * fix(traffic): commit a traffic tick even when a best-effort maintenance helper fails addTrafficLocked stages the inbound and client deltas, then runs three helpers (auto-renew, disable depleted clients, disable depleted inbounds) that are meant to log and continue. All three reused the function-scope err that the deferred commit/rollback inspects, so the last helper's error decided the whole tick: a failure in disableInvalidInbounds rolled back the already-staged traffic while AddTraffic reported success, and because xray had already advanced its counter baseline that traffic was lost for good. Give each best-effort helper its own error variable so only a genuine staging failure rolls the tick back. * fix(traffic): re-enable clients and serialize the write in Reset All Client Traffic ClientService.ResetAllTraffics zeroed up/down but, unlike every sibling reset path, never restored enable=true, so clients that had been auto-disabled for exceeding their quota stayed cut with zero usage after a reset. It also wrote client_traffics directly on the shared DB handle instead of through the serial traffic writer, reintroducing the cross-transaction lock-order deadlock the writer exists to prevent. Restore enable and run the reset inside submitTrafficWrite within one transaction. * fix(traffic): keep node reset propagation out of the serial traffic writer ResetAllTraffics and ResetInboundTraffic performed their remote-node reset HTTP calls inside submitTrafficWrite. Each call can block up to the remote timeout, and Reset All Traffics loops every node serially, so the single traffic-writer goroutine was held for seconds — long enough that the concurrent 5s traffic poll timed out submitting its own write and dropped the deltas it had already drained from xray. Do the DB reset inside the writer, then propagate to the nodes after it returns, matching how the mtproto quota reset is already sequenced. * fix(sub): stop the subscription from 500ing on valid-but-unusual stream settings The raw share-link generators used unchecked type assertions and unguarded array indexing: an empty Reality shortIds/serverNames array (random.Num(0) panics), a tcp-http header with no request block or an empty request.path, a grpc block missing its keys, empty stream settings, and a non-string Host header all panicked mid-generation. Because getSubs loops every client's link with no recover, one such client 500s the entire subscription for everyone. The sibling JSON, Clash and frontend generators already guard these; make the raw generators match with comma-ok assertions and length checks. * fix(sub): tolerate a hysteria inbound without hysteriaSettings in the JSON subscription genHy asserted stream["hysteriaSettings"].(map[string]any) without the comma-ok form, so a hysteria inbound whose StreamSettings omit the hysteriaSettings key (a valid, representable shape the raw generator renders fine) panicked and 500ed the entire JSON subscription. Use comma-ok; the downstream reads already guard each key, so a nil map degrades gracefully. * fix(sub): emit the pinned peer cert sha256 in Clash subscriptions The Clash stream builder computed tlsSettings["pin-sha256"] from the inbound's pinnedPeerCertSha256, but applySecurity's tls case never copied it onto the proxy, so it was written with no reader and silently dropped. Clash subscribers lost certificate pinning while JSON subscribers kept it. Surface pin-sha256 on the proxy in the tls case, matching the JSON emitter. * fix(link): parse the snake_case and extra-blob xhttp fields when importing a share link The panel's share-link emitters (Go and TS) carry advanced xhttp knobs as a snake_case x_padding_bytes plus an extra=<json> payload, but the Go parser's xhttp branch read only top-level camelCase params, so importing an xhttp link via the outbound-subscription feature dropped xPaddingBytes, scMaxEachPostBytes and the rest, silently reverting them to the stream defaults and producing a non-working outbound. Mirror the TS parser: read the snake_case alias, merge the extra JSON blob, then let explicit camelCase params win. * fix(frontend): decode URL-safe base64 when parsing an imported share link Base64.decode called window.atob directly, which rejects the base64url alphabet (- and _) and unpadded input. But the panel's own share-link emitter uses Base64.encode(x, true) (URL-safe, unpadded), and real SIP002 links do too, so importing a Shadowsocks link whose method:password encodes with a - or _ threw, fell back to the raw undecoded string, and produced a wrong method and garbage password (the vmess parser shared the same limitation). Normalize base64url and re-pad before atob so decode round-trips every emitted link. * fix(link): honor the vmess ws path and hysteria2 vcn params on import Two Go/TS parser parity gaps in the outbound share-link import path: parseVmess only applied a ws link's path when the inner JSON also carried a host key, so a generator that omits host dropped the path back to the default; and parseHysteria2 hardcoded verifyPeerCertByName to empty, ignoring the vcn param the panel emits, so a hysteria2 outbound with a decoy SNI and a distinct cert name failed TLS verification after import. The TS parser handles both; make the Go parser match. * fix(ui): stop the sniffing form island from clobbering unrendered fields antd's Form.useWatch only reports registered fields, so while the sniffing toggle was off the island emitted { enabled: false } upward and replaced the full Sniffing object in form state. Saving a VLESS reverse outbound then crashed in sniffingToWire on the missing ipsExcluded array; the loopback outbound and the inbound sniffing tab shared the same hole. Watch the store with preserve: true so unrendered fields keep their values, and seed a missing value from the schema defaults instead of an empty cast. * fix(sub): drop empty remark segments instead of leaving a stray separator expandSegment dropped a "|" segment only when its tokens rendered the unlimited mark, so a segment whose only token resolved to the empty string (a client with no comment, an unlimited client's expiry date) was kept as bare decoration, leaving a trailing "|" or a dangling emoji on every share link's remark. Drop a token-bearing segment whenever none of its tokens produce a real value, while still keeping pure-literal segments. * fix(xray): keep source- and domains-scoped routing rules when an inbound is deleted removeInboundTagFromRules drops a routing rule whose inboundTag list becomes empty only if the rule has no other matcher, but routingMatcherKeys omitted xray-core's canonical source and domains keys. A rule scoped by source or domains (common in hand-authored or imported configs) therefore lost its whole body — including a security-relevant block — when its single listed inbound was deleted, instead of just having the tag trimmed. Recognize source and domains as live matchers. * fix(xray): guard RemoveUser against an uninitialized handler client Every XrayAPI handler method returns an error when HandlerServiceClient is nil, except RemoveUser, which dereferenced it directly. A depletion sweep runs Init with the port ignored and, during a restart window where the fresh process's api port is still 0, Init fails and leaves the client nil — so RemoveUser panicked (recovered by the traffic writer, but re-thrown every poll) instead of returning an error. Add the same nil guard the siblings have. * fix(xray): do not revive a manually stopped Xray on a background restart RestartXray cleared isManuallyStopped unconditionally at its top, so the @30s pending-config cron (and warp/ldap/outbound reconcile jobs) that call RestartXray(false) resurrected an Xray the admin had deliberately stopped — unlike the crash-detector, which honors the manual-stop flag. Skip a non-forced restart while the stop flag is set; only an explicit forced restart clears it. * fix(xray): retry a failed pending-restart instead of dropping the config change The 30s cron consumed the need-restart flag with IsNeedRestartAndSetFalse before calling RestartXray and only logged a failure. If RestartXray failed early (a transient GetXrayConfig DB error) the old process kept running the old config, the crash detector saw a running process and never retried, and the flag stayed cleared — so an admin's saved change silently never reached the core. Move the consume/restart/retry into ApplyPendingRestart, which re-arms the flag on failure so the next tick retries. * fix(xray): synchronize the process version and apiPort fields Start writes p.version and p.apiPort (via refreshVersion/refreshAPIPort) after flipping the process to running, while GetXrayVersion and GetAPIPort read them lock-free from the status and traffic poll goroutines. The struct mutex deliberately excluded these fields, so a restart racing a poll was a real data race — a torn read of the version string header can crash. Extend the mutex to cover version and apiPort, doing the blocking version probe before taking the lock. * fix(settings): detect a wildcard listen collision between the web and sub ports The web/sub same-port check compared the two listen addresses as raw strings, so binding both on all interfaces with different spellings (webListen 0.0.0.0 vs an empty subListen) slipped past validation and only failed at startup with an opaque bind error. Treat any wildcard listen ('', 0.0.0.0, ::) as overlapping so the clash is reported up front, while still allowing two distinct specific addresses to share a port. * fix(db): mark the IP-limit cleanup seeder done on a fresh install ResetIpLimitNoFail2ban is a one-time migration that, on a host without fail2ban, zeroes every existing client's limitIp because the limit can't be enforced. It was missing from the fresh-install fast-path seeder list, so on a brand-new DB it did not run on the first boot but fired on the second — wiping any IP limits the admin had set in between. Add it to the fast-path so a truly fresh install marks it done up front (there is nothing to clean), leaving later admin-set limits intact. * fix(security): dial outbound subscriptions through the SSRF guard The outbound-subscription fetch validated the URL host once (resolving DNS and rejecting private targets) but then fetched with a plain HTTP client that re-resolves the host at dial time, so a subscription domain the attacker controls could pass validation as a public IP and rebind to 127.0.0.1 / a cloud metadata endpoint / an internal host for the actual dial — a blind SSRF into the panel's network. Route the direct fetch (and its redirects) through netsafe.SSRFGuardedDialContext, which resolves, checks and dials the same IP atomically, carrying the subscription's AllowPrivate flag on the request context; a configured egress proxy still dials its loopback bridge unguarded. * fix(security): bound the login-limiter attempts map The login rate limiter keys its records on the caller-supplied username and only evicted a record when that exact key was revisited or the login succeeded. An unauthenticated attacker replaying one CSRF token while rotating a fresh username per request seeded a record that was never revisited, growing the map without bound until the panel OOMs. Cap the map: before inserting a new record, reclaim records whose block has lapsed and whose failures aged out, and if the map is still at the ceiling under a broad flood, drop one so memory can never grow past the cap. * fix(tgbot): require admin for privileged callbacks, not just the first switch answerCallback wraps only its first callback switch in an isAdmin guard; the second switch (server usage, inbound/online enumeration, database backup export, ban logs, mass traffic reset, client creation) ran for every caller. Telegram delivers a callback with the tapping user's id, so a non-admin who can see an admin's inline keyboard — as when the bot runs in a group — could tap Backup and receive the full database and config, or reset all traffic. Default-deny before the second switch: a non-admin may only run the per-user client_* callbacks that resolve their own data from their Telegram id. * fix(eventbus): dispatch each subscriber in its own goroutine The fan-out loop called every subscriber's handler sequentially on the single dispatch goroutine. The email and Telegram notifiers block on network I/O for tens of seconds (or minutes when the remote is slow), so one slow subscriber stalled the whole loop: the 256-slot channel then filled and Publish silently dropped later events — including high-value xray.crash and node.down notifications unrelated to the slow handler. Hand each delivered event to every handler in its own goroutine so a blocking subscriber can no longer stall delivery to the others. safeCall already recovers panics, so a detached handler cannot take down the bus. * fix(integration): cap WARP API response body size doWarpRequest read the response with an unbounded io.ReadAll, unlike the sibling NordVPN client which already caps every read at maxResponseSize. A hostile panel egress proxy or a MITM on the Cloudflare WARP endpoint could stream an arbitrarily large body and force the panel into an unbounded allocation. Wrap the body in an io.LimitReader(maxResponseSize) to match the NordVPN client. * fix(email): bound every SMTP step with a connection deadline The "starttls"/"none" transport delivered through net/smtp.SendMail, which dials with an untimed net.Dial and never sets a socket deadline. When an SMTP server accepted the TCP connection but then stalled (or was a blackhole), the caller was released by Send's 30s select, but the sender goroutine and its socket stayed blocked until the OS TCP timeout — minutes per notification, leaking a goroutine and a connection each time. sendWithTLS dialed with a timeout but likewise armed no deadline on the protocol phase, and TestConnection (called synchronously from the settings handler, with no select guard) could hang the request indefinitely. Replace SendMail with sendPlain, which dials with smtpConnectTimeout and arms conn.SetDeadline(smtpDeadline) before the greeting read, preserving SendMail's opportunistic STARTTLS upgrade. Arm the same deadline in sendWithTLS and TestConnection so every SMTP step is bounded. * fix(server): guard access-log parser against malformed lines GetXrayLogs split each Xray access-log line on whitespace and then read fixed offsets — parts[1] for the timestamp and parts[i+1] after the "from", "accepted" and "email:" markers — without checking the line had that many fields. A truncated or malformed line (the logged destination is attacker-influenced) indexed past the slice and panicked; the panel handler returned a 500 via Gin's recovery. Extract the per-line field parsing into parseAccessLogFields and length guard every positional lookup so a short line yields a partial entry instead of panicking. * fix(server): guard xray key-generator output parsing GetNewX25519Cert, GetNewmldsa65 and GetNewmlkem768 parsed xray's stdout by reading lines[0], lines[1] and each line's second colon-separated field without any length check — unlike GetNewEchCert, which already guards its line count. If the xray binary printed fewer than two lines or reformatted its labels (a version change, or a silent failure that emitted nothing), the fixed slice index panicked and the handler 500'd. Extract the shared parsing into parseXrayKeyPairOutput, which length guards the line count and each label split and returns an error instead of panicking, then route all three generators through it. * fix(tgbot): stop auto-deleted messages from resetting wizard state SendMsgToTgbotDeleteAfter spawns a goroutine that, after the display delay, deleted the transient message and then unconditionally cleared the chat's conversation state. Every caller that ends a wizard step already clears the state synchronously, so that call was redundant — and harmful: if within the delay the user advanced to the next step (a callback sets a fresh awaiting_* state), the late goroutine wiped it, and the user's next message fell through unrecognized, silently dropping their input. Move the delayed deletion into deleteMessageAfterDelay, which only removes the message and no longer touches the conversation state. Guard deleteMessageTgBot against a nil bot so the deletion path is unit-testable. * fix(frontend): refetch a fresh CSRF token on 403 instead of reusing the stale meta tag On a 403 to an unsafe method the client cleared its cached CSRF token and called ensureCsrfToken to retry. But ensureCsrfToken prefers the <meta name="csrf-token"> tag baked into the page, which the production panel always injects, so the "refresh" re-read the same stale token and the /csrf-token refetch was never reached — the retry re-sent the token that had just been rejected and the save failed with an error toast. The token lives in the session and rotates when the session is regenerated (for example re-login in another tab), leaving the tab's baked-in meta token stale. Fetch the current token straight from /csrf-token in the 403 branch so the retry uses the authoritative server value. The existing tests only passed because they strip the meta tag; the new test keeps a stale tag present. * fix(frontend): surface backend error text from failed requests HttpUtil.get/post read the thrown HttpError body as response.data.message, but the backend error envelope (entity.Msg) serializes its text as msg. On any non-2xx JSON response the real reason was therefore dropped and the operator saw only the generic "Request failed with status N" toast. Read response.data.msg first (keeping message and the native error text as fallbacks). The sibling test had pinned the wrong body shape ({ message }); correct it to the real backend shape ({ success:false, msg }) so it exercises the actual envelope. * fix(frontend): share one WebSocket connection across bridge and hooks websocketBridge.ts and useWebSocket.ts each declared their own module-scoped sharedClient plus an identical getSharedClient, so the "shared" client was not shared between them: whenever a page using useWebSocket (Clients/Inbounds) mounted alongside the always-mounted bridge, the panel opened two sockets to /ws. The server then pushed every traffic/stats/nodes/inbounds snapshot to both, doubling WebSocket bandwidth and running two independent reconnect loops, and the hook's socket was never disconnected on unmount. Hoist a single getSharedWebSocketClient into api/websocket.ts and route both the bridge and the hook through it, so exactly one connection is opened. * fix(frontend): guard the outbounds WebSocket handler against non-array payloads onOutbounds wrote the raw WebSocket payload straight into the outboundsTraffic cache, unlike the sibling onNodes/onInbounds handlers which first check Array.isArray. A malformed non-array push (for example an object) would land in the cache with staleTime Infinity; consumers that call .find()/.map() on the outbounds list would then throw and crash the Outbounds tab. Add the same Array.isArray guard so a bad push is ignored. * fix(frontend): key the node table by the computed row key, not id The desktop node table used rowKey="id", but transitive sub-nodes (the read-only rows surfaced from downstream nodes) all carry id 0, so a topology with two or more transitive rows gave React duplicate keys. antd's rowKey prop overrides the row object's own computed `key` (`t-${guid}` for transitive rows, the numeric id otherwise), so the unique key the code already builds was ignored — causing row-state/DOM mis-association on any re-render (heartbeat refetch, address-eye toggle). The mobile card path already keyed by record.key. Key the table by "key" so transitive rows get their distinct t-${guid} identity; direct nodes keep key === id, so row selection (filtered to numeric keys) is unchanged. * fix(frontend): map routing row actions through the rule's real index The routing table hides balancer-loopback rules (`_bl_*`) but keeps each visible row's original index in `key`, then handed antd's positional row index straight to edit/delete/toggle/move/drag — all of which mutate the full, unfiltered routing.rules array. Once a hidden loopback rule precedes a visible one (e.g. a balancer whose fallback is another balancer, plus any rule added afterwards), the positional index no longer matches the array index, so deleting or editing a rule silently hit the wrong one — including destroying the loopback rule that keeps the balancer alive. Add originalRuleIndex to translate a positional row index back through the row's `key`, and route every mutating handler (openEdit, confirmDelete, toggleRule, moveUp/moveDown, drag) through it. When no loopback rows are hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): map outbound row actions through the outbound's real index The outbounds table hides balancer-loopback outbounds (`_bl_*`) but keeps each visible row's original index in `key`, then passed antd's positional row index to edit/delete/move and to the per-row probe (onTest) and its result lookup — all of which address the full, unfiltered outbounds array. Once a hidden loopback outbound precedes a visible one, the positional index diverges from the array index, so deleting or editing an outbound hit the wrong one (its deletion-impact plan and removal targeting the wrong entry), and the test button probed / showed results against the wrong outbound. Add originalOutboundIndex and route the mutating handlers through it; key the probe trigger and test-result columns by record.key. With no loopback rows hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): tolerate a malformed happyEyeballs value in the Xray Basics tab BasicsTab derived directHappyEyeballs by calling HappyEyeballsSchema.parse during render, guarding only against null/non-object. A wrong-typed field (e.g. happyEyeballs.tryDelayMs as a string) or any other shape mismatch — reachable via the Complete Template JSON editor or an imported config — threw straight out of render, white-screening the default Xray landing tab. Use safeParse and fall back to null so a bad value degrades to "no override" instead of crashing the page. * fix(frontend): preserve routing-rule fields the form does not surface The rule form rebuilt the rule from a fixed literal of only the fields it edits, and RoutingTab replaces the rule wholesale on confirm. Fields the form never exposes — localPort, localIP, process, ruleTag, webhook — are in the rule schema and can arrive via the advanced JSON editor or Import Rules; opening such a rule in the form and saving silently dropped them. Carry over every key of the original rule the form does not manage before applying the form-derived fields, so an edit only touches what it surfaces. * fix(frontend): re-sync the sniffing island when its value changes externally The sniffing config editor froze its seed value at mount and only watched its own inner AntD form, never reflecting a later change to the shared RHF `sniffing` path. Because the inbound form mounts every tab with forceRender, the friendly Sniffing tab and the Advanced JSON editor are live at once: editing sniffing in the JSON editor updated the RHF value but not the frozen island, so the next interaction with the friendly tab emitted the stale value and silently discarded the JSON edit. Add an effect that pushes an external value change into the inner form, guarded by the same lastEmitted marker the emit path uses so the island never re-seeds from its own echo and no update loop forms. * fix(frontend): don't drift a client's byte quota on a no-op save The quota field shows the total in GB rounded to two decimals; editing a client and saving converted that display value straight back to bytes. A byte total not aligned to 0.01 GB — one set via the API or an import — was therefore rewritten to the rounded value on any save that never touched the field, losing a few MB each time. Add resolveTotalBytes: keep the original byte total when the displayed GB still matches it, and only re-derive from GB when the user actually changed the field. * fix(eventbus): deliver events on a bounded per-subscriber worker The previous fix dispatched each event to every subscriber with a bare `go safeCall`. That unblocked the dispatch loop, but removed the bus's backpressure: under a login-attempt flood (which both notifier subscribers process without rate-limiting) with email/Telegram enabled, every attempt spawned handler goroutines that each block on network I/O for up to ~30s, with no bound — a goroutine and outbound-connection storm. It also let a subscriber's handler run concurrently with itself, racing the Telegram notifier's lazily-cached hostname. Give each subscriber its own bounded queue drained by a single worker goroutine. Dispatch does a non-blocking send per subscriber (dropping only that subscriber's event when its queue is full), so a slow subscriber still can't stall the others, concurrency is bounded to one in-flight handler per subscriber, per-subscriber event order is preserved, and Stop again waits for in-flight handlers to finish. * fix(frontend): map outbound mobile-card actions through the real index too The desktop outbounds table was keyed by the outbound's real index, but the mobile card list was left keying the probe trigger and every test-state lookup by the positional row index. With a hidden balancer-loopback outbound present, tapping Check on a mobile card probed the wrong outbound and the Test-All results landed on the wrong card. Key onTest and the testResult/isTesting reads by record.key, matching the desktop columns. * fix(frontend): meet WCAG AA contrast on the config-block link text The Storybook accessibility test flagged the share-link <code> block: with no explicit color it inherited a muted grey that renders as #888888 on the #f8f8f8 tertiary-fill background in CI's Chromium — a 3.33:1 contrast, below the 4.5:1 AA threshold. Set the text to the theme's primary text token so the colour is explicit and high-contrast in both light and dark themes instead of depending on an inherited value that varies by browser. * style(sub): simplify a negated conjunction to satisfy staticcheck QF1001 golangci-lint (staticcheck QF1001) flagged the `!(a && b)` guard in expandSegment. Rewrite it via De Morgan's law to the equivalent `!a || !b` form so the linter passes; behavior is unchanged. * fix: close panics and races the audit's own fixes left nearby Second-pass review of the 54-commit self-correcting audit. Each item below was confirmed by reading the surrounding source (and, where practical, the pre-fix code) before being changed; regression tests are included for every behavioral fix. Concurrency: - eventbus: Bus.Subscribe called wg.Add with no synchronization against a concurrent Bus.Stop's wg.Wait, a real "WaitGroup misuse" panic risk (e.g. a Telegram-bot settings save racing panel shutdown/restart). Stop now flips a mu-guarded `stopped` flag before waiting, and Subscribe checks it under the same lock, so Add and Wait can no longer race. Security: - login_limiter: evictForRoom's fallback eviction picked an arbitrary map key, including ones still under an active cooldown - an attacker flooding /login with fresh usernames could evict their own (or anyone's) blocked record and reset the lockout. The fallback now skips actively-blocked records, only falling back to an unconditional evict if the map is somehow entirely full of active blocks (preserves the hard memory cap). Subscription-endpoint panics (reachable by any client hitting /sub): - internal/sub/service.go: applyPathAndHostParams/Obj (ws/httpupgrade/xhttp with no path settings object) and the TLS alpn readers in three places used unchecked type assertions - exactly the bug class abab7cd0 patched elsewhere in the same switch statements, just not these call sites. - internal/sub/json_service.go, clash_service.go: the externalProxy loops in the JSON and Clash generators used unchecked assertions on a legacy/admin-supplied field (missing "port", non-object entry, etc.). - internal/sub/json_service.go: realityData's shortId/serverName selection could assert a non-string array element. Other correctness: - client_traffic.go: ResetAllTraffics (touched by 3eb214d0) still skipped clearing NodeClientTraffic node-sync baselines, unlike its sibling reset paths in the same file - a node's next sync would re-add pre-reset delta on top of the freshly-zeroed counter. - inbound_traffic.go: the traffic-tick tx's Commit/Rollback errors were silently discarded; now logged so a backend-level commit failure (e.g. an aborted Postgres tx from a best-effort helper) doesn't masquerade as a successful tick. - outbound_subscription.go: the new subscriptionFetchClient doc comment was wedged between fetchAndStore's existing comment and fetchAndStore itself, leaving fetchAndStore undocumented and the comment describing the wrong function. Convention cleanup: - Removed narrative // comments added by the audit that violate this repo's no-inline-comment rule (mostly narrating the specific bug/fix rather than a lasting contract, and mostly on new Test functions, which this repo's existing tests never comment) - calibrated against this exact codebase's own pre-existing comment style so legitimate godoc-style doc comments were left alone. --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-07-16 15:33:06 -07:00
return
}
if err := s.RestartXray(false); err != nil {
logger.Error("reconcile xray failed:", err)
if restart {
s.SetToNeedRestart()
}
if reconcile {
s.SetToNeedReconcile()
}
Repo-wide self-correcting audit: 54 verified bug fixes (#5970) * fix(email): resolve a name-addr smtpFrom into bare envelope address and display name The save-time validator accepts any RFC 5322 address form, so a value like '3x-ui Panel <panel(at)example.com>' passes validation, but Send and TestConnection fed that raw string to MAIL FROM, which strict servers reject with 501, and buildMessage mangled it into a quoted local part. Parse the configured sender at the point of use: the envelope gets the bare address and, when no explicit sender name is set, the display name embedded in the setting is used for the From header. * fix(email): report a missing sender address from the SMTP connection test TestConnection skipped the empty-from guard that Send enforces, so with no sender and no username configured the test issued the null reverse-path and could report success against a lenient relay while every real notification send kept failing with the missing-sender error. Guard the test path the same way and surface a dedicated translated message. * fix(sub): fall back to the raw subscription when an auto-detected format has no content With format auto-detection enabled, a client whose User-Agent matched the Clash or JSON regex was routed straight to that format handler. For a subscription whose entries convert to neither format (an MTProto-only subscription, for example) the handler returns an empty document and the request ended as 404, breaking a URL that served the raw list before the toggle. The auto-detect branches now serve the detected format only when it produces content and otherwise continue to the raw response; the explicit format endpoints keep answering 404 for empty documents. * fix(node): match prefixed central tags when filtering a selected-mode node snapshot FilterNodeSnapshot compared a node snapshot's inbound tags against the raw selected-tag list with an exact match, while its two siblings (SnapshotHasUnadoptedInbounds and the reconcile tagToCentral map) expand each selected tag to both its bare node-side form and its n<id>- prefixed central form. A panel-created node inbound is recorded in the selected list under the central prefixed tag but reported by the node under the bare tag, so the exact match dropped it from every snapshot and the orphan sweep then deleted its central row one tick after creation. Expand the allowed set with the same prefix flip the siblings use. * fix(client): refuse a bulk quota reduction that would fall to or below zero BulkAdjust clamped a client's new traffic limit with max(total+addBytes, 0). Because 0 is the unlimited sentinel, reducing a client's quota by more than it had left silently granted that client unlimited traffic. The sibling expiry branch already refuses an over-reduction; mirror it for quota so the adjustment is skipped with a clear reason instead of crossing the sentinel. * fix(client): persist a bulk adjustment's applied field even when the sibling field is skipped In a mixed BulkAdjust (both a days delta and a bytes delta), a per-field planning skip such as "unlimited expiry" or "unlimited traffic" was recorded in the same map that gated the client_traffics write. The applied field was already written to the inbound JSON and the clients table, but the enforcement row was left untouched, so the depletion job cut the client on the old limit while the panel showed the new one. Gate the traffic-row write on an actual inbound-processing failure rather than on any planning-phase skip note. * fix(inbound): always create in AddInbound instead of overwriting a row whose id was posted The add controller binds the inbound model's id form field and never clears it, and AddInbound persisted with GORM Save, which updates in place when the primary key is non-zero. A client that reused an existing id (for instance by duplicating an inbound fetched from /get and changing the port) silently overwrote that stored row instead of creating a new inbound. Zero the id at the top of AddInbound, matching how it already zeroes the client-stat ids. * fix(inbound): accept WireGuard clients when creating an inbound AddInbound's per-client validation switch had cases for every protocol except WireGuard, so a WireGuard client fell through to the default branch that requires a non-empty id. WireGuard clients are keyed by their public key and carry no id, so importing a WireGuard inbound or re-adding one to a reconciling node was rejected with "empty client ID". Add a wireguard case that validates the client key, mirroring addInboundClient. * fix(client): stop holding the inbound-lock registry mutex while waiting on one inbound lockInbound acquired the global registry mutex and then blocked on the per-inbound mutex without releasing the registry first. A slow client operation holding one inbound's mutex (for example a bulk delete pushing to an unreachable node) made the next waiter park on that inbound while still holding the registry mutex, which in turn blocked lockInbound for every other inbound — freezing client mutations panel-wide. Release the registry mutex before taking the per-inbound lock. * fix(client): honor keepTraffic when deleting a client that is attached to inbounds Delete, DeleteByEmail and BulkDelete all pass keepTraffic to their final cleanup transaction, but each called the per-inbound delete helper with a hardcoded false. That helper purges the client's traffic, IP and stat rows before the gated cleanup runs, so keepTraffic=true still destroyed all traffic history for any client actually attached to an inbound (the pinned test only covered a record with no inbound mappings). Thread the caller's keepTraffic through to the per-inbound helper at all three call sites. * fix(inbound): defer a local MTProto inbound edit's sidecar push until after commit UpdateInbound applied a local MTProto inbound change by calling the runtime UpdateInbound (which stops/starts the mtg sidecar or talks to it) from inside runSerializedTx. That runs process and network I/O on the single traffic-writer goroutine while a DB transaction is open, so a slow sidecar stalls traffic accounting and every concurrent client mutation, and a later step failing the transaction leaves the sidecar ahead of the rolled-back row. Move the push into the post-commit hook, matching the xray branch. Adds a SetLocalRuntimeOverride test seam mirroring the existing node override so the deferral is regression tested. * fix(client): delete external-link rows when bulk-deleting clients The single-client Delete path removes a client's client_external_links rows, but BulkDelete (and the DelDepleted reaper that routes through it) deleted the record, mappings and traffic while leaving the external-link rows keyed by the now-dead client id, so they accumulated as orphans. Delete them in the same cleanup transaction, keyed by client id like the single path. * fix(inbound): request an xray restart when toggling a routed MTProto inbound AddInbound, DelInbound and UpdateInbound all flag needRestart when an inbound routes MTProto through xray, so the egress SOCKS bridge is regenerated. Only SetInboundEnable's local path omitted it, so toggling a routed MTProto inbound off then on left the bridge out of the running config while the sidecar dialed its loopback port, blackholing that inbound until an unrelated restart. Flag the restart on the local enable path too. * fix(client): apply enable-by-email to every inbound a client is attached to ToggleClientEnableByEmail (Telegram bot) and SetClientEnableByEmail (LDAP sync) resolved a single inbound via the legacy client_traffics pointer and flipped enable only there. A client attached to several inbounds kept connecting through the siblings' running Xray after being disabled, and the next edit could re-enable it everywhere from a stale sibling. Route both through the applyClientFieldByEmail fan-out (the #5039 fix path) so the whole multi-inbound identity is toggled at once, dropping the circular Set/Toggle dependency. * fix(traffic): commit a traffic tick even when a best-effort maintenance helper fails addTrafficLocked stages the inbound and client deltas, then runs three helpers (auto-renew, disable depleted clients, disable depleted inbounds) that are meant to log and continue. All three reused the function-scope err that the deferred commit/rollback inspects, so the last helper's error decided the whole tick: a failure in disableInvalidInbounds rolled back the already-staged traffic while AddTraffic reported success, and because xray had already advanced its counter baseline that traffic was lost for good. Give each best-effort helper its own error variable so only a genuine staging failure rolls the tick back. * fix(traffic): re-enable clients and serialize the write in Reset All Client Traffic ClientService.ResetAllTraffics zeroed up/down but, unlike every sibling reset path, never restored enable=true, so clients that had been auto-disabled for exceeding their quota stayed cut with zero usage after a reset. It also wrote client_traffics directly on the shared DB handle instead of through the serial traffic writer, reintroducing the cross-transaction lock-order deadlock the writer exists to prevent. Restore enable and run the reset inside submitTrafficWrite within one transaction. * fix(traffic): keep node reset propagation out of the serial traffic writer ResetAllTraffics and ResetInboundTraffic performed their remote-node reset HTTP calls inside submitTrafficWrite. Each call can block up to the remote timeout, and Reset All Traffics loops every node serially, so the single traffic-writer goroutine was held for seconds — long enough that the concurrent 5s traffic poll timed out submitting its own write and dropped the deltas it had already drained from xray. Do the DB reset inside the writer, then propagate to the nodes after it returns, matching how the mtproto quota reset is already sequenced. * fix(sub): stop the subscription from 500ing on valid-but-unusual stream settings The raw share-link generators used unchecked type assertions and unguarded array indexing: an empty Reality shortIds/serverNames array (random.Num(0) panics), a tcp-http header with no request block or an empty request.path, a grpc block missing its keys, empty stream settings, and a non-string Host header all panicked mid-generation. Because getSubs loops every client's link with no recover, one such client 500s the entire subscription for everyone. The sibling JSON, Clash and frontend generators already guard these; make the raw generators match with comma-ok assertions and length checks. * fix(sub): tolerate a hysteria inbound without hysteriaSettings in the JSON subscription genHy asserted stream["hysteriaSettings"].(map[string]any) without the comma-ok form, so a hysteria inbound whose StreamSettings omit the hysteriaSettings key (a valid, representable shape the raw generator renders fine) panicked and 500ed the entire JSON subscription. Use comma-ok; the downstream reads already guard each key, so a nil map degrades gracefully. * fix(sub): emit the pinned peer cert sha256 in Clash subscriptions The Clash stream builder computed tlsSettings["pin-sha256"] from the inbound's pinnedPeerCertSha256, but applySecurity's tls case never copied it onto the proxy, so it was written with no reader and silently dropped. Clash subscribers lost certificate pinning while JSON subscribers kept it. Surface pin-sha256 on the proxy in the tls case, matching the JSON emitter. * fix(link): parse the snake_case and extra-blob xhttp fields when importing a share link The panel's share-link emitters (Go and TS) carry advanced xhttp knobs as a snake_case x_padding_bytes plus an extra=<json> payload, but the Go parser's xhttp branch read only top-level camelCase params, so importing an xhttp link via the outbound-subscription feature dropped xPaddingBytes, scMaxEachPostBytes and the rest, silently reverting them to the stream defaults and producing a non-working outbound. Mirror the TS parser: read the snake_case alias, merge the extra JSON blob, then let explicit camelCase params win. * fix(frontend): decode URL-safe base64 when parsing an imported share link Base64.decode called window.atob directly, which rejects the base64url alphabet (- and _) and unpadded input. But the panel's own share-link emitter uses Base64.encode(x, true) (URL-safe, unpadded), and real SIP002 links do too, so importing a Shadowsocks link whose method:password encodes with a - or _ threw, fell back to the raw undecoded string, and produced a wrong method and garbage password (the vmess parser shared the same limitation). Normalize base64url and re-pad before atob so decode round-trips every emitted link. * fix(link): honor the vmess ws path and hysteria2 vcn params on import Two Go/TS parser parity gaps in the outbound share-link import path: parseVmess only applied a ws link's path when the inner JSON also carried a host key, so a generator that omits host dropped the path back to the default; and parseHysteria2 hardcoded verifyPeerCertByName to empty, ignoring the vcn param the panel emits, so a hysteria2 outbound with a decoy SNI and a distinct cert name failed TLS verification after import. The TS parser handles both; make the Go parser match. * fix(ui): stop the sniffing form island from clobbering unrendered fields antd's Form.useWatch only reports registered fields, so while the sniffing toggle was off the island emitted { enabled: false } upward and replaced the full Sniffing object in form state. Saving a VLESS reverse outbound then crashed in sniffingToWire on the missing ipsExcluded array; the loopback outbound and the inbound sniffing tab shared the same hole. Watch the store with preserve: true so unrendered fields keep their values, and seed a missing value from the schema defaults instead of an empty cast. * fix(sub): drop empty remark segments instead of leaving a stray separator expandSegment dropped a "|" segment only when its tokens rendered the unlimited mark, so a segment whose only token resolved to the empty string (a client with no comment, an unlimited client's expiry date) was kept as bare decoration, leaving a trailing "|" or a dangling emoji on every share link's remark. Drop a token-bearing segment whenever none of its tokens produce a real value, while still keeping pure-literal segments. * fix(xray): keep source- and domains-scoped routing rules when an inbound is deleted removeInboundTagFromRules drops a routing rule whose inboundTag list becomes empty only if the rule has no other matcher, but routingMatcherKeys omitted xray-core's canonical source and domains keys. A rule scoped by source or domains (common in hand-authored or imported configs) therefore lost its whole body — including a security-relevant block — when its single listed inbound was deleted, instead of just having the tag trimmed. Recognize source and domains as live matchers. * fix(xray): guard RemoveUser against an uninitialized handler client Every XrayAPI handler method returns an error when HandlerServiceClient is nil, except RemoveUser, which dereferenced it directly. A depletion sweep runs Init with the port ignored and, during a restart window where the fresh process's api port is still 0, Init fails and leaves the client nil — so RemoveUser panicked (recovered by the traffic writer, but re-thrown every poll) instead of returning an error. Add the same nil guard the siblings have. * fix(xray): do not revive a manually stopped Xray on a background restart RestartXray cleared isManuallyStopped unconditionally at its top, so the @30s pending-config cron (and warp/ldap/outbound reconcile jobs) that call RestartXray(false) resurrected an Xray the admin had deliberately stopped — unlike the crash-detector, which honors the manual-stop flag. Skip a non-forced restart while the stop flag is set; only an explicit forced restart clears it. * fix(xray): retry a failed pending-restart instead of dropping the config change The 30s cron consumed the need-restart flag with IsNeedRestartAndSetFalse before calling RestartXray and only logged a failure. If RestartXray failed early (a transient GetXrayConfig DB error) the old process kept running the old config, the crash detector saw a running process and never retried, and the flag stayed cleared — so an admin's saved change silently never reached the core. Move the consume/restart/retry into ApplyPendingRestart, which re-arms the flag on failure so the next tick retries. * fix(xray): synchronize the process version and apiPort fields Start writes p.version and p.apiPort (via refreshVersion/refreshAPIPort) after flipping the process to running, while GetXrayVersion and GetAPIPort read them lock-free from the status and traffic poll goroutines. The struct mutex deliberately excluded these fields, so a restart racing a poll was a real data race — a torn read of the version string header can crash. Extend the mutex to cover version and apiPort, doing the blocking version probe before taking the lock. * fix(settings): detect a wildcard listen collision between the web and sub ports The web/sub same-port check compared the two listen addresses as raw strings, so binding both on all interfaces with different spellings (webListen 0.0.0.0 vs an empty subListen) slipped past validation and only failed at startup with an opaque bind error. Treat any wildcard listen ('', 0.0.0.0, ::) as overlapping so the clash is reported up front, while still allowing two distinct specific addresses to share a port. * fix(db): mark the IP-limit cleanup seeder done on a fresh install ResetIpLimitNoFail2ban is a one-time migration that, on a host without fail2ban, zeroes every existing client's limitIp because the limit can't be enforced. It was missing from the fresh-install fast-path seeder list, so on a brand-new DB it did not run on the first boot but fired on the second — wiping any IP limits the admin had set in between. Add it to the fast-path so a truly fresh install marks it done up front (there is nothing to clean), leaving later admin-set limits intact. * fix(security): dial outbound subscriptions through the SSRF guard The outbound-subscription fetch validated the URL host once (resolving DNS and rejecting private targets) but then fetched with a plain HTTP client that re-resolves the host at dial time, so a subscription domain the attacker controls could pass validation as a public IP and rebind to 127.0.0.1 / a cloud metadata endpoint / an internal host for the actual dial — a blind SSRF into the panel's network. Route the direct fetch (and its redirects) through netsafe.SSRFGuardedDialContext, which resolves, checks and dials the same IP atomically, carrying the subscription's AllowPrivate flag on the request context; a configured egress proxy still dials its loopback bridge unguarded. * fix(security): bound the login-limiter attempts map The login rate limiter keys its records on the caller-supplied username and only evicted a record when that exact key was revisited or the login succeeded. An unauthenticated attacker replaying one CSRF token while rotating a fresh username per request seeded a record that was never revisited, growing the map without bound until the panel OOMs. Cap the map: before inserting a new record, reclaim records whose block has lapsed and whose failures aged out, and if the map is still at the ceiling under a broad flood, drop one so memory can never grow past the cap. * fix(tgbot): require admin for privileged callbacks, not just the first switch answerCallback wraps only its first callback switch in an isAdmin guard; the second switch (server usage, inbound/online enumeration, database backup export, ban logs, mass traffic reset, client creation) ran for every caller. Telegram delivers a callback with the tapping user's id, so a non-admin who can see an admin's inline keyboard — as when the bot runs in a group — could tap Backup and receive the full database and config, or reset all traffic. Default-deny before the second switch: a non-admin may only run the per-user client_* callbacks that resolve their own data from their Telegram id. * fix(eventbus): dispatch each subscriber in its own goroutine The fan-out loop called every subscriber's handler sequentially on the single dispatch goroutine. The email and Telegram notifiers block on network I/O for tens of seconds (or minutes when the remote is slow), so one slow subscriber stalled the whole loop: the 256-slot channel then filled and Publish silently dropped later events — including high-value xray.crash and node.down notifications unrelated to the slow handler. Hand each delivered event to every handler in its own goroutine so a blocking subscriber can no longer stall delivery to the others. safeCall already recovers panics, so a detached handler cannot take down the bus. * fix(integration): cap WARP API response body size doWarpRequest read the response with an unbounded io.ReadAll, unlike the sibling NordVPN client which already caps every read at maxResponseSize. A hostile panel egress proxy or a MITM on the Cloudflare WARP endpoint could stream an arbitrarily large body and force the panel into an unbounded allocation. Wrap the body in an io.LimitReader(maxResponseSize) to match the NordVPN client. * fix(email): bound every SMTP step with a connection deadline The "starttls"/"none" transport delivered through net/smtp.SendMail, which dials with an untimed net.Dial and never sets a socket deadline. When an SMTP server accepted the TCP connection but then stalled (or was a blackhole), the caller was released by Send's 30s select, but the sender goroutine and its socket stayed blocked until the OS TCP timeout — minutes per notification, leaking a goroutine and a connection each time. sendWithTLS dialed with a timeout but likewise armed no deadline on the protocol phase, and TestConnection (called synchronously from the settings handler, with no select guard) could hang the request indefinitely. Replace SendMail with sendPlain, which dials with smtpConnectTimeout and arms conn.SetDeadline(smtpDeadline) before the greeting read, preserving SendMail's opportunistic STARTTLS upgrade. Arm the same deadline in sendWithTLS and TestConnection so every SMTP step is bounded. * fix(server): guard access-log parser against malformed lines GetXrayLogs split each Xray access-log line on whitespace and then read fixed offsets — parts[1] for the timestamp and parts[i+1] after the "from", "accepted" and "email:" markers — without checking the line had that many fields. A truncated or malformed line (the logged destination is attacker-influenced) indexed past the slice and panicked; the panel handler returned a 500 via Gin's recovery. Extract the per-line field parsing into parseAccessLogFields and length guard every positional lookup so a short line yields a partial entry instead of panicking. * fix(server): guard xray key-generator output parsing GetNewX25519Cert, GetNewmldsa65 and GetNewmlkem768 parsed xray's stdout by reading lines[0], lines[1] and each line's second colon-separated field without any length check — unlike GetNewEchCert, which already guards its line count. If the xray binary printed fewer than two lines or reformatted its labels (a version change, or a silent failure that emitted nothing), the fixed slice index panicked and the handler 500'd. Extract the shared parsing into parseXrayKeyPairOutput, which length guards the line count and each label split and returns an error instead of panicking, then route all three generators through it. * fix(tgbot): stop auto-deleted messages from resetting wizard state SendMsgToTgbotDeleteAfter spawns a goroutine that, after the display delay, deleted the transient message and then unconditionally cleared the chat's conversation state. Every caller that ends a wizard step already clears the state synchronously, so that call was redundant — and harmful: if within the delay the user advanced to the next step (a callback sets a fresh awaiting_* state), the late goroutine wiped it, and the user's next message fell through unrecognized, silently dropping their input. Move the delayed deletion into deleteMessageAfterDelay, which only removes the message and no longer touches the conversation state. Guard deleteMessageTgBot against a nil bot so the deletion path is unit-testable. * fix(frontend): refetch a fresh CSRF token on 403 instead of reusing the stale meta tag On a 403 to an unsafe method the client cleared its cached CSRF token and called ensureCsrfToken to retry. But ensureCsrfToken prefers the <meta name="csrf-token"> tag baked into the page, which the production panel always injects, so the "refresh" re-read the same stale token and the /csrf-token refetch was never reached — the retry re-sent the token that had just been rejected and the save failed with an error toast. The token lives in the session and rotates when the session is regenerated (for example re-login in another tab), leaving the tab's baked-in meta token stale. Fetch the current token straight from /csrf-token in the 403 branch so the retry uses the authoritative server value. The existing tests only passed because they strip the meta tag; the new test keeps a stale tag present. * fix(frontend): surface backend error text from failed requests HttpUtil.get/post read the thrown HttpError body as response.data.message, but the backend error envelope (entity.Msg) serializes its text as msg. On any non-2xx JSON response the real reason was therefore dropped and the operator saw only the generic "Request failed with status N" toast. Read response.data.msg first (keeping message and the native error text as fallbacks). The sibling test had pinned the wrong body shape ({ message }); correct it to the real backend shape ({ success:false, msg }) so it exercises the actual envelope. * fix(frontend): share one WebSocket connection across bridge and hooks websocketBridge.ts and useWebSocket.ts each declared their own module-scoped sharedClient plus an identical getSharedClient, so the "shared" client was not shared between them: whenever a page using useWebSocket (Clients/Inbounds) mounted alongside the always-mounted bridge, the panel opened two sockets to /ws. The server then pushed every traffic/stats/nodes/inbounds snapshot to both, doubling WebSocket bandwidth and running two independent reconnect loops, and the hook's socket was never disconnected on unmount. Hoist a single getSharedWebSocketClient into api/websocket.ts and route both the bridge and the hook through it, so exactly one connection is opened. * fix(frontend): guard the outbounds WebSocket handler against non-array payloads onOutbounds wrote the raw WebSocket payload straight into the outboundsTraffic cache, unlike the sibling onNodes/onInbounds handlers which first check Array.isArray. A malformed non-array push (for example an object) would land in the cache with staleTime Infinity; consumers that call .find()/.map() on the outbounds list would then throw and crash the Outbounds tab. Add the same Array.isArray guard so a bad push is ignored. * fix(frontend): key the node table by the computed row key, not id The desktop node table used rowKey="id", but transitive sub-nodes (the read-only rows surfaced from downstream nodes) all carry id 0, so a topology with two or more transitive rows gave React duplicate keys. antd's rowKey prop overrides the row object's own computed `key` (`t-${guid}` for transitive rows, the numeric id otherwise), so the unique key the code already builds was ignored — causing row-state/DOM mis-association on any re-render (heartbeat refetch, address-eye toggle). The mobile card path already keyed by record.key. Key the table by "key" so transitive rows get their distinct t-${guid} identity; direct nodes keep key === id, so row selection (filtered to numeric keys) is unchanged. * fix(frontend): map routing row actions through the rule's real index The routing table hides balancer-loopback rules (`_bl_*`) but keeps each visible row's original index in `key`, then handed antd's positional row index straight to edit/delete/toggle/move/drag — all of which mutate the full, unfiltered routing.rules array. Once a hidden loopback rule precedes a visible one (e.g. a balancer whose fallback is another balancer, plus any rule added afterwards), the positional index no longer matches the array index, so deleting or editing a rule silently hit the wrong one — including destroying the loopback rule that keeps the balancer alive. Add originalRuleIndex to translate a positional row index back through the row's `key`, and route every mutating handler (openEdit, confirmDelete, toggleRule, moveUp/moveDown, drag) through it. When no loopback rows are hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): map outbound row actions through the outbound's real index The outbounds table hides balancer-loopback outbounds (`_bl_*`) but keeps each visible row's original index in `key`, then passed antd's positional row index to edit/delete/move and to the per-row probe (onTest) and its result lookup — all of which address the full, unfiltered outbounds array. Once a hidden loopback outbound precedes a visible one, the positional index diverges from the array index, so deleting or editing an outbound hit the wrong one (its deletion-impact plan and removal targeting the wrong entry), and the test button probed / showed results against the wrong outbound. Add originalOutboundIndex and route the mutating handlers through it; key the probe trigger and test-result columns by record.key. With no loopback rows hidden the mapping is the identity, so ordinary configs are unaffected. * fix(frontend): tolerate a malformed happyEyeballs value in the Xray Basics tab BasicsTab derived directHappyEyeballs by calling HappyEyeballsSchema.parse during render, guarding only against null/non-object. A wrong-typed field (e.g. happyEyeballs.tryDelayMs as a string) or any other shape mismatch — reachable via the Complete Template JSON editor or an imported config — threw straight out of render, white-screening the default Xray landing tab. Use safeParse and fall back to null so a bad value degrades to "no override" instead of crashing the page. * fix(frontend): preserve routing-rule fields the form does not surface The rule form rebuilt the rule from a fixed literal of only the fields it edits, and RoutingTab replaces the rule wholesale on confirm. Fields the form never exposes — localPort, localIP, process, ruleTag, webhook — are in the rule schema and can arrive via the advanced JSON editor or Import Rules; opening such a rule in the form and saving silently dropped them. Carry over every key of the original rule the form does not manage before applying the form-derived fields, so an edit only touches what it surfaces. * fix(frontend): re-sync the sniffing island when its value changes externally The sniffing config editor froze its seed value at mount and only watched its own inner AntD form, never reflecting a later change to the shared RHF `sniffing` path. Because the inbound form mounts every tab with forceRender, the friendly Sniffing tab and the Advanced JSON editor are live at once: editing sniffing in the JSON editor updated the RHF value but not the frozen island, so the next interaction with the friendly tab emitted the stale value and silently discarded the JSON edit. Add an effect that pushes an external value change into the inner form, guarded by the same lastEmitted marker the emit path uses so the island never re-seeds from its own echo and no update loop forms. * fix(frontend): don't drift a client's byte quota on a no-op save The quota field shows the total in GB rounded to two decimals; editing a client and saving converted that display value straight back to bytes. A byte total not aligned to 0.01 GB — one set via the API or an import — was therefore rewritten to the rounded value on any save that never touched the field, losing a few MB each time. Add resolveTotalBytes: keep the original byte total when the displayed GB still matches it, and only re-derive from GB when the user actually changed the field. * fix(eventbus): deliver events on a bounded per-subscriber worker The previous fix dispatched each event to every subscriber with a bare `go safeCall`. That unblocked the dispatch loop, but removed the bus's backpressure: under a login-attempt flood (which both notifier subscribers process without rate-limiting) with email/Telegram enabled, every attempt spawned handler goroutines that each block on network I/O for up to ~30s, with no bound — a goroutine and outbound-connection storm. It also let a subscriber's handler run concurrently with itself, racing the Telegram notifier's lazily-cached hostname. Give each subscriber its own bounded queue drained by a single worker goroutine. Dispatch does a non-blocking send per subscriber (dropping only that subscriber's event when its queue is full), so a slow subscriber still can't stall the others, concurrency is bounded to one in-flight handler per subscriber, per-subscriber event order is preserved, and Stop again waits for in-flight handlers to finish. * fix(frontend): map outbound mobile-card actions through the real index too The desktop outbounds table was keyed by the outbound's real index, but the mobile card list was left keying the probe trigger and every test-state lookup by the positional row index. With a hidden balancer-loopback outbound present, tapping Check on a mobile card probed the wrong outbound and the Test-All results landed on the wrong card. Key onTest and the testResult/isTesting reads by record.key, matching the desktop columns. * fix(frontend): meet WCAG AA contrast on the config-block link text The Storybook accessibility test flagged the share-link <code> block: with no explicit color it inherited a muted grey that renders as #888888 on the #f8f8f8 tertiary-fill background in CI's Chromium — a 3.33:1 contrast, below the 4.5:1 AA threshold. Set the text to the theme's primary text token so the colour is explicit and high-contrast in both light and dark themes instead of depending on an inherited value that varies by browser. * style(sub): simplify a negated conjunction to satisfy staticcheck QF1001 golangci-lint (staticcheck QF1001) flagged the `!(a && b)` guard in expandSegment. Rewrite it via De Morgan's law to the equivalent `!a || !b` form so the linter passes; behavior is unchanged. * fix: close panics and races the audit's own fixes left nearby Second-pass review of the 54-commit self-correcting audit. Each item below was confirmed by reading the surrounding source (and, where practical, the pre-fix code) before being changed; regression tests are included for every behavioral fix. Concurrency: - eventbus: Bus.Subscribe called wg.Add with no synchronization against a concurrent Bus.Stop's wg.Wait, a real "WaitGroup misuse" panic risk (e.g. a Telegram-bot settings save racing panel shutdown/restart). Stop now flips a mu-guarded `stopped` flag before waiting, and Subscribe checks it under the same lock, so Add and Wait can no longer race. Security: - login_limiter: evictForRoom's fallback eviction picked an arbitrary map key, including ones still under an active cooldown - an attacker flooding /login with fresh usernames could evict their own (or anyone's) blocked record and reset the lockout. The fallback now skips actively-blocked records, only falling back to an unconditional evict if the map is somehow entirely full of active blocks (preserves the hard memory cap). Subscription-endpoint panics (reachable by any client hitting /sub): - internal/sub/service.go: applyPathAndHostParams/Obj (ws/httpupgrade/xhttp with no path settings object) and the TLS alpn readers in three places used unchecked type assertions - exactly the bug class abab7cd0 patched elsewhere in the same switch statements, just not these call sites. - internal/sub/json_service.go, clash_service.go: the externalProxy loops in the JSON and Clash generators used unchecked assertions on a legacy/admin-supplied field (missing "port", non-object entry, etc.). - internal/sub/json_service.go: realityData's shortId/serverName selection could assert a non-string array element. Other correctness: - client_traffic.go: ResetAllTraffics (touched by 3eb214d0) still skipped clearing NodeClientTraffic node-sync baselines, unlike its sibling reset paths in the same file - a node's next sync would re-add pre-reset delta on top of the freshly-zeroed counter. - inbound_traffic.go: the traffic-tick tx's Commit/Rollback errors were silently discarded; now logged so a backend-level commit failure (e.g. an aborted Postgres tx from a best-effort helper) doesn't masquerade as a successful tick. - outbound_subscription.go: the new subscriptionFetchClient doc comment was wedged between fetchAndStore's existing comment and fetchAndStore itself, leaving fetchAndStore undocumented and the comment describing the wrong function. Convention cleanup: - Removed narrative // comments added by the audit that violate this repo's no-inline-comment rule (mostly narrating the specific bug/fix rather than a lasting contract, and mostly on new Test functions, which this repo's existing tests never comment) - calibrated against this exact codebase's own pre-existing comment style so legitimate godoc-style doc comments were left alone. --------- Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com>
2026-07-16 15:33:06 -07:00
}
}
2025-09-20 00:35:50 -07:00
// DidXrayCrash checks if Xray crashed by verifying it's not running and wasn't manually stopped.
func (s *XrayService) DidXrayCrash() bool {
return !s.IsXrayRunning() && !isManuallyStopped.Load()
}
feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) (#5506) * feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) xray-core v26.6.22 (PR #6258) renamed the XHTTP session config keys sessionPlacement/sessionKey to sessionIDPlacement/sessionIDKey (no fallback kept in core) and added sessionIDTable (predefined charset name or literal ASCII) and sessionIDLength (range, e.g. 16-32, lower bound > 0). Panel changes: - Schema (xhttp.ts): rename the two keys, add sessionIDTable/sessionIDLength, and a z.preprocess that lifts legacy keys off stored configs so an upgraded panel never silently drops a saved session setting. - Wire normalize + share-link build/parse: rename keys, emit the two new fields, and accept legacy sessionPlacement/sessionKey from old share links. - Inbound + outbound XHTTP forms: rename field paths, add a sessionIDTable autocomplete (9 predefined tables + free ASCII) and a sessionIDLength range input shown only when a table is set, with light client validation (ASCII table, length min > 0; xray enforces the room-size minimum server-side). - Subscription (service.go) and Clash (clash_service.go) builders: emit the renamed + new keys, with a legacy fallback for not-yet-resaved inbounds. - Locales: add sessionIDTable/sessionIDLength labels + hints in all 13 files. Two sibling v26.6.22 XHTTP commits need no panel change and are covered by the core bump alone: #6332 (XHTTP/3 closes QUIC/UDP) and #6320 (udpHop honors the existing dialerProxy). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(xhttp): add Session ID Table to inbound form-blocks snapshot The new sessionIDTable input renders by default in the inbound XHTTP form, so its label joins the field-structure snapshot. sessionIDLength stays conditional (only shown when a table is set), so it does not appear here. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xhttp): migrate legacy session keys in the running xray config The Zod preprocess plus the subscription/Clash fallbacks only covered the panel UI and share-link output. The config handed to the running xray-core process is built from the raw stored streamSettings in GetXrayConfig, which did not rewrite the renamed XHTTP session keys — so a pre-upgrade inbound (or template outbound) stored with a non-default sessionPlacement was emitted unchanged and dropped by xray-core v26.6.22, until the admin re-saved it. Lift sessionPlacement/sessionKey onto sessionIDPlacement/sessionIDKey at config-generation time, in the existing inbound stream-rewrite block (next to the tls/reality/externalProxy handling) and across template outbounds. The lift is idempotent and leaves unchanged configs byte-identical so the hot-reload diff never sees a spurious change. Also tighten validateSessionIDLength to reject an inverted range (e.g. 32-16) in addition to the existing lower-bound > 0 check. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xray): avoid summed-capacity allocation in mergeSubscriptionOutbounds CodeQL go/allocation-size-overflow flagged the pre-sized make() whose capacity was a sum of three slice lengths. Grow the slice via append on a nil slice instead; same result, no overflow-prone capacity expression.
2026-06-23 08:38:16 -07:00
// liftXhttpSessionIDKeys renames the legacy XHTTP session keys
// (sessionPlacement/sessionKey) to the v26.6.22 #6258 names
// (sessionIDPlacement/sessionIDKey) inside a streamSettings map. xray-core kept
// no fallback for the old names, so a config stored before the rename would be
// silently ignored by the engine. Returns true if it changed anything.
func liftXhttpSessionIDKeys(stream map[string]any) bool {
xhttp, ok := stream["xhttpSettings"].(map[string]any)
if !ok {
return false
}
changed := false
for legacy, renamed := range map[string]string{
"sessionPlacement": "sessionIDPlacement",
"sessionKey": "sessionIDKey",
} {
v, has := xhttp[legacy]
if !has {
continue
}
if _, exists := xhttp[renamed]; !exists {
xhttp[renamed] = v
}
delete(xhttp, legacy)
changed = true
}
return changed
}
feat(xray): update xray-core to v26.10.10 and adapt panel Move the three Xray binary pins (DockerInit.sh, both release.yml fetches) to v26.10.10, in lockstep with the xtls/xray-core module that 82c77111 already bumped to 701af60772cd. xdns finalmask (#7090): a domain now takes a "names" list instead of a single "name", and a resolver is an "addrs" list of udp:// or tcp:// URLs instead of {type, settings:{addr}}. The loader ignores the old keys rather than failing, so every mask the 26.9.30 seeder wrote would load with no domain and no resolver and the tunnel would carry nothing. maskcompat.UpgradeLegacyXdns now lifts both the pre-26.9.30 string lists and the 26.9.30 objects straight to names/addrs, dropping entries the old loader refused. A new seeder (XdnsFinalmaskNamesAddrsFix) reruns it over every stored finalmask, and the save path, GetXrayConfig, both link importers and the mask editor's mount lift go through it. Template outbounds now get the build-time heal too: a 26.9.30 template pasted after the seeder passes the save check, since the core loads that shape without error. The editor edits names and addrs as tag lists and no longer requires record types, which the core now defaults (TXT on a client, A/CNAME/TXT/AAAA on a server). The wire format did not change, but clients on an older core cannot read the new config shape. Lua scripts (#6823): dns.script, routing.script and dns servers[].id are new. The xray page parsed the dns block with a strict schema, so loading it dropped script and id and the next save deleted them; the schema now keeps both and the DNS server modal carries the id through an edit. The hot routing apply built routing in-process, and that build now resolves the script file from the panel's working directory, so a script beside the xray binary failed every rule change into a full restart and broke the node bridge. RoutingService cannot change the script anyway, so the panel no longer resolves it there. TLS client roots (#7105): the core now verifies against its bundled Mozilla roots instead of the OS store unless tlsSettings.useSystemCA is set. The outbound TLS form gets a switch for it, so an outbound to a server signed by a privately installed CA can opt back in.
2026-10-10 22:55:12 -07:00
// healOutboundStreams applies every heal to each outbound's streamSettings, returning the
// original bytes when none changed anything so the hot-reload diff sees no edit.
func healOutboundStreams(raw json_util.RawMessage, heals ...func(stream map[string]any) bool) json_util.RawMessage {
feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) (#5506) * feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) xray-core v26.6.22 (PR #6258) renamed the XHTTP session config keys sessionPlacement/sessionKey to sessionIDPlacement/sessionIDKey (no fallback kept in core) and added sessionIDTable (predefined charset name or literal ASCII) and sessionIDLength (range, e.g. 16-32, lower bound > 0). Panel changes: - Schema (xhttp.ts): rename the two keys, add sessionIDTable/sessionIDLength, and a z.preprocess that lifts legacy keys off stored configs so an upgraded panel never silently drops a saved session setting. - Wire normalize + share-link build/parse: rename keys, emit the two new fields, and accept legacy sessionPlacement/sessionKey from old share links. - Inbound + outbound XHTTP forms: rename field paths, add a sessionIDTable autocomplete (9 predefined tables + free ASCII) and a sessionIDLength range input shown only when a table is set, with light client validation (ASCII table, length min > 0; xray enforces the room-size minimum server-side). - Subscription (service.go) and Clash (clash_service.go) builders: emit the renamed + new keys, with a legacy fallback for not-yet-resaved inbounds. - Locales: add sessionIDTable/sessionIDLength labels + hints in all 13 files. Two sibling v26.6.22 XHTTP commits need no panel change and are covered by the core bump alone: #6332 (XHTTP/3 closes QUIC/UDP) and #6320 (udpHop honors the existing dialerProxy). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(xhttp): add Session ID Table to inbound form-blocks snapshot The new sessionIDTable input renders by default in the inbound XHTTP form, so its label joins the field-structure snapshot. sessionIDLength stays conditional (only shown when a table is set), so it does not appear here. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xhttp): migrate legacy session keys in the running xray config The Zod preprocess plus the subscription/Clash fallbacks only covered the panel UI and share-link output. The config handed to the running xray-core process is built from the raw stored streamSettings in GetXrayConfig, which did not rewrite the renamed XHTTP session keys — so a pre-upgrade inbound (or template outbound) stored with a non-default sessionPlacement was emitted unchanged and dropped by xray-core v26.6.22, until the admin re-saved it. Lift sessionPlacement/sessionKey onto sessionIDPlacement/sessionIDKey at config-generation time, in the existing inbound stream-rewrite block (next to the tls/reality/externalProxy handling) and across template outbounds. The lift is idempotent and leaves unchanged configs byte-identical so the hot-reload diff never sees a spurious change. Also tighten validateSessionIDLength to reject an inverted range (e.g. 32-16) in addition to the existing lower-bound > 0 check. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xray): avoid summed-capacity allocation in mergeSubscriptionOutbounds CodeQL go/allocation-size-overflow flagged the pre-sized make() whose capacity was a sum of three slice lengths. Grow the slice via append on a nil slice instead; same result, no overflow-prone capacity expression.
2026-06-23 08:38:16 -07:00
if len(raw) == 0 {
return raw
}
var outbounds []map[string]any
if err := json.Unmarshal(raw, &outbounds); err != nil {
return raw
}
changed := false
for _, ob := range outbounds {
if stream, ok := ob["streamSettings"].(map[string]any); ok {
feat(xray): update xray-core to v26.10.10 and adapt panel Move the three Xray binary pins (DockerInit.sh, both release.yml fetches) to v26.10.10, in lockstep with the xtls/xray-core module that 82c77111 already bumped to 701af60772cd. xdns finalmask (#7090): a domain now takes a "names" list instead of a single "name", and a resolver is an "addrs" list of udp:// or tcp:// URLs instead of {type, settings:{addr}}. The loader ignores the old keys rather than failing, so every mask the 26.9.30 seeder wrote would load with no domain and no resolver and the tunnel would carry nothing. maskcompat.UpgradeLegacyXdns now lifts both the pre-26.9.30 string lists and the 26.9.30 objects straight to names/addrs, dropping entries the old loader refused. A new seeder (XdnsFinalmaskNamesAddrsFix) reruns it over every stored finalmask, and the save path, GetXrayConfig, both link importers and the mask editor's mount lift go through it. Template outbounds now get the build-time heal too: a 26.9.30 template pasted after the seeder passes the save check, since the core loads that shape without error. The editor edits names and addrs as tag lists and no longer requires record types, which the core now defaults (TXT on a client, A/CNAME/TXT/AAAA on a server). The wire format did not change, but clients on an older core cannot read the new config shape. Lua scripts (#6823): dns.script, routing.script and dns servers[].id are new. The xray page parsed the dns block with a strict schema, so loading it dropped script and id and the next save deleted them; the schema now keeps both and the DNS server modal carries the id through an edit. The hot routing apply built routing in-process, and that build now resolves the script file from the panel's working directory, so a script beside the xray binary failed every rule change into a full restart and broke the node bridge. RoutingService cannot change the script anyway, so the panel no longer resolves it there. TLS client roots (#7105): the core now verifies against its bundled Mozilla roots instead of the OS store unless tlsSettings.useSystemCA is set. The outbound TLS form gets a switch for it, so an outbound to a server signed by a privately installed CA can opt back in.
2026-10-10 22:55:12 -07:00
for _, heal := range heals {
if heal(stream) {
changed = true
}
feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) (#5506) * feat(xhttp): support sessionID* rename + sessionIDTable/Length (xray v26.6.22) xray-core v26.6.22 (PR #6258) renamed the XHTTP session config keys sessionPlacement/sessionKey to sessionIDPlacement/sessionIDKey (no fallback kept in core) and added sessionIDTable (predefined charset name or literal ASCII) and sessionIDLength (range, e.g. 16-32, lower bound > 0). Panel changes: - Schema (xhttp.ts): rename the two keys, add sessionIDTable/sessionIDLength, and a z.preprocess that lifts legacy keys off stored configs so an upgraded panel never silently drops a saved session setting. - Wire normalize + share-link build/parse: rename keys, emit the two new fields, and accept legacy sessionPlacement/sessionKey from old share links. - Inbound + outbound XHTTP forms: rename field paths, add a sessionIDTable autocomplete (9 predefined tables + free ASCII) and a sessionIDLength range input shown only when a table is set, with light client validation (ASCII table, length min > 0; xray enforces the room-size minimum server-side). - Subscription (service.go) and Clash (clash_service.go) builders: emit the renamed + new keys, with a legacy fallback for not-yet-resaved inbounds. - Locales: add sessionIDTable/sessionIDLength labels + hints in all 13 files. Two sibling v26.6.22 XHTTP commits need no panel change and are covered by the core bump alone: #6332 (XHTTP/3 closes QUIC/UDP) and #6320 (udpHop honors the existing dialerProxy). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(xhttp): add Session ID Table to inbound form-blocks snapshot The new sessionIDTable input renders by default in the inbound XHTTP form, so its label joins the field-structure snapshot. sessionIDLength stays conditional (only shown when a table is set), so it does not appear here. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xhttp): migrate legacy session keys in the running xray config The Zod preprocess plus the subscription/Clash fallbacks only covered the panel UI and share-link output. The config handed to the running xray-core process is built from the raw stored streamSettings in GetXrayConfig, which did not rewrite the renamed XHTTP session keys — so a pre-upgrade inbound (or template outbound) stored with a non-default sessionPlacement was emitted unchanged and dropped by xray-core v26.6.22, until the admin re-saved it. Lift sessionPlacement/sessionKey onto sessionIDPlacement/sessionIDKey at config-generation time, in the existing inbound stream-rewrite block (next to the tls/reality/externalProxy handling) and across template outbounds. The lift is idempotent and leaves unchanged configs byte-identical so the hot-reload diff never sees a spurious change. Also tighten validateSessionIDLength to reject an inverted range (e.g. 32-16) in addition to the existing lower-bound > 0 check. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(xray): avoid summed-capacity allocation in mergeSubscriptionOutbounds CodeQL go/allocation-size-overflow flagged the pre-sized make() whose capacity was a sum of three slice lengths. Grow the slice via append on a nil slice instead; same result, no overflow-prone capacity expression.
2026-06-23 08:38:16 -07:00
}
}
}
if !changed {
return raw
}
if rewritten, err := json.Marshal(outbounds); err == nil {
return rewritten
}
return raw
}
feat(xray): update xray-core to v26.10.10 and adapt panel Move the three Xray binary pins (DockerInit.sh, both release.yml fetches) to v26.10.10, in lockstep with the xtls/xray-core module that 82c77111 already bumped to 701af60772cd. xdns finalmask (#7090): a domain now takes a "names" list instead of a single "name", and a resolver is an "addrs" list of udp:// or tcp:// URLs instead of {type, settings:{addr}}. The loader ignores the old keys rather than failing, so every mask the 26.9.30 seeder wrote would load with no domain and no resolver and the tunnel would carry nothing. maskcompat.UpgradeLegacyXdns now lifts both the pre-26.9.30 string lists and the 26.9.30 objects straight to names/addrs, dropping entries the old loader refused. A new seeder (XdnsFinalmaskNamesAddrsFix) reruns it over every stored finalmask, and the save path, GetXrayConfig, both link importers and the mask editor's mount lift go through it. Template outbounds now get the build-time heal too: a 26.9.30 template pasted after the seeder passes the save check, since the core loads that shape without error. The editor edits names and addrs as tag lists and no longer requires record types, which the core now defaults (TXT on a client, A/CNAME/TXT/AAAA on a server). The wire format did not change, but clients on an older core cannot read the new config shape. Lua scripts (#6823): dns.script, routing.script and dns servers[].id are new. The xray page parsed the dns block with a strict schema, so loading it dropped script and id and the next save deleted them; the schema now keeps both and the DNS server modal carries the id through an edit. The hot routing apply built routing in-process, and that build now resolves the script file from the panel's working directory, so a script beside the xray binary failed every rule change into a full restart and broke the node bridge. RoutingService cannot change the script anyway, so the panel no longer resolves it there. TLS client roots (#7105): the core now verifies against its bundled Mozilla roots instead of the OS store unless tlsSettings.useSystemCA is set. The outbound TLS form gets a switch for it, so an outbound to a server signed by a privately installed CA can opt back in.
2026-10-10 22:55:12 -07:00
feat(xray): update xray-core to v26.10.21 and adapt panel Bump xtls/xray-core to 1b1647cd9ddf (v26.10.21) and move the three binary pins (DockerInit.sh, both release.yml fetches) in lockstep. The v26.10.10 release was pulled upstream, so the old pins already 404. XDRIVE (#7182): the core now runs XDRIVE only for VLESS inside VLESS encryption, with no flow, and an outbound only over Mux; it refuses the whole config otherwise. The panel offers XDRIVE only for VLESS, refuses to save an inbound without decryption, never emits a client flow on an XDRIVE stream (config, JSON subscription, forms), and turns Mux on when an outbound picks XDRIVE. An inbound or template outbound stored before the upgrade that the core would refuse is left out of the generated config with a warning, so it no longer stops every other inbound. The service "Google Drive" is now "google-drive" (old name read as a deprecated alias); saves, the schema and both form adapters store the new name. XDRIVE and MASQUE gain a client xmux; both schemas keep it instead of stripping it on an inbound save. Lua scripts (#7124, #7136): the core refuses a routing script beside rules or domainStrategy, and a dns script beside the fallback/parallel switches or a domain-scoped server. The panel always keeps its api routing rule, so a template save now refuses any routing script and a dns script in a refused combination, and a stored one is dropped from the generated config so the core still starts. Noise masks gain a Lua script that the loader compiles inside Build: the in-process build (template save check, hot add) now resolves a relative script the way the xray child does, beside the xray binary, instead of failing it. The mask editor gets a Script field that replaces the noise list. XDNS (#7140): a resolver takes https:// DoH addresses with worker, tlsConfig and quicConfig. The mask editor registered only addrs, so a save dropped the rest; it now edits all four. Finalmask filter (#7206): a new TCP/UDP mask screens an inbound's source IPs. The inbound mask editor offers it with its allow and block lists; it is a no-op on a dialer and an older client core refuses the type, so the share link and JSON subscription drop it like the Go link already does.
2026-10-22 08:10:11 -07:00
// dropInboundsRefusedByXDrive keeps out an XDRIVE inbound stored before xray-core 26.10.17 made it
// VLESS-encryption only (#7182); the core would refuse the whole config and stop every inbound.
func dropInboundsRefusedByXDrive(inbounds []xray.InboundConfig) []xray.InboundConfig {
return slices.DeleteFunc(inbounds, func(ib xray.InboundConfig) bool {
err := validateXDriveInbound(ib.Protocol, string(ib.Settings), string(ib.StreamSettings))
if err != nil {
logger.Warningf("Inbound %q: not started, %v", ib.Tag, err)
}
return err != nil
})
}
// dropXDriveOutboundsRefusedByCore keeps out an XDRIVE outbound stored before xray-core 26.10.17
// required VLESS encryption and mux on it (#7182), which would stop the whole core at startup.
func dropXDriveOutboundsRefusedByCore(raw json_util.RawMessage) json_util.RawMessage {
var outbounds []json.RawMessage
if len(raw) == 0 || json.Unmarshal(raw, &outbounds) != nil {
return raw
}
kept := make([]json.RawMessage, 0, len(outbounds))
for _, ob := range outbounds {
var probe struct {
Tag string `json:"tag"`
StreamSettings struct {
Network string `json:"network"`
} `json:"streamSettings"`
}
if json.Unmarshal(ob, &probe) == nil && probe.StreamSettings.Network == "xdrive" {
if err := xray.ValidateOutboundConfig(ob); err != nil {
logger.Warningf("Outbound %q: not started, xray-core refuses it: %v", probe.Tag, err)
continue
}
}
kept = append(kept, ob)
}
if len(kept) == len(outbounds) {
return raw
}
if rewritten, err := json.Marshal(kept); err == nil {
return rewritten
}
return raw
}
// routingScriptRefused mirrors xray-core 26.10.17's RouterConfig.Build (#7136): a routing
// "script" replaces rules and domainStrategy, so the core refuses a config holding both.
func routingScriptRefused(routing map[string]any) bool {
script, _ := routing["script"].(string)
rules, _ := routing["rules"].([]any)
return script != "" && (routing["domainStrategy"] != nil || len(rules) > 0)
}
// dnsScriptRefused mirrors xray-core 26.10.17's DNSConfig.Build (#7136): a dns "script" refuses the
// fallback and parallel switches, and any server scoped by domains, skipFallback or finalQuery.
func dnsScriptRefused(dns map[string]any) bool {
if script, _ := dns["script"].(string); script == "" {
return false
}
for _, key := range []string{"disableFallback", "disableFallbackIfMatch", "enableParallelQuery"} {
if on, _ := dns[key].(bool); on {
return true
}
}
servers, _ := dns["servers"].([]any)
return slices.ContainsFunc(servers, func(entry any) bool {
server, _ := entry.(map[string]any)
skipFallback, _ := server["skipFallback"].(bool)
finalQuery, _ := server["finalQuery"].(bool)
return hasDNSDomains(server["domains"]) || skipFallback || finalQuery
})
}
func hasDNSDomains(domains any) bool {
switch v := domains.(type) {
case []any:
return len(v) > 0
case string:
return v != ""
}
return false
}
// dropRefusedScript drops a section's Lua "script" the core would refuse, logging why: the
// panel's own rules and DNS settings stay, since its stats path depends on the api rule.
func dropRefusedScript(raw json_util.RawMessage, section string, refused func(map[string]any) bool) json_util.RawMessage {
var parsed map[string]any
if len(raw) == 0 || json.Unmarshal(raw, &parsed) != nil || !refused(parsed) {
return raw
}
logger.Warningf("%s: ignoring \"script\" %v, xray-core refuses it beside this %s section's other settings", section, parsed["script"], section)
delete(parsed, "script")
if rewritten, err := json.Marshal(parsed); err == nil {
return rewritten
}
return raw
}
feat(xray): update xray-core to v26.10.10 and adapt panel Move the three Xray binary pins (DockerInit.sh, both release.yml fetches) to v26.10.10, in lockstep with the xtls/xray-core module that 82c77111 already bumped to 701af60772cd. xdns finalmask (#7090): a domain now takes a "names" list instead of a single "name", and a resolver is an "addrs" list of udp:// or tcp:// URLs instead of {type, settings:{addr}}. The loader ignores the old keys rather than failing, so every mask the 26.9.30 seeder wrote would load with no domain and no resolver and the tunnel would carry nothing. maskcompat.UpgradeLegacyXdns now lifts both the pre-26.9.30 string lists and the 26.9.30 objects straight to names/addrs, dropping entries the old loader refused. A new seeder (XdnsFinalmaskNamesAddrsFix) reruns it over every stored finalmask, and the save path, GetXrayConfig, both link importers and the mask editor's mount lift go through it. Template outbounds now get the build-time heal too: a 26.9.30 template pasted after the seeder passes the save check, since the core loads that shape without error. The editor edits names and addrs as tag lists and no longer requires record types, which the core now defaults (TXT on a client, A/CNAME/TXT/AAAA on a server). The wire format did not change, but clients on an older core cannot read the new config shape. Lua scripts (#6823): dns.script, routing.script and dns servers[].id are new. The xray page parsed the dns block with a strict schema, so loading it dropped script and id and the next save deleted them; the schema now keeps both and the DNS server modal carries the id through an edit. The hot routing apply built routing in-process, and that build now resolves the script file from the panel's working directory, so a script beside the xray binary failed every rule change into a full restart and broke the node bridge. RoutingService cannot change the script anyway, so the panel no longer resolves it there. TLS client roots (#7105): the core now verifies against its bundled Mozilla roots instead of the OS store unless tlsSettings.useSystemCA is set. The outbound TLS form gets a switch for it, so an outbound to a server signed by a privately installed CA can opt back in.
2026-10-10 22:55:12 -07:00
func upgradeStreamLegacyXdns(stream map[string]any) bool {
return maskcompat.UpgradeLegacyXdns(stream["finalmask"])
}
func upgradeStreamLegacyUDPHop(stream map[string]any) bool {
return maskcompat.UpgradeLegacyUDPHop(stream["finalmask"])
}
func normalizeStreamUDPHop(stream map[string]any) bool {
return maskcompat.NormalizeUDPHop(stream["finalmask"])
}
// forceUserspaceWireGuardTun moves every WireGuard outbound to the gVisor TUN (#6796) and
// warns for each one it moves off a kernel TUN.
func forceUserspaceWireGuardTun(raw json_util.RawMessage) json_util.RawMessage {
rewritten, moved := xray.ForceUserspaceWireGuardTun(raw)
for _, tag := range moved {
logger.Warningf("Outbound %q: kernel TUN cannot be set up here, running WireGuard on the userspace TUN instead (#6796)", tag)
}
return rewritten
}