mirror of
https://github.com/MHSanaei/3x-ui.git
synced 2026-09-16 15:17:14 +00:00
f51b0040cf88c30dbc1515ceb0e5b0d79948bf61
3479 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 mrchatam
|
f51b0040cf |
fix(link): map vcn to verifyPeerCertByName in applySecurity (#6479)
Go share-link importer dropped verifyPeerCertByName for VLESS/Trojan/SS TLS while the TS parser and Hysteria2 path already kept it. Fixes #6477. Co-authored-by: mrchatam <287639636+mrchatam@users.noreply.github.com> |
||
|
Sanaei
|
dd46a06761 |
Update deps and fix AntD Space API
This change bumps the frontend and Go dependency set to newer patch/minor releases, including Vite, react-hook-form, zod, and the x/* Go modules. It also fixes a compatibility issue in the settings UI by switching Ant Design's Space usage from the deprecated `direction` prop to the current `orientation` prop. |
||
|
Sanaei
|
19a692e074 |
fix(install): stop copying tuic-server over /usr/local/bin
install_tuic_server copied the downloaded sidecar onto /usr/local/bin/tuic-server with cp -f, replacing an operator's own tuic-server install. The panel resolves bin/tuic-server first (GetBinaryPath in internal/tuic/process.go) and never reads that copy, so the copy could only ever clobber a foreign binary. |
||
|
Sanaei
|
5815254fc3 |
fix(tuic): evict the oldest relay flow instead of refusing new clients
udpRelay.flowFor returned "max relay flows reached" once the table held maxRelayFlows entries, and only the idle sweep (every minute, two-minute cutoff) freed slots. One host sending a single datagram from each of 4096 source ports therefore locked every new TUIC client out of the inbound for up to two minutes, repeatably. A full table now evicts the flow last seen longest ago, which under such a flood is one of the junk flows, and the newcomer is admitted. TestUDPRelayFullTableAdmitsNewClient fails on the refusing code with a read timeout for the third client. |
||
|
amae
|
6d96accd63 |
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> |
||
|
Lucas
|
0a2cd789ba | fix(sub): enable ML-KEM for Mihomo REALITY subscriptions (#6451) | ||
|
Timur Chernykh
|
9f07951ba7 |
feat(outbounds): support custom subscription user agents (#6398)
Some subscription providers require a client-specific User-Agent before returning outbound links. Persist an optional value per subscription and use it for refreshes and previews while preserving the existing default for blank values. |
||
|
Namso9
|
89ee1242bd |
feat(sub): add read-only HWID device-slot status endpoint (#6380)
* feat(sub): add read-only HWID device-slot status endpoint Closes #6357 A client with an HWID limit had no way to tell a subscriber how many device slots were left: /{subPath}/{subId} only exposes the gate as a boolean through X-Hwid-* headers on a 404, and ?format=info carries no limitHwid or registered count. Every "why can't I connect on my new phone" case therefore had to be answered by the operator by hand. GET /{subPath}/{subId}/hwid-status now returns the aggregate counters: {"active":true,"limit":2,"registered":1,"remaining":1,"full":false} - SELECT-only. It never registers an hwid, never touches last_seen and never calls the enforcement path, so asking about a slot cannot spend one. - Counters only: no hwid value or hash, no email, no device metadata, no IP, no User-Agent, and none of the X-Hwid-* gate headers. - The subscription id is already the bearer secret for /{subPath}/{subId}, so no admin token and no new auth mechanism. - Unknown and disabled subscriptions both answer a bare 404, with identical status, headers and body, so the route cannot be used to probe which subscription ids exist. - No HWID limit configured returns {"active":false,"limit":0,...}. - No schema change and no migration. Scoped to enabled clients exactly like effectiveHwidLimitForSubID, so the reported limit is always the limit the gate enforces on a shared sub_id, and remaining clamps at zero when the effective limit drops below the number of registered devices. A separate route leaves /{subPath}/{subId}, ?format=info and the JSON/Clash routes byte-for-byte unchanged. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * fix(sub): document hwid-status as the bare object it returns The OpenAPI operation for GET /{subPath}/{subId}/hwid-status inherited the {success,msg,obj} panel envelope from build-openapi.mjs's default 200 response, while the handler writes the HwidSlotStatus struct bare. A client generated from the spec would read `obj` and never find the counters, and the description prose contradicted the schema with a hand-written example. HwidSlotStatus now sits in openapigen's StructAllow with example: tags, the entry references the generated schema through a `responses` block, and build-openapi.mjs attaches the generated example to any `responses` entry that $refs a generated schema, so no example is hand-written. The HEAD variant the controller registers is documented like its siblings, and the summary follows the "path prefix is configured by subPath" wording now that fresh panels randomise the prefix. Regenerated frontend/public/openapi.json, docs/public/openapi.json and the subscription-server MDX. openapi-runtime-contracts.test.ts pins the bare schema, the generated example and the HEAD operation. --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> |
||
|
YoungReckless4
|
8f162994ef |
feat(clients): let admins set PersistentKeepalive on tunnel clients (#6377)
* feat(clients): let admins set PersistentKeepalive on tunnel clients
model.Client already carries KeepAlive, and every AmneziaWG/WireGuard client
config emitter already writes PersistentKeepalive when it is above zero -- but
nothing in the UI could set it, so it stayed 0 and the line was never emitted.
Without it a peer that goes quiet has nothing to trigger a handshake: WireGuard
only initiates when it has data to send. An idle client stays disconnected
after any interruption -- a NAT mapping timing out, a device sleeping, the
panel restarting -- until the user generates traffic themselves.
New clients default to 25, the conventional value, which also keeps the NAT
mapping open. Existing clients keep whatever they have, and 0 remains valid and
means "do not send keepalives".
* fix(clients): let an explicit 0 actually disable PersistentKeepalive
Addresses review feedback on the previous commit.
UpdateInboundClient carries a stored keepalive forward whenever the incoming
one is zero, so the settings JSON and the running peer survive a metadata-only
edit that omits the field. That was a 0 -> 0 no-op while no UI could set a
nonzero value. Now that the client form can, the carry-forward became reachable
in the other direction: a client created at the form's default of 25 could
never be returned to 0, and the hint text shipped to all 13 locales -- "0
disables it" -- described something the backend silently refused. The save even
reported success, because a settings blob that came back byte-identical skips
the transaction entirely.
The zero value cannot carry that distinction, so model.Client.KeepAlive becomes
*int: nil means the field was never sent, &0 means "send no keepalives". The
pointer survives the internal marshal in ClientService.Update, which is where an
explicit 0 was being erased by omitempty before UpdateInboundClient ever saw it.
ClientRecord.KeepAlive stays a plain int -- it is the stored column, where
"unset" has no meaning -- and the conversions bridge the two.
Two tests, both red before this change in the direction they cover: an explicit
0 must reach wg_keep_alive, and an update that omits the field must still leave
a stored 25 alone.
Also adds the output transform every other numeric field in the client form
already has, so a cleared box sends 0 rather than null.
* fix(clients): repair the keepalive pointer conversion after the main merge
Merging main brought buildAmneziaWGProxy (#6326) in beside the
Client.KeepAlive int -> *int change without reconciling the new call site,
so internal/sub stopped compiling and took every package importing it with
it. The two sides touched different lines, so git merged them without a
conflict -- the green `make verify` on
|
||
|
duqigit
|
64b6e43e2b |
feat(sub): add legacy Clash subscription endpoint (#6338)
* feat(sub): add legacy Clash subscription endpoint * fix(deps): update js-yaml to patched release Raise the Swagger UI js-yaml override to 4.3.2 and refresh the lockfile to resolve GHSA-2883-xcg3-v3hh without changing Swagger UI. * fix(sub): preserve client detection and normalize legacy cipher Keep the original Clash/Mihomo auto-detection default so existing subscription URLs continue returning YAML. Normalize the panel-supported chacha20-poly1305 alias when generating legacy Clash profiles, and cover both regressions through HTTP endpoint tests. * refactor(sub): drop an unreachable guard and make the alias test assert Review of the legacy Clash subscription endpoint left three LOW findings, all introduced by the change: - The comment above the routing merge ran to three lines, over CLAUDE.md's two-line cap. - validateClashRouteGraph on the legacy path could never fail: the legacy branch skips the routing merge, so it validated the literal config built a few lines above against itself. Dead code that reads as a guard. - TestClashAliasesSkipConfiguredPathConflicts asserted nothing — it could only fail on an escaping gin panic, so a regression that registered the alias handler on the configured path went unnoticed. It now drives each collision through the router and asserts which format answers each path. --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> |
||
|
Sanaei
|
3f1e52f09e |
refactor(panel): drop two duplicated helpers
SettingService.GetDefaultJSONConfig was a byte-identical copy of GetDefaultXrayConfig with no callers anywhere in the tree. amneziawgnet.normalizeDNSServer re-implemented the exported amneziawg.NormalizeDNSServer line for line, in a file that already imports that package for EffectiveMTU three lines above it. Its two callers now use the exported one, so the bare-IP-to-host:port rule has a single definition. |
||
|
Sanaei
|
0fbdf0f9bf |
fix(ui): keep the empty-group placeholder legible in dark mode
The "no group" em dash in the clients table and in two client-picker modals was drawn with an inline color: rgba(0,0,0,0.45). The panel renders every page under antd's darkAlgorithm as well, so on a dark container that near-black placeholder is effectively invisible. All three now use Typography.Text type="secondary", the idiom the rest of the panel already uses for muted text — GroupAddClientsModal itself uses it 58 lines further down. Mid-greys such as #888 elsewhere in the panel stay legible in both themes and are deliberately left alone. |
||
|
Sanaei
|
02f2a63c53 |
refactor(tgbot): extract the shared numeric keypad builder
Six callback flows — limit_traffic, reset_exp, ip_limit and their add_client_* counterparts — each carried the same 28-line inline keyboard: cancel, confirm, the 1-9 grid, clear, 0 and backspace. Only the callback prefix, the threaded email argument, the cancel target and the confirm label key ever differed, and the labels had already drifted apart between otherwise identical flows. numericKeypad now builds that grid from a numericKeypadSpec, cutting 168 lines to 6. The callback data is unchanged: every one of the 84 strings and all 6 label keys were diffed against the pre-refactor router and are byte-identical, which matters because encodeQuery hashes any query over 64 chars and buttons already sitting in a user's chat carry these strings. |
||
|
Sanaei
|
8b9cf260b6 |
perf(clients): batch the client record lookup in bulk operations
BulkResetTraffic resolved every address with its own GetRecordByEmail call, one SELECT per email, purely to find the disabled clients it has to re-enable. Resetting 30 clients issued 30 queries before the batched transaction even started, while BulkAdjust, BulkDelete and BulkSetEnable next to it already loaded their rows with a single chunked IN query. Those three carried a verbatim copy each of both the trim/dedupe loop and the chunked record load, so the reuse is the fix: trimmedUniqueEmails now delegates to the existing uniqueNonEmptyStrings, and clientRecordsByEmail holds the one chunked lookup all four call sites share. A DB failure during the lookup now aborts the reset instead of being swallowed per email; a missing row is still skipped, as before. The new test drives BulkResetTraffic with 3 and with 30 emails and fails unless both issue the same number of SELECTs against clients. |
||
|
MRVX
|
fc08b53395 |
feat(ui): add global command palette (Ctrl+K) for fast navigation and search (#6352)
* feat(ui): add global command palette (Ctrl+K) for fast navigation and search * fix(ui): address review feedback for shortcut listener, i18n parity, and search deep links * fix(ui): resolve search routing, translation keys, and palette state reset * fix(ui): improve command palette styling and sidebar transitions * fix(ui): address review feedback for typecheck, codegen, debouncing, and state reset * fix(ui): resolve effect state update warning and debounce reset in command palette * fix(ui): address review feedback for stale client search results and theme action * style(ui): apply oxfmt formatting to command palette and tests * fix(deps): update js-yaml override to resolve audit advisory * docs(api): sync the docs OpenAPI copy with the new InboundOption fields Adding Network/Security to InboundOption regenerated frontend/public/openapi.json, but docs/public/openapi.json is a hand-kept copy of that file and nothing checks it: make verify never reaches docs/, and docs-ci.yml fires only on docs/**. The two files were byte-identical on main and had diverged here, so the published API reference described a response shape the panel no longer returns. Regenerating the MDX under docs/content/docs/en/reference/api/ produced no change — the schema is read from the JSON at render time. * fix(ui): unnest the command palette row control and label its shortcut The palette row was a <button> wrapping the copy-subscription <button>. Nested interactive content is invalid HTML and React 19 logs two errors for it on every client result. The row is now a role="button" div using activateOnKey, the pattern the rest of the panel already uses, with line-height pinned so dropping the UA button style does not grow every row. Its keydown handler ignores events bubbling from the nested button: activateOnKey preventDefaults Enter, which would otherwise cancel the browser's Enter-to-click on the copy button and navigate instead. The sidebar chip hardcoded the Mac glyph while the handler accepts Ctrl as well, so Linux and Windows operators were shown a key they do not have; it now picks the modifier from the platform. Also restores the comment on ClientsPage's debouncedSearch that the deep-link change removed — the code it explains is unchanged. |
||
|
DIMFLIX
|
2dd903ea8e |
feat(sub): bake Happ/INCY routing profiles into the JSON subscription (#6402)
* feat(sub): parse generic Happ/INCY routing payloads for the JSON subscription Accepts the routing-rules format emitted for Happ and INCY (inline JSON, happ:// or incy:// deeplink, or a remote https:// URL resolved through the existing remote routing cache). The JSON subscription will bake these rules into its documents so header-ignoring clients still get routing. * feat(sub): bake Happ/INCY routing profiles into JSON subscription documents When subJsonRoutingRules is set, every emitted document (per-inbound and balancer alike) carries the profile's dns and routing rules baked in, so header-ignoring clients like Happ and INCY still get routing; the legacy simple-rules merge only applies when no profile is set. The balancer document builder keeps rewriting proxy-tag rules to the balancer. * feat(sub): add the subJsonRoutingRules setting Plumbed from the settings store through the subscription server into SubJsonService, so admins can set a routing profile once and every JSON subscription document carries it. * chore(api): regenerate OpenAPI artifacts for subJsonRoutingRules * feat(web): routing profile editor for the JSON subscription A textarea inside the JSON card accepts the routing profile (inline JSON, happ/incy deeplink, or https URL) with a remote-source badge; the badge helper moves to a shared module. Keys added to all 13 locales. * fix(sub): warm and lazily resolve the baked JSON routing source The routing profile was resolved once at service construction: a remote URL that was cold at that moment baked default routing forever, and the cron job never warmed it. The job now warms the subJsonRoutingRules URL, and the profile resolves per request with an in-memory memo (a failed resolve is not cached), so a warmed cache takes effect without a restart. * feat(sub): fall back to the JSON routing profile for the Routing header Happ and INCY download the geo files a routing profile references through the Routing response header. When the Happ header setting was blank the header stayed unset, and clients fetched no geo files even though a JSON routing profile was configured. A blank setting now falls back to the JSON profile: happ/incy deeplinks pass through, inline JSON and remote URLs are normalized to a happ:// deeplink; an unusable or oversized value leaves the header unset. Locale captions mention the fallback. * fix(sub): pass routingRules arg at call sites added by main Main gained four NewSubJsonService call sites after this branch forked; update them to the five-arg signature so internal/sub builds again. * fix(sub): address code review findings on the baked JSON routing The memoised baked template never invalidated, so an edited remote profile kept serving the superseded dns/routing subtrees until a panel restart; bakedTemplate now re-resolves the spec per request and rebuilds only when the payload actually changed (regression-tested). subJsonRoutingRules shared the happ persistence row with subRoutingRules, so only the last-written setting survived a restart; it now resolves under its own jsonhapp kind with the same validation and size caps. The setting also joins validateSettingsURLs, so remote values are canonicalised and bad URLs are rejected on save. Also: drop the unreachable half of the remote-source guard, cut the overlong comment blocks to the two-line convention, and deduplicate remoteSourceBadge in the General tab. Merges upstream/main (call sites for the widened NewSubJsonService signature). * style(sub): gofumpt the json_routing imports * fix(sub): accept happ add/ deeplinks and bound the routing warning The baked-JSON routing parser only recognised happ://routing/onadd/, but normalizeHappRouting treats happ://routing/add/ as an equally valid routing deeplink. An operator pasting the add/ form got the Routing header set, so the panel looked configured, while every JSON subscription document silently carried the default routing instead of their profile. resolveJsonRoutingSpec logged one warning per call and bakedTemplate calls it once per emitted document, so a single fetch of an unusable profile wrote one identical warning per document. On the public subscription server that floods the 10240-entry buffer the panel's log view reads, evicting real entries. Log only when the message changes, and reset on a successful resolve so a profile that recovers and fails again is still reported. Also resolve the template once in buildBalancerConfig: two resolves could straddle a profile refresh and pair one revision's dns with the other's routing. |
||
|
Rouzbeh†
|
ed5465d0f2 |
feat(clients): support setting HWID limit and MTProto ad-tag in bulk adjust (#6399)
* feat(clients): support setting HWID limit and MTProto ad-tag in bulk adjust Add HWID device limit and Telegram MTProto sponsor channel (ad-tag) support to the bulk client adjustment flow in both the panel API and frontend ClientBulkAdjustModal. Co-Authored-By: Claude Code <noreply@anthropic.com> * fix(clients): gate adTag to MTProto inbounds and avoid inbound rewrite for limitHwid Co-Authored-By: Claude Code <noreply@anthropic.com> * fix(clients): stamp updated_at only on the clients a bulk adjust changed The updated_at write was gated on hasInboundChanges, which accumulates over the whole inbound instead of describing the client in hand. Once any client in the settings array changed, every client after it was re-stamped as well, so whether an untouched client kept its own updated_at depended on its position in the array. That field feeds node-snapshot conflict resolution, where a spurious bump lets a stale snapshot value win over the stored record. Track the change per client and fold it into the inbound-level flag where the stamp is written, so the early return still skips a save whose settings JSON would be unchanged. Also condenses the BulkAdjust doc comment back to the two-line maximum. * docs(api): regenerate the bulkAdjust reference for limitHwid and adTag frontend/public/openapi.json was copied to docs/public/, but pnpm gen:api was never re-run, so the API reference page's heading, anchor id and search index still described bulkAdjust without limitHwid or adTag. docs-ci.yml fires only on docs/**, and that path had been touched, so nothing flagged the stale MDX. The externalLinks hunks are the generator rewrapping lines main had left stale, not a content change. * fix(i18n): stop enumerating fields in the bulk-adjust empty-form message bulkAdjustNothing listed the fields the form accepts, so it went stale every time one was added: only en-US ever gained "flow", leaving the other twelve locales describing days and traffic alone, and limitHwid and adTag would have repeated that. Say that one field is required instead of naming which, so the message cannot drift again. |
||
|
Pejman Yousefi
|
1456658028 |
feat(sub): add Happ client integration, routing presets, and app management (#6434)
* feat(sub): add Happ client integration, routing presets, and app management Implement comprehensive Happ proxy client integration according to official developer specifications. - Fix header emission on disabled routing and hidden settings to send explicit '0' headers rather than omitting, allowing Happ clients to reset cached settings. - Add support for 'happ://routing/off' deeplink in routing validation. - Preserve '?serverDescription=' query parameters in link fragments without escaping to support Happ server subtitles across VMess, VLESS, Trojan and SS. - Add Happ application management headers: ProviderID, New-Url, Fallback-Url, Sub-Info banners, Sub-Expire notifications, No-Limit mode, hardware ID enforcement, TUN modes/types, route exclusions, APNS exclusions, and per-app proxy settings. - Add curated routing presets (Iran Bypass, China Direct, AdBlock, Global) and interactive visual rule generator in frontend settings. - Synchronize all 13 translation locales with native Persian, Russian, and Chinese translations. * fix(sub): keep Happ header overrides behind the auto-detect opt-in The Routing-Enable/Hide-Settings off values were emitted on the User-Agent alone, so every panel that upgraded would push "Routing-Enable: 0" — documented by happ.su as disabling routing globally — to every Happ client without the operator enabling anything. They now ride subHappAutoDetect like every other Happ header. Two further mismatches against the vendor spec: - serverDescription was written as a key of the VMess base64 JSON object. happ.su documents it as a "#Title?serverDescription=<base64>" link parameter or a JSON "meta" entry, so the caption never reached Happ while every other VMess consumer received an unknown key. Dropped rather than moved: emitting the documented form is unsafe here because our own parser base64-decodes the whole VMess body (internal/util/link/outbound.go). - The TUN Mode dropdown stored the literal "default", forwarded as "Tun-Mode: default", where happ.su documents system|gvisor only. It now stores the unset value so no header is sent. TUN Type "default" is a documented value and is unchanged. Each fix carries a test that fails without it. |
||
|
Rouzbeh†
|
d5ab84e8d5 |
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> |
||
|
VibeProgramm
|
876497db6e |
feat(sub): add AmneziaWG proxy generation for Clash subscriptions (#6326)
* feat(sub): add AmneziaWG proxy generation for Clash subscriptions Add buildAmneziaWGProxy to generate mihomo-compatible wireguard proxy entries with amnezia-wg-option sub-block for AmneziaWG inbounds. Previously, AmneziaWG inbounds were silently skipped in Clash subscriptions (buildProxy returned nil), making them unusable with mihomo/Clash clients. The new function reuses the wireguard proxy base structure and adds: - v1.0 obfuscation fields (jc/jmin/jmax/s1-s4/h1-h4/i1-i5) - v1.5 fields (s3/s4/i1-i5) - v3 fields (header-protection-key, content-padding-addition, timing, random-trailers, disable-cookies) with automatic version: 3 tagging Closes #6310 * fix review nits: doc comment and test call * fix(sub): use the AmneziaWG inbound's own tunnel address in Clash proxies buildAmneziaWGProxy took the peer address from model.Client.AllowedIPs, which matchingClients resolves out of the shared clients.wg_allowed_ips column. That column holds one address per identity, so a client attached to both a WireGuard and an AmneziaWG inbound gets the other protocol's address written into its Clash proxy - an unroutable peer, since the running interface accepts only the AllowedIPs InstanceFromInbound derives from the inbound's own settings JSON. Read the address from settings.clients[] and fall back to the shared column. Also emit remote-dns-resolve alongside dns: mihomo gates its whole `dns` list on that flag (adapter/outbound/wireguard.go, NewWireGuard), so the panel's primaryDns/secondaryDns were inert in Clash while the vpn:// .conf turned them into a real DNS line. Restricted to bare IPs - mihomo aborts the entire config when dns.ParseNameServer rejects an entry, and nothing validates those fields. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * fix(sub): emit the AmneziaWG effective MTU in Clash proxies main's EffectiveMTU landed while this branch was open, so the Clash builder was the one AmneziaWG emitter left omitting mtu when the operator set none. The running interface uses EffectiveMTU (internal/amneziawgnet/device.go), as do the vpn:// .conf and both TS builders; mihomo instead falls back to its own 1408, which sits above the tunnel once s4 passes 12 and fragments every packet the client sends. GenerateObfuscation31 draws s4 from 12..27, so that is the default. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * fix(sub): reject a zoned DNS address before opting into remote-dns-resolve netip.ParseAddr accepts "fe80::1%eth0", but mihomo's parsePureDNSServer brackets it into "udp://[fe80::1%eth0]" and url.Parse rejects "%et" as a bad escape, so parseNameServer errors and parseProxies aborts the entire config - the whole subscription's Clash profile, not just this proxy (#4641 class). |
||
|
Sanaei
|
8076d5edfa |
chore(frontend): bump React and Zod deps
Update frontend dependencies to React/ReactDOM 19.3.0, Zod 4.6.1, and matching React type packages. Also replace Storybook addon-vitest override placeholders with explicit Vitest/browser-playwright versions to keep dependency resolution stable. |
||
|
BlindMaster24
|
87420e3bb1 |
fix(tgbot): close stale-inbound TOCTOU and contain handler panics (#6442)
Tapping an old get_clients_for_* inline keyboard re-fetched the inbound after the keyboard lookup, discarding the error; if the row vanished between the two reads, the second GetInbound returned nil and inbound.Remark panicked. The callback handler runs on a bare goroutine with no recover(), so that panic killed the whole panel process. Fetch the inbound once in a shared chooseInboundClient helper that answers an error callback on a missing row, and pass the row down to getInboundClientsFor instead of re-reading the DB, removing the between-reads window. Route all three OnReceive handler paths through a recover() barrier so no handler panic can take down the process, and log the GetInbound failure instead of silently swallowing it. |
||
|
Sanaei
|
33e6c2ec0c | chore(deps): raise the swagger-ui-react js-yaml override to 4.3.2 | ||
|
Sanaei
|
6ee74f2032 |
fix(amneziawgnet): wait for the client netstack goroutines before closing its device
TestPortForwardRoundTripTCPAndUDP flakes in the race job: closing the test's client WireGuard device races the goroutines still writing into its netstack. amneziawg-go's device.Close() calls tun.Close() before it stops the routine draining the tun, and netTun.Close() closes the unbuffered incomingPacket channel that WriteNotify sends on. A goroutine still inside a netstack write when the deferred clientDev.Close() runs therefore closes and sends on the same channel -- reported as a data race, and on a bad interleaving a "send on closed channel" panic. The TCP echo listener, its per-connection copies and the UDP echo all write into clientNet, and teardown only closed the two listeners before the device: nothing waited for the goroutines themselves. A WaitGroup deferred right after clientDev.Close() supplies the missing edge, since LIFO then puts the wait between the listener closes and the device close. Confirmed by flooding the existing UDP echo goroutine under GOMAXPROCS=1 and 2, which failed 3/6 and 2/6 runs with the stack CI reported and 0/12 with the fix. |
||
|
Sanaei
|
d0edbcec81 |
feat(xray): update xray-core to v26.9.9 and follow the udpHop move
Bump xtls/xray-core to 52a412d9e2f5 (v26.9.9) and the three binary pins in DockerInit.sh and release.yml in lockstep. Upstream moved UDP port hopping out of finalmask.quicParams.udpHop and into a standalone "udphop" UDP mask with a different shape (mode / interval / remotePorts / remoteIPs). The old key is gone from QuicParams, and since the config loader ignores unknown fields it is now silently dropped rather than rejected — port hopping just stops. The panel adapts where that key was live: - Both link importers rebuilt quicParams.udpHop from the standard mport param, so an imported hysteria2 link produced an outbound that no longer hops. They now emit a udphop mask in intervalremote mode, which is what the old key did. The mode is required: UDPHop.Build() rejects an empty or unknown one. - validFinalMaskUDPTypes and UdpMaskTypeSchema learn "udphop", otherwise the Go link generator strips the mask from every link and sub, and Zod strips it on the next form round trip. - mport generation (Go and frontend) reads the mask first and keeps reading the legacy key, so inbounds stored before the upgrade still advertise their range. On an inbound the old key was always inert — only hysteria's dialer consumed it — so nothing regresses server-side and no migration is needed. udphop stays out of the mask dropdown on purpose: it is client-only in core, which refuses to wrap a server socket, and that form is shared with the inbound editor. |
||
|
Sanaei
|
cfd596a489 |
fix(amneziawg): let a cleared header protection key reach a running device
amneziawg-go reads an absent UAPI line as "keep the current value", and addressFingerprint keys only on the addresses and MTU, so an obfuscation-only edit reconfigures in place rather than rebuilding. Clearing headerProtectionKey therefore never took effect: the device kept protecting headers with the old key. The stale key also keeps the S1-S4 minimum in force, so lowering S3/S4 in the same edit made every later IpcSet fail with -22 — after replace_peers had already dropped the peers. Send the all-zero key when the field is empty, which is how the UAPI expresses "disabled"; an empty value would be rejected, since it decodes to zero bytes. |
||
|
Sanaei
|
efc603f59c |
fix(settings): show the SMTP failure reason instead of a raw i18n key
classifySMTPError returned keys already carrying "pages.settings.", while the
four keys TestConnection returns directly do not, and the alert renders every
Message under that one prefix. Any classified failure therefore looked up
pages.settings.pages.settings.smtpErrorAuth, which does not exist, so the panel
printed the key instead of "Authentication failed — check username and
password". The unknown case was worse: it appended the raw error to the key, and
its own text interpolated {{ .Error }}, Go template syntax the frontend's
i18next never fills.
Return the keys unprefixed like the rest, and point the unknown case at the
panel log, which already carries the underlying error.
|
||
|
Sanaei
|
c392f367e1 |
fix(dns): stop offering a port field that DoH entries discard
Xray ignores port for DoH/DoHL/DoQL, so valuesToWire deliberately stores none for an encrypted address and a non-standard port has to go inside the URL. The form kept offering the field anyway, pre-filled with the 53 from its own defaults: a port typed there was dropped on save and redrawn as 53 on reopen, which reads as the panel losing the value. Render the port field only where it is actually stored. DoT keeps it, since tls:// is not an encrypted-address scheme for this purpose. Closes #6403 |
||
|
Sanaei
|
705b291d34 |
fix(amneziawg): stop losing an inbound and its server keys on the API path
Two saves that the panel UI never makes, but the documented REST API does. A client whose allowedIPs normalized to empty passed validation, then InstanceFromInbound skipped the peer and dropped the whole instance when it was the only one. Nothing logged it, so an enabled inbound simply never opened its socket. Refuse an enabled peer with no address, naming the client, the way the injection and collision checks already do. The server keypair was regenerated whenever a payload omitted privateKey, which invalidates every client config already distributed, and a payload carrying only privateKey left publicKey empty so rendered configs got a blank "PublicKey =". An omitted key now means unchanged: the stored pair is carried forward, a half-supplied pair has its public half derived, and generation is reserved for an inbound that has no stored keys at all. UpdateInbound loads the stored row before normalizing so those keys are available. Closes #6407 |
||
|
Sanaei
|
4e423fa452 |
fix(sub): drop Reality parameters when a host forces plain TLS
A Host row may set Security to tls on an inbound whose own stream is Reality. The emitted link then carried security=tls next to pbk, sid, spx and the Reality dest as sni: the endpoint no longer performs a Reality handshake, so those describe a server the client will never reach, and clients that honour them fail to connect. Only the security key was rewritten at emit time, and the existing strip covered alpn/sni/fp/pcs for forceTls=none alone. Clear the Reality-only parameters before the endpoint's own TLS overrides are applied, so a host that supplies its own sni or fingerprint still wins. Closes #6424 |
||
|
Sanaei
|
65c5580e7d |
fix(clients): keep per-peer keys when a client spans several tunnel inbounds
A client attached to several WireGuard/AmneziaWG inbounds is that many independent peers, each with its own keypair, preshared key and tunnel address. The client edit form can only represent one peer, so Update's per-inbound loop stamped that single field set onto every attached inbound: every peer ended up with identical keys and one inbound's address, and the tunnels on all the other nodes stopped working with no way to recover the overwritten values from the panel. The only guard covered AllowedIPs, and only for AmneziaWG. When more than one tunnel inbound is in scope and the caller sent no per-inbound override, clear the shared peer fields so UpdateInboundClient's existing carry-forward preserves each inbound's own. A scoped update (?inboundIds=) still narrows to one inbound and edits it normally. Closes #6372 |
||
|
Sanaei
|
2004340d1d |
fix(inbounds): let a node-adopted inbound keep its own protocol on edit
UpdateInbound restores the stored NodeID before the node-eligibility check, so the payload can never introduce an assignment there — the check could only ever fire on a row that already had one. A node's MTProto inbound arrives on the master by adoption, which does not go through that check, so every later edit of it was refused with "mtproto inbounds cannot be assigned to a node". That made the share address of a node-managed MTProto inbound impossible to change from the panel that generates its subscription links. Refuse only a protocol change into an ineligible protocol, which is the one way an update can still strand a row the master's sidecar loops would never reconcile. Closes #6415 |
||
|
Sanaei
|
bc57548a35 |
fix(clients): withdraw the delete tombstone when the email is re-created
Deleting a client tombstones its email for 90s so a node snapshot captured before the deletion cannot resurrect it. Nothing withdrew that tombstone when the operator re-created the same email, so on a master with at least one node the next merge filtered the live client out of the snapshot and SyncInbound pruned its inbound link. The client reappeared only once the tombstone expired, which is the 90-120s detach window reported. Withdraw it on a successful create, single and bulk, so a tombstone can never outlive the identity it was meant to bury. A failed create still leaves it standing, which is what keeps the stale-snapshot guard intact. Closes #6370 |
||
|
Sanaei
|
246d9207a5 |
fix(sub): emit a bare host in Clash proxies
A Clash "server" is a bare host, not a URI authority, but the custom share address strategy stores an IPv6 literal with brackets so the address normalizer can hand it to the raw link generators. The Clash renderer copied that value into every proxy verbatim, so mihomo received server: "[2001:db8::1]" and failed to parse the node. Raw links were unaffected because joinHostPort strips the brackets and re-adds exactly one. Strip them once where the renderer takes the resolved dest, which is the single place all three proxy builders read the address from. Closes #6373 |
||
|
Sanaei
|
20d7f91c65 |
refactor(ci): add an adversarial pass and name the analyst briefing
REVIEW.md told the reviewer which repository rules to check but never to try
breaking the change, so the conditions this panel actually meets went
unexamined. "Try to break it" adds six, each tied to a mechanism here rather
than to a generic checklist: an upgrade over an operator's existing rows and
the rollback that reads them again, a restart that drops in-memory state under
the cron jobs, a sub-node racing the master on the same row, an operation
applied twice, an inbound or client at the empty and the thousand end, and a
dependency that is down. It closes with the gate that running a case is not
reporting it - each one still has to clear the verification bar below it, so
the section cannot become a licence for hypotheticals.
repo-context.md said nothing about which bot reads it. Only the issue analyst
does, since the review job's briefing moved inline in
|
||
|
Sanaei
|
acf3603dc8 |
refactor(ci): review pull requests with one senior-engineer role
The review job ran the official code-review plugin, which fans a pull request out to five Sonnet reviewers plus a Haiku scorer per finding and drops everything scored under 80, and the briefing file spent most of its lines overriding that plugin. Both are gone: the job hands one Senior Software Engineer prompt to the action inline, the way the issue analyst does, and denies the Agent tool so the single role is mechanical rather than a request. REVIEW.md moves from the emoji markers to CRITICAL/HIGH/MEDIUM/LOW with a pre-existing qualifier. The uncapped rule is scoped to findings the pull request introduced or worsened so it cannot collide with the cap of three pre-existing ones. "A finding is a report, not a patch" stays as it was. Workflow housekeeping: GH_TOKEN, REPO and PR live in the job env instead of six step copies; the skip gate is per pull request, so a head pushed after the automatic review is reviewed only on @claude review; the comment counters sum gh's per-page jq output, which read "0\n0" as a review on a pull request with more than 100 comments; --max-turns rises to 300 because every read now costs the single agent a turn instead of a subagent. |
||
|
DuQi
|
47d2303334 |
fix(inbounds): reject missing TLS certificates before saving (#6429)
An inbound could be saved with security "tls" and a certificate row carrying neither a file path nor inline content. Nothing rejected it, so the row reached xray-core, whose readFileOrString fails with "both file and bytes are empty" and takes the whole config build down with it — every other inbound included. Validate the credentials on both sides of the wire. validateInboundTLSCertificates follows xray's file-over-inline precedence, requires a private key for every non-verify certificate and insists on at least one server certificate, so a verify-only CA list no longer passes as a server config. The inbound form's Zod schema enforces the same rules per field and serializes only the editor mode the operator actually used, and a failed save jumps to the Security tab naming the certificate row that broke. On update the guard is scoped to a real TLS edit. A row already stored incomplete is grandfathered: it stays editable, and only a save that breaks a previously valid block is refused. A sub-node stores whatever the master pushes, and Remote.UpdateInbound falls back to AddInbound when the node does not yet hold the tag, so a grandfathered row could otherwise never be deployed or re-seeded — the rejection is swallowed to a logger.Debug line and the node stays on a stale config while the panel shows the client as cut off. The controller now marks a node-sync request (mTLS or a node-sync token) on a per-request copy of InboundService, and the guard steps aside for it on both add and update: the row was judged where the operator acted, and a node that refuses it only falls out of sync. Operator and admin-token saves are held to the guard as before. The security union is parameterised on its tlsSettings branch instead of copied, and tlsCertUsesFiles is the one file-vs-inline inference shared by the form schema and the adapter, so the mode the editor opens in and the pair of fields the save serializes cannot drift apart. |
||
|
Rouzbeh†
|
9f76a66dcf |
feat(sub): add dummy info node and status configs for subscriptions (#6412)
* feat(settings): add subInfoNodeEnable and status template settings * feat(sub): add dummy info node and status configs for raw links * feat(sub): support dummy info node in clash and json subscriptions * feat(ui): add subscription info node switch and status templates to settings * style: apply gofumpt formatting * fix(sub): address review feedback on subscription info node - Restore GetSubs contract to avoid unintended remark expansions on non-subscription-body calls. - Exclude dummy info node from Clash PROXY select group when active server nodes exist. - Track hasEnabledClient and set traffic.Enable in JSON and Clash paths so status tokens evaluate correctly. - Deterministically sort client emails across subscriptions before selecting primaryEmail. - Consolidate duplicated info-node evaluation logic into resolveInfoNodeRemark helper. - Remove redundant pure-getter test from setting_sub_info_node_test.go. Co-Authored-By: Claude Code <noreply@anthropic.com> --------- Co-authored-by: Claude Code <noreply@anthropic.com> |
||
|
Gleb Gudkov
|
b8597314f8 |
docs(api): mark collection responses nullable (#6430)
* docs(api): mark collection responses nullable Describe allLinks and panel log response objects as nullable string arrays so generated clients accept the existing nil-slice wire format. Pin both schemas with buildSpec regression assertions and regenerate the OpenAPI copies. * docs(api): include nullable Xray log responses Allow generated response arrays to opt into nullability while retaining their schema references and Go-derived examples. Apply this to Xray logs, whose nil slices already serialize as null, and pin the schema and example through buildSpec. |
||
|
YoungReckless4
|
3cd3836d77 |
fix(amneziawg): account for S4 junk in the default tunnel MTU (#6376)
* fix(amneziawg): account for S4 junk in the default tunnel MTU
amneziawg prepends S4 random bytes to every transport packet
(device.NewOutboundElement) and, unlike content padding and random trailers,
never clamps them against the tunnel MTU. A full-size packet therefore lands on
the wire at MTU + 60 + S4 bytes: 20 IPv4 + 8 UDP + S4 + 16 transport header +
16 poly1305 tag.
With the 1420 default that overflows a 1500-byte link once S4 exceeds 20, and
GenerateObfuscation31 draws S4 from 12..27 inclusive -- so roughly 44% of newly
created inbounds fragment every full-size packet they send.
Measured on a live pair of interfaces, predicted against observed:
MTU 1380 S4 12 -> 1452 on the wire (fits)
MTU 1420 S4 12 -> 1492 (fits)
MTU 1420 S4 20 -> 1500 (exactly at the limit)
MTU 1420 S4 21 -> 1501 (fragments)
MTU 1420 S4 27 -> 1507 (fragments)
EffectiveMTU now subtracts S4 from the default; an explicit MTU is untouched.
Client configs carry the same number. They previously omitted the MTU line
whenever the server had no explicit value, which left the client on its own
1420 default and fragmented the client-to-server direction even after the
server side was fixed -- silently, and only in one direction. All three
emitters (the Go subscription text and the two TypeScript ones) now agree,
which is what the existing parity test exists to protect.
* fix(amneziawg): rebuild the device when S4 changes the derived MTU
Addresses review feedback on the previous commit.
Deriving the default MTU from S4 made a construction-time-only property depend
on a hot-reloadable input, but addressFingerprint -- ensureLocked's only rebuild
trigger -- still hashed the raw inst.MTU. S4 is a UAPI field, so an S4-only edit
took the in-place IpcSet branch and the gVisor netstack kept the MTU derived
from the old S4 while all three client emitters already advertised the new one.
Every panel-created inbound leaves mtu unset, so that was the normal case, not
an edge one: with S4 raised far enough the fragmentation this fix exists to
remove came straight back, and stayed until a panel restart or an unrelated
address edit.
Folding EffectiveMTU into the fingerprint fixes it. An explicit MTU still takes
the in-place branch on an S4 edit, since it does not move the interface MTU.
Also trims four comment blocks to the 2-line cap in CLAUDE.md, and points
NewDevice's doc comment at EffectiveMTU instead of the deleted defaultMTU.
|
||
|
Amirmohammad Sadat Shokouhi
|
5a63d5d468 |
fix(mtproto): use hosts for public share links (#6369)
* fix(mtproto): use hosts for public share links Generate MTProto subscription, client, copy, QR, and export links from managed Hosts so reverse-proxied public ports are advertised correctly. Migrate the redundant legacy custom share address into a Host and keep old imports compatible. Closes #5126. * fix(mtproto): keep host share links lossless and consistent Address review on the MTProto hosts share-link change. The migration no longer drops a legacy custom share address: an unrelated (or disabled) Host stopped suppressing it, so only a Host already advertising the same address does. An imported address now clears the same validation the strict normalizer applies to every other protocol before it becomes a Host. Panel and subscription agree on the endpoint a Host advertises: a portless host string inherits the inbound port rather than the group's, and a port-only host inherits the inbound address instead of emitting server=%3A8443. LinksForClient prefers host endpoints for every protocol, the way getSubs and inboundLinks already do, so the client-links API no longer ignores managed hosts. * fix(mtproto): migrate legacy share address past unusable hosts The seeder skipped the conversion whenever any Host already carried the address, including one that is disabled or excludes the raw sub type. hostEndpoints drops those, so nothing advertised the address afterwards and the marker committed with no way back. The duplicate check now mirrors that same predicate. UpdateInbound cleared a legacy MTProto shareAddr without the Host conversion AddInbound runs, so re-applying an inbound definition through the API dropped the public address silently. Both paths share one capture helper now. Refresh the generated clients API reference for the summary reworded in the previous commit. * fix(inbounds): wait for the hosts list before building mtproto links The page destructured only `hosts` from useHostsQuery, and that list reads empty both while /panel/api/hosts/list is in flight and after it fails. withMtprotoHostEndpoints then returns the inbound untouched, so Copy, QR and Export advertise the internal listen port — the endpoint this branch exists to replace. It is worse than not fixing it: the seeder has already moved a legacy custom share address into a Host, so the fallback is the panel's own hostname instead of the operator's address, and the Go generators reading the same rows from the DB stay correct, so the two disagree for one inbound. Fold the query into the page's existing readiness gate, the same way useInbounds and HostsPage already consume that hook, so an empty list means "no hosts" rather than "not loaded yet". The error branch fires only when nothing is cached, so a refetch failing on window focus does not blank a page whose host rows are still perfectly usable. --------- Co-authored-by: Sanaei <ho3ein.sanaei@gmail.com> |
||
|
ilyusha
|
d2ac3b4d7a |
fix(cli): let -getApiToken name the token it regenerates (#6405)
* fix(cli): let -getApiToken name the token it regenerates The flag's help text said "Display current API token". It cannot display anything -- tokens are stored as SHA-256 hashes, and the command's own first two output lines say so. What it does is destroy and reissue a credential: GetApiToken calls RecreateByName on the hardcoded name "cli-fallback". The help therefore invited an operator to run a command they believed was read-only, and it revoked a token someone else was holding. Because that name is a single global slot, two callers silently invalidate each other, and the loser is left with a token that answers HTTP 404 with an empty body -- indistinguishable from a wrong base path, so the failure does not even say what happened. install.sh is one of those callers, at lines 1231 and 1325, so the collision already exists inside this repository. Add -tokenName, defaulting to cli-fallback so install.sh and every existing invocation behave exactly as before. -getApiToken stays a boolean on purpose: install.sh calls it as `x-ui setting -getApiToken true`, and a string flag would swallow that trailing argument and mint a token named "true". The name now reaches both branches of GetApiToken. On a database with no tokens the command used to create one called "install", which the CLI could then never rotate -- defeating the stated purpose of the cli-fallback constant, that -getApiToken cannot accumulate admin-equivalent credentials it never revokes. Both branches use the resolved name, so repeated calls rotate a single slot instead of leaving a permanent token behind. Also cap the name at 64 characters in RecreateByName. Create already enforces that limit on the same column; RecreateByName did not, and it now receives operator input. Assisted-by: Claude Code:claude-opus-5 (mostly) * fix(cli): keep the installer's token out of the rotated slot Folding both branches of GetApiToken onto one name made the bug worse in the exact case this change is about. install.sh records the token it gets on a fresh panel; with both branches on cli-fallback, the next bare -getApiToken rotated that very row and silently invalidated the credential written into the install-result file. Restore the split default -- "install" when the database has no tokens, cli-fallback when it does -- so nothing about an unnamed call changes. An explicit -tokenName still applies to both branches, which is what keeps the flag coherent: -tokenName ci-bot now yields ci-bot on a fresh panel too, rather than "install". Pin it with a test that reads the install row's id and hash before and after a rotation, since a name-only assertion would pass against a deleted-and- recreated row. * fix(cli): stop the `-getApiToken true` form from swallowing -tokenName Three corrections from review. Go's flag package stops parsing at the first non-flag argument, so the trailing `true` in install.sh's invocation does not merely get ignored -- it terminates parsing. An operator copying that documented shape and writing `x-ui setting -getApiToken true -tokenName ci-bot` left tokenName empty, so the command rotated cli-fallback: the shared-slot collision this change exists to remove, reachable through the one form the repository itself demonstrates. Verified against the built binary, which printed `The API token "cli-fallback" has been regenerated`. Drop the stray `true` from both install.sh call sites so the documented form no longer teaches the trap, and warn whenever `setting` is given positional arguments, naming what was ignored. A warning rather than an error, because an older install.sh in the wild still passes `true` and must keep working. Cover both branches in the help strings. They described only the rotation path, so on a fresh panel -h announced cli-fallback while the command actually mints `install`, and nothing is regenerated or invalidated there at all -- misleading help being the defect this change set out to remove. Assert the concrete error in the name-length test. It checked only that some error came back, which RecreateByName's empty-name guard and its transaction errors would satisfy just as well. |
||
|
Sanaei
|
2d151d7648 |
Update deps and simplify parsing
Bump frontend and Go dependencies, then modernize a few hot paths with newer Go string/range helpers. Also widen the client form quota/limit fields to improve the layout. |
||
|
Sanaei
|
2ec6c73613 |
feat(xray): update xray-core to v26.9.8 and adapt panel
Bump xtls/xray-core to 37ceb8b4b6 (v26.9.8) and the three binary pins (DockerInit.sh, release.yml Linux + Windows) in lockstep. No deleted symbols; the impact is entirely on the JSON config surface. Outbound "proxySettings" is now refused by the config loader (moved to streamSettings.sockopt.dialerProxy) and a freedom outbound rejects sockopt.addressPortStrategy. Either key in a stored template would keep the core from starting after the upgrade, so a new OutboundRemovedKeysFix seeder rewrites xrayTemplateConfig once: proxySettings.tag becomes sockopt.dialerProxy (an existing dialerProxy wins) and addressPortStrategy is dropped from freedom outbounds. Template saves and outbound subscriptions already run through the vendored loader, so the new refusals surface there with the core's own message. REALITY no longer applies a built-in minClientVer (26.3.27) when the field is empty. The form placeholder and the min/max hints in all 13 locales now say that empty means no minimum. New upstream keys the Zod schemas would otherwise strip, with form support where a sibling field already had it: - blackhole response type "custom" with base64 customResponseData - realm finalmask ipMode (dual/v4/v6) and portMapping (UPnP / NAT-PMP) - quicParams brutalDisableLossCompensation, disableChromeParrot, disableGSO, disableStatelessReset - hysteria masquerade proxy xForwarded - wireguard outbound remoteDNS - routing rule localOS freedom.domainStrategy is only deprecated upstream (auto-migrated to sockopt.domainStrategy with a warning) and is left untouched. |
||
|
Sanaei
|
a5e68f410f |
perf(node): bound the per-client node push and fan out the traffic reset
An operator with several nodes reported that editing a client or resetting
its traffic takes more than ten seconds on the master. Measured against real
Remote HTTP (fake node servers, one client per node), the healthy case is
already fast — 3 nodes: update 51ms, delete 51ms; 5 nodes: 102ms / 103ms —
but two things were not:
- ResetTrafficByEmail still walked its inbounds one node round-trip after
another: 152ms at 3 nodes, 253ms at 5, linear in node count.
- Every per-client op blocked on the SLOWEST node's push. With one node
answering in 3s, update/delete/reset all took 3003ms regardless of node
count. A node that answers the 4s heartbeat probe but hangs on the push
stays "online", so every edit waited on it up to remoteHTTPTimeout — the
ten seconds in the report. More nodes only raise the odds one is sick.
The push is an immediacy optimisation, not the source of truth: every one of
these ops calls MarkNodeDirtyTx inside the transaction that commits the
change, before it pushes, and the node reconcile job converges a dirty node on
its next 5s tick by re-sending the inbound whose fingerprint was not advanced.
So bound the synchronous push with nodeClientPushTimeout = 4s — the budget the
heartbeat and traffic-sync jobs already treat as "responsive" — at the eight
node-branch push sites. A node that does not answer in time is left dirty and
converged a few seconds later instead of stalling the request; the tag-cache
list fetch inside resolveRemoteID shares the same budget.
Once one push in a batch has timed out, the rest of that inbound's batch now
stops pushing too, as AddInboundClient already did: the node is dirty and one
reconcile converges the whole inbound. Deleting three clients on one hung node
went from 30.08s (three remote timeouts) to 4.06s; at the 32-client push
threshold that is 128s of deadlines saved per inbound.
Fan the reset out through fanoutInboundApplies like the other client ops. Its
node propagation is still attempted whatever the node's status flag says, as
before, because nothing replays a traffic reset — the reconcile pushes inbound
config, not counters — so a node still serving after being marked offline must
receive it now or never.
Trade-offs stated plainly: a node that would have answered in 4–10s now falls
to the reconcile's full-inbound push, which on the node is a delete+add of the
inbound and drops its sessions there — the same fallback a failed 10s push
already used, now reached sooner. The reset stays best-effort with no retry
path, which predates this change. The response still reports success while a
timed-out node catches up; the pending-node badge is keyed off node status by
design, so only the warning log records it.
Tests: a barrier test that a sequential reset cannot satisfy; two tests against
a real runtime.Remote and an httptest node that hangs on the push, pinning that
an edit returns at the deadline (exactly one push reached the node, the node
is left dirty) and that a bulk delete stops after its first timed-out push.
All red without the change; the two hung-node tests pay their 4s deadline on
every run.
|
||
|
Sanaei
|
e9e2e30278 |
perf(clients): push a bulk client change to every node at once
|
||
|
Sanaei
|
f2cf589947 |
fix(node): flag every hosting node before a client edit applies
A client edit fans out one transaction per inbound, and each one renames the single shared clients row but calls MarkNodeDirtyTx for only its OWN node. So between the first and the last commit the record already carries the new email while every other node hosting that client is still config_dirty = false. setRemoteTrafficLocked gates the snapshot merge on that flag, so a merge landing in the gap is accepted, sees a pre-rename snapshot, finds no record for the old email and inserts one through syncInboundClients' CreateInBatches — the only place in the panel that creates a client record. The ghost is never in any later merge's perInboundOld, so markSyncOrphan never fires and ReapSyncOrphans never collects it: the operator is left with a permanent second client under the old name. The same stale merge reverts an expiry-only edit instead of duplicating it. Mark every node hosting the client dirty in one serialized write before the fanout starts, so a merge queued behind it skips the node instead of merging a half-applied edit. The nodes were going to be marked by their own applies anyway; doing it up front only moves it earlier, and a client on local-only inbounds never reaches the writer at all. The set is the client's FULL attachment list, taken before the inboundIds filter narrows it: the rename rewrites the one shared record, so an inbound the filter excluded goes stale too. Two tests, both red without the change. The first pins the ordering rather than the end state — it reads the watched node's flag from inside another inbound's push, so moving the marking after the fanout turns it red. The second pins that the filtered path still covers the excluded node. This narrows the window rather than closing it everywhere. A reconcile tick can still clear the flag mid-fanout, and on a filtered edit the excluded inbound keeps the old email in its settings for good, so its next merge duplicates again. The case-drift path — a node reporting another case of a known email — is untouched and still duplicates. |
||
|
Sanaei
|
f072d0448d |
fix(clients): flag the restart a partly-applied edit or delete still needs
|
||
|
Sanaei
|
33058c8eed |
fix(tgbot): use a token telego accepts in the edit-message tests
telego.NewBot validates the token against `^\d+:[\w-]{35}$` before any
option is applied, so the "test-token" literal in the two
not-modified tests failed with "telego: invalid token format" and the
go-test and race jobs went red on every run since #6340. Use a
placeholder token that matches the format; the tests now reach the
mock API server, pass with the guard in place and fail without it.
|
||
|
BlindMaster24
|
8e13f8b172 |
fix(tgbot): suppress 'message not modified' warnings in Telegram edit calls (#6340)
When users click Refresh buttons in the Telegram bot (usage_refresh, client_refresh, ips_refresh, onlines_refresh), editMessageText and editMessageReplyMarkup are always called even when the content has not changed. Telegram returns a 400 "message is not modified" error which was logged as Warning, cluttering the logs on every refresh click. Add isTelegramNotModifiedError helper that detects this specific Telegram API error and logs it at Debug level instead of Warning. |