v3.3.1 removed the Panel Proxy URL field from the UI but left the stored
panelProxy/tgBotProxy values in the DB. The Telegram bot still reads
tgBotProxy directly, so a stale value masked the panelOutbound egress
fallback. Add a one-off seeder to drop both rows.
Closes#5266
The paged client list is sorted/paginated server-side but fetched with staleTime: Infinity, so the WS client_stats patch only refreshed traffic on already-visible rows — newly connected clients never appeared and the sort order went stale until a manual refresh.
Add a 5s refetchInterval so the current page tracks reality, and drive the table overlay off isPlaceholderData so the background poll does not flash it.
The .online-dot uses vertical-align: middle, which in inline layout
aligns to baseline + half x-height — visibly off-centre inside the Ant
Tag's line box. Add a .dot-tag utility (inline-flex, align-items:
center) and apply it to the Online tag so the dot and label share one
centred axis. Other dot usages (Nodes page Space, card heads, stat
rows) already sit in flex containers and are unaffected.
The #5035 change tagged node-hosted entries with the node name to
disambiguate multi-node subscriptions, but the node name is
panel-internal and leaked into the profile names end users see in
their client apps. Drop the suffix entirely — remarks are the
admin-set inbound remark again.
The vlessenc fix (#5185) enabled flow on XHTTP only in the security=none
branch of genVlessLink, and the Clash builder still gated flow on
network==tcp. With XHTTP+REALITY+vlessenc the panel accepts and stores
the flow (inboundCanEnableTlsFlow passes), but subscriptions dropped it,
so clients received configs without xtls-rprx-vision.
Add vlessFlowAllowed mirroring inboundCanEnableTlsFlow — tcp with
tls/reality, or xhttp with vlessenc regardless of security layer — and
use it in both the vless:// link generator and the Clash proxy builder.
Updating geo files printed raw curl progress meters showing 0 bytes when
files were already current (curl -z conditional download), claimed success
unconditionally, and restarted xray even when nothing was downloaded —
confusing enough to be reported as a bug (#5230).
Now each file reports updated / already up to date / download failed,
failures no longer print a success message, and the restart (which drops
live connections) only happens when a file actually changed. Same for the
non-interactive 'x-ui update-all-geofiles' command.
Revert formatInboundLabel to the pre-#5151 behavior: display the inbound
remark when set, otherwise the inbound tag, instead of "tag (remark)".
Affects the Attach clients / Attached inbounds views and client lists.
Routing keeps its own tag (remark) formatting.
Bump several Go indirect dependencies (golang.org/x/exp, golang.org/x/tools, google.golang.org/genproto) and update go.sum accordingly. Regenerate frontend/package-lock.json to record updated npm package versions (including Ant Design and related rc-component packages and other transitive updates).
- Split the Happ and Clash/Mihomo routing sections out of Information into
their own dedicated tabs.
- Extract the profile/branding fields (title, support URL, profile page,
announcement, theme dir) out of the mislabeled "Subscription Title"
divider into a new Profile tab.
- Move the Update Interval setting into Information and drop the
single-field Intervals tab.
- Add the "profile" tab label across all locales.
Client add/update/remove also rewrite settings.clients on each attached
inbound, so the Xray config query could go stale. Invalidate it alongside
the clients and inbounds buckets.
- Replace the Telegram "Notification Time" free-text field with a guided
cron builder: @every + number + unit (s/m/h), the @hourly/@daily/@weekly/
@monthly macros, and a Custom option that seeds a valid 6-field crontab
(cron runs with seconds enabled) as an escape hatch.
- Move "Restart Xray After Auto Disable" from the External Traffic tab to
Panel Settings, where it belongs.
- Add a "Template guide" link to the Sub Theme Directory setting pointing at
docs/custom-subscription-templates.md.
- Localize all new strings across every locale.
Bump row action icons to 18px across the clients, inbounds, groups and
nodes tables for better visibility.
In the clients table, cap the Client column at 220px and give Duration a
fixed width so the Traffic column becomes the flexible one that absorbs
the remaining horizontal space instead of Client growing oversized.
Add a per-inbound "Route through Xray" toggle (off by default) plus an
optional outbound picker on MTProto inbounds. mtg only supports a SOCKS5
upstream, so when enabled the panel injects a loopback SOCKS bridge into
the generated Xray config — tagged with the inbound's own tag — and mtg
dials Telegram through it via a [network] proxies upstream. The router
then governs Telegram egress: matchable in the Routing tab, or forced to a
chosen outbound/balancer via the picker.
- mtproto: Instance carries RouteThroughXray + XrayRoutePort (in the
fingerprint); InstanceFromInbound parses them; renderConfig emits the
socks5 [network] upstream; freeLocalPort exported as FreeLocalPort.
- xray.go: injectMtprotoEgress appends the loopback SOCKS bridge and
prepends an optional inboundTag->outbound/balancer rule, hot-appliable
like injectPanelEgress.
- inbound.go: backend-owned egress port persisted in settings, allocated
once and carried across edits (stored value wins); stripped with the
inert outboundTag when routing is off; allocation failure fails the save;
routed add/update/del force a config regen.
- mtproto_job: skip folding mtg metrics for routed inbounds (the bridge,
carrying the inbound tag, is metered by xray_traffic_job) to avoid
double-counting.
- frontend: toggle + outbound/balancer Select (useOutboundTags) on the
MTProto form; i18n keys for all locales.
Replace the per-outbound burstObservatory polling (one temp xray spawn +
up to 15s of /debug/vars polling per outbound, serialised) with one
shared temp xray instance per batch: every tested outbound gets its own
loopback SOCKS inbound plus an inboundTag->outboundTag routing rule, and
the panel times a real HTTP request through each one in parallel. The
probe returns as soon as the response lands and records the HTTP status
plus an httptrace breakdown (proxy connect / TLS via outbound / first
byte) shown in the result popover.
New POST /panel/api/xray/testOutbounds endpoint (array in, results in
input order, max 50); the legacy /testOutbound endpoint now delegates to
the same engine. Test All chunks HTTP probes 16 per request, and a batch
whose shared process never comes up (one structurally-broken outbound
poisons the config) retries each item in an isolated instance so the
broken outbound reports xray's real error while the rest still test.
Split the group traffic summary into two inbound-style cards: a "Total
upload / download" card with up/down arrow icons and a "Total Usage" card
with the pie icon. Add the totalUpDown label across all locales.
Switching the transport to mKCP auto-seeds a mkcp-legacy entry into
finalmask.udp, but switching back to another transport only dropped the
kcpSettings blob and left the mask behind. It survived downstream pruning
(finalmask.udp was non-empty) and bled into every client share link.
Strip auto-seeded mkcp-legacy entries from finalmask.udp whenever the
network changes away from kcp, leaving user-authored masks intact.
Fixes#5221
Add per-group up/down to GroupSummary (backend + schema), surface them
as Upload/Download columns in the groups table, and fold upload/download
into the Total traffic summary card. Rename the group "Clients in group"
column to just "Clients" across all locales.
Accept the @-prefixed abstract socket form (e.g. @xray/in.sock) for an
inbound listen address, not just path-based sockets. The form now allows
Port 0 for both, and the Address help text documents the @ form across
all locales. The backend already treated both prefixes as unix sockets.
@
An authenticated admin could set xrayTemplateConfig.log.access/error to an
arbitrary path (via the raw Xray editor or a wholesale DB import), making the
supervised Xray process write its log there — an arbitrary file write as the
Xray user (root in many deployments). resolveXrayLogPaths now reduces any log
path to its base filename under config.GetLogFolder(), so absolute paths and
".." traversal can no longer escape the log folder; "" and "none" still
disable logging.
Importing a second inbound whose clients overlap an already-imported inbound
failed with "UNIQUE constraint failed: client_traffics.email". The import path
carries exported ClientStats, and tx.Save(inbound) cascaded that has-many
association as INSERTs whose ON CONFLICT targets only the primary key, so a
shared email (already owning a row from the first import) tripped the global
unique constraint.
Omit the ClientStats association on save and insert the carried stats ourselves
with the same OnConflict{email, DoNothing} guard AddClientStat already uses:
new clients keep their imported counters, shared emails reuse the existing row.
Then run an idempotent AddClientStat pass over all clients so any client present
in settings but missing from the stats payload still gets a traffic row (else it
would escape quota/expiry accounting), and propagate insert errors so the tx
rolls back instead of committing a partial state.
Add a subSortIndex field to inbounds that controls the order of links
in subscription output only: the raw sub body, the HTML sub page, and
the JSON/Clash formats (all served from the same query). Lower values
come first; ties keep id order. The panel inbound list is unaffected.
The value is editable in the inbound form next to the share-address
fields, propagates to nodes via wireInbound, and follows the usual
node-sync rules (copied on import, mirrored while not dirty, never a
structural change).
Rescoped from #5214 by @Ponywka.
- Rename tabs: "Basic" → "Basics", "Config" → "Credentials"
- Move reverseTag field from Credentials tab to Basics tab
- Move IP log button inline with limitIp field (tooltip button, edit mode only)
- Hide random email button when editing an existing client
- Add tooltips to totalGB and limitIp fields with descriptive hints
- Rename labels: "Total Sent/Received (GB)" → "Traffic Limit (GB)", "Duration" → "Duration (days)"
- Add renewDays translation key for auto-renew label with unit hint
- Remove redundant filterOption and width style from AutoComplete group selectors
- Update all 15 locale files with new and renamed translation keys
Multi-node panels had no way to narrow the inbounds or clients lists to
a single node. Add a node filter to both pages:
- Inbounds: a toolbar select (All / Local / each node) that filters the
list client-side; shown only when the panel has nodes or node-attached
inbounds.
- Clients: a Nodes multi-select in the filter drawer. Node selections
are mapped onto inbound IDs client-side and fed through the existing
inbound CSV paging parameter, so the paging backend is untouched; an
impossible id (-1) is sent when no inbound matches so the filter
yields an honest empty result. InboundOption now carries nodeId to
make the mapping possible.
The local panel is selectable via a 0 sentinel (inbounds without a
nodeId). New i18n keys in all 13 locales.
WG peers were only identifiable by their keys. Add an optional panel-side
comment per peer: editable in the inbound form (echoed next to "Peer N"
in the section header), stored in the settings JSON alongside the
panel-only privateKey (xray-core ignores unknown peer fields), and
appended to the share link / .conf remark so the device is identifiable
in client apps too.
GET /panel/api/clients/get/:email returned the quota (totalGB) but not
the bytes the client has actually used, forcing API consumers to scrape
it elsewhere. Add a sibling "usedTraffic" field (up+down, including the
cross-node global overlay) next to "client" and "inboundIds".
The fragment "packets" field was a locked dropdown (tlshello / 1-3 / 1-5)
in both the finalmask TCP-mask form and the Freedom outbound form, while
xray-core accepts any "n-m" packet range. Replace both with an
AutoComplete that keeps the presets as suggestions and validates free
input as "tlshello" or a numeric range.
The settings page validates the whole AllSetting object before saving, so a
tgCpu value that isn't an integer in 0-100 (left over from an older or corrupt
setting) failed validation with "tgCpu: Invalid input" and blocked saving every
other setting too. Clamp/round tgCpu to a valid integer in the model
constructor, defaulting to 80 when it isn't a finite number.
NodeFormSchema required inboundTags, but the inboundTags Form.Item is only
mounted when inboundSyncMode is "selected" - antd onFinish omits unmounted
fields, so saving with the default "all" mode failed schema validation with
Zod generic "Invalid input" (regression from #5178; same class as the
earlier pinnedCertSha256 fix).
Also tolerate null inboundTags (Go nil slice) for nodes saved before #5178,
both in the form schema and NodeRecordSchema, and normalize edit-mode values.
The panel seeded xhttp configs with scMaxEachPostBytes=1000000 and
scMinPostsIntervalMs=30 — xray-core''s own defaults — and emitted them
into every generated config and share link. The literal
scMinPostsIntervalMs=30 is a stable DPI fingerprint that Russia''s TSPU
keys on to block connections on mobile networks.
New configs no longer seed these values (empty schema/template defaults,
so xray-core applies its internal defaults). For configs already stored
with the old defaults, the link/subscription builders now drop values
equal to xray-core''s defaults instead of advertising them — covering
panel share links, the raw subscription, and the JSON subscription
without requiring every inbound to be re-saved. Non-default values the
user set deliberately are still emitted.
- Issue acme.sh HTTP-01 over IPv4 unless the host has no global IPv4
address: the hardcoded --listen-v6 started a v6-only standalone
listener, so validation of a domain whose A record points at this
host always failed (#4994).
- Add a custom cert/key path option to the "Set Cert paths" menu so
certificates living outside /root/cert (e.g. certbot under
/etc/letsencrypt) can be wired to the panel from the CLI (#5010).
- Derive the displayed Access URL from the certificate's actual SAN
list instead of the cert folder name, list the other covered names,
and show the panel's custom-path certificate in "Show Existing
Domains" (#5070). Also silence find when /root/cert doesn't exist.
applyClientFieldByEmail patched only the first inbound that the
client_traffics row pointed at. For a multi-inbound client the sibling
inbounds kept the old expiryTime/totalGB/limitIp in their settings JSON,
and the next SyncInbound over a stale sibling reverted the edit in the
normalized records — the Telegram bot's expiry change appeared to apply
and then sprang back. Patch the field on every inbound linked to the
email, falling back to the legacy single-inbound lookup for clients that
were never normalized.
An inbound pushed to nodes keeps the same remark on every copy, so a
multi-node subscription (and the panel's per-client link view) listed
several identically-named entries differing only by address. Append the
node name to the remark of node-hosted inbounds unless the admin already
included it.
The inbound/client context menus hold a dozen items; when antd flips a
tall menu upward near the screen edge it overflowed the top of the
viewport, hiding the first entries. Cap the overlay height and scroll.
The schema and form inputs allowed any value >= 1, but xray-core rejects
UdpIdleTimeout outside 2-600 seconds at startup, so an out-of-range value
silently killed the whole config.
The fallbacks card only renders for VLESS/Trojan over RAW with TLS or
Reality security, and a new inbound starts at security=none — so the Add
Inbound page looked like it had lost fallback support entirely. Show an
inline hint in that state pointing at the Security tab.
The panel egress is injected as a routing rule, so a routing balancer is
a valid target for it (unlike the geodata download, which dials a forced
outbound tag and bypasses the router). Surface routing balancers in the
panel outbound picker as a separate group, and emit balancerTag instead
of outboundTag in the injected egress rule when the configured tag names
a balancer, so the panel's own traffic load-balances across its members.
The `node` share-address strategy resolves to an address only when the
inbound can live on a node; for a local inbound it is always empty and
behaves like `listen`. Drop the `node` option from the picker unless an
enabled, node-eligible node exists, and coerce the value to `listen`
otherwise so the Select never shows an option that does nothing.
The v3 outbound test spins up a temp xray that probes the outbound via
burstObservatory. Two regressions made it report "Failed" for healthy
outbounds on high-latency / tunnel-routed boxes (e.g. default route over
an OpenVPN tun device to a remote proxy), even though client traffic over
the same outbound works:
- Each probe disables keep-alive, so every attempt is a cold round-trip
(redial + re-handshake). The 5s per-probe timeout was too tight for such
paths and every probe timed out. Restore the ~10s budget the pre-v3
SOCKS-based test gave a cold connection (timeout 5s -> 10s) and widen the
poll window 12s -> 15s so one full probe can complete and surface alive.
- The temp config set log error to "none", discarding the real failure
reason, so "Failed" was undiagnosable. Route error logs to stderr ("")
like the production template does, so the probe error (DNS lookup
failure, connection refused, deadline exceeded, TLS error, ...) is
captured into the panel/Xray log, and point the operator there in the
generic timeout messages.
Multi-node sync/import drift can leave the same client twice inside an
inbound's legacy settings.clients JSON while the normalized
client_inbounds table stays clean (SyncInbound dedupes the rows it
writes but never rewrites the JSON). All three subscription builders
iterated that JSON verbatim, so every duplicate entry became a
duplicate profile in the raw, Clash, and JSON output.
Filter and dedupe by email in one shared helper (link generation keys
purely on inbound + email, so same-email entries are pure duplicates
and dropping them is lossless). The clash/json services' own
inboundService copies became unused and are removed.
The generated WARP outbound used domainStrategy ForceIP, which may pick
the AAAA record for engage.cloudflareclient.com; on a host with
half-configured IPv6 the handshake then blackholes with nothing in the
logs. ForceIPv4v6 prefers IPv4 and still falls back to IPv6 on
v6-only hosts, matching the official WARP client's behavior.
It also set noKernelTun: false, so with root privileges the real
outbound used kernel TUN — a path that needs CAP_NET_ADMIN plus fwmark
routing and fails silently on many VPS setups — while the panel's
connectivity probe always tests with noKernelTun: true. The status
check and real traffic exercised different data paths and could
disagree. Generate WARP and NordVPN outbounds with the userspace TUN
so both follow the path the probe validates.
Only affects newly added/reset outbounds; existing templates keep
their saved settings.
Subscriptions resolved a node-managed inbound's address to the node's
panel address unconditionally, so an inbound bound to a specific public
IP advertised an endpoint clients could not reach. The shareAddrStrategy
field added in #5162 only applied to panel share/QR links by design.
resolveInboundAddress now follows the same order as the panel's link
builder: 'listen' prefers a routable bind, 'custom' prefers shareAddr,
and the default 'node' keeps the existing node-first behavior, so output
is unchanged for inbounds that never set the field. Applies to raw,
JSON, and Clash subscriptions, which all resolve through this path.
Help text in all locales updated to drop the 'subscriptions are not
affected' caveat.
The orphan sweeps in setRemoteTrafficLocked deleted the (node, email)
baseline row unconditionally whenever an email was missing from one
inbound's snapshot stats — even though baselines are keyed per node, not
per inbound. For a client attached to two inbounds of the same node whose
stats the node reports under only one of them, the sweep for the other
inbound deleted the baseline at the end of every sync cycle. Depending on
inbound order, the baseline written earlier in the same transaction was
wiped each time, so the next cycle computed delta against a missing
baseline (zero) and the client's traffic froze permanently.
Scope both sweeps to the union of emails across the whole snapshot: a
baseline is only dropped when the email left the node entirely.
The DNS server table columns were memoized with only [t] as deps, so
they permanently captured the first render's openEditServer callback,
which closed over the initial (null) dns settings. Clicking Edit then
resolved the server to null and the modal fell back to default values.
Stabilize openEditServer/deleteServer (and the fakedns equivalents)
with useCallback and include them in the column memo deps so the
columns refresh whenever the servers list changes.
sysctl --system re-applies every sysctl file on the host, surfacing
unrelated "Invalid argument" errors from the distro's own defaults
(e.g. Ubuntu 22.04's 50-default.conf on kernels 5.14+). Apply only
/etc/sysctl.d/99-bbr-x-ui.conf on enable, and drop the redundant
re-apply on disable since sysctl -w already restores the live values.
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.
A client attached to several panels has one aggregated row on each
master, but a node only ever saw its local share: the node UI
under-reported usage, and the node kept serving a client whose
cross-panel total had already exceeded its quota — the master's disable
push doesn't kill established connections unless the node restarts xray
itself.
Masters now push their aggregated per-client counters to each node from
NodeTrafficSyncJob (throttled, scoped to the clients that node hosts).
The node stores them in the new client_global_traffics side table keyed
by (masterGuid, email), overwritten on every push so a master-side
reset propagates, and:
- overlays max(local, pushed) onto UI read paths (slim inbound list,
inbound detail, clients list, WS stats, per-email lookups). The full
/panel/api/inbounds/list stays un-overlaid on purpose: it doubles as
the traffic snapshot masters poll, and overlaying it would corrupt
every master's delta accounting;
- trips disableInvalidClients when any master's pushed total exceeds
the client's quota, so the existing RestartXrayOnClientDisable flow
disconnects the client locally;
- clears the side rows on traffic reset, auto-renew, and client
delete, keeping a renewed quota window clean.
Supersedes #5204, which folded pushed globals into client_traffics and
compensated with read-back baselines — that double-counted first-sight
emails and could not work with several masters sharing one node.
client_traffics is the per-email accumulator shared across every inbound
and node the client is attached to. setRemoteTrafficLocked deleted it
unguarded in two sweeps — when a node inbound vanished from the snapshot
(node reinstall, tag change, another master's reconcile on a shared
node) and when an email left one inbound's stats — even though the
email was still attached elsewhere. The next sync then re-seeded the
row with that node's counter alone, so the panel showed the last
changed panel's number instead of the summed total.
Guard both sweeps with emailUsedByOtherInbounds, matching what the
manual-edit path (updateClientTraffics) already does. Truly removed
clients are still cleaned up by the zero-attachment sweep.
The mobile card rendered a static antd Badge for every bucket. When the
client is enabled, online, and not depleted, render the same animated
online-dot span the desktop Online column and the nodes list use.
The auto-disable job flips client.enable off in the settings JSON when a
client expires or exhausts its traffic, so the inbounds-page rollup filed
every ended client under the gray Disabled badge (and double-counted it
in Depleted when stats were present). Classify with depleted-first
priority, matching computeClientsSummary and the client info modal.
Also backfill cross-inbound client_traffics rows in GetInboundsSlim:
the row is keyed on email and only preloads on the inbound the client
was created on, so on every other attached inbound the depleted/expiring
checks could never fire.
When update.sh regenerates a short webBasePath, it writes the new path to the
database after the panel is already running with the old path loaded in memory.
Without a restart the server keeps serving the old path while the UI shows the new
one, making the new path unreachable.
Instead of requiring a manual SOCKS5/HTTP URL, the panel now lets the
admin pick an Xray outbound from a dropdown (same UX as Geodata
Auto-Update). At runtime, injectPanelEgress appends a loopback SOCKS
inbound (tag: panel-egress) and prepends a routing rule so the panel's
own HTTP traffic — version checks, Telegram, normal geo-file updates —
is routed through the chosen outbound. Xray-native Geodata Auto-Update
is unaffected (it uses its own geodata.outbound inside Xray). Blackhole
outbounds are excluded from both picker dropdowns since routing any
download through one just drops it. Translations updated for all 13
locales.
Add a hot-apply layer that computes a diff between the old and new
generated config and applies only the changed parts through the Xray
gRPC HandlerService and RoutingService, avoiding a full process restart
whenever possible. A restart is still performed when sections that have
no reload API (log, dns, policy, observatory, ...) actually change.
Key additions:
- internal/xray/hot_diff.go: ComputeHotDiff with canonical-JSON
comparison (sorted keys, null=absent, full number precision) so UI
reformatting never triggers a spurious restart
- internal/xray/api.go: AddOutbound/DelOutbound, ApplyRoutingConfig,
GetBalancerInfo, SetBalancerTarget, TestRoute gRPC wrappers
- internal/web/service/xray.go: tryHotApply, ensureAPIServices,
GetBalancersStatus, OverrideBalancer, TestRoute service methods
- internal/web/controller/xray_setting.go: balancerStatus,
balancerOverride, routeTest API endpoints
- frontend: BalancersTab live-status/override columns, RouteTester
component, Restart button removed (Save now hot-applies)
- balancer-helpers.ts: syncObservatories never creates observatory
sections for random/roundRobin balancers (no reload API → restart)
- i18n: balancerLive/Override/routeTester keys added to all 13 locales
Remove the panel-side custom geo download feature (service, controller,
/panel/api/custom-geo/* endpoints, CustomGeoResource model, UI tab) in
favor of Xray-core's native geodata section
(https://xtls.github.io/config/geodata.html).
- pass the top-level "geodata" key through xray.Config so it survives
the template round-trip into the generated config
- add a Geodata Auto-Update section to the Xray Updates modal that
edits geodata (cron schedule, download outbound, asset list) in the
config template and restarts Xray on save
- previously downloaded geo files in the bin folder keep working in
ext: routing rules; the orphaned custom_geo_resources table is left
in place so existing source URLs stay recoverable
* 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.
Delete/update located the client in an inbound's settings JSON by the
record's credential (uuid/password/auth). When that credential drifted
from the inbound JSON -- e.g. a rotated UUID left behind, or duplicated
by a past partial-update bug -- the lookup failed with "Client Not Found
In Inbound For ID: <uuid>" and aborted the whole operation, making the
client impossible to remove from the panel.
Key every delete/update/detach path on email, the client's stable
identity. This survives credential drift and heals duplicate-email
entries by removing all of them.
- Delete/DeleteByEmail/Detach/DetachByEmailMany -> DelInboundClientByEmail
- delInboundClients / bulkDelInboundClients: match settings by email
- UpdateInboundClient: locate the entry to replace by email
(param clientId -> oldEmail); update all callers to pass the email
- bulkAdjustInboundClients: match by email
- writeBackClientSubID: pass email; drop unused sourceProtocol param
- make per-inbound deletion idempotent via ErrClientNotInInbound
- remove now-orphaned DelInboundClient, clientKeyForProtocol and
getClientPrimaryKey; scale test deletes by email
HTTP-pinging a subscription outbound always reported "Probe timed out".
The frontend sends only the template outbounds as allOutbounds, but
subscription outbounds are injected at runtime and aren't in that list,
so burstObservatory had no outbound matching the tag to probe.
Append the tested outbound when its tag is missing instead of only when
allOutbounds is empty, so the probe always has a target while preserving
the template outbounds that back dialerProxy chains.
Expose mtg's [domain-fronting] section (ip/port/proxy-protocol) plus
proxy-protocol-listener, prefer-ip, and debug on MTProto inbounds. Each
key is written to the generated mtg-<id>.toml only when set, so mtg's own
defaults apply otherwise. The instance fingerprint now covers these
fields, so editing an option restarts the sidecar.
Since MTProto is mtg-served (not Xray), sniffing does not apply: hide the
Sniffing tab and the Advanced sniffing sub-editor, drop it from the
Advanced "All" JSON view, and emit empty sniffing in the wire payload,
all gated by a new canEnableSniffing predicate.
* fix(logs): render journalctl output in the SysLog viewer
The log viewer's parseLogLine only understood the app-log format
(2006/01/02 15:04:05 LEVEL - body). With SysLog ticked the backend
returns journalctl lines (Mon DD HH:MM:SS host ident[pid]: LEVEL - body),
so the parser mistook the journal time for the level and dropped the
body, leaving only timestamps. Detect and strip the journald prefix,
keep the journal timestamp as the stamp, then parse the real level and
body from the remainder.
* feat(mtproto): surface mtg output and add status reporting
mtg's stdout/stderr was captured by a writer that kept only the last
line and showed it nowhere, so the reason a proxy could not reach
Telegram was invisible. Stream mtg output line-by-line into the x-ui
log, tagged per inbound, so it appears in the panel log viewer and
journald.
Also fix mangled log lines: logger.Info uses fmt.Sprint, which drops
the space between adjacent string operands, producing output like
'inbound3on0.0.0.0:8443'. Switch the affected mtproto calls to the
formatted (*f) variants.
Add show_mtproto_status to x-ui.sh so 'x-ui status' reports each
mtproto inbound's mtg process state and bind address.
* fix(logs): parse all journalctl message shapes in SysLog viewer
Real journalctl output mixes four message shapes after the
'Mon DD HH:MM:SS host ident[pid]:' prefix: go-logging 'LEVEL - msg'
(x-ui/xray), Go std-log with an embedded date (net/http, runtime),
telego's '[timestamp] LEVEL msg', and systemd lines. The viewer only
understood the first, so std-log and telego lines — which never contain
' - ' — collapsed to a bare timestamp (e.g. the 8s telego 409 spam).
Extract the parser into a pure, testable module and teach it the other
shapes: strip the redundant Go std-log date, lift the level out of
telego brackets, and always keep the message body. Add a unit test
covering each shape with real captured lines.
* fix(mtproto): reap orphaned mtg sidecars so a stale one can't break new clients
On Linux x-ui does not kill its mtg children when it dies (no kill-on-exit,
unlike the Windows job object). After a crash, OOM, kill -9, or update, a
stale mtg keeps holding the inbound port with an OLD secret, so new clients
fail the FakeTLS handshake and get silently domain-fronted to the fakeTLS
domain instead of proxied to Telegram (a few MB of traffic, never connects).
Sweep orphans at startup: on the first reconcile, before x-ui starts any of
its own mtg, scan /proc and SIGKILL any process whose executable is our
mtg-<goos>-<goarch> binary. x-ui is the sole owner of mtg, so anything alive
then is an orphan. Runs once per process (swept guard), survives the
binary-deleted-during-update case via /proc/<pid>/cmdline, and is a no-op on
Windows (job object) and other platforms.
Also clear stray mtg in update.sh/install.sh after stopping x-ui, anchored to
the 'mtg-linux-<arch> run ' invocation so the pattern can't match unrelated
command lines (e.g. x-ui.sh's own 'grep mtg-linux').
* fix(logs): drop dead body initializer flagged by eslint no-useless-assignment
* fix(mtproto): drop remark fragment from tg://proxy export link
The mtproto export link appended the inbound remark as a URL fragment
(tg://proxy?server=...&port=...&secret=...#remark). Telegram Desktop
rejects a proxy deep link with a trailing fragment as 'This proxy link
is invalid', breaking one-click import, and a remark is meaningless for
proxy links across clients. Stop adding it in both the panel link
(genMtprotoLink) and the subscription service. Fixes#5105.
* fix(x-ui.sh): remove unused check_mtproto_status helper
show_mtproto_status does its own process check, so check_mtproto_status
was dead code. Drop it (per Copilot review on #5107).
Update frontend package.json and refresh dependencies for a new release (frontend version -> 0.3.0). Regenerated lockfile and upgraded multiple JS packages (notably @swagger-api/apidom family, @rc-component packages, codemirror, etc.) and added libc metadata where applicable. Also update Go module dependencies (go.mod and go.sum) as part of routine dependency maintenance.
Sum up+down across each group's clients via a LEFT JOIN on email in
ListGroups, expose it as trafficUsed on GroupSummary, and render it as a
new column plus a "Total traffic" summary card. Drops the unused "Empty
groups" card and its translation key.
Importing an inbound that was exported from another panel (or whose node
was later deleted) carried a non-zero nodeId referencing a node that does
not exist on the importing panel. AddInbound -> nodePushPlan -> NodeSyncState
looked that node up and returned gorm's "record not found", so the import
failed with "Something went wrong Failed: record not found".
Node IDs are panel-local and not portable, so the import handler now drops a
nodeId that does not exist on this panel (new NodeService.NodeExists helper),
importing it as a local inbound instead of erroring. This mirrors the existing
nodeId==0 normalization in the same handler.
The SSL Certificate Management "Revoke" option only called acme.sh --revoke,
leaving local files under /root/cert/<domain>, acme.sh renewal state under
~/.acme.sh/<domain>, and the panel cert paths in the DB. Renamed to
"Revoke & Remove": it now also drops acme.sh tracking, deletes the local cert
directory and acme.sh state dirs (RSA + ECC), resolves the real IP address(es)
for IP certs so their renewal state is torn down too, and clears the panel cert
paths (x-ui cert -reset) + restarts when the deleted domain was in use.
The Telegram bot's config (enable flag, token, proxy, API server) was read
only during a full server.Start(). The in-process "Restart Panel" path
(StopPanelOnly/StartPanelOnly -> stop/start with the bot flag false) skipped
the bot entirely, so enabling the bot or changing its proxy did not take
effect until a full OS-level `systemctl restart x-ui`.
Cycle the Telegram bot in the panel-only restart path while still leaving
Xray and the traffic writer untouched (preserving the #4265 freeze fix), so
"Restart Panel" reconciles the bot with the latest settings.
Fixes#5033
* 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
Settings and Xray config endpoints now live at /panel/api/setting/* and /panel/api/xray/*, registered under the existing /panel/api group so they inherit the same Bearer-or-session auth (checkAPIAuth) as the rest of the API. An API token is a full-admin credential, so this just makes the surface consistent. The SPA page routes /panel/settings and /panel/xray are unchanged.
BREAKING CHANGE: the old /panel/setting/* and /panel/xray/* paths are removed. External callers must switch to the /panel/api/ prefix. Frontend call sites, API docs, the dev proxy, and the route-documentation test are updated to match.
A new emit_jsonschema.go walks the same allow-listed structs as the zod/types/examples emitters and writes generated/schemas.ts (SCHEMAS). build-openapi mounts it under components.schemas and points each typed response obj at a $ref instead of an untyped {} blob, so Swagger renders real models and openapi-generator can emit clients.
Also add a vitest guard that safeParses every EXAMPLES entry against its generated zod schema, reviving the previously unused generated/zod.ts and catching drift between the example and schema emitters.
ServeOpenAPISpec shipped servers:[{url:"/"}], so Swagger UI "Try it out" and external generators hit the origin root and ignored a non-root webBasePath. Inject the runtime base path into the single servers entry at serve time, touching only that field via json.RawMessage so the rest of the spec is preserved verbatim.
Stop hand-writing OpenAPI response examples, which kept drifting from the real payloads (clients/traffic missing fields, inbounds/list exposing userId which is json:"-", the fictional inbound-443 tag instead of the real in-<port>-<transport> form).
tools/openapigen now emits frontend/src/generated/examples.ts: a per-struct example instance built from type defaults, validate oneof/min bounds, and example: struct tags, with nested-ref expansion and a cycle guard. build-openapi.mjs composes the {success,obj} envelope from it for any endpoint annotated with responseSchema (+ responseSchemaArray for lists); the hand-written response is dropped for those. Service DTOs InboundOption/ApiTokenView/ProbeResultUI are added to the walker.
#4996: client password regeneration now produces a valid Shadowsocks 2022 PSK (correct base64 length per cipher) when an SS2022 inbound is attached, in both the single and bulk client forms; backend surfaces ssMethod on /inbounds/options so the UI can pick the right length.
Also: Swagger UI persists the Authorization token across reloads (persistAuthorization).
A fragment TCP finalmask with an empty length (the form's default for a
newly added mask) serializes to a 0-0 range, and xray-core rejects
LengthMin == 0 with a fatal config error that aborts the whole process,
taking every inbound offline. Default a new fragment mask to length
100-200 and add a form validator rejecting an empty value or a zero
minimum range before save. Verified against xray 26.6.1 (#4998).
URL-safe base64 (-/_ with stripped padding) broke Shadowrocket import: it decodes the add/sub path segment as standard base64 and rejects -/_, so the subscription was silently not added. Revert to plain btoa() output as originally shipped in #3489.
Commit 6ed6f57b (#4988) added tcpWindowClampHint, the four realityTarget* keys, and the three FreedomHappyEyeballs* keys to en-US only. Fill in the other 12 locales so the new sockopt hint, REALITY target validation messages, and Freedom Happy Eyeballs options are localized. Technical tokens (REALITY, Xray-core, IPv4/IPv6, Happy Eyeballs, port examples, ms) are kept literal.
* feat(nodes): add stable panel GUID identity (multi-hop phase 0)
Per-panel autoincrement node ids are meaningless one hop away, so in a chained topology (Node1 -> Node2 -> Node3) the master cannot attribute online clients or inbounds to the physical node that hosts them (#4983).
Introduce a stable self-identifier: each panel generates and persists a panelGuid (settings table, mirroring GetSecret), returns it in panel/api/server/status, and the master learns it per node via the heartbeat into a new Node.Guid column. Guarded so an old-build node or a failed probe never clears a known GUID. No behavior change yet - this is the identity foundation Phases 1-2 key on.
Refs #4983
* feat(nodes): attribute inbounds to their origin node by GUID (multi-hop phase 1)
Add Inbound.OriginNodeGuid: the GUID of the panel that physically hosts an inbound. Empty means this panel's own xray; set means it was synced from a node. SetRemoteTraffic now fills it per synced inbound - keeping a non-empty value the node forwarded from its own sub-node (so a transitive inbound stays attributed to the deepest node across hops), and otherwise attributing the node's own local inbounds to that node's GUID. Empty (old-build node without a GUID) leaves the existing node_id-based attribution untouched.
The field rides the existing inbound JSON, so /list propagates it up the chain with no serve-side change. Phase 2 will key per-node online off this instead of the panel-local node_id.
Refs #4983
* feat(nodes): key online status by node GUID end-to-end (multi-hop phase 2)
Replace the panel-local node-id keying of per-node online status with the stable panelGuid, so a client several hops down a node chain is attributed to the node that physically hosts it instead of the intermediate node it syncs through (#4983).
xray/process.go stores each direct node's reported GUID-keyed subtree and merges them (correct at any depth); the service assembles GetOnlineClientsByGuid (own clients under this panel's GUID + every node under its GUID). FetchTrafficSnapshot fetches the new /clients/onlinesByGuid, falling back to the flat /onlines for old-build nodes (keyed under the node's GUID or a master-local synthetic id). The node rollup, the WS onlineByGuid/activeInbounds fields, and the inbounds-page rollup all scope by GUID; local inbounds get their OriginNodeGuid filled with the panel's GUID at serve time so the frontend keys uniformly.
Old-build nodes degrade to the prior flat behaviour via the synthetic node:<id> key. Refs #4983
Refs #4983
* feat(nodes): surface transitive sub-nodes on the master (multi-hop phase 3a)
Each panel publishes read-only summaries of the nodes it manages via GET /panel/api/server/descendants (node API token). The heartbeat job caches each direct node's summaries; GetNodeTree merges them as transitive model.Node projections (Id 0, Transitive=true, ParentGuid = their parent node's GUID) and recomputes InboundCount/OnlineCount/DepletedCount per origin GUID so a direct node shows only its own inbounds and each sub-node shows its own (#4983).
The Nodes-page list endpoint and the heartbeat broadcast now return the tree; GetAll stays direct-only for probing/syncing. One transitive level is surfaced (covers Node1->Node2->Node3); deeper recursion is a follow-up. Backend only - the Nodes-page nested UI lands next.
Refs #4983
* feat(nodes): render transitive sub-nodes nested + read-only on the Nodes page (multi-hop phase 3b)
The Nodes page now shows a node's downstream sub-nodes (learned via the descendants tree) as indented, read-only rows ordered right under their parent: no enable toggle, probe, edit, delete, update, selection, or history expander - just a 'Sub-node' tag whose tooltip names the parent it is reached through. Desktop table and mobile cards both handle it. Transitive rows are keyed by GUID (their Id is 0) so they don't collide with real nodes (#4983).
Rows nest by parentGuid rather than AntD tree-children to avoid clashing with the existing per-row history expander. New labels added to en-US (other locales fall back until translated). Refs #4983
Refs #4983
* i18n(nodes): translate subNode/subNodeTip across all locales
Phase 3b added these two Nodes-page keys (read-only sub-node tag + tooltip) only to en-US; fill in the other 12 locales so the multi-hop sub-node UI is fully localized. The {parent} placeholder is preserved in every translation.
Refs #4983
The IP-limit job scrapes the Xray access log, which keeps lines tagged with a client's old email for up to a log-rotation cycle after a rename or delete. For each such email getInboundByEmail (settings LIKE %email%) found nothing, so the job logged 'failed to fetch inbound settings: record not found' every run and recreated an inbound_client_ips row for the dead email (rows reappeared even after manual deletion).
processLogFile now resolves the inbound once per email: if it maps to no inbound (gorm.ErrRecordNotFound) it logs at Debug, drops any orphan tracking row, and skips - so stale entries self-heal instead of spamming ERROR. The resolved inbound is passed into updateInboundClientIps, removing its internal lookup. updateClientTraffics also calls DelClientIPs alongside DelClientStat so a full inbound edit that drops an email doesn't leave a ghost row.
Closes#4963
A standalone inbound bound to a public/wildcard listen that still carried a stale inbound_fallbacks row had its share/subscription link rewritten with the master's port + Reality/TLS settings (keeping only its own transport), producing an unusable link that silently fails - the client connects but no traffic flows. The leak hit every backend link surface: subscription URL, JSON sub, Clash sub, and the panel Client Information link.
Gate projectThroughFallbackMaster on reachability: only project a child that is not directly reachable on its own listen (loopback or a unix-domain socket). A public or wildcard inbound advertises its own port + security regardless of any fallback row. Legit loopback/socket fallback children still project as before.
Closes#4987
Long TLS-tab labels overflowed their field in locales with wider strings (e.g. Russian 'Pinned Peer Cert SHA-256'). Add AntD labelWrap to the inbound and outbound form modals so any over-long label wraps onto a second line instead of overflowing, and shorten the Russian pinnedPeerCertSha256 label to fit.
Closes#4986
The per-client traffic merge built enable = CASE WHEN ? = 0 THEN 0 ELSE enable END, mixing an integer literal with the boolean enable column. PostgreSQL rejects this with SQLSTATE 42804, aborting every node traffic merge transaction every 5s and freezing all up/down/last_online accounting on Postgres main panels. Replace with enable AND ?, which is type-safe on Postgres (boolean AND boolean) and identical in semantics on SQLite: the node may only disable a client, never re-enable one the panel already disabled.
Closes#4964
customTables is the plural array form of customTable, so default it to [""] and edit it with a tags Select instead of binding a text Input to an array value.
* feat(x-ui.sh): add migrateDB command and menu for SQLite .db <-> .dump
Adds an "x-ui migrateDB <file>" subcommand and a PostgreSQL-menu option (9)
that convert between a SQLite .db and a portable .dump file. Direction is
auto-detected from the extension and delegated to the bundled binary
(x-ui migrate-db --dump/--restore), so no external sqlite3 client is needed.
Depends on the matching binary support, so it is only usable from the next
panel release.
* fix(x-ui.sh): address review feedback on migrateDB
Per Copilot review on PR #4910:
- Probe the bundled binary for migrate-db --dump support and fail with a clear
upgrade message instead of a raw "flag not defined" error on old builds.
- Prompt before overwriting an existing .dump in dump mode (parity with restore).
- Refuse to restore into the live database path while x-ui is running, to avoid
corrupting the running panel.
- Fix the usage/synopsis strings to show input is optional ([file] not <file>).
The vless/trojan link parser's TLS branch read only sni/fp/alpn, so the
ech (echConfigList) and pcs (pinnedPeerCertSha256) query params were
dropped on import even though buildStream allocates both fields. Read
them in applySecurityParams to match the inbound link generator and the
hysteria2 parser.
The frontend section still described the old multi-page app. Rewrite it for the current React Router SPA (single index.html bundle), TanStack Query server state, the Zod source-of-truth model plus generated types, and link logic under src/lib/xray. Update the "adding a page" flow to the route-based approach and drop the stale MIGRATED_ROUTES / "no React Router" notes.
Correct the Vite pin (was 8.0.13 "never bump", now exact 8.0.16) and add "npm run test" to the PR checklist.
Document the new "Run 3x-ui (Postgres)" launch profile and fix the gitignore claim: .vscode/launch.json is checked in, not gitignored.
The external proxy "Host" field was bound to dest (the connection address that becomes the link host) but labeled "Host", misleading users into thinking it set a transport host header. Relabel it to "Address" to match what it actually controls.
Add per-entry ECH (echConfigList) to the external proxy schema, form (shown under Force TLS = TLS), the TS link generator, and the Go sub services: ech is emitted on share links and vmess objects, and written into the stream so the JSON subscription picks it up via the existing tlsData reader.
Node-backed client and inbound edits no longer hard-fail when the backing node is offline or disabled. Edits commit to the panel DB immediately and reconcile to the node when it reconnects (eventual consistency); the panel is the single source of truth for desired config.
- Add Node.ConfigDirty/ConfigDirtyAt; mark a node dirty when an edit commits without reaching it (cleared via CAS on ConfigDirtyAt after a full reconcile).
- nodePushPlan() reads node state fresh from the DB, skips the push for offline/disabled nodes (no 10s hang), and treats push failures as non-fatal across every mutation path (client add/update/del + bulk + attach/detach; inbound add/update/del/toggle/resetTraffic).
- ReconcileNode() pushes the panel's desired config to a node on reconnect (refreshing the remote tag cache first) and prunes node-side orphans; runs before the traffic pull in the node sync job.
- While a node is dirty the traffic pull applies only up/down deltas and node-initiated disables, never overwriting desired config from a stale node snapshot.
- Surface a non-blocking 'saved; will sync on reconnect' warning to the UI.
Validated with a two-panel Docker E2E: client delete/update, attach/detach, and inbound add/delete all reconcile correctly offline -> reconnect.
client_traffics is keyed by email (one shared row per client across every
inbound it is attached to). addClientTraffic filtered with
`inbound_id NOT IN (node inbounds)`, so when a client was attached to both a
node inbound and the mother inbound and the node inbound was attached first,
the shared row carried the node inbound's id (AddClientStat uses OnConflict
DoNothing and never refreshes it) and the local xray's traffic for that client
was dropped entirely. The client showed online but its usage stayed at zero
unless the mother inbound happened to be attached first.
Match purely by email instead. The reported emails come only from the local
xray, which only knows local-attached clients, so the query is still correctly
scoped, and this also repairs already-broken rows that a per-row AddClientStat
fix alone could not.
echConfigList was stored under tlsSettings.settings but the share-link
and JSON-subscription generators only read fingerprint and
pinnedPeerCertSha256 from that bag, silently dropping ECH from VLESS,
Trojan and VMess links. Read echConfigList alongside them and flatten it
into tlsSettings.echConfigList for the JSON subscription.
Closes#4933
Drive every client/inbound/group endpoint at 100k-200k clients on PostgreSQL and fix the latent issues found in previously-unbenchmarked paths:
- enrichClientStats: chunk the email IN lookup (was an unchunked bind that crashed past 65535 clients without traffic rows, taking down GetInbounds/GetInboundDetail/GetAllInbounds)
- GetOnlineClients: add the missing nil-process guard its siblings already have, so ListPaged no longer panics before xray starts
- GetClientTrafficByEmail: read UUID/subId from the indexed clients table instead of parsing the inbound's full settings JSON (439ms to ~1.5ms, flat in N)
- BulkResetTraffic: replace the per-email serialized loop with one chunked bulk UPDATE in a single transaction
- DelDepleted: delegate to the already-batched BulkDelete instead of deleting each depleted client one by one
Adds a postgres-gated full endpoint sweep plus an A/B benchmark, and SQLite correctness tests for the changed methods.
Bulk client operations bound their entire working set in a single
WHERE x IN (...) clause, which exceeds PostgreSQL's 65535-parameter limit
(and SQLite's 32766) and gives the planner a pathological query, so they
failed outright on inbounds/selections larger than the limit. Every such
query is now chunked at 400 items:
- BulkDelete / delete-all-clients: six IN queries chunked, and the
per-row delete tombstone (which swept the whole in-memory map on every
call, O(N^2)) replaced with a single bulk sweep.
- BulkAdjust: record and inbound-mapping lookups chunked.
- AddToGroup / RemoveFromGroup (bulk add/remove to group): three IN
queries chunked.
- replaceGroupValue (rename/delete group): inbound-mapping lookup chunked.
- List (all-clients listing): link and traffic lookups chunked.
Measured on PostgreSQL 16: delete-all-clients on a 100k-client inbound
now completes in ~7s (previously crashed at the parameter limit); bulk
add/remove to group ~6s and full client list ~1s at 100k.
sync_scale_postgres_test.go adds skip-gated benchmarks for delete-all,
group add/remove, and list.
Follow-up to the SyncInbound bulk rewrite, fixing the remaining O(M*N)
and O(M)-round-trip behaviour in the add/delete and bulk paths that made
them time out on large inbounds (worst case minutes), especially on
PostgreSQL.
- compactOrphans: chunk the "email IN (...)" lookup (400/batch) instead
of binding every email at once. A single huge IN exceeded PostgreSQL's
65535-parameter limit (and SQLite's) and made the planner pathological,
so add/delete failed outright past ~100k clients.
- emailsUsedByOtherInbounds: new batched form used by delInboundClients
(BulkDetach) and bulkDelInboundClients (BulkDelete), replacing a
per-email global JSON scan (O(M*N)) with one scan, and skipped entirely
when keepTraffic is set.
- BulkCreate: rewritten to validate/dedup in one pass, then group clients
by inbound and add them in a single addInboundClient call per inbound
(one getAllEmailSubIDs, one settings rewrite, one SyncInbound) instead
of running the full single-create pipeline per client.
- Bulk delete/adjust: batch DelClientStat/DelClientIPs with IN deletes
and wrap the settings Save + SyncInbound in one transaction, so the
per-row writes share a single fsync instead of one per row.
Measured on PostgreSQL 16 (one inbound, M=2000 affected clients):
- create: 8m35s (M=500) -> ~1-5s
- detach: 52s -> ~4s (flat in N)
- delete: ~16s -> ~1-4s
- adjust: ~20s -> ~7-10s
add/delete of a single client on a 200k-client inbound stays in seconds.
sync_scale_postgres_test.go adds skip-gated benchmarks (XUI_DB_TYPE=
postgres) for the single add/delete and the five bulk operations.
Every client mutation funnels through SyncInbound, which ran O(n) DB
round-trips per call: one SELECT per client, a Save+UpdateColumn per
client, and a per-row junction INSERT. Toggling a single client on a
large inbound issued thousands of queries and timed out, badly so on
PostgreSQL where each round-trip pays TCP latency.
SyncInbound now:
- loads existing records with a single chunked SELECT ... email IN (...)
instead of one query per client
- writes only the records that actually changed (skips no-op Saves), so
toggling/editing one client writes one row, not all of them
- batch-creates new records and batch-inserts the junction rows
Merge and sticky-field semantics are unchanged. Measured on PostgreSQL
16: a single-client toggle on a 50k-client inbound drops from ~8m54s to
~0.9s, and seeding 50k clients from ~2m48s to ~1.6s; 200k clients sync
in seconds.
A skip-gated benchmark (web/service/sync_scale_postgres_test.go, run
with XUI_DB_TYPE=postgres) reproduces and verifies the scaling.
- Cleanup on issuance/install failure now also removes the acme.sh
${domain}_ecc (and ${ip}_ecc) directory, not just ${domain}, so a
failed run no longer leaves partial state behind.
- The 'existing certificate' check only reuses a cert when its
fullchain.cer and key files are actually present and non-empty;
otherwise the broken state is removed and issuance proceeds. This
fixes the 0-byte fullchain.pem produced by reusing failed state.
- Menu option 5 (set cert paths) now registers the acme.sh --installcert
hook with --reloadcmd 'x-ui restart' when acme.sh knows the domain, so
auto-renewal copies the renewed cert and reloads the panel.
MHSanaei