This introduces a disabled-by-default allow-remote-jobs setting that
controls whether the management server may run jobs (such as debug
bundles) on a peer. The flag propagates end to end: through client
configuration, the daemon SetConfig and Login requests, authentication,
and system info, up to management, where it is stored on the peer and
exposed on the peers API as remote_jobs_allowed. The client refuses any
management-requested job unless the peer has opted in. Because enabling
remote jobs crosses the user-to-root boundary, turning it on requires
privilege, mirroring the SSH-server gate. Administrators can enforce the
setting through MDM policy on both macOS and Windows, and MDM can also
override the debug-bundle upload URL. The change ships policy
documentation and generated profile templates, and adds configuration,
conflict, and enforcement tests covering the opt-in, privilege, and MDM
paths.
Delegating Agent Network today means handing out full account admin, and
regular users cannot see their own usage or how to connect a local tool.
Add two roles on top of the existing agent_network permission
submodules. agent_network_admin owns the whole area (providers,
policies, guardrails, budgets, usage, logs, settings) with read-only
users, groups, peers, and account info needed to build policies, and
nothing else in the account. usage_viewer is the regular User baseline
plus read on the aggregated usage and cost overview: no provider
configuration, no policies, no request-level logs, which can contain
captured prompts. billing_admin gets a proper permission-map entry with
the User baseline so role resolution stops failing with role-not-found;
its plan and invoice permissions stay enforced cloud-side.
Add the self-service endpoints behind the "My Agent Network" view,
available to every authenticated user because both answers are scoped
strictly to the caller. GET /api/agent-network/me/setup returns the
account endpoint plus the providers and models the caller's own groups
authorize, computed with the same rules the proxy enforces: policy
filtering as in policy selection, model allowlist union intersected
with declared models, orphan and disabled providers omitted. Not set up
and no access are deliberately indistinguishable, and the response
carries display metadata only. GET /api/agent-network/me/consumption
returns the caller's own user-dimension counters.
* [client] Reuse the profile's account for iOS SSO logins
Android reads the profile's stored account and passes it as the OIDC
login_hint, and records it again after a successful login. iOS did neither: it
called GetOAuthFlow with an empty hint, so a re-login was resolved by whatever
session the browser's cookie jar held rather than by the account the profile
belongs to. With a non-ephemeral browser session that is the wrong account as
soon as more than one is signed in.
Mirror client/android/login.go: hint from mobile.ReadProfileEmail before the
flow, mobile.WriteProfileEmail after Login succeeds. Storing after Login and
not before keeps a rejected token from leaving a hint that points at an
account which cannot be used.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* [client] Persist the account email on tvOS and on the device flow
Two paths left a profile with no account bound, so every later login went out
without a login_hint — the case this change exists to remove.
WriteProfileEmail went through util.WriteJsonWithRestrictedPermission, which
writes a temp file and renames it over the target. The tvOS App Group sandbox
blocks exactly that, which is why the config sitting next to this file is
written with DirectWriteOutConfig. On tvOS the email write therefore failed and
was dropped with a warning. Use DirectWriteJson: the file is rewritten whole
from a single key, so the only thing atomicity buys here is surviving a crash
mid-write, and a torn file reads back as "no email" and is replaced by the next
login.
The device authorization flow never populated TokenInfo.Email, unlike the PKCE
flow, so a client driven through it — Android TV and tvOS — bound no account at
all. Parse the ID token there too.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* [client] Report a failed close from DirectWriteJson
The deferred close assigned its error to err, but the return value was not
named, so the assignment went nowhere: a close that failed was logged and the
function still returned nil. The write is only durable once the file closes
cleanly, so every caller — the management config, the profile configs and the
profile account email — could be told the data landed when it had not.
Name the return so the assignment does what its shape always intended, and
report the failure once. When the body succeeded the close error is returned and
the caller logs it. When the body already failed, that error is the one that
explains the failure and is what the caller gets, which leaves the deferred log
as the only place the close failure can surface — at debug, per the logging
rules for close errors on writes.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
* [client] Resolve profiles for the sudo invoking user instead of root
The SSH server flags force `netbird up` through sudo, but the CLI resolved
every per-user path with the process user. As root that reads root's own
(empty) local state, so a `sudo netbird up` silently switched the daemon from
the user's profile to the default one — cancelling any login already waiting
in the browser — and then ran an SSO login for the default profile's config.
Whichever account that login returned, the default profile's peer belongs to
someone else, so every attempt ended in "peer is already registered by a
different User or a Setup Key", with nothing telling the user why.
Resolve the acting user through SUDO_USER when running as root: the active
profile, the profile config paths and the stored account email now come from
the invoking user's directories. Privilege decisions are untouched — they stay
on the kernel credentials of the daemon connection, which an environment
variable can never influence; a forged SUDO_USER only selects a profile root
could select anyway.
The invoking user's directories are strictly read-only under sudo. Anything
root wrote there would be root-owned and break the user's own runs, so instead
of chowning files back, the local writes are skipped: the active-profile
bookkeeping and the account-email state simply do not update from a sudo run
(the daemon records the switch on its side; a skipped email write costs at
most one extra account prompt later).
Plain root — no sudo context — has no user to act for, so the ambiguity is
refused instead of guessed at: when the daemon's active profile differs from
what root resolves and no --profile was given, up fails with a message naming
both profiles, instead of silently switching the daemon and failing later with
the ownership error.
* [client] Act on the daemon-resolved profile and fail closed in the root guard
Under sudo the local active-profile mirror is not updated, so up/login
re-reading it after a profile switch acted on the previous profile; use
the daemon-resolved ID directly instead. The plain-root guard now runs
after the readiness wait, denies on lookup errors and empty responses,
and matches the owning username as well; an unowned profile (fresh
install) and a daemon predating the RPC stay allowed. Write-skip
decisions key off the sudo environment alone so a transient user lookup
failure cannot turn a run into writing root-owned files into the user's
directory, and RemoveProfileState honors the read-only rule too.
* [client] Return a wrapped error instead of double-reporting the dial failure
* [client] Read the profile from the daemon when the local mirror is not authoritative
Under sudo without --profile, `up` took the active profile from the invoking
user's local active_profile.txt mirror and drove the daemon to it. But that
mirror is never written under sudo (the SwitchProfile write is a no-op), so it
goes stale after any --profile run and silently switches the daemon back to the
mirror's default. The plain-root guard was meant to refuse exactly this
ambiguity but only ran for plain root, never for the sudo case the fix targets.
When there is no --profile and the mirror is not authoritative (sudo or plain
root), take the profile the daemon already holds for the invoking user instead
of the stale mirror: stay on the user's current profile when the daemon owns it
(or it is unowned, as on a fresh install), and refuse with a --profile hint when
the daemon is on another user's profile. A daemon predating the RPC keeps the
mirror-derived profile.
Reproduce (before this change):
1. As a non-root user misha, with the daemon installed and running:
sudo netbird up --profile work
misha connects on the `work` profile.
2. Because the local mirror write is skipped under sudo,
~misha/.config/netbird/active_profile.txt still says `default` (or is still
absent, which also resolves to `default`).
3. Run a bare:
sudo netbird up
The CLI reads `default` from the frozen mirror and sends ProfileName=default;
the daemon silently switches away from `work` and brings the tunnel up on
`default` — a different account/peer than the one last chosen, with no
warning. After this change step 3 stays on `work`.
* [client] Return a sentinel error instead of nil-nil for the missing daemon RPC
* [client] Load the extend-session hint from the resolved profile
* [client] Fail closed instead of reading root's config when the sudo user lookup fails
* [client] Fail closed in InvokingUser when the sudo user lookup fails
A previous change made baseConfigDir fail closed when SUDO_USER cannot be
resolved, but InvokingUser still fell through to user.Current(). Those two
guards disagreed: the active-profile mirror and the email state refused to
read root's directory, while every profile-path caller happily resolved as
root.
The consequence of a transient NSS failure under sudo was that
Profile.FilePath resolved through getConfigDirForUser("root"), creating
/var/lib/netbird/root and reading the profile JSON from there, and the CLI
sent Username "root" to the daemon in SetConfig and ListProfiles, so the
daemon resolved the same phantom namespace. The invoking user was silently
moved onto a root-owned profile instead of being told the lookup failed.
Fail closed at the single source of the fallback. getConfigDirForUser is
left alone on purpose: it is a pure path helper that also serves
daemon-supplied usernames, and under sudo with a successful lookup it must
still create the invoking user's own profile directory.
The error already says what went wrong: an invalid base64 payload
reports the offending byte offset, and a wrong key size reports the
length. Passing the key itself adds nothing an operator can act on, and
the line is emitted at Error level, so it reaches every log sink and
every debug bundle.
* [client] Allow logging out of the active profile when profiles are disabled
A profile-addressed logout was refused outright when the profiles feature is
disabled: handleProfileLogout ran validateProfileOperation, which returned
Unavailable ("profiles are disabled, you cannot use this feature without
profiles enabled") before looking at which profile was targeted.
The desktop UI always addresses logout by profile — both the profile menu and
the session-expiration dialog send the active profile's ID — so a client with
profiles disabled could not log out at all; only a plain `netbird logout`,
which takes the profile-less path, still worked. Logging out of the profile the
daemon is already running is a deregistration, not profile management, and with
profiles disabled there is a single profile anyway, so every profile-addressed
logout is by definition an active-profile logout.
Replace validateProfileOperation with validateProfileLogout, which skips the
profiles-disabled check when the target is the active profile and keeps gating
logout of any other profile. This mirrors switchProfileIfNeeded, which already
gates only the branch that actually manages profiles. The dropped
allowActiveProfile parameter was always true, leaving canRemoveProfile
unreachable, so both are removed.
* [client] Compare the username and propagate state errors on profile logout
Review follow-ups on the logout gate:
Propagate the GetActiveProfileState failure instead of discarding it. A failed
lookup made the target look non-active, so a caller with profiles disabled got
"profiles are disabled" in place of the real error.
Compare the username along with the ID when deciding whether the target is the
active profile, matching switchProfileIfNeeded. Legacy profile IDs are display
names, so two users can hold the same ID in their own profile directories, and
an ID-only match let one user's logout pass the gate against the other user's
active profile. The default profile is shared and carries no username, so it
keeps matching on the ID alone.
Re-read the active profile before the connection teardown rather than reusing
the pre-flight snapshot. Login switches profiles under guardedConfigMu, which
the logout path does not hold, so a login that landed while the deregistration
was in flight would otherwise lose its fresh connection to a stale flag.
* [client] Address review on the profile logout gate
Pass the username down to logoutFromProfile and reuse the running config only
when the target is the active profile for that username. On an ID-only match a
legacy profile ID shared between two users made the connected-client path
deregister the active peer while its connection stayed up, which the gate fix
alone did not cover.
Split the setup-key-less branch of Login into beginSSOLogin, with the
reuse-the-pending-flow decision in pendingOAuthFlowResponse. Login's cognitive
complexity drops from 37 to 21 (gocognit), clearing the SonarQube report on
this file with no behaviour change.
Point the test fixture at an https URL, since the profiles a gated logout must
not touch only need to be unreachable, not plaintext.
This extends the management-requested remote debug-bundle job with two
new, optional parameters. anonymize_level selects how aggressively the
bundle is scrubbed: "default" keeps internal (private) IP ranges
readable, while "strict" also anonymizes private, CGNAT and link-local
addresses; the value is trimmed and lowercased, and an unknown level is
rejected at creation. upload_url lets an operator point the peer at a
specific upload service instead of the default one; it must be a
well-formed https URL with a host, and an empty value falls back to the
default upload server. Both fields flow through the job workload API and
are surfaced in the create-debug-job modal on the dashboard. Validation
is shared so the client executor and the management boundary agree on
what a valid upload URL is, preventing drift between the two checks.
* [client] Keep the route selection when every requested ID is unavailable
A non-append SelectRoutes() wipes the current selection before applying
the requested one, but it validated the requested IDs only afterwards,
while already mutating. A request naming no available route at all left
every route deselected and returned an error - so a typo in a route ID
silently dropped the user's exit node, and the routes stayed applied
while the selector claimed nothing was selected.
Validate first and bail out before touching any state when nothing in
the request is available. A request with at least one available route
keeps applying the valid part and reporting the rest, and an empty
request still deselects everything, since that is the caller asking for
exactly that rather than a failed lookup.
* [client] Trim the new comments to the contributing guide's length budget
CONTRIBUTING.md caps comments at 90 characters per line and roughly 250
per comment. The three comments added by this PR were over both limits.
The test comments also restated their own test names, so they lose that
half and keep only the why.
* [client] Apply the route selection even when some IDs are unknown
SelectRoutes and DeselectRoutes returned the error before TriggerSelection,
so a request mixing valid and unknown network IDs changed the selector but
never reached the routing table. The valid routes read as selected while
`ip route` showed nothing.
Trigger the selection first and return the error afterwards. The inner
selectRoutes already applied the valid part of a partial request, only the
outer layer dropped it.
* [client] Publish the network selection event on a partial failure
Returning early on error was correct while an error meant nothing had
happened. A partial failure now changes the selection and the routing
table, so returning first left the change with no trace in the event log
or the UI, even though the new state had already been broadcast.
* [client] Cover the append and deselect-all paths of the selection guard
The append path was never destructive and behaves the same with or without
the early return, so that case is characterization rather than a regression
test. The deselect-all case is a real guard: the early return also skips
resetting deselectAll, so a typo no longer drops the "nothing selected,
including future networks" policy.
* [client] Pin that a fully invalid selection disturbs nothing
The selection is now applied on every request, including one where no ID is
known and the selector is left untouched. Nothing may be torn down or
reinstalled on that path.
* Revert "[client] Publish the network selection event on a partial failure"
This reverts commit 26219592.
The event would lie on the opposite path: when no requested ID is available
the selector is left untouched, so an unconditional publish reports a change
that never happened. Telling that case from a partial failure needs the
manager to report whether anything was applied, which is a new signal in its
API and does not belong in a PR about the selector guard. Follow-up instead.
---------
Co-authored-by: Riccardo Manfrin <3090891+riccardomanfrin@users.noreply.github.com>
* Bump go version to 1.26 and go-quic to v0.62.0
* Replace deprecated ecdsa public key assembly and add tests for jwt
* Update goversioninfo
* Pin go toolchain to 1.26.7
* [client] Drop agentConnecting whenever ICE session state clears
Closing a WorkerICE raced a blocked dial goroutine: Close released the
agent while connect() was still inside Dial, and the goroutine's own
cleanup skipped its flag reset because w.agent no longer matched. With
agentConnecting stuck on true, evalConnStatus read the peer as
connected, the reconnection guard stopped sending offers and
same-session offers were dropped, so the peer could not recover without
a restart. An aborted recreate in OnNewOffer reaches the same wedged
state without any race.
Route every teardown path through one abandonNegotiation helper so the
agent and flag fields always clear together; Close now also cleans up
residual state left by an aborted recreate.
* [client] Drive the ICE teardown race test through the real dial goroutine
The regression test simulated the stale goroutine by calling closeAgent
directly, so it pinned the symptom rather than the mechanism. Rework it
to start a real negotiation, tear it down mid-flight and let the actual
goroutine run its own cleanup: with no remote responder the dial can
only fail once Close cancels it, so the interleaving stays deterministic
without sleeps or injection points.
Assert the full idle state that abandonNegotiation owns (agent nil,
connecting false, remote session ID empty) instead of only InProgress,
and make the stale-cleanup ownership test verify that the newer session
survives field by field.
* [client] Assert live remote session ID after stale ICE cleanup
The stale-cleanup test compared a snapshot captured before closeAgent
ran, so clearing the field during cleanup would have gone unnoticed.
Read the field under the mutex after the cleanup instead.
* [client] Give the ICE race tests a no-op signal client
The candidate callback fires from a real gather and dereferences the
signaler, so a nil one crashes the test package intermittently when
gather wins the race against Close. Build the worker with a stub
signal.Client instead.
* [client] Read the ICE dial cancel func from an argument in connect
The error paths read w.agentDialerCancel without holding muxAgent while
OnNewOffer rewrites the field for a newer negotiation, a data race the
new teardown test trips under -race. Reading a stale value also let an
old goroutine cancel another session's dial. Capture the cancel func at
goroutine spawn, like the dial context already is.
* [client] Guard the ICE dial success path against stale negotiations
The stale-cleanup guard in closeAgent only protected teardown. Its
success-path counterpart was missing: an older negotiation could complete
agentDial after a newer one replaced w.agent, then clear the newer
session's agentConnecting, record lastSuccess and publish its dead
connection via onICEConnectionIsReady.
Verify ownership under muxAgent twice: right after the dial returns, so a
stale goroutine drops its connection before touching a closed agent, and
again at the state-commit point, atomic with the agentConnecting and
lastSuccess writes, so a replacement arriving in the meantime cannot get
its state clobbered. Both paths close the stale connection and return
without modifying worker state. A regression test holds session A's dial
open until session B is installed, then releases it; the stale connection
must be discarded and B's agent, connecting flag and remote session ID
must survive.
* [client] Fix ICE teardown test leak and document the stale delivery window
A code review of the stale-negotiation guard found a leftover resource
leak in TestWorkerICE_StaleCloseAgentKeepsCurrentSession: session B is
never closed, so its ICE sockets and blocked dial goroutine live as long
as the test process. Register t.Cleanup(w.Close).
The delivery race flagged after the success-path guard is pre-existing
and self-correcting - the newer negotiation overwrites the transient
endpoint - so document it in the existing todo instead of locking the
callback, which would invert lock order against Conn.Close. Adjust the
teardown test comment to match the now-synchronous Close flag clearing.
* [client] Close the session-expiration dialog only on an actual session renewal
The dialog auto-closed on any Connected status snapshot, but the daemon
emits Connected periodically regardless of session state, so the warning
popup disappeared on the next snapshot (~30s) with no chance to
re-authenticate. Close only when the snapshot's session deadline jumps
past the one the dialog was opened for, meaning the session was renewed
from another surface (tray action, CLI, main window).
* [client] Compare session renewals against the exact deadline in the expiration dialog
The dialog reconstructed its reference deadline from the relative seconds
URL parameter, which carries up to a second of truncation and mount
latency, forcing a renewal-detection margin wide enough to miss a renewal
made shortly after the previous login. Pass the absolute deadline (unix
ms) from both tray call sites - the extend flow's cached deadline and the
final warning's event metadata - so any forward jump in the snapshot
deadline closes the dialog; the seconds-derived fallback with a small
tolerance remains for an unknown deadline.
* [client] Derive the expiration dialog countdown from the deadline
The per-second decrement assumed the interval fires once a second, but
the webview's timers get suspended for tens of seconds under App Nap /
hidden-window throttling, leaving the displayed countdown behind the
wall clock by the suspended time. Recompute the remaining time from the
absolute deadline on every tick so the first tick after a suspension
shows the correct value.
* [client] Tolerate the warning deadline's second precision in the renewal check
The final-warning metadata formats the deadline as RFC3339 truncated to
whole seconds while the status snapshot keeps millisecond precision, so
an unchanged deadline could appear up to 999 ms newer than the exact URL
value and close the dialog on the first snapshot. Allow a sub-second
tolerance on the exact path; any real renewal jumps by at least seconds.
Which applications the tunnel carries belongs with the rest of a
profile's preferences rather than on the Android side, so the choice
follows the profile the user is on.
Adds a split tunnel store beside the SSH session store, over a
"split-tunnel" namespace holding the mode and both selections. The two
selections are kept apart because the platform applies an allow list or
a deny list and never both, and so that switching mode does not throw
away the picks made in the other one.
Only the store itself needs the android build tag; the rest stays
untagged so it is covered by the package's host tests.
* fix(mobile): stop the client synchronously so a restart cannot inherit a cancelled context
Original finding
----------------
A user reported that leaving home and switching from wifi to cellular killed
all Internet traffic until NetBird was turned off. A debug bundle captured the
failure (iOS, CLI 0.75.0, self-hosted management, generated 2026-08-18 01:17;
the incident is at 2026-08-17 22:37:38-51 UTC).
The bundle shows the whole sequence:
22:37:38.255 management sync stream drops (keepalive ACK timeout)
22:37:43.670 Swift: "Network type changed: wifi -> cellular" -> schedules a
restart with a 1s debounce
22:37:44.737 Go: "ensuring wg interface is removed, Netbird engine context
cancelled" - engineCtx dies, every peer gets context canceled
22:37:49.910 iface.go:238 "failed to remove WireGuard interface utun6:
timeout when waiting for interface utun6 to be removed"
-> the teardown stretches out for ~5s
22:37:50.710 Swift: "restartClient: starting client", needsLogin=false
(so this is NOT a login expiry)
22:37:51.013 Go: connect.go:476 "exiting client retry loop due to
unrecoverable error: context canceled" - the OLD run dies here
22:37:51.333 Go: grpc.go:135 "failed creating connection to Management
Service: context canceled" - the NEW start, 2ms after the old
run finally exited
22:37:51.334 Swift: "restartClient: start failed" -> widget disconnected
then nothing for 15 minutes
The tunnel stayed installed with no engine behind it, so every packet was
black-holed. status.txt, generated ~14 hours later, still reads Management:
Disconnected / Signal: Disconnected / Peers count: 0/0 - the client never
recovered on its own.
Root cause
----------
Client.Stop() cancelled a shared ctxCancel field and returned immediately,
without waiting for the run loop to exit. The Swift stop{} completion handler
therefore fired while the Go teardown was still running (stretched out by the
utun6 removal timeout), and the start that followed landed on a context that
the outgoing run was about to cancel.
Two further paths wrote the same shared field. IsLoginRequired() and
LoginForMobile() each overwrote c.ctxCancel, so any call to them during a live
session discarded the running engine's cancel function. restartClient() calls
needsLoginCached() on exactly this path.
Changes
-------
- Stop() now drives the stored ConnectClient: ConnectClient.Stop() cancels the
run context and blocks on runExited, so the caller's completion handler only
fires once the run loop has really finished. The ctxCancel path stays as a
fallback for when no ConnectClient exists yet (e.g. during LoginForMobile).
- Run() owns its cancel in a local variable, so a concurrent call that
overwrites the shared field can no longer cancel this run's context through
the deferred cleanup.
- IsLoginRequired() and LoginForMobile() use local cancels and leave the shared
field alone. LoginForMobile's cancel moves into the deferred cleanup of the
goroutine that outlives the call, so the OAuth token wait is not cut short.
- The Android SDK gets the same treatment. The structural defect is identical
there, but the trigger is absent: Android has no automatic engine restart on
a network type change, and no interface-removal timeout to stretch the
teardown. This part is preventive, not a fix for an observed failure.
* fix(mobile): do not let a superseded startup publish its client
Review found a window the previous commit left open. Run stored its cancel
function and only published the ConnectClient later, after loading config and
constructing the client. A Stop landing inside that window found no
ConnectClient, cancelled the run and returned immediately. A new Run could then
publish its own client, and the cancelled older run — still executing — would
overwrite it with a client that was already being torn down. The next Stop
stopped that stale client and left the live one running with nothing tracking
it.
Runs now carry a generation. Run claims one before doing any work and publishes
its client only while the generation is still current; a superseded run returns
without touching the shared state. Stop bumps the generation, so any startup
still in flight is invalidated, then cancels it and waits for the run to exit
before returning (20s cap so a wedged teardown cannot block the caller
forever).
setState is gone: publishState replaces it at both call sites on each platform.
* fix(ios): add a non-waiting Stop for callers on a deadline
Stop now waits for the run loop to exit, which is what a restart needs but
wrong for stopTunnel: iOS gives NEPacketTunnelProvider only a few seconds
there before it kills the extension, and the wait can run to its 20s cap.
Waiting past the deadline earns a SIGKILL, so the next start inherits a dirty
state instead of the orderly shutdown the wait was meant to buy.
StopWithoutWait tears the client down and returns. ConnectClient.Stop blocks on
runExited with no cap of its own, so the non-waiting path runs it detached
rather than only skipping the runDone wait.
Android keeps a single blocking Stop: it has no equivalent deadline.
* fix(mobile): guard the run lifecycle with a single lock
Stop and beginRun each touched the same lifecycle state across two locks in
sequence: take stateMu, release it, then take ctxCancelLock. A run starting in
that gap installed its own cancel before Stop reached it, so Stop cancelled the
fresh run and left its own target running — the same class of defect this branch
exists to fix, this time in the locking rather than the state.
ctxCancel moves into the stateMu group, and both sides take their snapshot in
one critical section. ctxCancelLock then guarded nothing and is gone.
* fix(mobile): drop the run-generation machinery for a serialized lifecycle
The platform callers (Swift/Kotlin) always stop before starting and coalesce
restarts, so the generation counter guarded against call patterns that cannot
occur. Replace it with a single-run contract:
- startRun refuses a second Run while the previous one has not exited
- finishRun clears the published state on every exit path, including errors
- Stop cancels and waits for the run loop with a bounded timeout; it no
longer calls ConnectClient.Stop, whose wait is unbounded
- concurrent Stops wait on the same exit channel instead of returning early
- a superseded startup no longer reports a clean nil exit
* revert(android): drop the run lifecycle changes
Android does not have the defect this PR fixes. On ux/ios-style-redesign the
EngineRestarter is gone: network changes are handled as events instead of an
engine restart, so nothing stops the client and starts it again.
The remaining stop() callers are all final teardowns on the main thread with a
framework deadline - the stop-engine broadcast receiver, onDestroy, onRevoke and
the binder's stopEngine. A Stop that waits for the run loop would risk an ANR
there for a race that cannot occur, so the fix stays iOS-only.
* fix(ios): make loginComplete race-free
The OAuth goroutine spawned by LoginForMobile sets loginComplete after the
call has returned to Swift, while the Swift side polls IsLoginComplete and
later calls ClearLoginComplete from its own thread. The plain bool made all
three unsynchronized: the store may never become visible to the poller, and
a Clear racing the store can be lost, leaving a stale true that makes the
next login look already complete.
Switch the field to atomic.Bool. It is a standalone flag rather than part of
the run lifecycle that stateMu guards, and it has to stay readable while the
login goroutine is still in flight.
* [client] Add catch-all NRPT rule when NetBird is the primary DNS resolver
* Remove obvious comments
* Install the catch-all rule where the adapter's DNS is set
addDNSSetupForAll makes us the peer's main DNS forwarder, and the catch-all NRPT
rule is the other half of that same job: without it the adapter's NameServer only
adds one more resolver to the set Windows queries in parallel. Having the two in
one place says that, where a separate block at the end of applyDNSConfig read as
an afterthought.
The block could not simply move up: removeDNSMatchPolicies deletes the catch-all
key too, so installing the rule before it ran would have had the rule deleted
moments later. The cleanup now runs first, which is what it was always for - it
clears what the previous apply installed before this one installs anything - and
keeps being unconditional, so a leftover rule from an earlier run cannot survive
into a config that no longer wants it.
* Name the escape hatch after the behaviour it restores
NB_DISABLE_DNS_CATCHALL_NRPT described the mechanism it switches off. What an
operator reaching for it wants is the behaviour they had before, so name it that:
NB_USE_LEGACY_DNS_RESOLUTION, matching NB_USE_LEGACY_ROUTING, the only other
legacy switch in the client.
Not NB_WIN_LEGACY_FULL_TUNNEL_DNS_RESOLVE, as first suggested: the rule follows a
primary nameserver group, not a full tunnel, and putting FULL_TUNNEL in a public
variable name would carry that confusion for as long as the variable lives. No
OS prefix either, since nothing else in the client has one and this switch is
inert anywhere but Windows by construction.
Behaviour and default are unchanged: the catch-all rule is on unless the variable
says otherwise.
* Exempt .local from the catch-all rule
RFC 6762 reserves .local for multicast DNS and says unicast resolvers must not
answer for it. The catch-all rule hands it to us anyway, we forward it to
whatever upstream the primary nameserver group points at, and the answer comes
back NXDOMAIN for hosts that do exist - printers, NAS boxes, anything
announcing itself on the link. Confirmed on a Win11Pro VM: laptop.local resolves
with the client down and returns "Nome DNS inesistente" with it up, and the
client log shows the query arriving on the catch-all handler and being forwarded
to 1.1.1.1.
An NRPT rule that names a namespace and lists no servers is an exemption: the
DNS client resolves those names as it would with no rule at all. What that looks
like in the registry is not what it sounds like. Writing no server value and
clearing ConfigOptions produces a rule Windows treats as a no-op - it never
appears in Get-DnsClientNrptPolicy -Effective and the catch-all keeps the query.
The value has to be present and empty, with ConfigOptions still 0x8: the flag
says the server list is the meaningful part of the rule, and an empty list then
means "no server, resolve normally". Verified both encodings on the VM.
Installed together with the catch-all, since without one nothing captures .local
in the first place, and removed with it.
Exclusivity is unaffected elsewhere, and a more specific rule still wins - a
match domain under .local keeps resolving through NetBird, which is what a
legacy Active Directory domain named corp.local needs. Verified separately that
a match domain does take precedence over the catch-all: declaring fritz.box
against the local router restored laptop.fritz.box while the catch-all was in
force.
* Treat the root namespace as a match domain, not a special case
The catch-all had a function, a registry key and a call site of its own, which
made it look like a different mechanism. It is not: "." is an NRPT namespace like
any other, it just happens to match every name. So it goes into the match domain
list, and addDNSMatchPolicy writes it along with the rest — batching, GPO
variant, volatile keys and cleanup all come for free.
The .local exemption stays a rule of its own, and not for symmetry: it is the one
rule with a different server list, an empty one. Putting it in the same Name
value would give it our resolver and exempt nothing.
Windows expands a rule's Name value into one effective namespace each, so a rule
carrying {.example.com, .} still shows both as separate rows in
Get-DnsClientNrptPolicy -Effective. Nothing is lost for diagnosis by dropping the
dedicated key.
Suggested by Vik in review.
* Do not report a failed NRPT cleanup as success
removeRegistryKeyFromDNSPolicyConfig returned nil for every OpenKey error, so a
permission or registry failure was indistinguishable from a key that was never
there. Cleanup then reported success while the rule stayed in force — which is
how a rule outlives the interface it points at and keeps sending every query to
an address that no longer answers.
Distinguish the two, the way listNRPTRuleKeys already does for the policy store
root: a missing key is nothing to do, anything else reaches the caller.
restoreHostDNS now propagates that error instead of logging it. applyDNSConfig
keeps logging on purpose: there we are about to write fresh rules over whatever
survived, while restore is the path where a rule left behind is the whole
problem.
Also addresses review nits on the tests: doc comments on the two added cases,
reported Close and DeleteKey errors so a failed cleanup cannot contaminate the
next registry test, and a context message on the exemption's namespace assertion.
* Run lazy connection manager for rosenpass peers
* Treat forward-target peers as normal lazy connections
* Run Rosenpass in permissive mode on the embedded proxy
* [client] Add iOS NetBirdSDK profile manager binding
Mirror the Android profile manager in the iOS gomobile binding so the
core's ID-based profilemanager.ServiceManager owns profile state on iOS
too, instead of a parallel Swift reimplementation.
Adds client/ios/NetBirdSDK/profile_manager.go (//go:build ios): an
ID-based ProfileManager wrapping ServiceManager with iOS-specific path
handling (default profile at the container-root netbird.cfg, others as
profiles/<id>.json) and a gomobile-friendly API: List/Add/Switch/Rename/
Logout/Remove plus active config/state path accessors. The default
profile keeps the reserved "default" id and is never assigned a hex id.
* fix(ios): preserve profile name when saving config during auth
NewAuth built a fresh in-memory config from only the management URL, so
the SSO/setup-key save (DirectWriteOutConfig) overwrote the profile config
file the profile manager had just written, wiping the display name to ""
and forcing the UI to fall back to the profile ID. Load the existing config
when present and override only the management URL, keeping the name and keys.
* [client] Extract the mobile profile manager into client/mobile
The Android and iOS gomobile bindings carried two near-identical copies of
the profile manager. Move the shared implementation into a new client/mobile
package and reduce both bindings to thin adapters that only translate to
gomobile-friendly types (gomobile binds per package, so the Profile /
ProfileArray wrappers have to stay platform-side).
Also bring the account-email layer over to the shared package: an SSO login
records the account under <stem>.account.json so the next login can pass it
as an OIDC login_hint. Logout keeps it, profile removal drops it. The suffix
deliberately differs from .state.json, which the engine's state manager owns
in the same directory on mobile.
Adds profilemanager.Prefs (namespaced per-profile preference store) and its
cleanup in ServiceManager.RemoveProfile, exposed through the shared manager
as ProfilePrefs.
* Support per-peer lazy connection state and default proxy peers to lazy
* Classify forward targets from incoming config in lazy exclusion
* Set IsUserspaceBind mock so lazy manager starts in engine test
* Skip lazy exclude reconciliation when the set is unchanged
* Keep cached lazy flag when a sync carries no peer config
Losing the last network only flipped the availability state: the dead management, signal and relay sockets stayed silently connected until their own timeouts, so the client kept reporting Connected with no network at all.
Introduce client/netevents with a Manager that ties the availability state, the connection sweeper and the status recorder together, and move the netstate and netsweep packages under it (netsweep renamed to sweep). SetNetworkAvailable(false) now also sweeps the registered connections so their owners redial and the listener reaches the NoNetwork state.
The Android and iOS bindings own a Manager instance and inject it through the constructors; consumers hold the concrete *Manager whose nil zero value reports always-online and never sweeps, with interfaces kept only as parameter contracts. The relay guard settle wait moved into the Manager as WaitSettled, removing the netevents import from the relay package.
* fix a nil-ptr error occuring in sendInitialSync when the peer being synced is deleted
Signed-off-by: Dmitri Dolguikh <dmitri.external@netbird.io>
* handle a nil ptr in GetPeerNetworkMapComponents
Signed-off-by: Dmitri Dolguikh <dmitri.external@netbird.io>
---------
Signed-off-by: Dmitri Dolguikh <dmitri.external@netbird.io>
The main and Settings windows were created at startup and kept alive hidden on close, so an idle tray held two webview processes for surfaces the user may never open. Both are now built on first show and destroyed on close, which takes the idle footprint on macOS from ~160 MB to ~74 MB.
The WindowManager owns creation: it rebuilds the main window on the next show and hands out live pointers, since a stored one goes stale. Every show is deferred until the frontend reports it has rendered, so a freshly created window is never on screen empty, with a timeout so a frontend that never reports cannot strand a window hidden.
[proxy,management] Serve Bedrock model discovery from the control plane
A Bedrock provider could never answer a model-discovery request. The router
sent GET /inference-profiles to the record's upstream, which has to be
bedrock-runtime.<region> for InvokeModel to work, and that host does not
implement the operation. ListInferenceProfiles is a control-plane operation on
bedrock.<region>.amazonaws.com, and one provider record carries one upstream,
so the two hosts genuinely differ.
The route now carries a discovery host, taken from the catalog's declaration
with the region read back out of the configured upstream, and the listing — and
only the listing — goes there. Inference is untouched. A proxied or self-hosted
Bedrock endpoint gets no discovery host at all rather than a guessed one, since
inventing a host would send the operator's credential somewhere they never
configured.
Two things had to follow for the listing to be usable once it arrives. The
response filter only understood OpenAI's {"data":[{"id":…}]}, so a Bedrock
listing fell through it untouched, offering every profile in the account
whatever the policy said. And discoverableModels intersected by exact string,
so a record registering the raw profile id while a guardrail names the catalog
key intersected to nothing — bounding a working provider's listing down to
empty.
Normalisation is the third. The geography in front of a cross-region profile
was matched against a hardcoded list of four, so every profile issued under jp,
au, ca, sa or us-gov carried its prefix into the pricing key, matched no
catalog entry and metered at zero. It is now recognised by either the geography
or the vendor that follows it, so an id has to be new on both axes at once to
slip through — a live eu-central-1 listing returned "global.xai.grok-4.6" days
after the vendor list was first written.
[management] Offer a provider's live model list in the config form
Adds POST /api/agent-network/catalog/providers/models, which asks a vendor
which models an operator's own credential can actually reach, so the provider
form can offer a live list instead of only the compiled-in catalog. The catalog
goes stale, and it cannot see an account: which OpenAI models an org is
entitled to, which Bedrock inference profiles an account and region hold, which
Vertex models a project has enabled.
The endpoints, auth headers and response shapes come from probing the live APIs
(#7244); each vendor invented its own envelope and none can be guessed from the
request. Bedrock shaped the design: its listing lives on the control plane
while inference must go to the runtime host, so Discovery carries its own host
rather than reusing the record's upstream, and profile ids are taken verbatim
because the region prefix is what AWS requires at invoke time.
A caller supplies either the key they are typing or the id of a saved record
whose stored credential is reused — never both, since accepting both would run
an arbitrary credential under the identity of a record the caller may only be
permitted to read. Gated on Create rather than Read, because this spends the
operator's credential against a third party.
Management has not made outbound calls on an operator's behalf before and it
holds a credential for every provider, so every resolved address must be public
— covering loopback, RFC1918, the cloud metadata address and NetBird's own
100.64/10 range — and redirects are not followed, since a redirect moves the
request to a host the check never saw.
The vendor is authoritative for the id; the catalog stays authoritative for
pricing. A discovered model the shipped table cannot price returns
pricing_known: false so the operator must set rates rather than being
registered at a silent zero.
[proxy,management] Bound model discovery to the caller's own policies
GET /v1/models was bounded by the provider record's enumerated models, which is
the right bound only while a single policy reaches a provider. Where two teams
share one provider under different allowlists, every caller was offered the
union — each model outside their own policy being a request the guardrail
refuses a moment later. A gateway record enumerating nothing was worse: it
offered the upstream's entire catalogue however narrow the policy was.
Each route now carries one rule per authorising policy — its source groups and
the models it permits — instead of a single flattened list. At request time the
router keeps the rules whose groups intersect the caller's, unions their
models, and intersects that with what the provider serves.
nil and [] stay distinct end to end: a policy setting no allowlist reaches the
router as nil and lifts the restriction for the groups it binds, while an
enabled allowlist with no models arrives as [] and permits nothing. Collapsing
them would let a listing that should offer nothing fall open to everything.
The guardrail's own per-provider allowlist is untouched. It is a fail-closed
backstop that cannot tell who is asking, so discovery is now narrower than the
backstop rather than wider.
[proxy,management] Conform the Agent Network endpoint to the LLM gateway protocol
Reviewed the proxy against Claude Code's published gateway contract. The
transport layer already held up; fourteen gaps sat one layer up, in the model
catalog and in the non-inference endpoints clients call.
Two of them cost money. The catalog carried no claude-opus-5 or
claude-sonnet-5, so an operator could not authorise the models coding agents
default to — those requests denied as not-routable, or priced at zero where a
catch-all carried them. And gateway records pin ParserID "openai" while the
same record serves /v1/messages, so Anthropic responses were read with the
OpenAI parser, which never looks at message_start where input tokens live:
input metered as roughly zero on every stream and cost was skipped entirely.
The rest fix requests refused for structural rather than policy reasons: model
discovery denied for every account with a model allowlist, token counting
denied on Bedrock and mis-parsed on Vertex, startup probes refused and written
into the access log at every session start, and denials rendered in a shape no
LLM client parses. Two changes are additive by design — the deny body keeps
every field it had and adds the vendor's error object alongside, and body-level
identity injection is now gated on the request's dialect so it stops sending
OpenAI-shape fields into Anthropic bodies that reject them.
The end-to-end work turned up one more: the discovery filter treated any slash
in a model id as a gateway prefix, which would have dropped every self-hosted
"Qwen/..." model from the picker.
* [misc] Let gomobile init fetch the toolchain gobind needs
The previous commit's CI run confirmed the failure on a comment-only diff
off main, so the cause is not any branch's changes:
Android / Build failure
iOS / Build failure
`gomobile init` re-installs gobind from x/mobile@latest whatever gomobile is
pinned to, and setup-go sets GOTOOLCHAIN=local, so the install dies the
moment @latest declares a newer Go than go.mod does:
gomobile: go install golang.org/x/mobile/cmd/gobind@latest failed: exit status 1
go: golang.org/x/mobile@v0.0.0-20260821190718-4776eadac327
requires go >= 1.26.0 (running go 1.25.12; GOTOOLCHAIN=local)
GOTOOLCHAIN=auto on that step alone lets the install fetch what it asks for.
Scoped to the step deliberately: the repo's Go version and every build below
it stay on go.mod's toolchain, so this buys the mobile jobs nothing except
the ability to run gobind.
Pinning gobind next to gomobile does not work — init re-installs @latest
regardless. A durable fix is to stop `init` reaching the network at all, or
to track x/mobile's Go requirement in go.mod; both are larger changes than a
red CI warrants right now.
