mirror of
https://github.com/netbirdio/netbird.git
synced 2026-10-07 14:09:07 +02:00
d56e6fc5f4ae07e352ada103fd6354d95d1715c3
4
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Riccardo Manfrin
|
a8dff998ef |
[client] Gate settings updates on value, not on field presence (#7398)
* [client] Gate settings updates on value, not on field presence
The update-settings kill switch (--disable-update-settings /
NB_DISABLE_UPDATE_SETTINGS / the MDM DisableUpdateSettings key) forbids
changing settings, but it decided what a "change" was by looking at
whether a field was present in the request. The CLI fills the whole
config surface of SetConfigRequest and LoginRequest from its flags and
environment on every `netbird up` (setupSetConfigReq in cmd/up.go), so a
client configured by environment restates its own configuration on every
start and tripped the gate every time.
SetConfig only warned about that, but Login carries the same fields and
was gated the same way, and Login runs inside the CLI's backoff loop: the
daemon answered every attempt with codes.Unavailable, `netbird up` never
completed, and a container with NB_DISABLE_UPDATE_SETTINGS plus any
config env var (NB_MANAGEMENT_URL, for one) could not come up at all.
Both gates now compare values. Config.WouldChange is the dry-run half of
UpdateConfig: it runs the very same diff logic (Config.apply) against a
copy of the stored config, so the gate cannot drift from what an actual
update would do, nor go stale when a field is added. A request that
restates what the profile already holds changes nothing and is allowed; a
request that diverges is refused exactly as before, and a dry run that
cannot be evaluated fails closed. A profile with no config on disk yet is
judged against the config the daemon would create for it.
For Login, the compared input comes from loginOverridesInput, which
persistLoginOverrides also uses to perform the write, so the gate judges
precisely the two fields a login can persist (management URL, pre-shared
key) and no field it ignores.
Two adjacent defects surfaced while making the comparison exact:
- Config.apply compared URLs as raw strings, so the same endpoint spelled
without its default port ("https://api.netbird.io" vs
"https://api.netbird.io:443") counted as a new value and rewrote the
config. It now compares the parsed forms.
- UpdateConfig did not collapse the redacted pre-shared key, unlike
UpdateOrCreateConfig and DirectUpdateConfig, so a UI round-trip of the
mask replaced the stored key with asterisks.
The CLI warning for a refused SetConfig said the method was not available
in the daemon, which sent people looking for a version mismatch that was
not there; it now reports the refusal.
* [client] Do not write the profile config while only reading it to decide
The update-settings gate needs the stored config to decide whether a request
changes anything, so the previous commit moved that read ahead of the refusal.
The read is not side-effect free: profilemanager.GetConfig writes the config
back whenever apply() has to fill in a default the file was missing. A request
that the gate then refuses had therefore already rewritten the profile file.
PeekConfig is GetConfig without that write-back. The returned config is still
normalized in memory, which is what the decision needs; the file is left
exactly as it was found. Every caller of storedConfigAtPath feeds a gate that
can refuse, so they all peek.
Note for reviewers: the daemon still normalizes the file on startup and on
every real update, so nothing depends on a read performing that write.
* [client] Compare service URLs as endpoints, not as strings
Three places in one request path each had their own notion of "same
management URL": the config layer compared the parsed URLs as strings, the
privileged-change gate compared scheme + host + effective port, and the MDM
conflict check compared strings after filling in the default port. Only the
middle one was right.
A string comparison answers the wrong question. "https://api.netbird.io",
"https://api.netbird.io/" and "https://API.netbird.io:443" are one endpoint
written three ways, so a client restating its own management URL with a
trailing slash — a normal way to write it — was still read as a client asking
to be repointed, and the update-settings gate refused it. The MDM check had
the same flaw against the enforced value.
profilemanager.SameServiceURL is now the single comparison: same scheme, same
host case-insensitively as DNS names are, same effective port. The config
layer, the privileged-change gate and the MDM conflict check all defer to it,
so there is one answer to "did this URL change?" instead of three.
* [client] Stop the config dry run from generating throwaway keys
The dry run's baseline for a profile with no config file yet went through
createNewConfig, and apply() generates a WireGuard and an SSH key whenever it
finds those fields empty. The baseline is compared against and discarded, so
every evaluation minted a keypair it threw away — and logged "generated new
Wireguard key". The CLI retries Login in a backoff loop, so a first `netbird up`
on a fresh profile filled the daemon log with what reads like peer-key rotation.
The baseline now starts from the shared skeleton with placeholder keys, so
apply() has nothing to generate. No ConfigInput field maps to either key, so
the comparison is unaffected.
* [client] Cover the login the update-settings gate used to refuse
The gate's decision procedure was tested directly, but no test drove the Login
RPC that the refusal actually broke: the CLI retries Login in a backoff loop,
so a refused no-op login is what kept a client configured by environment from
ever coming up. The handler-level coverage stopped at the refusal case, which
passes on the pre-fix code too.
This test fails on the pre-fix daemon with "update settings are disabled" and
passes now. Past the gate the handler does real work the test does not stand
up, so it asserts only that the refusal did not happen.
* [client] Re-take the update-settings decision under the config lock
Login checks twice on purpose: the first check refuses the ordinary case
early, and authorizeAndPrepareLogin re-takes the authoritative one under
guardedConfigMu because the first is unsynchronized against a concurrent
privileged request. The update-settings decision is now equally
value-dependent — it compares the request against the stored config — but it
was taken only in the first, unlocked check.
So a login that was a no-op when it was checked could be written after a
concurrent writer had repointed the profile, which is exactly the window the
lock exists to close. The decision is now re-taken alongside the privilege one,
which also makes it the last read before persistLoginOverrides writes.
The test drives that interleaving through the existing afterLoginPreCheck seam
and fails without the re-check.
* [client] Drop an unreachable guard and fix two stale comments
- loginOverridesInput's nil-message guard cannot be reached: Login
dereferences the message well before it, in storedLoginConfig.
- The docstring above afterLoginPreCheck described persistLoginOverrides,
which lives further down the file and now carries its own.
- UpdateConfig's comment named DirectUpdateConfig; the function is
DirectUpdateOrCreateConfig.
* [client] Make config reads pure and provision the identity explicitly
Reading a config wrote it back. profilemanager.readConfig persisted whatever
apply() had filled in, and ReadConfig created and wrote the file outright when
it was absent, so every reader was quietly a writer: a gate deciding whether to
refuse a request, a UI listing profiles, a mobile getter reading one preference.
The previous commit worked around that with a PeekConfig variant, which left
two read functions with opposite side effects and the antipattern still there
for everyone else.
Only one thing in a read genuinely had to be persisted: apply() generated the
WireGuard and SSH keys when it found them empty, and a generated key cannot be
recomputed — losing it means the peer comes back with a different identity and
registers again. Everything else apply() fills in is a deterministic default
that the next read recomputes anyway.
So identity provisioning is now its own step, Config.EnsureIdentity, and the
callers that provision write the result out themselves, in the open:
- Server.getConfig, the daemon's provisioning point;
- the CLI's foreground login, which is about to dial management;
- update() / directUpdate(), the config write paths — a stored profile can
legitimately carry no identity, since a mobile logout clears the keys in
place, and the next write is what has to mint a new one.
ReadConfig and GetConfig no longer write anything, PeekConfig is gone, and the
dry-run baseline no longer needs placeholder keys to keep apply() from minting
real ones.
One deliberate leftover: readConfig still calls util.EnforcePermission, which
chmods a config file whose permissions are too broad. It changes no content and
is idempotent, and dropping it would leave a legacy file world-readable until
its first write.
* [client] Name the two config readers for what they do
ReadConfig and GetConfig differed in one thing — what happens when the file is
absent — and neither name said which was which:
- ReadConfig -> ReadOrGenerateConfig (reads it, or generates one in memory)
- GetConfig -> GetExistingConfig (reads it, or fails)
Three comments went with them:
- GetConfig's said "return with Config and if it was created. Errors out if it
does not exist", which described a bool it does not return and a creation it
never performs.
- ReadConfig's explained that it does not write, which is what a reader is
supposed to do anyway.
- Server.getConfig's said it "errors out if it does not exist", which it does
not — it resolves a default config, and now provisions the identity too.
* [client] Do not panic on a config with no sync message version
apply() wrote the incoming sync message version through the stored pointer,
without checking it was there: a config that carries no version yet made it
dereference nil. Reachable from the update-settings dry run, which runs inside
a request handler — where failing closed is the worst acceptable outcome, and a
panic is not one.
The field is now reassigned like every other optional one, which also means
apply() no longer mutates anything the caller still holds through a pointer, so
the dry run's copy has one less field to detach.
Reported by cubic-dev-ai on PR #7398.
* [client] Compare the client certificate paths before reporting a change
apply() assigned the incoming mTLS certificate and key paths and set updated
unconditionally, without comparing them to what the config already held. It is
the same presence-instead-of-value mistake this branch set out to fix, one
layer down: a caller restating its own certificate paths was reported as
changing them, which trips the value-aware update-settings gate.
Reported by cubic-dev-ai on PR #7398.
* [client] Address the remaining bot findings on PR #7398
- Login logged the active-profile-state error and returned the same cause; the
repo's guidelines call for one or the other, and the wrapped error is the one
that carries context. (CodeRabbit)
- `netbird up` reported a codes.Unavailable SetConfig failure as "the daemon
refused the settings update", but that code also covers a daemon that became
unreachable. It now reports what the daemon said without asserting why.
(cubic-dev-ai)
- TestLogin_ChangingTheManagementURLIsRefused asserted the error and nothing
else, while "refused before it can touch daemon state" is the contract. It now
checks the stored management URL, the in-progress login and the active profile,
matching its SetConfig counterpart. (cubic-dev-ai)
* [client] Keep the peer identity out of a read that finds no file
ReadOrGenerateConfig resolves a default config when the profile has no file
yet, and createNewConfig was minting the WireGuard and SSH keys while doing so.
That defeated the provisioning pair it was meant to serve: the CLI's foreground
login calls EnsureIdentity to find out whether it has to persist the keys, got
generated == false because the read had already generated them, and so never
wrote them out. The login then dialed management with an identity that only
existed in memory, and the next login registered a second peer.
createNewConfig no longer provisions. createProvisionedConfig is the variant
that does, and the callers whose contract is "usable as it comes back" use it:
CreateInMemoryConfig, whose callers connect with the result, and the two
create-and-write branches. A read gets a config with no identity, so the
caller's own EnsureIdentity reports the work and triggers the write.
Reported by CodeRabbit and cubic-dev-ai on PR #7398, both on the same defect.
* [client] Stop the gate test from dialing the real management server
TestLogin_RestatingTheStoredConfigPassesTheGate asserts that the gate lets a
no-op login through, and the handler then went on to do the login for real:
isLoginRequired builds an auth client when isLoginRequiredFn is unset, so the
test dialed the profile's management URL — api.netbird.io:443. It took 1.05s
locally and would hang on a runner with no egress, for a fact about the gate
that needs no network at all.
Stubbed like the login_outcome tests do. The test now runs in 0.00s.
Reported by cubic-dev-ai on PR #7398.
* [client] Keep the admin panel path part of its identity
The endpoint comparison introduced for the management URL was applied to the
admin URL too, and that one is opened in a browser rather than dialed over
gRPC: a panel served under /netbird is not the panel served at the root. So a
config whose admin URL differed only by path reported no change, and the new
path was never persisted — a custom panel URL could not be updated at all.
SameServiceURLIncludingPath adds what a URL carries past its endpoint (path,
query, fragment, userinfo) while still treating equivalent spellings as equal:
a missing path and "/" are the same root, and so is a trailing slash. The
management URL keeps the endpoint-only comparison, since only the endpoint is
ever dialed.
Ports are also normalized numerically now, so ":0443" and ":443" are one port.
Reported by cubic-dev-ai on PR #7398 (two findings).
* [client] Treat a profile with no identity as already deregistered
Two findings on the same consequence of pure reads: a profile can legitimately
carry no keys, because logging out clears them in place.
- sendLogoutRequestWithConfig went straight to wgtypes.ParseKey and failed with
"incorrect key size: 0" on the second logout of the same profile. There is
nothing to deregister for a peer that was never registered, so it returns
cleanly. Before pure reads this case was hidden: the read minted a key and
the daemon dialed management with one it had never seen.
- The mobile logout read the config with the generating reader right after
checking the file exists. The two are not atomic, so a profile removed in
between was resolved from the defaults and recreated by the write that
follows. It uses the existing-file reader now.
Reported by cubic-dev-ai and CodeRabbit on PR #7398.
* [client] Fail `netbird up` when the daemon refuses the settings update
With the update-settings kill switch on, `netbird up --enable-rosenpass`
connected and said almost nothing: SetConfig refused the change, the CLI
downgraded that to a warning, and Login carries no rosenpass field to apply, so
the flag was silently dropped. The setting stayed disabled, which is the point
of the switch, but the caller was never told their request had been ignored.
The refusal now travels as codes.FailedPrecondition instead of
codes.Unavailable, and the CLI fails on it. Unavailable means "the daemon
cannot serve this call", which is why the CLI downgraded it and why
client/ui/services reads it as an unreachable daemon — both wrong for a daemon
that answered and refused. FailedPrecondition also matches what the MDM gate
already returns for a managed field, so both refusals are now one class of
error, and it is added to the login backoff's early-exit codes so a refused
login stops instead of retrying for 30s.
This does not put the container back in the deadlock: with the value-aware
gate, a client restating its own configuration is not refused at all, so
nothing reaches this path unless a real change was asked for.
* [client] Name the reader storedConfigAtPath actually calls
The purity note still said profilemanager.GetConfig, which the rename two
commits later turned into GetExistingConfig.
Reported by cubic-dev-ai on PR #7398.
* [client] Restore the gofmt alignment of the error constants
The comment added above errUpdateSettingsDisabled in the previous commit split
the const block's alignment group, so gofmt wants the two constants above it
re-aligned. CI runs gofmt, so this would have failed the lint job.
* [client] Let an unprivileged caller log out a profile with no identity
The empty-key check sat behind requirePrivilegeForDeregistration, so an
unprivileged logout of an identity-less profile was refused with
PermissionDenied instead of completing as the no-op it is. And it was refused
for most profiles, not a corner case: the gate arms whenever the SSH server is
enabled, and sshServerEnabled reads an absent ServerSSHAllowed as enabled, so
every legacy profile qualifies.
The check now runs first. What the gate protects against is handing this
machine's registered key to another management server; with no key there is
nothing to hand over and nothing to protect.
Reported by CodeRabbit and cubic-dev-ai on PR #7398, both on the same defect.
* [client] Stop `netbird login` from retrying a refusal for 30 seconds
`netbird up` and `netbird login` both run Login through the backoff cycle, and
each carried its own copy of the list of codes that end it. Only up.go learned
about codes.FailedPrecondition, so a refused `netbird login` kept retrying and
then reported "login backoff cycle failed" instead of what the daemon said.
terminalLoginError is now that list, once, next to WithBackOff — the duplicated
copies are what let the two commands disagree in the first place.
Reported by cubic-dev-ai on PR #7398.
* [client] Answer terminalLoginError's nil case on its own terms
A successful Login reaches terminalLoginError with a nil error, and nothing
covered that. It happens to work on grpc v1.80.0 — gstatus.FromError(nil)
answers (nil, true), and Status.Code tolerates a nil receiver by returning
codes.OK, which is not in the terminal set — but that is a chain of internal
details to be relying on for the common path, and none of it was asserted.
Now the nil error is handled where it is obvious, and the table covers it.
Reported by CodeRabbit on PR #7398, which called it a panic; measured on
v1.80.0 it is not one. The gap was the untested reliance, not a crash.
* [client] Treat an unset optional field as its default when diffing a config
Seven Config fields mean "the effective default" when they hold no value:
the five SSH toggles, the SSH JWT cache TTL, and the network monitor. Every
consumer already reads a nil as that default, but apply() diffed them by
presence — `config.X == nil || *input.X != *config.X` — so an input restating
the default counted as a change.
That made the update-settings gate refuse `netbird up` outright. The CLI
sends every flag whose value came from an environment variable
(SetFlagsFromEnvVars goes through pflag's FlagSet.Set, which marks the flag
Changed), and the config a plain login writes leaves all seven unset, so a
container configured with, say, NB_ENABLE_SSH_ROOT=false restated a default
the file held as null on every start and was answered with
FailedPrecondition.
apply() now resolves the seven up front, the way it already did for
ServerSSHAllowed and RemoteJobsAllowed, which also repairs such a profile on
its next write. With the values named, the comparisons below diff values
instead of presence, so their nil branches are gone.
The network monitor keeps its platform default — on for windows and darwin —
and naming it as false elsewhere is what createEngineConfig already read a
nil to be. getJWTCacheTTL reaches the same 0 through its own default, and
Android's GetEnableSSH* getters already answered nil with false.
* [client] Normalize the config before diffing it in WouldChange
apply() reports two different things through one bool: an input that changed
a value, and a field it had to fill in because the config carried none. The
update-settings gate reads that bool as "the caller asked for a change", so
any config still missing a default answered a request that asks for nothing
with a refusal.
Readers already hand out normalized configs — readConfig applies an empty
input for exactly this reason — which is why the gate got away with it. But a
handler that refuses a request must not depend on where its caller obtained
the config, and it must not start reading "this profile predates a field" as
"the caller asked for a change" the day someone adds one with a default.
WouldChange now runs the filling-in as a pass of its own and discards its
verdict, so the pass that answers the caller measures only what the input
did.
* [client] Stop the last config write that skipped normalization
Every path that creates or updates a profile config goes through apply(),
which resolves an optional field to its default — except RenameProfile,
which read the file with a bare json.Unmarshal, set the name, and wrote it
straight back. That copied whatever the file held, so a config written by a
client that stored these fields as null kept them null. It could not
introduce a null, only carry one forward, but renaming a profile is a poor
place to leave a half-resolved config behind. It now reads through
GetExistingConfig, which normalizes what it hands out.
The tests state the invariant the fix completes, over the *bool fields of
Config listed by reflection so a field added later is covered without
touching them: none may come out of apply() unset, and no write may store
one as null. An optional bool that can be nil, true or false forces every
reader to invent the meaning of nil, and makes a diff of the config compare
presence rather than value — which is exactly what refused `netbird up` for
a client restating its own defaults.
SyncMessageVersion stays a genuine three-state field and is not covered: it
is an *int whose absence means the client pins no version, and it travels to
management that way.
* [client] Refuse a serialized config that carries no peer identity
ConfigFromJSON still promised a "fully initialized" config after this PR
moved key generation out of apply() into EnsureIdentity, but identity stopped
being one of the defaults it applies. Its two callers both connect with what
they get back: the iOS SDK's Client.SetConfigFromJSON keeps it as the
preloaded config Run() uses on tvOS, and Auth.SetConfigFromJSON as the config
it authenticates with.
No caller feeds it a document without keys today — every stored document
comes from Auth.GetConfigJSON, whose config is provisioned by
DirectUpdateOrCreateConfig or CreateInMemoryConfig, and the tvOS app only
ever edits fields of a document it already has. This is a safety net for the
next caller, not a live bug.
Provisioning the identity here would be the wrong net. Neither caller can
hand a generated key back to the store the document came from — Client
exports no config at all — so the peer would connect under an identity
nothing persists and register anew on every launch, which is the failure the
EnsureIdentity split exists to prevent. A document with no identity means
nobody has logged in yet, and saying so is the only useful answer.
Both keys are required because both are dead ends when missing: an empty
WireGuard key fails the management login on its size, and an empty SSH key
fails ssh.GeneratePublicKey in ConnectClient before the engine starts.
* [client] Say that the null-on-disk fixture is synthesized, not written
The test comment described the null state in the present tense — "the config
a plain login writes leaves every one of them unset" — which was true before
this branch and is not any more: apply() now resolves those fields, so a
login writes them set. unsetOnDisk puts the null state back deliberately, to
stand in for a profile an older client wrote. Comments only.
* [client] Gather the optional-field defaults into one function
Resolving an unset optional field was spread over five places: the two
values newConfigSkeleton pre-sets, the block this branch added for the SSH
toggles, the network monitor's own if, the `else if` tails of
ServerSSHAllowed and RemoteJobsAllowed, and a trailing if for
DisableNotifications several hundred lines further down. Reading apply() left
no single answer to "what does this field default to, and who decides".
They now live in Config.resolveUnsetDefaults, which apply() calls before it
compares anything — the ordering being the point, since it is what lets
every comparison below diff values instead of presence. The comparisons for
ServerSSHAllowed, RemoteJobsAllowed and DisableNotifications lose their
`config.X == nil ||` clauses accordingly, as the other six already had.
newConfigSkeleton keeps its two, and that is the one asymmetry worth naming:
ServerSSHAllowed defaults to false for a new profile and to true for a
legacy one, and it only works because the skeleton runs first. The doc
comment says so, where before it was implied by the order of two distant
blocks.
Pure refactor. Verified as one: for the four fields whose branches moved,
plus two that did not and the JWT TTL, all 63 combinations of stored value
(nil/false/true) against input value (absent/false/true) produce byte-
identical resolved values and `updated` verdicts before and after.
* [client] Resolve the merge conflicts left in the tree
|
||
|
Maycon Santos
|
ebc259e30b |
[management,client] Gate remote jobs behind an admin opt-in with MDM support (#7153)
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. |
||
|
Viktor Liu
|
63c26be72f | [client] Add local Prometheus metrics endpoint (#6689) | ||
|
Viktor Liu
|
0f5d2d91fb | [client] Authorize daemon IPC callers by their local identity (#6967) |