Compare commits

..

33 Commits

Author SHA1 Message Date
Viktor Liu
0f593450d7 Prefer an available OAuth flow instead of hard-coding device code 2026-08-12 17:27:52 +02:00
Viktor Liu
db9fcf39ef [client] Gate IPv6 forwarding on overlay v6 and preserve host RA acceptance (#6221) 2026-08-12 16:07:00 +02:00
Lamera
52faa202b2 [client] fall back to per-IP ACL rules when ipset is unavailable (#6332) 2026-08-12 14:37:48 +02:00
Viktor Liu
f5ce0bc65a [client] Fix macOS DNS panic on malformed scutil output (#7180) 2026-08-12 13:25:12 +02:00
Maycon Santos
77e5ac776b [infrastructure] Let a suite outside this repo use the e2e harness (#7176)
e2e/harness documents itself as feature-agnostic, but three details
assumed the caller lives in this repo, so the terraform provider's
acceptance suite would otherwise carry a second harness for the same
product.

repoRoot took the first module root above the working directory as the
Docker build context, which from another module is the caller's own
root, with no combined/Dockerfile.multistage in it. It now requires that
ancestor to be this module, and otherwise asks the go tool for the
source: for a dependent, the extracted directory of the version it pins,
so the server matches the client library it was compiled against. That
lookup uses -mod=readonly, since automatic vendor mode otherwise reports
an empty Dir.

Geolocation was disabled unconditionally. Agent-network ingest does not
use it, but location-based posture checks need the database, and a rule
management cannot evaluate fails rather than passing.
StartClient pinned one network alias and set no hostname, so a second
agent could not start and a peer's name was arbitrary. Management
records that hostname, making it the peer's name in the API.
The client entrypoint is copied with an explicit mode: git tracks it
100755, but the module cache extracts 0444, so a dependent's build
produced a container exiting with "permission denied".

Adds CombinedOption, WithGeolocation, WithServerEnv, ClientOption and
WithClientName.
2026-08-12 11:19:25 +02:00
Maycon Santos
12546e231c [client] adjust gtk3 version release job (#7163)
- Align default names and reuse same environment variables

- With the uploads now targeting the same stable/yum paths as the GTK4
packages, two packages named netbird-ui with the same version and arch
would collide in the repo indexes. Give the GTK3 variant its own
package name and mark the two as conflicting alternatives.

---------

Co-authored-by: Zoltan Papp <zoltan.pmail@gmail.com>
2026-08-12 10:34:34 +02:00
Viktor Liu
052cf5a748 [client] Derive Windows SSH privilege checks from the token and group membership (#6966) 2026-08-11 18:16:37 +02:00
Viktor Liu
95a458801c [doc] Point translation contributions at Crowdin (#7161) 2026-08-11 16:02:09 +02:00
Viktor Liu
14f9f8ce22 Fix Crowdin export paths and align import and export with repo bundle state (#7162) 2026-08-11 16:01:33 +02:00
Maycon Santos
f805c149d9 [management] Record reverse proxy usage for activity accounting (#7116)
People who only ever reach private services through the reverse proxy were
invisible to activity accounting. Active users are counted from user.LastLogin
or from the LastSeen of a peer they own, and neither column was written on the
proxy paths — so a person signing in via SSO to a proxied service, or a peer
serving one over the mesh, never showed up in the 24 hour numbers.
Both writes now happen where the proxy already authenticates:
- GenerateSessionToken stamps LastLogin after the session token is signed,
  the same column and the same way the dashboard and device login paths do.
- ValidateTunnelPeer stamps the calling peer's LastSeen, the column its owner
  activates through.
The policy lives in a new reverseproxy/activity manager rather than in the gRPC
service, matching the module layout the other reverse proxy domains use. It
skips what can never count — service users, embedded proxy peers and WASM
clients — and throttles peer writes to once an hour, well inside the window
accounting asks about and far above the proxy's five minute tunnel cache.
The peer write is a single indexed UPDATE that touches only
peer_status_last_seen. Connected and SessionStartedAt are left alone so the
session-ownership fencing MarkPeerConnectedIfNewerSession relies on is never
disturbed, and the timestamp comes from the database clock rather than the
caller, for the same reason the other status writers take it from there. The
caller's cutoff travels into the statement's WHERE, so concurrent requests for
one peer collapse into a single write instead of each acting on its own stale
read, and a peer that was never seen — NULL last seen, since Status is an
embedded pointer — still records its first activity.
Nothing outside the reverse proxy changes behaviour: the only addition
elsewhere is the RefreshPeerLastSeen store method the manager calls.
2026-08-11 15:54:39 +02:00
Viktor Liu
99048e2bf2 [infrastructure] Add Crowdin configuration for UI translation sync (#7155) 2026-08-11 14:24:39 +02:00
Brad Ison
27b2d3f351 [management] Add a proxy-connect authorizer seam (#7136)
At proxy connect time, the declared cluster address is validated for
shape and checked for availability (`IsClusterAddressAvailable`), and
from then on the declaration is what routes the cluster's mappings to
the connection. Deployments that embed management through the
integrations seam may need a policy on that claim — deciding which
credential is allowed to declare which address.

This adds an optional `ProxyConnectAuthorizer` hook on
`ProxyServiceServer`, following the pattern of the existing `Set*` seams
(`SetServiceManager`, `SetAgentNetworkSynthesizer`,
`SetAgentNetworkLimitsService`, `SetProxyController`):

- A nil-able interface field plus `SetProxyConnectAuthorizer`, guarded
by the existing mutex.
- One call at the end of `validateProxyConnect`, so both
`GetMappingUpdate` and `SyncMappings` are covered by a single site.
- **Nothing installs it by default** — with the hook unset (always, in
this repo), behavior is byte-for-byte unchanged, which the tests pin.

Design details:

- The authorizer runs **last** — after input validation and the
availability check — and **outside** the account-scoped branch, so
management-wide tokens and token-less connects are also presented to it
rather than bypassing policy.
- The authorizer receives the presented `*types.ProxyAccessToken` (nil
when none), the proxy ID, and the declared address. Everything it needs
is already in the request/context; no proto or schema change.
- A plain error from the authorizer surfaces as `PermissionDenied`,
keeping an authorization rejection distinguishable from the
`AlreadyExists` used for address conflicts in proxy logs. A status error
passes through unchanged so implementations can pick their own code.
2026-08-10 19:50:00 +02:00
Brad Ison
ebfdf7d7b8 [management] Rework Agent Network endpoint identity and settings bootstrap (#7085)
Store the per-account gateway endpoint as {domain, proxy_address} with a
global unique index on the full hostname; dedicated = (domain ==
proxy_address). Bootstrap becomes an explicit POST carrying exactly one
of proxy_address (server allocates an adjective-noun label beneath it)
or endpoint (claimed verbatim, address-first); provider create loses its
bootstrap side effect. PUT is a full replace with every field required —
the immutable identity fields must be echoed unchanged and a mismatch is
rejected with 422. A guarded DELETE releases the endpoint: refused with
412 while providers exist or a proxy is actively serving the endpoint
hostname (matched case-insensitively); re-creating bootstraps fresh. A
self-addressed pin excludes its address from the account's cluster allow
list, and the live mapping update path now addresses the serving proxy
from the synthesized service. Existing rows are migrated on all three
store engines.
2026-08-10 19:06:55 +02:00
Viktor Liu
e8671a811d [client, relay] Migrate relay QUIC tracer to qlog and bump quic-go to 0.59.1 (#7124) 2026-08-10 17:41:31 +02:00
Zoltan Papp
1ca26d8faa [client] Bump the netbirdio/wails fork to drop the native WebView2 (#7128)
## Describe your changes

The native_webview2loader build tag embedded Microsoft's
WebView2Loader.dll via //go:embed. Those DLLs are gitignored in the fork
and go mod vendor resolves embed patterns regardless of build
constraints, so vendoring the module failed on the missing files when
packaging for openSUSE.

The fork now removes that branch along with the go-winloader dependency,
which drops out of the module graph here. The default GoWebView2Loader
path is unaffected.



## Issue ticket number and link

<!--
Required for anything that changes behavior. Link the issue (or the
validated
discussion it came from) that the NetBird team already agreed on. See

https://github.com/netbirdio/netbird/blob/main/CONTRIBUTING.md#ticket-first-pr-second
-->

## Stack

<!-- branch-stack -->

### Checklist
- [x] Is it a bug fix
- [ ] Is a typo/documentation fix
- [ ] Is a feature enhancement
- [ ] It is a refactor
- [ ] Created tests that fail without the change (if possible)
- [ ] I ran and tested this change locally — I did not rely on CI to
find out whether it works
- [ ] This PR has a single purpose (not a fix + refactor + feature in
one)
- [ ] This change is a trivial fix, **OR** it links an issue the NetBird
team agreed on beforehand. Changes to the public API, gRPC protocols,
functionality behavior, CLI / service flags, or new features always need
that agreement first. See
[CONTRIBUTING.md](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTING.md#ticket-first-pr-second).

> By submitting this pull request, you confirm that you have read and
agree to the terms of the [Contributor License
Agreement](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTOR_LICENSE_AGREEMENT.md).

## Documentation
Select exactly one:

- [ ] I added/updated documentation for this change
- [x] Documentation is **not needed** for this change (explain why)

### Docs PR URL (required if "docs added" is checked)
Paste the PR link from https://github.com/netbirdio/docs here:

https://github.com/netbirdio/docs/pull/__


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Chores**
  * Updated the application framework revision.
  * Removed an unused dependency.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-08-10 16:41:50 +02:00
Viktor Liu
f9abe2727f [client] Do not misroute WireGuard packets to the STUN handler (#7059) 2026-08-10 13:41:09 +02:00
Zoltan Papp
664a3d026c [client] Fix credentials for the gtk3 package uploads (#7125) 2026-08-10 12:58:50 +02:00
Zoltan Papp
d2c961f67c [client] Declare the xdg-utils dependency for the netbird-ui packages (#7126)
## Describe your changes

The UI shells out to xdg-open to launch the external browser for the SSO
verification page, which the embedded webview cannot open inline, and to
reveal the debug bundle in the file manager.

client/ui/build/linux/nfpm/nfpm.yaml lists xdg-utils for every package
format, but the released packages are built from the goreleaser configs,
where it was missing: the GTK and WebKitGTK dependencies carried over
and xdg-utils did not. Add it to all four nfpm dependency lists.

## Issue ticket number and link

<!--
Required for anything that changes behavior. Link the issue (or the
validated
discussion it came from) that the NetBird team already agreed on. See

https://github.com/netbirdio/netbird/blob/main/CONTRIBUTING.md#ticket-first-pr-second
-->

## Stack

<!-- branch-stack -->

### Checklist
- [x] Is it a bug fix
- [ ] Is a typo/documentation fix
- [ ] Is a feature enhancement
- [ ] It is a refactor
- [ ] Created tests that fail without the change (if possible)
- [ ] I ran and tested this change locally — I did not rely on CI to
find out whether it works
- [ ] This PR has a single purpose (not a fix + refactor + feature in
one)
- [ ] This change is a trivial fix, **OR** it links an issue the NetBird
team agreed on beforehand. Changes to the public API, gRPC protocols,
functionality behavior, CLI / service flags, or new features always need
that agreement first. See
[CONTRIBUTING.md](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTING.md#ticket-first-pr-second).

> By submitting this pull request, you confirm that you have read and
agree to the terms of the [Contributor License
Agreement](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTOR_LICENSE_AGREEMENT.md).

## Documentation
Select exactly one:

- [ ] I added/updated documentation for this change
- [x] Documentation is **not needed** for this change (explain why)

### Docs PR URL (required if "docs added" is checked)
Paste the PR link from https://github.com/netbirdio/docs here:

https://github.com/netbirdio/docs/pull/__


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Bug Fixes**
* Added required desktop integration support to Debian and RPM packages.
* Ensured GTK3 packages include the same runtime support for opening
links and files through the system.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-08-10 11:28:15 +02:00
Viktor Liu
5584f8ef0a [client] Add strict anonymization level and MAC anonymization to debug bundles (#7102) 2026-08-10 11:27:20 +02:00
Maycon Santos
f65f7b347e [management] Deny reverse proxy access to pending and blocked users (#7105)
A user in the Pending Approval state could complete SSO and reach any
SSO-protected reverse proxy service distributed to a group they belong
to, including the All Users group. The reverse proxy authorization path
checked the session token signature, that the user exists, that the
user's account matches the service's account, and group membership —
never the user's account status. The REST API (`permissions/manager.go`)
and peer registration both gate on that state, but the proxy gRPC
service does not go through the permissions manager, so neither gate
applied. A pending user is persisted as blocked and pending approval, so
blocked users reached those services the same way.

`ValidateSession` now denies on account status, reporting
`pending_approval` or `user_blocked` so the proxy access log and the
denied page carry the cause rather than a generic refusal.
`GenerateSessionToken` refuses to mint a token for such a user at all,
so the browser never receives a session cookie and the OIDC callback can
tell the user why instead of showing "Service configuration error".
`ValidateUserGroupAccess` and `ValidateTunnelPeer` close the same gap;
for the tunnel path this covers a user blocked after their peer was
registered, since peer group membership alone kept mesh-origin access
open.

A single helper produces both the denied reason for the RPC responses
and the sentinel error for the error-returning callers, so the four
entry points cannot drift apart. A user the store cannot resolve is
denied rather than passed through.

One thing deliberately left out: session cookies are validated locally
by the proxy against the service public key with no management
round-trip, so a cookie issued before a user is blocked stays valid
until it expires (24h by default). That is a revocation-propagation
problem rather than this authorization gap, and every option for it
(per-request validation with a cache, short-lived tokens with refresh,
push-based revocation) changes the proxy hot path or the
proxy/management protocol. Worth its own ticket.
2026-08-08 20:48:34 +09:00
Maycon Santos
179e8f6e13 [infrastructure] add grafana dashboard for licensed management (#7095) 2026-08-08 15:04:06 +09:00
Pascal Fischer
2ee21d2b5c [management] Affected peers for user updates (#7099) 2026-08-07 18:07:53 +02:00
Riccardo Manfrin
eb619fc7e3 [client] disambiguate the connection_type metric tag (#7043)
## Describe your changes

`recordConnectionMetrics` mapped only `conntype.Relay` to `relay` and
let a `default` branch
record everything else as `ice`. That silently included `ICETurn` — an
ICE connection through
a TURN server, which
[`conn.isRelayed`](https://github.com/netbirdio/netbird/blob/main/client/internal/peer/conn.go#L788-L795)
itself counts as relayed — and `None`, the transient state set when the
relay drops

([conn.go:632](https://github.com/netbirdio/netbird/blob/main/client/internal/peer/conn.go#L632))
or the peer state is reset
([conn.go:757](https://github.com/netbirdio/netbird/blob/main/client/internal/peer/conn.go#L757)).
Both were reported as direct peer-to-peer, so the `ice` share overstated
direct connections on
every platform.

The mapping now lists every priority explicitly and emits `ice_p2p`,
`ice_turn`, `relay` or
`unknown`. The new values deliberately do not reuse `ice` to avoid
ambuguity.


## Issue ticket number and link

No public issue. Found while reviewing the first production sample of
client metrics: 38% of iOS
connection events were tagged `ice` on a platform that forces relay by
default, which traced back
to the `default` branch at
[client/internal/peer/conn.go#L963-L968](https://github.com/netbirdio/netbird/blob/main/client/internal/peer/conn.go#L963-L968).

## Stack

<!-- branch-stack -->

### Checklist
- [x] Is it a bug fix
- [ ] Is a typo/documentation fix
- [ ] Is a feature enhancement
- [ ] It is a refactor
- [x] Created tests that fail without the change (if possible)

> By submitting this pull request, you confirm that you have read and
agree to the terms of the [Contributor License
Agreement](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTOR_LICENSE_AGREEMENT.md).

## Documentation
Select exactly one:

- [] I added/updated documentation for this change
- [x] Documentation is **not needed** for this change (explain why)

Internal metrics documentation only, in
`client/internal/metrics/infra/README.md`: the four
`connection_type` values with their derivation, and a note that pre-fix
`ice` samples are not
comparable with `ice_p2p`. No public API, CLI or configuration change,
so no netbirdio/docs PR.

### Docs PR URL (required if "docs added" is checked)
Paste the PR link from https://github.com/netbirdio/docs here:

N/A

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

## Summary by CodeRabbit

* **New Features**
* Connection metrics now distinguish direct peer-to-peer, TURN-assisted,
relay, and unknown connection types.
  * Metrics include clearer connection and peer identification details.

* **Documentation**
* Updated connection timing metric values, traffic semantics, priority
behavior, and historical data guidance.

* **Bug Fixes**
* Unset or unrecognized connection priorities are no longer incorrectly
classified as peer-to-peer.
* Unknown-transport metrics are skipped to prevent misleading connection
data.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-08-07 17:11:05 +02:00
Nicolas Frati
8632a0d215 [infrastructure] Detect community GHCR images during enterprise migration (#7101)
The migration wizard detected the community deployment only by the
Docker Hub image prefix (netbirdio/netbird-server), so deployments
installed from the ghcr.io mirror failed with "Could not find a service
running netbirdio/netbird-server*".

This broadens the server and dashboard detection to also accept
ghcr.io/netbirdio/... images. The regexes are anchored at the tag/digest
separator so Enterprise images (netbird-server-cloud, dashboard-cloud)
are still rejected, an already-migrated deployment must not be detected
as a community one. Error messages updated to mention both forms.
2026-08-07 17:04:58 +02:00
Riccardo Manfrin
f63fd21e0c [client] peer: re-arm the WireGuard watcher after a lazy wake (#7091)
## Describe your changes

The Conn struct is reused across lazy-connection deactivate/activate.
Close
cancels the WireGuard watcher (via wgWatcherCancel, and ctxCancel also
tears
down its context) but left conn.wgWatcher pointing at the stopped
instance.
enableWgWatcherIfNeeded skips while conn.wgWatcher is non-nil, so the
next Open
never started a fresh watcher: once a lazy connection had idled and
woken, the
peer ran with no watcher at all — no WireGuard handshake-timeout
detection and
none of the escalation that depends on it.

Clear conn.wgWatcher and conn.wgWatcherCancel in Close so the next Open
re-arms
a fresh watcher.

## Issue ticket number and link

<!--
Required for anything that changes behavior. Link the issue (or the
validated
discussion it came from) that the NetBird team already agreed on. See

https://github.com/netbirdio/netbird/blob/main/CONTRIBUTING.md#ticket-first-pr-second
-->

## Stack

<!-- branch-stack -->

### Checklist
- [X] Is it a bug fix
- [ ] Is a typo/documentation fix
- [ ] Is a feature enhancement
- [ ] It is a refactor
- [ ] Created tests that fail without the change (if possible)
- [ ] I ran and tested this change locally — I did not rely on CI to
find out whether it works
- [ ] This PR has a single purpose (not a fix + refactor + feature in
one)
- [ ] This change is a trivial fix, **OR** it links an issue the NetBird
team agreed on beforehand. Changes to the public API, gRPC protocols,
functionality behavior, CLI / service flags, or new features always need
that agreement first. See
[CONTRIBUTING.md](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTING.md#ticket-first-pr-second).

> By submitting this pull request, you confirm that you have read and
agree to the terms of the [Contributor License
Agreement](https://github.com/netbirdio/netbird/blob/main/CONTRIBUTOR_LICENSE_AGREEMENT.md).

## Documentation
Select exactly one:

- [ ] I added/updated documentation for this change
- [X] Documentation is **not needed** for this change (explain why)

### Docs PR URL (required if "docs added" is checked)
Paste the PR link from https://github.com/netbirdio/docs here:

https://github.com/netbirdio/docs/pull/__


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Bug Fixes**
* Improved connection cleanup by fully releasing WireGuard watcher
resources when a connection closes.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-08-07 17:01:08 +02:00
Pascal Fischer
524b8b9718 [management] prewarm a posture check cache on network map generation (#7093) 2026-08-07 15:03:40 +02:00
Zoltan Papp
2ce6323602 [client] Update the wails fork reference to the integration branch head (#7087) 2026-08-07 09:57:52 +02:00
Viktor Liu
9a05a1c698 [client] Reword the firewalld package comment (#7081) 2026-08-06 20:36:13 +02:00
Viktor Liu
5dd914782a [client] Stop the macOS UI on pkg upgrade (#7079) 2026-08-06 16:50:17 +02:00
Ben
8c19b7a30a [infrastructure] Generate a session cookie encryption key on fresh self-hosted installs (#7056)
The community and enterprise bootstrap scripts now generate a dedicated
`server.auth.sessionCookieEncryptionKey` from 32 random bytes for fresh
installations.
The key is Base64-encoded, persisted independently from the datastore
encryption key, and reused from the generated configuration after
restarts.

The community script now also creates `config.yaml` with mode `0600`
before writing it, matching the existing enterprise behavior.

The server already supports the session cookie encryption key, but the
bootstrap scripts left it unset.
This adds defense-in-depth for newly generated deployments while
preserving the existing server-side nonce validation.
Because this only changes newly generated configuration, existing
installations and sessions are unchanged.

A focused regression test checks key presence, decoded length,
separation from the datastore key, YAML placement, and file mode for
both scripts.
It is included in the infrastructure workflow.
2026-08-06 13:55:43 +02:00
Zoltan Papp
1e2a7aa571 [client] Add a UI setting to stay connected after quitting (#7078)
Quitting the GUI from the tray always sent a Down RPC, dropping the VPN
connection with it. Some users want the connection to survive the UI.
2026-08-06 12:56:27 +02:00
Zoltan Papp
98f8a24108 [client] Ship a legacy GTK3 UI package for distros without WebKitGTK 6.0 (#7040)
Ship a legacy GTK3 UI package for distros without WebKitGTK 6.0
2026-08-06 12:13:20 +02:00
Viktor Liu
b2ae98f0fd [misc] Expand agent guidelines with security, type-safety and lifecycle conventions (#7076) 2026-08-06 11:38:24 +02:00
341 changed files with 21392 additions and 27088 deletions

View File

@@ -475,6 +475,132 @@ jobs:
path: dist/
retention-days: 3
release_ui_gtk3:
# Legacy GTK3/WebKit2GTK 4.1 UI build for distros without WebKitGTK 6.0
# (Ubuntu 22.04, Debian 12, RHEL 9, Fedora <=39). Runs on ubuntu-22.04 so
# the binary links against the oldest supported glibc.
runs-on: ubuntu-22.04
outputs:
release_ui_gtk3_artifact_url: ${{ steps.upload_release_ui_gtk3.outputs.artifact-url }}
steps:
- name: Checkout
uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
with:
fetch-depth: 0 # It is required for GoReleaser to work properly
persist-credentials: false
- name: Parse semver string
id: semver_parser
uses: netbirdio/shared-actions/actions/parse-semver@be5df6047383da2236e02243cceb857d8567c27e # v0.0.2
- name: Set snapshot flag
if: ${{ !startsWith(github.ref, 'refs/tags/v') }}
run: |
echo "flags=--snapshot" >> $GITHUB_ENV
- name: Set build vars
if: ${{ startsWith(github.ref, 'refs/tags/v') }}
run: |
if [[ "x-${{ steps.semver_parser.outputs.prerelease }}" == "x-" && "x-${{ github.repository }}" == "x-netbirdio/netbird" ]]; then
echo "x-${{ github.repository }}"
echo "x-${{ steps.semver_parser.outputs.prerelease }}"
echo "SKIP_PUBLISH=false" >> $GITHUB_ENV
else
echo "x-${{ github.repository }}"
echo "x-${{ steps.semver_parser.outputs.prerelease }}"
fi
- name: Set up Go
uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
with:
go-version-file: "go.mod"
cache: false
- name: Cache Go modules
# Restore-only from the release_ui cache written by trusted runs; the
# module cache is identical (same go.sum) and stale build-cache
# entries just miss.
uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
with:
path: |
~/go/pkg/mod
~/.cache/go-build
key: ${{ runner.os }}-ui-go-releaser-${{ hashFiles('**/go.sum') }}
restore-keys: |
${{ runner.os }}-ui-go-releaser-
- name: Install modules
run: go mod tidy
- name: check git status
run: git --no-pager diff --exit-code
- name: Set up Node.js
uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020 # v4
with:
node-version: '22'
- name: Set up pnpm
uses: pnpm/action-setup@a3252b78c470c02df07e9d59298aecedc3ccdd6d # v3.0.0
with:
version: 11
- name: Install dependencies
run: sudo apt update && sudo apt install -y -q libgtk-3-dev libwebkit2gtk-4.1-dev
- name: Decode GPG signing key
if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
env:
GPG_RPM_PRIVATE_KEY: ${{ secrets.GPG_RPM_PRIVATE_KEY }}
run: |
echo "$GPG_RPM_PRIVATE_KEY" | base64 -d > /tmp/gpg-rpm-signing-key.asc
echo "GPG_RPM_KEY_FILE=/tmp/gpg-rpm-signing-key.asc" >> $GITHUB_ENV
- name: Install wails3 CLI
# Version derived from go.mod so the binding generator always matches
# the wails runtime the binary links against.
# -tags gtk3: the CLI links the wails runtime's cgo packages, and the
# default tags request gtk4/webkitgtk-6.0 pkg-config entries that do
# not exist on ubuntu-22.04.
run: |
WAILS_VERSION=$(go list -m -f '{{.Version}}' github.com/wailsapp/wails/v3)
go install -tags gtk3 github.com/wailsapp/wails/v3/cmd/wails3@$WAILS_VERSION
- name: Run GoReleaser
uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
with:
version: ${{ env.GORELEASER_VER }}
args: release --config .goreleaser_ui_gtk3.yaml --clean ${{ env.flags }}
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
UPLOAD_DEBIAN_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
UPLOAD_YUM_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
GPG_RPM_KEY_FILE: ${{ env.GPG_RPM_KEY_FILE }}
NFPM_NETBIRD_UI_RPM_GTK3_PASSPHRASE: ${{ secrets.GPG_RPM_PASSPHRASE }}
SKIP_PUBLISH: ${{ env.SKIP_PUBLISH }}
- name: Verify RPM signatures
run: |
docker run --rm -v $(pwd)/dist:/dist fedora:41 bash -c '
dnf install -y -q rpm-sign curl >/dev/null 2>&1
curl -sSL https://pkgs.netbird.io/yum/repodata/repomd.xml.key -o /tmp/rpm-pub.key
rpm --import /tmp/rpm-pub.key
echo "=== Verifying RPM signatures ==="
for rpm_file in /dist/*.rpm; do
[ -f "$rpm_file" ] || continue
echo "--- $(basename $rpm_file) ---"
rpm -K "$rpm_file"
done
'
- name: Clean up GPG key
if: always()
run: rm -f /tmp/gpg-rpm-signing-key.asc
- name: upload non tags for debug purposes
id: upload_release_ui_gtk3
uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a #v7.0.1
with:
name: release-ui-gtk3
path: dist/
retention-days: 3
release_ui_darwin:
runs-on: macos-latest
outputs:
@@ -688,7 +814,7 @@ jobs:
comment_release_artifacts:
name: Comment release artifacts
runs-on: ubuntu-latest
needs: [release, release_ui, release_ui_darwin]
needs: [release, release_ui, release_ui_gtk3, release_ui_darwin]
if: ${{ always() && github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository }}
permissions:
contents: read
@@ -700,12 +826,14 @@ jobs:
env:
RELEASE_RESULT: ${{ needs.release.result }}
RELEASE_UI_RESULT: ${{ needs.release_ui.result }}
RELEASE_UI_GTK3_RESULT: ${{ needs.release_ui_gtk3.result }}
RELEASE_UI_DARWIN_RESULT: ${{ needs.release_ui_darwin.result }}
RELEASE_ARTIFACT_URL: ${{ needs.release.outputs.release_artifact_url }}
LINUX_PACKAGES_ARTIFACT_URL: ${{ needs.release.outputs.linux_packages_artifact_url }}
WINDOWS_PACKAGES_ARTIFACT_URL: ${{ needs.release.outputs.windows_packages_artifact_url }}
MACOS_PACKAGES_ARTIFACT_URL: ${{ needs.release.outputs.macos_packages_artifact_url }}
RELEASE_UI_ARTIFACT_URL: ${{ needs.release_ui.outputs.release_ui_artifact_url }}
RELEASE_UI_GTK3_ARTIFACT_URL: ${{ needs.release_ui_gtk3.outputs.release_ui_gtk3_artifact_url }}
RELEASE_UI_DARWIN_ARTIFACT_URL: ${{ needs.release_ui_darwin.outputs.release_ui_darwin_artifact_url }}
GHCR_IMAGES_MARKDOWN: ${{ needs.release.outputs.ghcr_images }}
with:
@@ -728,6 +856,7 @@ jobs:
['Windows packages', process.env.WINDOWS_PACKAGES_ARTIFACT_URL, process.env.RELEASE_RESULT],
['macOS packages', process.env.MACOS_PACKAGES_ARTIFACT_URL, process.env.RELEASE_RESULT],
['UI artifacts', process.env.RELEASE_UI_ARTIFACT_URL, process.env.RELEASE_UI_RESULT],
['UI GTK3 artifacts', process.env.RELEASE_UI_GTK3_ARTIFACT_URL, process.env.RELEASE_UI_GTK3_RESULT],
['UI macOS artifacts', process.env.RELEASE_UI_DARWIN_ARTIFACT_URL, process.env.RELEASE_UI_DARWIN_RESULT],
];
@@ -784,7 +913,7 @@ jobs:
trigger_signer:
runs-on: ubuntu-latest
needs: [release, release_ui, release_ui_darwin, test_windows_installer]
needs: [release, release_ui, release_ui_gtk3, release_ui_darwin, test_windows_installer]
if: startsWith(github.ref, 'refs/tags/')
steps:
- name: Trigger binaries sign pipelines

View File

@@ -257,6 +257,15 @@ jobs:
with:
persist-credentials: false
- name: Verify fresh-install session cookie key hardening
run: |
grep -Fxq ' SESSION_COOKIE_ENCRYPTION_KEY=$(openssl rand -base64 32)' infrastructure_files/getting-started.sh
grep -Fxq ' sessionCookieEncryptionKey: "$SESSION_COOKIE_ENCRYPTION_KEY"' infrastructure_files/getting-started.sh
grep -Fxq ' install -m 600 /dev/null config.yaml' infrastructure_files/getting-started.sh
grep -Fxq ' openssl rand -base64 32' infrastructure_files/getting-started-enterprise.sh
grep -Fxq ' NETBIRD_SESSION_COOKIE_ENCRYPTION_KEY=$(rand_b64_key)' infrastructure_files/getting-started-enterprise.sh
grep -Fxq ' sessionCookieEncryptionKey: "${NETBIRD_SESSION_COOKIE_ENCRYPTION_KEY}"' infrastructure_files/getting-started-enterprise.sh
- name: Verify Dex retirement notice
run: |
if infrastructure_files/getting-started-with-dex.sh >stdout.txt 2>stderr.txt; then

View File

@@ -96,6 +96,7 @@ nfpms:
- netbird (>= 0.75.0)
- libgtk-4-1 (>= 4.14)
- libwebkitgtk-6.0-4
- xdg-utils
- maintainer: Netbird <dev@netbird.io>
description: Netbird client UI.
@@ -119,6 +120,7 @@ nfpms:
- netbird >= 0.75.0
- (gtk4 >= 4.14 or libgtk-4-1 >= 4.14)
- (webkitgtk6.0 or libwebkitgtk-6_0-4)
- xdg-utils
rpm:
signature:

136
.goreleaser_ui_gtk3.yaml Normal file
View File

@@ -0,0 +1,136 @@
version: 2
env:
- SKIP_PUBLISH={{ if index .Env "SKIP_PUBLISH" }}{{ .Env.SKIP_PUBLISH }}{{ else }}true{{ end }}
project_name: netbird-ui
before:
hooks:
# Bindings are gitignored; regenerate before the frontend build so
# the @wailsio/runtime Vite plugin can resolve them (vite refuses to
# build without them).
# -f '-tags gtk3': the generator type-checks client/ui, whose cgo imports
# would otherwise resolve gtk4/webkitgtk-6.0 pkg-config entries that do
# not exist on ubuntu-22.04.
- sh -c 'cd client/ui && wails3 generate bindings -clean=true -ts -f "-tags gtk3"'
- sh -c 'cd client/ui/frontend && pnpm install --frozen-lockfile && pnpm build'
builds:
# Legacy GTK3 / WebKit2GTK 4.1 build for distros without WebKitGTK 6.0
# (Ubuntu 22.04, Debian 12, RHEL 9, Fedora <=39). The gtk3 tag flips the
# Wails Linux backend to the GTK3 stack and swaps our GTK4-only XEmbed
# tray host for the pure-Go stub (client/ui/xembed_host_gtk3_linux.go).
# Must be built on the oldest supported glibc (ubuntu-22.04 runner).
- id: netbird-ui-gtk3
dir: client/ui
binary: netbird-ui
env:
- CGO_ENABLED=1
goos:
- linux
goarch:
- amd64
ldflags:
- -s -w -X github.com/netbirdio/netbird/version.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}} -X main.builtBy=goreleaser
mod_timestamp: "{{ .CommitTimestamp }}"
tags:
- production
- gtk3
archives:
- id: linux-gtk3-arch
name_template: "{{ .ProjectName }}-linux-gtk3_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
builds:
- netbird-ui-gtk3
nfpms:
# Mutually-exclusive alternative to the GTK4 netbird-ui package -- both
# ship the same /usr/bin/netbird-ui from the shared stable/yum repos, so
# this one carries its own name and conflicts with the GTK4 package.
- maintainer: Netbird <dev@netbird.io>
description: Netbird client UI.
homepage: https://netbird.io/
license: BSD-3-Clause
vendor: NetBird
id: netbird_ui_deb_gtk3
package_name: netbird-ui-gtk3
file_name_template: "{{ .PackageName }}_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
builds:
- netbird-ui-gtk3
formats:
- deb
scripts:
postinstall: "release_files/ui-post-install.sh"
contents:
- src: client/ui/build/linux/netbird.desktop
dst: /usr/share/applications/org.wails.netbird.desktop
- src: client/ui/build/appicon.png
dst: /usr/share/pixmaps/netbird.png
conflicts:
- netbird-ui
replaces:
- netbird-ui
dependencies:
- netbird (>= 0.75.0)
- libgtk-3-0
- libwebkit2gtk-4.1-0
- xdg-utils
- maintainer: Netbird <dev@netbird.io>
description: Netbird client UI.
homepage: https://netbird.io/
license: BSD-3-Clause
vendor: NetBird
id: netbird_ui_rpm_gtk3
package_name: netbird-ui-gtk3
file_name_template: "{{ .PackageName }}_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
builds:
- netbird-ui-gtk3
formats:
- rpm
scripts:
postinstall: "release_files/ui-post-install.sh"
contents:
- src: client/ui/build/linux/netbird.desktop
dst: /usr/share/applications/org.wails.netbird.desktop
- src: client/ui/build/appicon.png
dst: /usr/share/pixmaps/netbird.png
# No `replaces` here: nfpm maps it to rpm Obsoletes, which would make
# dnf swap installed GTK4 netbird-ui packages for this one on upgrade.
conflicts:
- netbird-ui
dependencies:
- netbird >= 0.75.0
- (gtk3 or libgtk-3-0)
- (webkit2gtk4.1 or libwebkit2gtk-4_1-0)
- xdg-utils
rpm:
signature:
key_file: '{{ if index .Env "GPG_RPM_KEY_FILE" }}{{ .Env.GPG_RPM_KEY_FILE }}{{ end }}'
# The GTK4 UI job shares project_name, so the default checksum file name would
# collide with it on the shared GitHub release.
checksum:
name_template: "{{ .ProjectName }}_gtk3_checksums.txt"
changelog:
disable: true
uploads:
- name: debian
skip: "{{ .Env.SKIP_PUBLISH }}"
ids:
- netbird_ui_deb_gtk3
mode: archive
target: https://pkgs.wiretrustee.com/debian/pool/{{ .ArtifactName }};deb.distribution=stable;deb.component=main;deb.architecture={{ if .Arm }}armhf{{ else }}{{ .Arch }}{{ end }};deb.package=
username: dev@wiretrustee.com
method: PUT
- name: yum
skip: "{{ .Env.SKIP_PUBLISH }}"
ids:
- netbird_ui_rpm_gtk3
mode: archive
target: https://pkgs.wiretrustee.com/yum/{{ .Arch }}{{ if .Arm }}{{ .Arm }}{{ end }}
username: dev@wiretrustee.com
method: PUT

317
AGENTS.md
View File

@@ -14,20 +14,22 @@ in this file, not duplicated there.
## Contents
- [NetBird Agent Guidelines](#netbird-agent-guidelines)
- [Contents](#contents)
- [STOP and ask the user before](#stop-and-ask-the-user-before)
- [Quick reference](#quick-reference)
- [Structure](#structure)
- [Where to look](#where-to-look)
- [Repo-wide principles](#repo-wide-principles)
- [Error handling](#error-handling)
- [Comments](#comments)
- [Testing](#testing)
- [Pitfalls](#pitfalls)
- [Commits, PRs, releases](#commits-prs-releases)
- [After you push: CI and review bots](#after-you-push-ci-and-review-bots)
- [Discussion and support](#discussion-and-support)
- [STOP and ask the user before](#stop-and-ask-the-user-before)
- [Quick reference](#quick-reference)
- [Structure](#structure)
- [Where to look](#where-to-look)
- [Security](#security)
- [Agent conventions](#agent-conventions)
- [Repo-wide principles](#repo-wide-principles)
- [Type safety](#type-safety)
- [Concurrency and lifecycle](#concurrency-and-lifecycle)
- [Error handling](#error-handling)
- [Comments](#comments)
- [Testing](#testing)
- [Pitfalls](#pitfalls)
- [Commits, PRs, releases](#commits-prs-releases)
- [After you push: CI and review bots](#after-you-push-ci-and-review-bots)
- [Discussion and support](#discussion-and-support)
## STOP and ask the user before
@@ -157,11 +159,125 @@ netbird/
| LLM routing / Agent Network | `proxy/internal/llm/`, `agent-network/` |
| End-to-end tests | `e2e/` |
## Security
### Never fail open
When a security check — access control, an IP restriction, an auth decision —
hits an error such as an unparseable value, an unavailable lookup, or a state it
does not recognize, it must **deny**. Never skip the check or allow the request
through because the check itself failed, and make the `default` and unknown cases
of a security-related `switch` deny rather than fall through.
### Daemon RPC input is untrusted
The agent runs as root (LocalSystem on Windows), so a daemon RPC crosses a
privilege boundary: treat every field as untrusted input rather than as something
the UI or CLI validated on the way in.
When you add or change an RPC, ask what the handler does with caller input while
running as root. If the answer touches a filesystem path, a URL or host, or a
privileged state change, it needs a gate **in the handler** — a check in the client
that normally calls it is not a check at all.
- **A caller-supplied path the daemon opens.** Never `os.Open` it as root.
Constrain it, then open it *as the caller* with `ipcauth.OpenOwnedFile`, which
opens `O_NOFOLLOW`, requires a regular file, and refuses a file the caller does
not own — so a symlink or hardlink aimed at a root-only file is rejected.
- **A caller-supplied URL or host the daemon fetches.** Restrict the scheme and
allow only known hosts for unprivileged callers. Prefer a lexical host
allowlist plus TLS verification over "resolve the host, then reject private
IPs": the resolve-then-trust pattern has a DNS-rebinding race (public IP at
check time, attacker IP at connect time), while a name allowlist has no IP
check to race. Never accept `http://` where `https://` is expected.
- **A privileged state change** (SSH root login, management URL, deregistration)
gates on the caller identity from `ipcauth.CallerIdentity(ctx)`.
Caller identity comes from the kernel — `SO_PEERCRED`, `LOCAL_PEERCRED`, or the
named-pipe client token — and never from an RPC field. When
`ipcauth.CallerIdentity` reports that it could not determine an identity, **deny**;
do not fall back to treating the caller as the transport peer.
## Agent conventions
### Three networking modes
Where packets actually flow depends on the mode the agent is running in. The
three are not interchangeable, so establish which one a change applies to — and
what it should do in the other two — before you write it.
- **kernel mode** (Linux only): in-kernel WireGuard®. The kernel handles both
peer-to-peer and routed traffic, and ACLs are iptables or nftables rules. The
client programs kernel facilities but never sees the traffic itself.
- **userspace mode** (wireguard-go with a TUN): wireguard-go runs in-process. The
kernel handles peer-to-peer traffic once it leaves the TUN, while routed traffic
— exit nodes and network routes — goes through the userspace forwarder, which
terminates the connection and re-establishes it over OS sockets. Used on
platforms without kernel WireGuard® or when the user opts out.
- **netstack mode**: wireguard-go in-process with no TUN and no kernel
networking. The forwarder does all routing by stitching userspace sockets, and
listeners such as the embedded SSH and DNS servers bind on a gVisor netstack.
Used where the process cannot create a TUN device, such as the embedded client
(`client/embed/`) and the WASM build.
### The overlay interface is not "WireGuard"
Do not put "WireGuard" in identifiers or comments unless the code is genuinely
coupled to WireGuard® specifically — a wireguard-go call, a handshake field, a
kernel WireGuard® netlink attribute. For the interface, the host, peers, or
traffic in general, say "the NetBird interface", "the interface", or "the overlay".
Most firewall, routing, and DNS code is transport-agnostic, so a WireGuard®
reference there is simply inaccurate and rots as the transports change.
### IPv6 is a soft feature
The IPv6 overlay is opt-in dual-stack, and capability can change at runtime. Treat
it as soft rather than a requirement:
- Gate local v6 paths on the interface accessor (`wgIface.Address().HasIPv6()`),
not on raw state fields, and skip the v6 path when the host has no v6 rather
than returning an error.
- Treat an empty or unparseable peer v6 address as "no v6 for that peer" and skip
it, keeping the v4 path working.
- Never let a missing v6 break v4. Fail-closed is for security checks; a
capability mismatch skips the v6 work and carries on.
### Environment variables
Name the variable in a constant and parse booleans with `strconv.ParseBool` rather
than comparing strings inline, so an unexpected value is logged instead of
silently meaning false:
```go
const EnvDisableFeature = "NB_DISABLE_FEATURE"
func isDisabledByEnv() bool {
val := os.Getenv(EnvDisableFeature)
if val == "" {
return false
}
disabled, err := strconv.ParseBool(val)
if err != nil {
log.Warnf("failed to parse %s: %v", EnvDisableFeature, err)
return false
}
return disabled
}
```
### Validating against protocol specs
When a change depends on what a protocol actually mandates, read the specification
text from the [IETF datatracker](https://datatracker.ietf.org/) rather than a
summary, and check that you have the current RFC — the widely cited one for a
protocol is often superseded. Cite the section, not just the document, so a
reviewer can jump straight to the rule.
## Repo-wide principles
1. **Run `go fmt` on every modified Go file.** Formatting is not optional.
2. **Zero unaddressed diagnostics.** Fix IDE and linter warnings on code you
touch, and delete imports, helpers, and parameters your refactor orphaned.
2. **Zero unaddressed linter warnings.** Fix what `golangci-lint` reports on code
you touch, and delete imports, helpers, and parameters your refactor orphaned.
Exception: unused parameters in shared code may be consumed by builds outside
this repository — do not remove them, ask instead.
3. **Function comments are mandatory for exported functions**, written as full
@@ -175,9 +291,12 @@ netbird/
7. **Avoid LLM-slop tells:** em dashes, hedging narration, restating the diff in
prose, trailing summaries. Defaults, not absolute bans. Applies to code,
comments, commit messages, and PR descriptions alike.
8. **Concurrency: do a two-pass race analysis after every change** that adds
shared state. Guard maps and slices with a mutex, keep critical sections
short, and run `go test -race` on the touched packages.
8. **Concurrency: do a two-pass race analysis after every change** that touches
shared state, including reads of existing maps and slices. Guard them with a
mutex (or an atomic or channel where that fits better), keep critical
sections short, and run `go test -race` on the touched packages. See
[Concurrency and lifecycle](#concurrency-and-lifecycle) for the failure modes
to check for.
9. **Cross-platform builds must keep working.** The agent targets Linux, macOS,
Windows, FreeBSD, Android, and iOS. When you add a platform-specific file,
add the counterpart or a build-tagged fallback for the others.
@@ -185,6 +304,93 @@ netbird/
11. **Never log secrets** — private keys, setup keys, tokens, PAT values — and
keep peer IPs and hostnames out of logs above debug level.
## Type safety
**No bare primitives for domain concepts.** A `string` parameter for an account
ID next to a `string` parameter for a peer ID is two bugs waiting to happen,
because the compiler cannot catch the swap. Declare the type once and use it
throughout, converting only at the boundaries where data enters or leaves —
protobuf, gRPC, HTTP, an external library.
```go
type ServiceID string
type AccountID string
// Internal: typed all the way through
func (r *Router) RemoveRoute(host SNIHost, svcID ServiceID) { ... }
// Proto boundary: convert once, on the way in and on the way out
svcID := ServiceID(mapping.GetId())
req.ServiceId = string(svcID)
```
- **IP addresses are `netip.Addr`**, not `string` and not `net.IP`. Parse at the
boundary and pass the typed value inward.
- **Always `Unmap()`** after parsing an address, after converting from `net.IP`,
and after extracting one from `RemoteAddr()`. This normalizes a v4-mapped v6
address (`::ffff:10.1.2.3`) to plain v4 so IPv4 rules match it. A stored or
compared mapped address silently fails to match those rules.
- **Ports are `uint16`** internally; use `int` only where a library forces it and
convert immediately.
- **Enums are a typed string with constants**, so the valid set is discoverable
and a typo fails to compile.
- **Map keys follow the same rule**, and must be a real type (`type ServiceID
string`) rather than an alias (`type serviceID = string`) — an alias silently
accepts bare strings.
## Concurrency and lifecycle
Beyond the mutex hygiene in the principles above, check for these failure
modes.
- **Never read a struct field inside a goroutine** when another goroutine may nil
or reassign it. Pass the value as a parameter, or capture it into a local before
launching. This matters most when `Stop()` nils a field without waiting for the
goroutine to finish.
```go
go func(ifaceName string) { // good: passed in, cannot be nilled underneath
m.Start(ctx, ifaceName)
}(iface.Name())
```
- **Never wait on a channel while holding a lock the sender needs.** Copy what you
need out from under the lock, release it, then wait.
```go
func (m *Manager) Stop() {
m.mu.Lock()
cancel, done := m.cancel, m.done
m.mu.Unlock()
if cancel != nil {
cancel()
<-done
}
}
```
- **`Stop`/`Close` must be idempotent** — guard on an already-stopped flag or a
nil cancel — and must release the state they guarded. Clear maps and caches;
a cancelled goroutine holding a live map still pins that memory. Note that a
nil map only panics on writes; reads and iteration behave like an empty map,
so where post-close use must be rejected, check the stopped flag explicitly.
- **Publish coupled state only after every fallible step succeeds.** When several
fields form an invariant, build them into locals and assign them to the receiver
at the end. Assigning as you go leaves the object half-initialized when a later
step fails, so a readiness predicate reports ready while a coupled field is nil.
If an earlier step already had an external side effect — a created chain, an
opened handle, an inserted rule — roll it back before returning the error.
- **Clean up what you own on constructor error paths.** Once a constructor has
started something, every later error path must undo it: cancel a goroutine and
wait for it to exit, stop a ticker, close a watcher. The object is never
returned, so its `Close` will never run.
- **A failed `Start` must undo everything it started.** When a component brings up
several subsystems in sequence — connection manager, watchers, routing, DNS,
flow, persisted state — a failure partway through has to tear down the ones
already running, not just close the handle the error came from. Put the
already-started guard *before* that teardown path, so a rejected second `Start`
cannot dismantle the one that is running.
## Error handling
Use single-assignment form when the error is only needed inside the `if`:
@@ -248,6 +454,45 @@ Log the errors you choose not to act on:
- Close errors may be ignored for read-only operations; log them at debug for
writes.
**Do not log and return the same error.** It gets reported twice, from two places,
and the second reader cannot tell whether it happened once or twice. Return it and
let the caller decide. The exception is an API handler that has already written a
response. Internal helpers return errors rather than logging and swallowing them.
**Never return a typed nil as an error.** A nil `*MyError` stored in an `error`
interface is not nil, so `err != nil` is true and callers take the failure path on
success. Return the error only where it is actually set:
```go
if _, err := conn.Write(buf); err != nil { // good
return err
}
return nil
```
**Accumulate with `multierror` when an operation should continue past individual
failures** — teardown, cleanup, or setup where partial success is acceptable.
`client/errors.FormatErrorOrNil` returns nil for an empty accumulator, so callers
still see a plain nil on full success:
```go
func (m *Manager) Cleanup() error {
var merr *multierror.Error
for _, r := range m.resources {
if err := r.Close(); err != nil {
merr = multierror.Append(merr, fmt.Errorf("close %s: %w", r.Name, err))
}
}
return nberrors.FormatErrorOrNil(merr)
}
```
| Scenario | Approach | Why |
| --------------------- | --------------------- | ----------------------------------------- |
| Cleanup / teardown | Accumulate | Clean up as much as possible |
| Setup with rollback | Abort on first error | Partial state is invalid; undo what stuck |
| Setup with partial OK | Accumulate | Degraded operation is still useful |
## Comments
Comment the **why**, never the **what**. Default to no comment, and add one only
@@ -269,10 +514,14 @@ checksum = updateChecksum(checksum, oldPort, newPort)
### Length budget
- **90 characters per line.** Wrap the comment, do not run past it.
- **250 characters per comment**, roughly three wrapped lines. Doc comments on
exported identifiers may exceed it when the API genuinely needs the
explanation; inline comments inside a function body may not.
Neither of these is linter-enforced, so they are conventions the surrounding code
mostly follows rather than hard limits:
- **Around 90 characters per line.** Wrap the comment rather than running well past
it.
- **Roughly 250 characters per comment**, about three wrapped lines. Doc comments
on exported identifiers may exceed it when the API genuinely needs the
explanation; inline comments inside a function body rarely should.
The budget is a smell detector, not a rule to game. Do not compress a needed
explanation into cryptic shorthand to fit — if a block of code needs more than
@@ -329,6 +578,19 @@ up, and the 250-character budget does not apply to them.
otherwise.
- **Message guidance:** optional for `NoError`/`Error`; always give context for
comparison, boolean, and collection assertions.
- **Reproduce a bug before fixing it.** Write the test, watch it fail *for the
reason you expect* — a test that fails for an unrelated reason proves nothing —
then apply the fix and confirm it passes. Add the thin surrounding cases while
you are there.
- **Use `t.Setenv`** rather than `os.Setenv` so the previous value is restored on
cleanup. To test the unset case, call `t.Setenv` first to register the restore,
then `os.Unsetenv`.
- **Prefer `t.Cleanup` over `defer`** in any test with parallel subtests: the
parent function returns, running its `defer`s, while parallel subtests are
still suspended. Sequential subtests finish inside `t.Run`, so `defer` is safe
there, but `t.Cleanup` works in both cases.
- **Explanatory comments in tests are welcome.** Describe the scenario being set
up; the comment budget below does not apply to them.
```go
server, err := StartTestServer()
@@ -380,7 +642,8 @@ assert.Equal(t, expectedResult, result, "Result should match expected")
than replacing it with your own summary: describe the change, link the issue,
tick the checklist honestly (including "ran locally" and "single purpose"),
and complete the documentation section. Do not tick a box you have not
verified, and do not delete rows that do not apply.
verified, and do not delete rows that do not apply — the docs gate in CI reads
that section and fails when it is missing.
- **Keep the PR description short.** Under 1000 words on top of the template's
own text, and usually far less — a few paragraphs. Reviewers read the diff;
@@ -439,6 +702,12 @@ assert.Equal(t, expectedResult, result, "Result should match expected")
on their own. Propose that split to the user rather than opening one large PR
and hoping.
Prefer GitHub's stacked pull requests for such a sequence, rather than
hand-managing base branches: open each PR against the branch below it instead of
`main`, so every PR's diff shows only its own change. Merging a layer retargets
the PRs above it, and branch protections and required checks on the base branch
still apply to each one.
- **User-facing changes need a docs PR** in
[netbirdio/docs](https://github.com/netbirdio/docs), linked from the PR
description.

View File

@@ -112,6 +112,7 @@ aligns with our security standards and design expectations.
- [Test suite](#test-suite)
- [Checklist before submitting a PR](#checklist-before-submitting-a-pr)
- [When we close a PR](#when-we-close-a-pr)
- [Translations](#translations)
- [Other project repositories](#other-project-repositories)
- [Contributor License Agreement](#contributor-license-agreement)
@@ -612,6 +613,17 @@ A closed PR is not a rejected idea. Take it back to the
[discussion](https://github.com/netbirdio/netbird/discussions), settle the
approach, and reopen the work from there.
## Translations
Desktop UI translations are not contributed through pull requests. Translate on
[Crowdin](https://crowdin.com/project/netbird) instead: no ticket needed, just
join the project and pick your language. Crowdin syncs with this repository and
opens the service PRs itself, so hand-edited locale files would conflict with
the next sync. Style, terminology, and review guidance live in
[client/ui/i18n/TRANSLATING.md](client/ui/i18n/TRANSLATING.md). To request a
language the project does not offer yet, ask on the Crowdin project page or in
a [discussion](https://github.com/netbirdio/netbird/discussions).
## Other project repositories
NetBird project is composed of 3 main repositories:

View File

@@ -15,6 +15,7 @@ import (
log "github.com/sirupsen/logrus"
nbAnonymize "github.com/netbirdio/netbird/client/anonymize"
"github.com/netbirdio/netbird/client/iface/device"
"github.com/netbirdio/netbird/client/internal"
"github.com/netbirdio/netbird/client/internal/debug"
@@ -32,6 +33,13 @@ import (
types "github.com/netbirdio/netbird/upload-server/types"
)
// AnonymizeLevelDefault and AnonymizeLevelStrict are the accepted
// anonymizeLevel values for DebugBundle.
const (
AnonymizeLevelDefault = nbAnonymize.LevelDefaultString
AnonymizeLevelStrict = nbAnonymize.LevelStrictString
)
// ConnectionListener export internal Listener for mobile
type ConnectionListener interface {
peer.Listener
@@ -278,8 +286,10 @@ func (c *Client) GetTunSettings() (*TunSettings, error) {
}
// DebugBundle generates a debug bundle, uploads it, and returns the upload key.
// It works both with and without a running engine.
func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (string, error) {
// It works both with and without a running engine. anonymizeLevel is "default"
// or "strict"; strict also anonymizes internal IP ranges, peer names, and
// WireGuard public keys, and implies anonymize.
func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool, anonymizeLevel string) (string, error) {
cfg, cacheDir, cc := c.stateSnapshot()
// If the engine hasn't been started, load config from disk
@@ -298,6 +308,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
InternalConfig: cfg,
StatusRecorder: c.recorder,
TempDir: cacheDir,
StatePath: platformFiles.StateFilePath(),
}
if cc != nil {
@@ -321,6 +332,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
deps,
debug.BundleConfig{
Anonymize: anonymize,
AnonymizeLevel: nbAnonymize.ParseLevel(anonymizeLevel),
IncludeSystemInfo: true,
},
)

View File

@@ -2,6 +2,7 @@ package anonymize
import (
"crypto/rand"
"encoding/base64"
"fmt"
"math/big"
"net"
@@ -15,13 +16,88 @@ import (
const anonTLD = ".domain"
// Level selects how much the anonymizer redacts. Levels are ordered: a higher
// level redacts strictly more. On the wire (protos, flags) levels travel as
// their string form.
type Level int
const (
// LevelDefault anonymizes public IP addresses, IPv6 ULA, domains, and MAC
// addresses. Internal IPv4 ranges (RFC 1918, CGNAT, link-local) are
// preserved so support can reason about the real topology.
LevelDefault Level = iota
// LevelStrict additionally anonymizes internal IP ranges, peer names, and
// WireGuard public keys.
LevelStrict
)
// LevelDefaultString and LevelStrictString are the wire forms of the levels,
// for boundaries that pass levels as strings (flags, protos, mobile bindings).
const (
LevelDefaultString = "default"
LevelStrictString = "strict"
)
// ParseLevel maps s to a Level. Empty means LevelDefault; anything
// unrecognized maps to LevelStrict so an unknown request never yields less
// anonymization than intended.
func ParseLevel(s string) Level {
switch strings.ToLower(s) {
case "", LevelDefaultString:
return LevelDefault
default:
return LevelStrict
}
}
// String returns the wire form of the level: "default" or "strict".
func (l Level) String() string {
if l >= LevelStrict {
return LevelStrictString
}
return LevelDefaultString
}
// protectedDomains are NetBird-operated suffixes that stay recognizable in an
// anonymized bundle. At LevelStrict the labels in front of them (the peer
// name) are still replaced, except under netbird.io, which only hosts
// NetBird infrastructure (api, signal, flow), never peer names.
var protectedDomains = []string{"netbird.io", "netbird.selfhosted", "netbird.cloud", "netbird.stage"}
const infraDomain = "netbird.io"
var (
macColonRegex = regexp.MustCompile(`\b[0-9a-fA-F]{2}(?::[0-9a-fA-F]{2}){5}\b`)
macDashRegex = regexp.MustCompile(`\b[0-9a-fA-F]{2}(?:-[0-9a-fA-F]{2}){5}\b`)
wgKeyRegex = regexp.MustCompile(`\b[A-Za-z0-9+/]{43}=`)
)
type Anonymizer struct {
ipAnonymizer map[netip.Addr]netip.Addr
domainAnonymizer map[string]string
currentAnonIPv4 netip.Addr
currentAnonIPv6 netip.Addr
startAnonIPv4 netip.Addr
startAnonIPv6 netip.Addr
// domainOrder caches the keys of domainAnonymizer sorted longest-first
// for AnonymizeString; it is rebuilt when the map gains entries.
domainOrder []string
labelAnonymizer map[string]string
labelAnonymized map[string]struct{}
labelCounter uint32
macAnonymizer map[string]string
macCounter uint32
wgKeyAnonymizer map[string]string
wgKeyAnonymized map[string]struct{}
currentAnonIPv4 netip.Addr
currentAnonIPv6 netip.Addr
startAnonIPv4 netip.Addr
startAnonIPv6 netip.Addr
// LevelStrict also anonymizes internal ranges (RFC 1918, CGNAT,
// link-local), replacing them from the dedicated internal pools below so
// a reader can still tell an internal address from a public one.
level Level
currentAnonInternalIPv4 netip.Addr
currentAnonInternalIPv6 netip.Addr
startAnonInternalIPv4 netip.Addr
startAnonInternalIPv6 netip.Addr
domainKeyRegex *regexp.Regexp
}
@@ -32,25 +108,50 @@ func DefaultAddresses() (netip.Addr, netip.Addr) {
return netip.AddrFrom4([4]byte{198, 51, 100, 0}), netip.MustParseAddr("2001:db8:ffff::")
}
// InternalAddresses returns the pool starts used in strict mode for internal
// ranges. Both are reserved ranges that cannot collide with real addressing:
// 198.18.0.0 (RFC 2544 benchmarking), 2001:db8:1:: (RFC 3849 documentation).
func InternalAddresses() (netip.Addr, netip.Addr) {
return netip.AddrFrom4([4]byte{198, 18, 0, 0}), netip.MustParseAddr("2001:db8:1::")
}
func NewAnonymizer(startIPv4, startIPv6 netip.Addr) *Anonymizer {
internalIPv4, internalIPv6 := InternalAddresses()
return &Anonymizer{
ipAnonymizer: map[netip.Addr]netip.Addr{},
domainAnonymizer: map[string]string{},
labelAnonymizer: map[string]string{},
labelAnonymized: map[string]struct{}{},
macAnonymizer: map[string]string{},
wgKeyAnonymizer: map[string]string{},
wgKeyAnonymized: map[string]struct{}{},
currentAnonIPv4: startIPv4,
currentAnonIPv6: startIPv6,
startAnonIPv4: startIPv4,
startAnonIPv6: startIPv6,
level: LevelDefault,
currentAnonInternalIPv4: internalIPv4,
currentAnonInternalIPv6: internalIPv6,
startAnonInternalIPv4: internalIPv4,
startAnonInternalIPv6: internalIPv6,
domainKeyRegex: regexp.MustCompile(`\bdomain=([^\s,:"]+)`),
}
}
// SetLevel selects the anonymization level. The zero value of a new
// Anonymizer is LevelDefault.
func (a *Anonymizer) SetLevel(level Level) {
a.level = level
}
func (a *Anonymizer) AnonymizeIP(ip netip.Addr) netip.Addr {
// Normalize 4-in-6 addresses so ::ffff:192.168.1.1 classifies and maps
// like 192.168.1.1.
ip = ip.Unmap()
if ip.IsLoopback() ||
ip.IsLinkLocalUnicast() ||
ip.IsLinkLocalMulticast() ||
ip.IsInterfaceLocalMulticast() ||
(ip.Is4() && ip.IsPrivate()) ||
ip.IsUnspecified() ||
ip.IsMulticast() ||
isWellKnown(ip) ||
@@ -59,18 +160,100 @@ func (a *Anonymizer) AnonymizeIP(ip netip.Addr) netip.Addr {
return ip
}
if isInternal(ip) && a.level < LevelStrict {
return ip
}
if _, ok := a.ipAnonymizer[ip]; !ok {
if ip.Is4() {
a.ipAnonymizer[ip] = a.currentAnonIPv4
a.currentAnonIPv4 = a.currentAnonIPv4.Next()
} else {
a.ipAnonymizer[ip] = a.currentAnonIPv6
a.currentAnonIPv6 = a.currentAnonIPv6.Next()
}
a.ipAnonymizer[ip] = a.nextAnonIP(ip)
}
return a.ipAnonymizer[ip]
}
func (a *Anonymizer) nextAnonIP(ip netip.Addr) netip.Addr {
// At the strict level, internal addresses (including IPv6 ULA, matched
// by IsPrivate) come from the internal pools so they remain recognizable
// as internal without disclosing the real values.
if a.level >= LevelStrict && (isInternal(ip) || ip.IsPrivate()) {
if ip.Is4() {
anon := a.currentAnonInternalIPv4
a.currentAnonInternalIPv4 = a.currentAnonInternalIPv4.Next()
return anon
}
anon := a.currentAnonInternalIPv6
a.currentAnonInternalIPv6 = a.currentAnonInternalIPv6.Next()
return anon
}
if ip.Is4() {
anon := a.currentAnonIPv4
a.currentAnonIPv4 = a.currentAnonIPv4.Next()
return anon
}
anon := a.currentAnonIPv6
a.currentAnonIPv6 = a.currentAnonIPv6.Next()
return anon
}
// AnonymizeMAC replaces a MAC address with a consistent placeholder from the
// locally administered range starting at 02:00:00:00:00:01, at every
// anonymization level. Broadcast, multicast, all-zero, and already assigned
// placeholder addresses are preserved. The colon and dash spellings of the
// same address share one placeholder; the output keeps the input's separator.
func (a *Anonymizer) AnonymizeMAC(mac string) string {
hw, err := net.ParseMAC(mac)
if err != nil || len(hw) != 6 {
return mac
}
if isWellKnownMAC(hw) || a.isAnonymizedMAC(hw) {
return mac
}
key := hw.String()
anon, ok := a.macAnonymizer[key]
if !ok {
a.macCounter++
anon = fmt.Sprintf("02:00:00:%02x:%02x:%02x", byte(a.macCounter>>16), byte(a.macCounter>>8), byte(a.macCounter))
a.macAnonymizer[key] = anon
}
if strings.Contains(mac, "-") {
anon = strings.ReplaceAll(anon, ":", "-")
}
return anon
}
// isAnonymizedMAC reports whether hw is a placeholder this anonymizer already
// handed out, so a second pass over anonymized output leaves it unchanged.
func (a *Anonymizer) isAnonymizedMAC(hw net.HardwareAddr) bool {
if hw[0] != 0x02 || hw[1] != 0 || hw[2] != 0 {
return false
}
value := uint32(hw[3])<<16 | uint32(hw[4])<<8 | uint32(hw[5])
return value <= a.macCounter
}
// AnonymizeWGKey replaces a WireGuard public key with a consistent random
// placeholder of the same shape. Keys are only anonymized at LevelStrict;
// placeholders already handed out pass through unchanged.
func (a *Anonymizer) AnonymizeWGKey(key string) string {
if a.level < LevelStrict || !looksLikeWGKey(key) {
return key
}
if _, ok := a.wgKeyAnonymized[key]; ok {
return key
}
anon, ok := a.wgKeyAnonymizer[key]
if !ok {
anon = generateAnonymousKey()
a.wgKeyAnonymizer[key] = anon
a.wgKeyAnonymized[anon] = struct{}{}
}
return anon
}
func (a *Anonymizer) AnonymizeUDPAddr(addr net.UDPAddr) net.UDPAddr {
// Convert IP to netip.Addr
ip, ok := netip.AddrFromSlice(addr.IP)
@@ -89,12 +272,12 @@ func (a *Anonymizer) AnonymizeUDPAddr(addr net.UDPAddr) net.UDPAddr {
// isInAnonymizedRange checks if an IP is within the range of already assigned anonymized IPs
func (a *Anonymizer) isInAnonymizedRange(ip netip.Addr) bool {
if ip.Is4() && ip.Compare(a.startAnonIPv4) >= 0 && ip.Compare(a.currentAnonIPv4) <= 0 {
return true
} else if !ip.Is4() && ip.Compare(a.startAnonIPv6) >= 0 && ip.Compare(a.currentAnonIPv6) <= 0 {
return true
if ip.Is4() {
return inPoolRange(ip, a.startAnonIPv4, a.currentAnonIPv4) ||
inPoolRange(ip, a.startAnonInternalIPv4, a.currentAnonInternalIPv4)
}
return false
return inPoolRange(ip, a.startAnonIPv6, a.currentAnonIPv6) ||
inPoolRange(ip, a.startAnonInternalIPv6, a.currentAnonInternalIPv6)
}
func (a *Anonymizer) AnonymizeIPString(ip string) string {
@@ -118,14 +301,17 @@ func (a *Anonymizer) AnonymizeDomain(domain string) string {
baseDomain = domain[:len(domain)-1]
}
if strings.HasSuffix(baseDomain, "netbird.io") ||
strings.HasSuffix(baseDomain, "netbird.selfhosted") ||
strings.HasSuffix(baseDomain, "netbird.cloud") ||
strings.HasSuffix(baseDomain, "netbird.stage") ||
strings.HasSuffix(baseDomain, anonTLD) {
if strings.HasSuffix(baseDomain, anonTLD) {
return domain
}
if suffix := protectedSuffix(baseDomain); suffix != "" {
if a.level < LevelStrict || baseDomain == suffix || suffix == infraDomain {
return domain
}
return withTrailingDot(a.anonymizePeerName(baseDomain, suffix), hasDot)
}
parts := strings.Split(baseDomain, ".")
if len(parts) < 2 {
return domain
@@ -141,12 +327,53 @@ func (a *Anonymizer) AnonymizeDomain(domain string) string {
}
result := strings.Replace(baseDomain, baseForLookup, anonymized, 1)
if hasDot {
result += "."
if a.level >= LevelStrict && len(parts) > 2 {
prefix := strings.TrimSuffix(baseDomain, "."+baseForLookup)
result = a.anonymizeLabels(prefix, "host") + "." + anonymized
// The full mapping feeds AnonymizeString so seeded FQDNs are caught
// in log lines as a whole, labels included.
a.domainAnonymizer[baseDomain] = result
}
return withTrailingDot(result, hasDot)
}
// anonymizePeerName replaces the labels in front of a protected suffix with
// numbered peer placeholders, keeping the suffix, and records the full
// mapping for string replacement in logs. The numbering keeps a peer
// recognizable across the whole bundle without disclosing its name.
func (a *Anonymizer) anonymizePeerName(baseDomain, suffix string) string {
prefix := strings.TrimSuffix(baseDomain, "."+suffix)
result := a.anonymizeLabels(prefix, "peer") + "." + suffix
if result != baseDomain {
a.domainAnonymizer[baseDomain] = result
}
return result
}
// anonymizeLabels replaces each dot-separated label with a consistent
// numbered placeholder ("<placeholder>-<n>"). Wildcard labels and
// placeholders already handed out pass through unchanged.
func (a *Anonymizer) anonymizeLabels(prefix, placeholder string) string {
labels := strings.Split(prefix, ".")
for i, label := range labels {
if label == "*" {
continue
}
if _, ok := a.labelAnonymized[label]; ok {
continue
}
anon, ok := a.labelAnonymizer[label]
if !ok {
a.labelCounter++
anon = fmt.Sprintf("%s-%d", placeholder, a.labelCounter)
a.labelAnonymizer[label] = anon
a.labelAnonymized[anon] = struct{}{}
}
labels[i] = anon
}
return strings.Join(labels, ".")
}
func (a *Anonymizer) AnonymizeURI(uri string) string {
u, err := url.Parse(uri)
if err != nil {
@@ -181,16 +408,70 @@ func (a *Anonymizer) AnonymizeString(str string) string {
str = ipv4Regex.ReplaceAllStringFunc(str, a.AnonymizeIPString)
str = ipv6Regex.ReplaceAllStringFunc(str, a.AnonymizeIPString)
for domain, anonDomain := range a.domainAnonymizer {
str = strings.ReplaceAll(str, domain, anonDomain)
for _, domain := range a.sortedDomains() {
str = strings.ReplaceAll(str, domain, a.domainAnonymizer[domain])
}
str = a.AnonymizeSchemeURI(str)
str = a.AnonymizeDNSLogLine(str)
// MAC handling runs after the IP passes so preserved IPv6 addresses are
// already out of the way; the separator guard skips matches embedded in a
// longer colon- or dash-separated sequence (such as an IPv6 tail).
str = a.anonymizeMACsInString(str, macColonRegex, ':')
str = a.anonymizeMACsInString(str, macDashRegex, '-')
if a.level >= LevelStrict {
str = wgKeyRegex.ReplaceAllStringFunc(str, a.AnonymizeWGKey)
}
return str
}
// sortedDomains returns the domain mappings longest-first, so a full-FQDN
// mapping (strict level) is applied before the base-domain mapping it
// contains. The order is rebuilt only when domainAnonymizer has grown.
func (a *Anonymizer) sortedDomains() []string {
if len(a.domainOrder) == len(a.domainAnonymizer) {
return a.domainOrder
}
a.domainOrder = a.domainOrder[:0]
for domain := range a.domainAnonymizer {
a.domainOrder = append(a.domainOrder, domain)
}
slices.SortFunc(a.domainOrder, func(x, y string) int {
if d := len(y) - len(x); d != 0 {
return d
}
return strings.Compare(x, y)
})
return a.domainOrder
}
// anonymizeMACsInString replaces MAC addresses matched by re, skipping
// matches that directly adjoin another sep so a six-group run inside a longer
// separated sequence is left alone.
func (a *Anonymizer) anonymizeMACsInString(str string, re *regexp.Regexp, sep byte) string {
matches := re.FindAllStringIndex(str, -1)
if len(matches) == 0 {
return str
}
var b strings.Builder
last := 0
for _, m := range matches {
if (m[0] > 0 && str[m[0]-1] == sep) || (m[1] < len(str) && str[m[1]] == sep) {
continue
}
b.WriteString(str[last:m[0]])
b.WriteString(a.AnonymizeMAC(str[m[0]:m[1]]))
last = m[1]
}
b.WriteString(str[last:])
return b.String()
}
// AnonymizeSchemeURI finds and anonymizes URIs with ws, wss, rel, rels, stun, stuns, turn, and turns schemes.
func (a *Anonymizer) AnonymizeSchemeURI(text string) string {
re := regexp.MustCompile(`(?i)\b(wss?://|rels?://|stuns?:|turns?:|https?://)\S+\b`)
@@ -239,10 +520,79 @@ func isWellKnown(addr netip.Addr) bool {
"128.0.0.0", "8000::", // 2nd split subnet for default routes
}
if slices.Contains(wellKnown, addr.String()) {
return slices.Contains(wellKnown, addr.String())
}
// isInternal reports whether ip identifies a host only within the local
// network: IPv4 private (RFC 1918), CGNAT (RFC 6598), and link-local (v4 and
// v6). These are preserved at the default level so support can reason about
// the real topology, and replaced from the internal pools at the strict
// level. IPv6 ULA is deliberately not internal: its random global ID uniquely
// fingerprints the network, so it is anonymized at every level.
func isInternal(ip netip.Addr) bool {
return (ip.Is4() && ip.IsPrivate()) ||
ip.IsLinkLocalUnicast() ||
isCGNAT(ip)
}
func inPoolRange(ip, start, current netip.Addr) bool {
return ip.Compare(start) >= 0 && ip.Compare(current) <= 0
}
// isWellKnownMAC reports whether hw carries no stable host identity: all-zero
// or a group address (broadcast and multicast).
func isWellKnownMAC(hw net.HardwareAddr) bool {
if hw[0]&1 == 1 {
return true
}
for _, b := range hw {
if b != 0 {
return false
}
}
return true
}
// looksLikeWGKey reports whether s has the shape of a WireGuard key:
// 44 base64 characters decoding to 32 bytes.
func looksLikeWGKey(s string) bool {
if len(s) != 44 || s[43] != '=' {
return false
}
decoded, err := base64.StdEncoding.DecodeString(s)
return err == nil && len(decoded) == 32
}
func generateAnonymousKey() string {
buf := make([]byte, 32)
if _, err := rand.Read(buf); err != nil {
return strings.Repeat("A", 43) + "="
}
return base64.StdEncoding.EncodeToString(buf)
}
// protectedSuffix returns the protected NetBird suffix baseDomain ends with,
// or empty. The match is label-anchored so an unrelated domain that merely
// ends in the same characters is not preserved.
func protectedSuffix(baseDomain string) string {
for _, d := range protectedDomains {
if baseDomain == d || strings.HasSuffix(baseDomain, "."+d) {
return d
}
}
return ""
}
func withTrailingDot(domain string, hasDot bool) string {
if hasDot {
return domain + "."
}
return domain
}
// isCGNAT reports whether addr is in 100.64.0.0/10 (RFC 6598), the range
// NetBird assigns overlay peer addresses from.
func isCGNAT(addr netip.Addr) bool {
cgnatRangeStart := netip.AddrFrom4([4]byte{100, 64, 0, 0})
cgnatRange := netip.PrefixFrom(cgnatRangeStart, 10)

View File

@@ -1,8 +1,11 @@
package anonymize_test
import (
"bytes"
"encoding/base64"
"net/netip"
"regexp"
"strings"
"testing"
"github.com/stretchr/testify/assert"
@@ -44,6 +47,301 @@ func TestAnonymizeIP(t *testing.T) {
}
}
func TestParseLevel(t *testing.T) {
tests := []struct {
input string
expect anonymize.Level
}{
{"", anonymize.LevelDefault},
{"default", anonymize.LevelDefault},
{"DEFAULT", anonymize.LevelDefault},
{"strict", anonymize.LevelStrict},
{"STRICT", anonymize.LevelStrict},
// Unknown values must never yield less anonymization than requested.
{"garbage", anonymize.LevelStrict},
}
for _, tc := range tests {
t.Run("input="+tc.input, func(t *testing.T) {
assert.Equal(t, tc.expect, anonymize.ParseLevel(tc.input), "parsed level should match")
})
}
}
func TestAnonymizeIP_DefaultLevelInternalRanges(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
tests := []struct {
name string
ip string
expect string
}{
{"RFC1918 10/8", "10.1.2.3", "10.1.2.3"},
{"RFC1918 172.16/12", "172.16.5.5", "172.16.5.5"},
{"RFC1918 192.168/16", "192.168.1.1", "192.168.1.1"},
{"CGNAT", "100.64.0.5", "100.64.0.5"},
{"IPv4 link-local", "169.254.1.1", "169.254.1.1"},
{"IPv6 link-local", "fe80::1", "fe80::1"},
// ULA is anonymized even at the default level: its random global ID
// uniquely fingerprints the network, unlike shared RFC 1918 space.
{"IPv6 ULA", "fd12:3456:789a::1", "2001:db8:ffff::"},
// 4-in-6 addresses classify like their unmapped IPv4 form.
{"4-in-6 RFC1918", "::ffff:192.168.1.1", "192.168.1.1"},
{"4-in-6 CGNAT", "::ffff:100.64.0.5", "100.64.0.5"},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
result := anonymizer.AnonymizeIP(netip.MustParseAddr(tc.ip))
assert.Equal(t, tc.expect, result.String(), "default level should preserve internal ranges except ULA")
})
}
}
func TestAnonymizeIP_StrictLevel(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(anonymize.LevelStrict)
// Order matters: internal pool addresses are assigned sequentially.
tests := []struct {
name string
ip string
expect string
}{
{"RFC1918 192.168/16", "192.168.1.1", "198.18.0.0"},
{"Second RFC1918", "192.168.1.2", "198.18.0.1"},
{"Repeated RFC1918", "192.168.1.1", "198.18.0.0"},
{"RFC1918 10/8", "10.1.2.3", "198.18.0.2"},
{"RFC1918 172.16/12", "172.16.5.5", "198.18.0.3"},
{"CGNAT", "100.64.0.5", "198.18.0.4"},
{"IPv4 link-local", "169.254.1.1", "198.18.0.5"},
{"Public IPv4 uses public pool", "1.2.3.4", "198.51.100.0"},
{"IPv6 link-local", "fe80::1", "2001:db8:1::"},
{"IPv6 ULA", "fd12:3456:789a::1", "2001:db8:1::1"},
{"Public IPv6 uses public pool", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
{"Loopback IPv4", "127.0.0.1", "127.0.0.1"},
{"Loopback IPv6", "::1", "::1"},
{"Unspecified", "0.0.0.0", "0.0.0.0"},
{"Multicast", "224.0.0.251", "224.0.0.251"},
{"Well known resolver", "8.8.8.8", "8.8.8.8"},
{"Well known split marker", "128.0.0.0", "128.0.0.0"},
{"In internal pool range", "198.18.0.3", "198.18.0.3"},
{"In public pool range", "198.51.100.0", "198.51.100.0"},
{"4-in-6 repeated RFC1918", "::ffff:192.168.1.1", "198.18.0.0"},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
result := anonymizer.AnonymizeIP(netip.MustParseAddr(tc.ip))
assert.Equal(t, tc.expect, result.String(), "strict level should replace internal ranges from the internal pools")
})
}
}
func TestAnonymizeString_StrictInternalIPs(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(anonymize.LevelStrict)
input := "route 10.20.30.0/24 via 192.168.1.1 dev eth0 src 100.64.0.7"
firstPass := anonymizer.AnonymizeString(input)
secondPass := anonymizer.AnonymizeString(firstPass)
assert.NotContains(t, firstPass, "10.20.30.0", "private network address should be anonymized")
assert.NotContains(t, firstPass, "192.168.1.1", "private gateway should be anonymized")
assert.NotContains(t, firstPass, "100.64.0.7", "CGNAT address should be anonymized")
assert.Contains(t, firstPass, "/24", "prefix length should be preserved")
assert.Equal(t, firstPass, secondPass, "second pass should not further anonymize the string")
}
func TestAnonymizeMAC(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
first := anonymizer.AnonymizeMAC("aa:bb:cc:dd:ee:0f")
assert.Equal(t, "02:00:00:00:00:01", first, "first MAC should get the first placeholder")
assert.Equal(t, first, anonymizer.AnonymizeMAC("aa:bb:cc:dd:ee:0f"), "repeated MAC should map to the same placeholder")
assert.Equal(t, first, anonymizer.AnonymizeMAC("AA:BB:CC:DD:EE:0F"), "case should not affect the mapping")
assert.Equal(t, "02-00-00-00-00-01", anonymizer.AnonymizeMAC("AA-BB-CC-DD-EE-0F"), "dash form should keep its separator but share the mapping")
second := anonymizer.AnonymizeMAC("10:22:33:44:55:66")
assert.Equal(t, "02:00:00:00:00:02", second, "second distinct MAC should get the next placeholder")
tests := []struct {
name string
mac string
}{
{"Broadcast", "ff:ff:ff:ff:ff:ff"},
{"IPv4 multicast", "01:00:5e:00:00:fb"},
{"IPv6 multicast", "33:33:00:00:00:01"},
{"All zero", "00:00:00:00:00:00"},
{"Assigned placeholder", "02:00:00:00:00:01"},
{"Invalid", "not-a-mac"},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
assert.Equal(t, tc.mac, anonymizer.AnonymizeMAC(tc.mac), "should be preserved")
})
}
}
func TestAnonymizeString_MACAddresses(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
tests := []struct {
name string
input string
expect string
}{
{
name: "nftables ether rule",
input: "ether saddr aa:bb:cc:dd:ee:ff drop",
expect: "ether saddr 02:00:00:00:00:01 drop",
},
{
name: "Windows dash form",
input: "Physical Address : AA-BB-CC-DD-EE-FF",
expect: "Physical Address : 02-00-00-00-00-01",
},
{
name: "IPv6 address tail is not treated as MAC",
input: "addr fe80:0:11:22:33:44:55:66 scope link",
expect: "addr fe80:0:11:22:33:44:55:66 scope link",
},
{
name: "broadcast MAC preserved",
input: "dst ff:ff:ff:ff:ff:ff type ARP",
expect: "dst ff:ff:ff:ff:ff:ff type ARP",
},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
result := anonymizer.AnonymizeString(tc.input)
assert.Equal(t, tc.expect, result, "MAC addresses should be anonymized at every level")
assert.Equal(t, result, anonymizer.AnonymizeString(result), "second pass should not change the result")
})
}
}
func TestAnonymizeWGKey(t *testing.T) {
key := base64.StdEncoding.EncodeToString(bytes.Repeat([]byte{0x42}, 32))
t.Run("default level preserves keys", func(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
assert.Equal(t, key, anonymizer.AnonymizeWGKey(key), "default level should not touch WireGuard keys")
})
t.Run("strict level replaces keys", func(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(anonymize.LevelStrict)
anon := anonymizer.AnonymizeWGKey(key)
assert.NotEqual(t, key, anon, "strict level should replace the key")
assert.Regexp(t, `^[A-Za-z0-9+/]{43}=$`, anon, "placeholder should keep the WireGuard key shape")
assert.Equal(t, anon, anonymizer.AnonymizeWGKey(key), "repeated key should map to the same placeholder")
assert.Equal(t, anon, anonymizer.AnonymizeWGKey(anon), "an assigned placeholder should pass through unchanged")
assert.Equal(t, "not-a-key", anonymizer.AnonymizeWGKey("not-a-key"), "non-key values should be preserved")
})
}
func TestAnonymizeString_WGKeys(t *testing.T) {
key := base64.StdEncoding.EncodeToString(bytes.Repeat([]byte{0x42}, 32))
input := "peer " + key + " handshake completed"
t.Run("default level preserves keys", func(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
assert.Equal(t, input, anonymizer.AnonymizeString(input), "default level should not touch WireGuard keys in strings")
})
t.Run("strict level replaces keys", func(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(anonymize.LevelStrict)
firstPass := anonymizer.AnonymizeString(input)
assert.NotContains(t, firstPass, key, "the key should not survive strict anonymization")
assert.Equal(t, anonymizer.AnonymizeWGKey(key), extractKey(t, firstPass), "string replacement should be consistent with AnonymizeWGKey")
assert.Equal(t, firstPass, anonymizer.AnonymizeString(firstPass), "second pass should not change the result")
})
}
func extractKey(t *testing.T, logLine string) string {
t.Helper()
fields := strings.Fields(logLine)
require.Len(t, fields, 4, "log line should keep its structure")
return fields[1]
}
func TestAnonymizeDomain_StrictLevel(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(anonymize.LevelStrict)
t.Run("netbird peer name", func(t *testing.T) {
result := anonymizer.AnonymizeDomain("my-laptop.netbird.cloud")
assert.Regexp(t, `^peer-\d+\.netbird\.cloud$`, result, "peer name should be anonymized, suffix kept")
assert.NotContains(t, result, "my-laptop", "the peer name should not survive")
assert.Equal(t, result, anonymizer.AnonymizeDomain("my-laptop.netbird.cloud"), "repeated domain should map consistently")
assert.Equal(t, result, anonymizer.AnonymizeDomain(result), "an anonymized domain should pass through unchanged")
})
t.Run("bare netbird domain", func(t *testing.T) {
assert.Equal(t, "netbird.cloud", anonymizer.AnonymizeDomain("netbird.cloud"), "the bare protected suffix should be preserved")
})
t.Run("netbird infrastructure preserved", func(t *testing.T) {
assert.Equal(t, "api.netbird.io", anonymizer.AnonymizeDomain("api.netbird.io"),
"netbird.io hosts infrastructure, not peer names, and should stay readable")
})
t.Run("leading labels of other domains", func(t *testing.T) {
result := anonymizer.AnonymizeDomain("host1.corp.example.com")
assert.Regexp(t, `^host-\d+\.host-\d+\.anon-[a-zA-Z0-9]+\.domain$`, result, "every label should be anonymized")
for _, label := range []string{"host1", "corp", "example"} {
assert.NotContains(t, result, label, "no original label should survive")
}
assert.Equal(t, result, anonymizer.AnonymizeDomain("host1.corp.example.com"), "repeated domain should map consistently")
})
t.Run("same label maps consistently across domains", func(t *testing.T) {
first := anonymizer.AnonymizeDomain("shared.one.com")
second := anonymizer.AnonymizeDomain("shared.two.com")
assert.Equal(t, strings.Split(first, ".")[0], strings.Split(second, ".")[0], "the shared host label should get one placeholder")
})
t.Run("wildcard label preserved", func(t *testing.T) {
result := anonymizer.AnonymizeDomain("*.example.com")
assert.Regexp(t, `^\*\.anon-[a-zA-Z0-9]+\.domain$`, result, "the wildcard label should stay a wildcard")
})
}
func TestAnonymizeDomain_DefaultLevelKeepsPeerNames(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
assert.Equal(t, "my-laptop.netbird.cloud", anonymizer.AnonymizeDomain("my-laptop.netbird.cloud"),
"default level should preserve netbird FQDNs including the peer name")
assert.Regexp(t, `^sub\.anon-[a-zA-Z0-9]+\.domain$`, anonymizer.AnonymizeDomain("sub.example.com"),
"default level should keep subdomain labels")
}
func TestAnonymizeString_StrictPeerNames(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(anonymize.LevelStrict)
// Seed like the bundle generator does from the status: base first, then
// the full FQDN, so replacement must prefer the longer mapping.
anonBase := anonymizer.AnonymizeDomain("example.com")
anonPeer := anonymizer.AnonymizeDomain("peer1.netbird.cloud")
anonHost := anonymizer.AnonymizeDomain("host1.example.com")
logLine := "connected to peer1.netbird.cloud via host1.example.com endpoint"
firstPass := anonymizer.AnonymizeString(logLine)
assert.NotContains(t, firstPass, "peer1", "the peer name should not survive in logs")
assert.NotContains(t, firstPass, "host1", "the host label should not survive in logs")
assert.Contains(t, firstPass, anonPeer, "the seeded peer mapping should be applied")
assert.Contains(t, firstPass, anonHost, "the seeded host mapping should be applied, not just the base mapping")
assert.NotContains(t, firstPass, "host1."+anonBase, "the base mapping must not preempt the longer FQDN mapping")
assert.Equal(t, firstPass, anonymizer.AnonymizeString(firstPass), "second pass should not change the result")
}
func TestAnonymizeDNSLogLine(t *testing.T) {
anonymizer := anonymize.NewAnonymizer(netip.Addr{}, netip.Addr{})
tests := []struct {

View File

@@ -27,8 +27,8 @@ import (
const errCloseConnection = "Failed to close connection: %v"
var (
logFileCount uint32
systemInfoFlag bool
logFileCount uint32
systemInfoFlag bool
uploadBundleFlag bool
uploadBundleURLFlag string
uploadBundleInsecureFlag bool
@@ -156,6 +156,11 @@ func debugConfigDump(cmd *cobra.Command, _ []string) error {
// request. Returns an error if the RPC fails or if the daemon reports
// an upload failure reason.
func debugBundle(cmd *cobra.Command, _ []string) error {
anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
if err != nil {
return err
}
conn, err := getClient(cmd)
if err != nil {
return err
@@ -168,10 +173,11 @@ func debugBundle(cmd *cobra.Command, _ []string) error {
client := proto.NewDaemonServiceClient(conn)
request := &proto.DebugBundleRequest{
Anonymize: anonymizeFlag,
SystemInfo: systemInfoFlag,
LogFileCount: logFileCount,
CliVersion: version.NetbirdVersion(),
Anonymize: anonymizeEnabled,
AnonymizeLevel: anonymizeLevel.String(),
SystemInfo: systemInfoFlag,
LogFileCount: logFileCount,
CliVersion: version.NetbirdVersion(),
}
if uploadBundleFlag {
request.UploadURL = uploadBundleURLFlag
@@ -229,6 +235,11 @@ func runForDuration(cmd *cobra.Command, args []string) error {
return fmt.Errorf("invalid duration format: %v", err)
}
anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
if err != nil {
return err
}
conn, err := getClient(cmd)
if err != nil {
return err
@@ -368,10 +379,11 @@ func runForDuration(cmd *cobra.Command, args []string) error {
cmd.Println("Creating debug bundle...")
request := &proto.DebugBundleRequest{
Anonymize: anonymizeFlag,
SystemInfo: systemInfoFlag,
LogFileCount: logFileCount,
CliVersion: version.NetbirdVersion(),
Anonymize: anonymizeEnabled,
AnonymizeLevel: anonymizeLevel.String(),
SystemInfo: systemInfoFlag,
LogFileCount: logFileCount,
CliVersion: version.NetbirdVersion(),
}
if uploadBundleFlag {
request.UploadURL = uploadBundleURLFlag

View File

@@ -5,7 +5,6 @@ import (
"fmt"
"os"
"os/user"
"runtime"
"strings"
log "github.com/sirupsen/logrus"
@@ -121,7 +120,7 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
loginRequest := proto.LoginRequest{
SetupKey: providedSetupKey,
ManagementUrl: managementURL,
IsUnixDesktopClient: isUnixRunningDesktop(),
IsUnixDesktopClient: util.HasGraphicalSession(),
Hostname: hostName,
DnsLabels: dnsLabelsReq,
ProfileName: &handle,
@@ -189,7 +188,8 @@ func doExtendSession(ctx context.Context, cmd *cobra.Command) error {
client := proto.NewDaemonServiceClient(conn)
req := &proto.RequestExtendAuthSessionRequest{}
// the CLI runs in the user's session, the daemon does not: tell it what we can see
req := &proto.RequestExtendAuthSessionRequest{HasGraphicalSession: util.HasGraphicalSession()}
// Pre-fill the IdP login hint from the active profile so the user
// doesn't have to retype their email. Best-effort: we still proceed
// without a hint if the lookup fails.
@@ -408,8 +408,13 @@ func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *pro
hint = profileState.Email
}
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, isUnixRunningDesktop(), false, hint)
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, util.HasGraphicalSession(), false, hint)
if err != nil {
// enrolling a device is the one flow a setup key can replace
if auth.IsSSOUnavailable(err) {
return nil, fmt.Errorf("%w. Set this device up with a setup key instead: "+
"https://docs.netbird.io/how-to/register-machines-using-setup-keys", err)
}
return nil, err
}
@@ -458,14 +463,6 @@ func openURL(cmd *cobra.Command, verificationURIComplete, userCode string, noBro
}
}
// isUnixRunningDesktop checks if a Linux OS is running desktop environment
func isUnixRunningDesktop() bool {
if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
return false
}
return os.Getenv("DESKTOP_SESSION") != "" || os.Getenv("XDG_CURRENT_DESKTOP") != ""
}
func setEnvAndFlags(cmd *cobra.Command) error {
SetFlagsFromEnvVars(rootCmd)

View File

@@ -21,6 +21,7 @@ import (
"github.com/spf13/pflag"
"google.golang.org/grpc"
"github.com/netbirdio/netbird/client/anonymize"
daddr "github.com/netbirdio/netbird/client/internal/daemonaddr"
"github.com/netbirdio/netbird/client/internal/profilemanager"
)
@@ -69,6 +70,7 @@ var (
autoConnectDisabled bool
extraIFaceBlackList []string
anonymizeFlag bool
anonymizeLevelFlag string
dnsRouteInterval time.Duration
// lazyConnEnabled is the parse target for the deprecated --enable-lazy-connection
// flag. The flag is inert; the value is no longer read (use NB_LAZY_CONN instead).
@@ -156,7 +158,8 @@ func init() {
rootCmd.MarkFlagsMutuallyExclusive("setup-key", "setup-key-file")
rootCmd.PersistentFlags().StringVar(&preSharedKey, preSharedKeyFlag, "", "Sets WireGuard PreSharedKey property. If set, then only peers that have the same key can communicate.")
rootCmd.PersistentFlags().StringVarP(&hostName, "hostname", "n", "", "Sets a custom hostname for the device")
rootCmd.PersistentFlags().BoolVarP(&anonymizeFlag, "anonymize", "A", false, "anonymize IP addresses and non-netbird.io domains in logs and status output")
rootCmd.PersistentFlags().BoolVarP(&anonymizeFlag, "anonymize", "A", false, "anonymize public IP addresses, MAC addresses, and non-netbird.io domains in logs and status output; private, CGNAT, and link-local IP ranges are kept (see --anonymize-level strict)")
rootCmd.PersistentFlags().StringVar(&anonymizeLevelFlag, "anonymize-level", "", "anonymization level: \"default\" or \"strict\"; strict also anonymizes private, CGNAT, and link-local IP ranges, peer names, and WireGuard public keys. Setting this flag implies --anonymize")
rootCmd.PersistentFlags().StringVarP(&configPath, "config", "c", profilemanager.DefaultConfigPath, "Overrides the default profile file location")
rootCmd.AddCommand(upCmd)
@@ -293,6 +296,19 @@ var CLIBackOffSettings = &backoff.ExponentialBackOff{
Clock: backoff.SystemClock,
}
// effectiveAnonymize resolves the --anonymize and --anonymize-level flags:
// setting a level implies anonymization, and an invalid level is rejected.
func effectiveAnonymize() (bool, anonymize.Level, error) {
if anonymizeLevelFlag == "" {
return anonymizeFlag, anonymize.LevelDefault, nil
}
level := anonymize.ParseLevel(anonymizeLevelFlag)
if !strings.EqualFold(anonymizeLevelFlag, level.String()) {
return false, anonymize.LevelDefault, fmt.Errorf("invalid anonymize level %q: use %q or %q", anonymizeLevelFlag, anonymize.LevelDefault.String(), anonymize.LevelStrict.String())
}
return true, level, nil
}
func getSetupKey() (string, error) {
if setupKeyPath != "" && setupKey == "" {
return getSetupKeyFromFile(setupKeyPath)

View File

@@ -121,8 +121,14 @@ func statusFunc(cmd *cobra.Command, args []string) error {
sessionExpiresAt = ts.AsTime().UTC()
}
anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
if err != nil {
return err
}
var outputInformationHolder = nbstatus.ConvertToStatusOutputOverview(resp.GetFullStatus(), nbstatus.ConvertOptions{
Anonymize: anonymizeFlag,
Anonymize: anonymizeEnabled,
AnonymizeLevel: anonymizeLevel,
DaemonVersion: resp.GetDaemonVersion(),
DaemonStatus: nbstatus.ParseDaemonStatus(status),
StatusFilter: statusFilter,

View File

@@ -21,8 +21,8 @@ import (
"github.com/netbirdio/netbird/client/internal"
"github.com/netbirdio/netbird/client/internal/peer"
"github.com/netbirdio/netbird/client/internal/profilemanager"
"github.com/netbirdio/netbird/client/proto"
nbnet "github.com/netbirdio/netbird/client/net"
"github.com/netbirdio/netbird/client/proto"
"github.com/netbirdio/netbird/client/server"
"github.com/netbirdio/netbird/client/system"
"github.com/netbirdio/netbird/shared/management/domain"
@@ -421,12 +421,6 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
if cmd.Flag(serverSSHAllowedFlag).Changed {
req.ServerSSHAllowed = &serverSSHAllowed
}
if cmd.Flag(serverVNCAllowedFlag).Changed {
req.ServerVNCAllowed = &serverVNCAllowed
}
if cmd.Flag(disableVNCApprovalFlag).Changed {
req.DisableVNCApproval = &disableVNCApproval
}
if cmd.Flag(enableSSHRootFlag).Changed {
req.EnableSSHRoot = &enableSSHRoot
}
@@ -529,14 +523,30 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
if cmd.Flag(serverSSHAllowedFlag).Changed {
ic.ServerSSHAllowed = &serverSSHAllowed
}
if cmd.Flag(serverVNCAllowedFlag).Changed {
ic.ServerVNCAllowed = &serverVNCAllowed
}
if cmd.Flag(disableVNCApprovalFlag).Changed {
ic.DisableVNCApproval = &disableVNCApproval
if cmd.Flag(enableSSHRootFlag).Changed {
ic.EnableSSHRoot = &enableSSHRoot
}
applySSHFlagsToConfig(cmd, &ic)
if cmd.Flag(enableSSHSFTPFlag).Changed {
ic.EnableSSHSFTP = &enableSSHSFTP
}
if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
ic.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
}
if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
ic.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
}
if cmd.Flag(disableSSHAuthFlag).Changed {
ic.DisableSSHAuth = &disableSSHAuth
}
if cmd.Flag(sshJWTCacheTTLFlag).Changed {
ic.SSHJWTCacheTTL = &sshJWTCacheTTL
}
if cmd.Flag(interfaceNameFlag).Changed {
if err := parseInterfaceName(interfaceName); err != nil {
@@ -609,49 +619,6 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
return &ic, nil
}
func applySSHFlagsToConfig(cmd *cobra.Command, ic *profilemanager.ConfigInput) {
if cmd.Flag(enableSSHRootFlag).Changed {
ic.EnableSSHRoot = &enableSSHRoot
}
if cmd.Flag(enableSSHSFTPFlag).Changed {
ic.EnableSSHSFTP = &enableSSHSFTP
}
if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
ic.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
}
if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
ic.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
}
if cmd.Flag(disableSSHAuthFlag).Changed {
ic.DisableSSHAuth = &disableSSHAuth
}
if cmd.Flag(sshJWTCacheTTLFlag).Changed {
ic.SSHJWTCacheTTL = &sshJWTCacheTTL
}
}
func applySSHFlagsToLogin(cmd *cobra.Command, req *proto.LoginRequest) {
if cmd.Flag(enableSSHRootFlag).Changed {
req.EnableSSHRoot = &enableSSHRoot
}
if cmd.Flag(enableSSHSFTPFlag).Changed {
req.EnableSSHSFTP = &enableSSHSFTP
}
if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
req.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
}
if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
req.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
}
if cmd.Flag(disableSSHAuthFlag).Changed {
req.DisableSSHAuth = &disableSSHAuth
}
if cmd.Flag(sshJWTCacheTTLFlag).Changed {
ttl := int32(sshJWTCacheTTL)
req.SshJWTCacheTTL = &ttl
}
}
func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte, cmd *cobra.Command) (*proto.LoginRequest, error) {
loginRequest := proto.LoginRequest{
SetupKey: providedSetupKey,
@@ -659,7 +626,7 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
NatExternalIPs: natExternalIPs,
CleanNATExternalIPs: natExternalIPs != nil && len(natExternalIPs) == 0,
CustomDNSAddress: customDNSAddressConverted,
IsUnixDesktopClient: isUnixRunningDesktop(),
IsUnixDesktopClient: util.HasGraphicalSession(),
Hostname: hostName,
ExtraIFaceBlacklist: extraIFaceBlackList,
DnsLabels: dnsLabels,
@@ -681,14 +648,31 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
if cmd.Flag(serverSSHAllowedFlag).Changed {
loginRequest.ServerSSHAllowed = &serverSSHAllowed
}
if cmd.Flag(serverVNCAllowedFlag).Changed {
loginRequest.ServerVNCAllowed = &serverVNCAllowed
}
if cmd.Flag(disableVNCApprovalFlag).Changed {
loginRequest.DisableVNCApproval = &disableVNCApproval
if cmd.Flag(enableSSHRootFlag).Changed {
loginRequest.EnableSSHRoot = &enableSSHRoot
}
applySSHFlagsToLogin(cmd, &loginRequest)
if cmd.Flag(enableSSHSFTPFlag).Changed {
loginRequest.EnableSSHSFTP = &enableSSHSFTP
}
if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
loginRequest.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
}
if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
loginRequest.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
}
if cmd.Flag(disableSSHAuthFlag).Changed {
loginRequest.DisableSSHAuth = &disableSSHAuth
}
if cmd.Flag(sshJWTCacheTTLFlag).Changed {
sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
loginRequest.SshJWTCacheTTL = &sshJWTCacheTTL32
}
if cmd.Flag(disableAutoConnectFlag).Changed {
loginRequest.DisableAutoConnect = &autoConnectDisabled

View File

@@ -1,102 +0,0 @@
//go:build windows || (darwin && !ios)
package cmd
import (
"fmt"
"net"
"net/netip"
"os"
log "github.com/sirupsen/logrus"
"github.com/spf13/cobra"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
var (
vncAgentSocket string
vncAgentTargetUID uint32
)
func init() {
vncAgentCmd.Flags().StringVar(&vncAgentSocket, "socket", "", "Unix-domain socket path the agent listens on (required)")
vncAgentCmd.Flags().Uint32Var(&vncAgentTargetUID, "target-uid", 0, "uid the agent should drop privileges to before listening (darwin only; 0 = stay as current uid)")
rootCmd.AddCommand(vncAgentCmd)
}
// vncAgentCmd runs a VNC server inside the user's interactive session,
// listening on a Unix-domain socket. The NetBird service spawns it: on
// Windows via CreateProcessAsUser into the console session, on macOS via
// launchctl asuser into the Aqua session.
var vncAgentCmd = &cobra.Command{
Use: "vnc-agent",
Short: "Run VNC capture agent (internal, spawned by service)",
Hidden: true,
RunE: func(cmd *cobra.Command, args []string) error {
log.SetReportCaller(true)
log.SetFormatter(&log.JSONFormatter{})
log.SetOutput(os.Stderr)
if vncAgentSocket == "" {
return fmt.Errorf("--socket is required")
}
token := os.Getenv("NB_VNC_AGENT_TOKEN")
if token == "" {
return fmt.Errorf("NB_VNC_AGENT_TOKEN not set; agent requires a token from the service")
}
// Purge the token from env so it doesn't leak via /proc/<pid>/environ.
if err := os.Unsetenv("NB_VNC_AGENT_TOKEN"); err != nil {
log.Debugf("unset NB_VNC_AGENT_TOKEN: %v", err)
}
// Drop root privileges to the target console user BEFORE creating
// the listening socket: keeps a post-auth bug in the encoder /
// input / capture paths confined to the user's own privileges
// rather than escalating to host root, and makes the daemon's
// LOCAL_PEERCRED check see the right uid. No-op on Windows
// (both processes run as SYSTEM) and when --target-uid is 0.
if vncAgentTargetUID != 0 {
if err := dropAgentPrivileges(vncAgentTargetUID); err != nil {
return fmt.Errorf("drop privileges to uid %d: %w", vncAgentTargetUID, err)
}
}
if err := os.Remove(vncAgentSocket); err != nil && !os.IsNotExist(err) {
log.Debugf("remove stale socket %s: %v", vncAgentSocket, err)
}
ln, err := net.Listen("unix", vncAgentSocket)
if err != nil {
return fmt.Errorf("listen on %s: %w", vncAgentSocket, err)
}
if err := os.Chmod(vncAgentSocket, 0o600); err != nil {
log.Debugf("chmod %s: %v", vncAgentSocket, err)
}
ctx := cmd.Context()
capturer, injector, err := newAgentResources()
if err != nil {
_ = ln.Close()
return err
}
srv := vncserver.New(vncserver.Config{
Capturer: capturer,
Injector: injector,
DisableAuth: true,
AgentTokenHex: token,
Listener: ln,
})
if err := srv.Start(ctx, netip.AddrPort{}, netip.Prefix{}); err != nil {
return fmt.Errorf("start vnc server: %w", err)
}
log.Infof("vnc-agent listening on %s, ready", vncAgentSocket)
<-ctx.Done()
log.Info("vnc-agent context cancelled, shutting down")
return srv.Stop()
},
SilenceUsage: true,
}

View File

@@ -1,25 +0,0 @@
//go:build darwin && !ios
package cmd
import (
"fmt"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
func newAgentResources() (vncserver.ScreenCapturer, vncserver.InputInjector, error) {
// Ask for Screen Recording here and nowhere else. This process runs as the
// console user, which TCC requires for a user-scope service, and it is fresh
// per connection, which is what makes the dialog appear at all: TCC shows it
// once per process. The request blocks until the user answers, so it also
// keeps the Accessibility ask that follows the first input out of its way.
vncserver.RequestScreenRecording()
capturer := vncserver.NewMacPoller()
injector, err := vncserver.NewMacInputInjector()
if err != nil {
return nil, nil, fmt.Errorf("macOS input injector: %w", err)
}
return capturer, injector, nil
}

View File

@@ -1,77 +0,0 @@
//go:build darwin && !ios
package cmd
import (
"fmt"
"os"
"os/user"
"strconv"
"syscall"
)
// dropAgentPrivileges drops the vnc-agent process from root (its
// launchctl-asuser-inherited starting uid) to the target console user
// before any other initialisation runs. Without this the agent runs as
// root for the lifetime of the session; any post-auth memory-safety
// issue in the capture/input/encode paths would then be a root-level
// RCE on the host instead of a user-level one. Also makes the daemon's
// LOCAL_PEERCRED check correctly identify the agent as the console user,
// not as root.
//
// Returns an error when the agent is running as a non-root uid that
// differs from targetUID: non-root can only setuid to itself, so a
// mismatch here means the spawn went to the wrong session.
func dropAgentPrivileges(targetUID uint32) error {
if targetUID == 0 {
return fmt.Errorf("refusing to keep agent running as root (target uid 0)")
}
cur := uint32(os.Getuid())
if cur == targetUID {
return nil
}
if cur != 0 {
return fmt.Errorf("agent uid %d does not match expected %d and we lack root to fix it", cur, targetUID)
}
// Resolve the target user's real primary group rather than reusing
// targetUID as the gid: a user's primary group on macOS is typically
// staff(20), not gid==uid. Fail closed if the lookup fails.
targetGID, err := primaryGroupID(targetUID)
if err != nil {
return err
}
// Drop supplementary groups first: setgid alone doesn't touch the
// auxiliary group list, leaving root's groups attached would let the
// dropped process write to root-only group-writable files.
if err := syscall.Setgroups([]int{}); err != nil {
return fmt.Errorf("setgroups([]): %w", err)
}
if err := syscall.Setgid(targetGID); err != nil {
return fmt.Errorf("setgid(%d): %w", targetGID, err)
}
if os.Getgid() != targetGID || os.Getegid() != targetGID {
return fmt.Errorf("setgid verification: gid=%d egid=%d, expected %d", os.Getgid(), os.Getegid(), targetGID)
}
if err := syscall.Setuid(int(targetUID)); err != nil {
return fmt.Errorf("setuid(%d): %w", targetUID, err)
}
if uint32(os.Getuid()) != targetUID || uint32(os.Geteuid()) != targetUID {
return fmt.Errorf("setuid verification: uid=%d euid=%d, expected %d", os.Getuid(), os.Geteuid(), targetUID)
}
return nil
}
// primaryGroupID resolves the real primary group id of the user with the
// given uid. Fails closed: a lookup or parse error returns an error so the
// caller never falls back to using uid as the gid.
func primaryGroupID(targetUID uint32) (int, error) {
u, err := user.LookupId(strconv.Itoa(int(targetUID)))
if err != nil {
return 0, fmt.Errorf("look up uid %d: %w", targetUID, err)
}
gid, err := strconv.Atoi(u.Gid)
if err != nil {
return 0, fmt.Errorf("parse gid %q for uid %d: %w", u.Gid, targetUID, err)
}
return gid, nil
}

View File

@@ -1,55 +0,0 @@
//go:build darwin && !ios
package cmd
import (
"strings"
"testing"
)
// TestDropAgentPrivileges_RefusesRootTarget locks in the contract that
// dropAgentPrivileges must never be a no-op when asked to keep the
// agent as root (target uid 0). A future caller that passes 0 by
// mistake would otherwise leave the post-auth attack surface running
// with full root privileges.
func TestDropAgentPrivileges_RefusesRootTarget(t *testing.T) {
err := dropAgentPrivileges(0)
if err == nil {
t.Fatal("expected refusal for target uid 0, got nil")
}
if !strings.Contains(err.Error(), "root") {
t.Fatalf("error should mention root, got: %v", err)
}
}
// TestDropAgentPrivileges_NoOpWhenAlreadyTarget covers the dev path
// where the agent is launched by hand as the target user (no root
// available, no setuid needed). The helper must succeed silently
// instead of trying (and failing) a setuid to its current uid.
func TestDropAgentPrivileges_NoOpWhenAlreadyTarget(t *testing.T) {
// Skip when running as root: the early-return path we want to
// cover only fires when current uid == target uid.
uid := currentUIDForTest()
if uid == 0 {
t.Skip("test must not run as root; cannot exercise the no-op early-return")
}
if err := dropAgentPrivileges(uid); err != nil {
t.Fatalf("expected no-op when current uid == target, got: %v", err)
}
}
// TestDropAgentPrivileges_RefusesMismatchedNonRoot guards the "non-root
// caller tries to setuid to a different uid" path: setuid would fail
// with EPERM anyway, but the helper should surface a clear error
// before issuing the syscall so a misconfigured spawn (wrong --target-uid
// flag) is debuggable.
func TestDropAgentPrivileges_RefusesMismatchedNonRoot(t *testing.T) {
uid := currentUIDForTest()
if uid == 0 {
t.Skip("test must not run as root; covered case requires non-root caller")
}
err := dropAgentPrivileges(uid + 1)
if err == nil {
t.Fatal("expected refusal when non-root caller asks to setuid elsewhere")
}
}

View File

@@ -1,11 +0,0 @@
//go:build darwin && !ios
package cmd
import "os"
// currentUIDForTest exposes os.Getuid for the darwin dropprivs tests
// without leaking an os import into the test file itself.
func currentUIDForTest() uint32 {
return uint32(os.Getuid())
}

View File

@@ -1,14 +0,0 @@
//go:build windows
package cmd
// dropAgentPrivileges is a no-op on Windows: the agent and the daemon
// both run as SYSTEM (the daemon spawns the agent into the interactive
// session via CreateProcessAsUser with an impersonation token, but the
// resulting process still runs under SYSTEM, not under the user's
// account). The Windows path relies on the DACL-restricted socket
// directory, the unpredictable per-spawn socket name, the listen-readiness
// gate, and the per-spawn token for integrity instead.
func dropAgentPrivileges(_ uint32) error {
return nil
}

View File

@@ -1,15 +0,0 @@
//go:build windows
package cmd
import (
log "github.com/sirupsen/logrus"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
func newAgentResources() (vncserver.ScreenCapturer, vncserver.InputInjector, error) {
sessionID := vncserver.GetCurrentSessionID()
log.Infof("VNC agent running in Windows session %d", sessionID)
return vncserver.NewDesktopCapturer(), vncserver.NewWindowsInputInjector(), nil
}

View File

@@ -1,16 +0,0 @@
package cmd
const (
serverVNCAllowedFlag = "allow-server-vnc"
disableVNCApprovalFlag = "disable-vnc-approval"
)
var (
serverVNCAllowed bool
disableVNCApproval bool
)
func init() {
upCmd.PersistentFlags().BoolVar(&serverVNCAllowed, serverVNCAllowedFlag, false, "Allow embedded VNC server on peer")
upCmd.PersistentFlags().BoolVar(&disableVNCApproval, disableVNCApprovalFlag, false, "Disable per-connection user approval prompts for the embedded VNC server")
}

View File

@@ -11,30 +11,19 @@ import (
// bundle collector all share one definition.
const UILogFile = "gui-client.log"
var (
// StateDir holds persistent state (config, profiles, install metadata).
StateDir string
// RuntimeDir holds ephemeral artifacts that should not survive reboot,
// such as Unix sockets for daemon and per-session IPC. Empty on
// platforms without a conventional /var/run-style location.
RuntimeDir string
)
var StateDir string
func init() {
StateDir = os.Getenv("NB_STATE_DIR")
if StateDir != "" {
return
}
switch runtime.GOOS {
case "windows":
StateDir = filepath.Join(os.Getenv("PROGRAMDATA"), "Netbird")
case "darwin", "linux":
StateDir = "/var/lib/netbird"
RuntimeDir = "/var/run/netbird"
case "freebsd", "openbsd", "netbsd", "dragonfly":
StateDir = "/var/db/netbird"
RuntimeDir = "/var/run/netbird"
}
if v := os.Getenv("NB_STATE_DIR"); v != "" {
StateDir = v
}
if v := os.Getenv("NB_RUNTIME_DIR"); v != "" {
RuntimeDir = v
}
}

View File

@@ -2,8 +2,8 @@
// its wg interface into firewalld's "trusted" zone. This is required because
// firewalld's nftables chains are created with NFT_CHAIN_OWNER on recent
// versions, which returns EPERM to any other process that tries to insert
// rules into them. The workaround mirrors what Tailscale does: let firewalld
// itself add the accept rules to its own chains by trusting the interface.
// rules into them. Trusting the interface makes firewalld itself add the
// accept rules to its own chains instead.
package firewalld
// TrustedZone is the firewalld zone name used for interfaces whose traffic

View File

@@ -42,6 +42,7 @@ type aclManager struct {
optionalEntries map[string][]entry
ipsetStore *ipsetStore
v6 bool
ipsetSupported bool
stateManager *statemanager.Manager
}
@@ -60,6 +61,8 @@ func newAclManager(iptablesClient *iptables.IPTables, wgIface iFaceMapper) (*acl
func (m *aclManager) init(stateManager *statemanager.Manager) error {
m.stateManager = stateManager
m.ipsetSupported = m.probeIPSetSupport()
m.seedInitialEntries()
m.seedInitialOptionalEntries()
@@ -91,6 +94,12 @@ func (m *aclManager) AddPeerFiltering(
if m.v6 && ipsetName != "" {
ipsetName += "-v6"
}
// When the kernel lacks the required ipset hash module, fall back to
// per-IP iptables rules (pre-0.68 behavior) so ACLs keep working instead
// of silently leaving the chain empty.
if ipsetName != "" && !m.ipsetSupported {
ipsetName = ""
}
proto := protoForFamily(protocol, m.v6)
specs := filterRuleSpecs(ip, proto, sPort, dPort, action, ipsetName)
@@ -498,6 +507,40 @@ func transformIPsetName(ipsetName string, sPort, dPort *firewall.Port, action fi
}
}
// probeIPSetSupport checks whether the kernel can create the ipset type used for
// ACL rules. On kernels lacking the required ipset hash module, ipset creation
// fails (e.g. "invalid argument"), which would otherwise leave the ACL chain
// empty and silently drop all policy-permitted inbound traffic. When unsupported,
// the manager falls back to per-IP iptables rules.
func (m *aclManager) probeIPSetSupport() bool {
// Use a unique name so concurrent processes don't collide and we only ever
// destroy the set we created ourselves. ipset names are limited to 31 chars,
// so use a short random suffix.
probeName := "nb-probe-" + uuid.New().String()[:8]
opts := ipset.CreateOptions{
Replace: true,
}
if m.v6 {
opts.Family = ipset.FamilyIPV6
}
if err := ipset.Create(probeName, ipset.TypeHashNet, opts); err != nil {
log.Warnf("ipset is not available (failed to create probe set: %v); "+
"falling back to per-IP iptables ACL rules. Ensure the kernel provides "+
"the ipset hash:net module (ip_set_hash_net) for better performance with large rule sets", err)
return false
}
defer func() {
if err := ipset.Destroy(probeName); err != nil {
log.Debugf("destroy ipset probe set %q: %v", probeName, err)
}
}()
return true
}
func (m *aclManager) createIPSet(name string) error {
opts := ipset.CreateOptions{
Replace: true,

View File

@@ -0,0 +1,240 @@
//go:build privileged
package iptables
import (
"net/netip"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
fw "github.com/netbirdio/netbird/client/firewall/manager"
"github.com/netbirdio/netbird/client/iface"
"github.com/netbirdio/netbird/client/iface/wgaddr"
)
func iptRefcountIfaceV4() *iFaceMock {
return &iFaceMock{
NameFunc: func() string { return "wt-refcount" },
AddressFunc: func() wgaddr.Address {
return wgaddr.Address{
IP: netip.MustParseAddr("10.20.0.1"),
Network: netip.MustParsePrefix("10.20.0.0/24"),
}
},
}
}
func iptRefcountIfaceDual() *iFaceMock {
return &iFaceMock{
NameFunc: func() string { return "wt-refcount" },
AddressFunc: func() wgaddr.Address {
return wgaddr.Address{
IP: netip.MustParseAddr("10.20.0.1"),
Network: netip.MustParsePrefix("10.20.0.0/24"),
IPv6: netip.MustParseAddr("fd00::1"),
IPv6Net: netip.MustParsePrefix("fd00::/64"),
}
},
}
}
func newIptRefcountManager(t *testing.T, dual bool) *Manager {
t.Helper()
var ifMock *iFaceMock
if dual {
ifMock = iptRefcountIfaceDual()
} else {
ifMock = iptRefcountIfaceV4()
}
m, err := Create(ifMock, iface.DefaultMTU)
require.NoError(t, err, "create manager")
require.NoError(t, m.Init(nil), "init manager")
t.Cleanup(func() {
require.NoError(t, m.Close(nil), "close manager")
})
return m
}
func iptDnatV4(port uint16) fw.ForwardRule {
return fw.ForwardRule{
Protocol: fw.ProtocolTCP,
DestinationPort: fw.Port{Values: []uint16{port}},
TranslatedAddress: netip.MustParseAddr("10.20.0.2"),
TranslatedPort: fw.Port{Values: []uint16{80}},
}
}
func iptDnatV6(port uint16) fw.ForwardRule {
return fw.ForwardRule{
Protocol: fw.ProtocolTCP,
DestinationPort: fw.Port{Values: []uint16{port}},
TranslatedAddress: netip.MustParseAddr("fd00::2"),
TranslatedPort: fw.Port{Values: []uint16{80}},
}
}
// TestIptablesRouting_RepeatedEnableSingleReference verifies that EnableRouting
// (called on every network-map update) holds at most one reference per family
// and a single DisableRouting drops both back to zero.
func TestIptablesRouting_RepeatedEnableSingleReference(t *testing.T) {
m := newIptRefcountManager(t, true)
state := m.router.ipFwdState
require.NoError(t, m.EnableRouting(), "first enable")
require.NoError(t, m.EnableRouting(), "second enable")
require.NoError(t, m.EnableRouting(), "third enable")
v4, v6 := state.Counts()
assert.Equal(t, 1, v4, "repeated enable holds a single v4 reference")
assert.Equal(t, 1, v6, "repeated enable holds a single v6 reference")
require.NoError(t, m.DisableRouting(), "disable")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "single disable releases the v4 reference")
assert.Equal(t, 0, v6, "single disable releases the v6 reference")
}
// TestIptablesRouting_DisableKeepsDNATReference verifies that an unpaired
// DisableRouting does not release references held by active DNAT rules.
func TestIptablesRouting_DisableKeepsDNATReference(t *testing.T) {
m := newIptRefcountManager(t, true)
state := m.router.ipFwdState
r1, err := m.AddDNATRule(iptDnatV6(9095))
require.NoError(t, err, "add v6 dnat")
require.NoError(t, m.DisableRouting(), "unpaired disable")
_, v6 := state.Counts()
assert.Equal(t, 1, v6, "DNAT-held reference survives unpaired DisableRouting")
require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat")
_, v6 = state.Counts()
assert.Equal(t, 0, v6, "delete releases the DNAT reference")
}
// TestIptablesDNAT_RefcountBalancedV4 covers a Balanced Add/Delete pair on v4.
func TestIptablesDNAT_RefcountBalancedV4(t *testing.T) {
m := newIptRefcountManager(t, false)
state := m.router.ipFwdState
r1, err := m.AddDNATRule(iptDnatV4(7081))
require.NoError(t, err, "add v4 dnat 1")
v4, v6 := state.Counts()
assert.Equal(t, 1, v4, "v4 refcount after first add")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
r2, err := m.AddDNATRule(iptDnatV4(7082))
require.NoError(t, err, "add v4 dnat 2")
v4, v6 = state.Counts()
assert.Equal(t, 2, v4, "v4 refcount after second add")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
require.NoError(t, m.DeleteDNATRule(r1))
v4, v6 = state.Counts()
assert.Equal(t, 1, v4, "v4 refcount after first delete")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
require.NoError(t, m.DeleteDNATRule(r2))
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "v4 refcount after second delete")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
}
// TestIptablesDNAT_RefcountBalancedV6 checks the v6 path increments v6 only and
// decrements back to zero.
func TestIptablesDNAT_RefcountBalancedV6(t *testing.T) {
m := newIptRefcountManager(t, true)
require.NotNil(t, m.router6, "v6 router")
require.Same(t, m.router.ipFwdState, m.router6.ipFwdState, "shared state")
state := m.router.ipFwdState
r1, err := m.AddDNATRule(iptDnatV6(9081))
require.NoError(t, err, "add v6 dnat 1")
v4, v6 := state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 1, v6, "v6 refcount after first add")
r2, err := m.AddDNATRule(iptDnatV6(9082))
require.NoError(t, err, "add v6 dnat 2")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "v4 refcount unchanged")
assert.Equal(t, 2, v6, "v6 refcount after second add")
require.NoError(t, m.DeleteDNATRule(r1))
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "v4 refcount unchanged")
assert.Equal(t, 1, v6, "v6 refcount after first delete")
require.NoError(t, m.DeleteDNATRule(r2))
v4, v6 = state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 0, v6, "v6 refcount after second delete")
}
// TestIptablesDNAT_DuplicateAddNoLeak verifies the duplicate-rule path returns
// without bumping the refcount.
func TestIptablesDNAT_DuplicateAddNoLeak(t *testing.T) {
m := newIptRefcountManager(t, true)
state := m.router.ipFwdState
rule := iptDnatV4(7083)
r1, err := m.AddDNATRule(rule)
require.NoError(t, err)
v4, _ := state.Counts()
assert.Equal(t, 1, v4)
_, err = m.AddDNATRule(rule)
require.NoError(t, err, "duplicate add")
v4, _ = state.Counts()
assert.Equal(t, 1, v4, "duplicate add must not increment")
require.NoError(t, m.DeleteDNATRule(r1))
v4, _ = state.Counts()
assert.Equal(t, 0, v4, "single delete must drop to zero")
}
// TestIptablesDNAT_DeleteMissingNoUnderflow verifies Delete on an unknown rule
// neither errors nor releases the refcount.
func TestIptablesDNAT_DeleteMissingNoUnderflow(t *testing.T) {
m := newIptRefcountManager(t, true)
state := m.router.ipFwdState
phantom := iptDnatV4(7099)
require.NoError(t, m.DeleteDNATRule(&phantom), "delete missing v4")
v4, v6 := state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 0, v6)
phantom6 := iptDnatV6(9099)
require.NoError(t, m.DeleteDNATRule(&phantom6), "delete missing v6")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 0, v6)
r1, err := m.AddDNATRule(iptDnatV4(7100))
require.NoError(t, err)
v4, _ = state.Counts()
assert.Equal(t, 1, v4, "real add still increments after phantom delete")
require.NoError(t, m.DeleteDNATRule(r1))
}
// TestIptablesDNAT_DoubleDeleteNoUnderflow verifies a second Delete on the same
// rule is a no-op.
func TestIptablesDNAT_DoubleDeleteNoUnderflow(t *testing.T) {
m := newIptRefcountManager(t, true)
state := m.router.ipFwdState
r1, err := m.AddDNATRule(iptDnatV6(9083))
require.NoError(t, err)
_, v6 := state.Counts()
assert.Equal(t, 1, v6)
require.NoError(t, m.DeleteDNATRule(r1), "first delete")
_, v6 = state.Counts()
assert.Equal(t, 0, v6)
require.NoError(t, m.DeleteDNATRule(r1), "second delete must be no-op")
_, v6 = state.Counts()
assert.Equal(t, 0, v6, "double delete must not underflow")
}

View File

@@ -89,7 +89,7 @@ func (m *Manager) createIPv6Components(wgIface iFaceMapper, mtu uint16) error {
}
// Share the same IP forwarding state with the v4 router, since
// EnableIPForwarding controls both v4 and v6 sysctls.
// Forwarding refcounter is per-family but shared between v4 and v6 routers.
m.router6.ipFwdState = m.router.ipFwdState
m.aclMgr6, err = newAclManager(ip6Client, wgIface)
@@ -402,17 +402,12 @@ func (m *Manager) SetLogLevel(log.Level) {
}
func (m *Manager) EnableRouting() error {
if err := m.router.ipFwdState.RequestForwarding(); err != nil {
return fmt.Errorf("enable IP forwarding: %w", err)
}
return nil
// v6 only when the overlay actually has v6.
return m.router.ipFwdState.RequestRouting(m.router6 != nil)
}
func (m *Manager) DisableRouting() error {
if err := m.router.ipFwdState.ReleaseForwarding(); err != nil {
return fmt.Errorf("disable IP forwarding: %w", err)
}
return nil
return m.router.ipFwdState.ReleaseRouting()
}
// AddDNATRule adds a DNAT rule

View File

@@ -291,3 +291,40 @@ func TestIptablesCreatePerformance(t *testing.T) {
})
}
}
// TestIptablesACLIPSetFallback verifies that when the kernel lacks ipset support,
// the ACL manager falls back to per-IP iptables rules (-s <ip>) instead of
// silently leaving the chain empty. See discussion #6125.
func TestIptablesACLIPSetFallback(t *testing.T) {
ipv4Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
require.NoError(t, err)
// Use Create()/Init() so the router-owned chains (chainRTFWDIN/OUT) are
// created before the ACL manager's createDefaultChains() references them.
manager, err := Create(ifaceMock, iface.DefaultMTU)
require.NoError(t, err)
require.NoError(t, manager.Init(nil))
aclMgr := manager.aclMgr
// Simulate a kernel without the ipset hash module.
aclMgr.ipsetSupported = false
defer func() {
require.NoError(t, manager.Close(nil))
}()
ip := netip.MustParseAddr("10.20.0.42")
port := &fw.Port{Values: []uint16{22}}
rules, err := aclMgr.AddPeerFiltering(nil, ip.AsSlice(), "tcp", nil, port, fw.ActionAccept, "nb0000001")
require.NoError(t, err, "AddPeerFiltering should succeed via fallback")
require.NotEmpty(t, rules)
rule := rules[0].(*Rule)
require.Empty(t, rule.ipsetName, "fallback rule must not reference an ipset")
require.Contains(t, strings.Join(rule.specs, " "), "-s 10.20.0.42", "fallback rule must match by source IP")
require.NotContains(t, strings.Join(rule.specs, " "), "--match-set", "fallback rule must not use ipset matching")
// The rule must actually be present in the ACL chain (not silently dropped).
checkRuleSpecs(t, ipv4Client, rule.chain, true, rule.specs...)
}

View File

@@ -102,7 +102,7 @@ func newRouter(iptablesClient *iptables.IPTables, wgIface iFaceMapper, mtu uint1
wgIface: wgIface,
mtu: mtu,
v6: iptablesClient.Proto() == iptables.ProtocolIPv6,
ipFwdState: ipfwdstate.NewIPForwardingState(),
ipFwdState: ipfwdstate.NewIPForwardingState(wgIface.Name()),
}
r.ipsetCounter = refcounter.New(
@@ -770,10 +770,6 @@ func (r *router) updateState() {
}
func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
if err := r.ipFwdState.RequestForwarding(); err != nil {
return nil, err
}
ruleKey := rule.ID()
if _, exists := r.rules[ruleKey+dnatSuffix]; exists {
return rule, nil
@@ -840,18 +836,34 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
for key, ruleInfo := range rules {
if err := r.iptablesClient.Append(ruleInfo.table, ruleInfo.chain, ruleInfo.rule...); err != nil {
if rollbackErr := r.rollbackRules(rules); rollbackErr != nil {
log.Errorf("rollback failed: %v", rollbackErr)
}
r.cleanupFailedDNATAdd(rules)
return nil, fmt.Errorf("add rule %s: %w", key, err)
}
r.rules[key] = ruleInfo.rule
}
if err := r.ipFwdState.RequestForwarding(r.v6); err != nil {
r.cleanupFailedDNATAdd(rules)
return nil, fmt.Errorf("enable forwarding: %w", err)
}
r.updateState()
return rule, nil
}
// cleanupFailedDNATAdd removes the bookkeeping written by a partially applied
// AddDNATRule before rolling back the kernel rules, so no entries remain that
// never got a forwarding refcount. rollbackRules re-adds entries it failed to
// remove from the kernel.
func (r *router) cleanupFailedDNATAdd(rules map[string]ruleInfo) {
for key := range rules {
delete(r.rules, key)
}
if err := r.rollbackRules(rules); err != nil {
log.Errorf("rollback failed: %v", err)
}
}
func (r *router) rollbackRules(rules map[string]ruleInfo) error {
var merr *multierror.Error
for key, ruleInfo := range rules {
@@ -868,32 +880,47 @@ func (r *router) rollbackRules(rules map[string]ruleInfo) error {
}
func (r *router) DeleteDNATRule(rule firewall.Rule) error {
if err := r.ipFwdState.ReleaseForwarding(); err != nil {
log.Errorf("%v", err)
}
ruleKey := rule.ID()
_, hadDNAT := r.rules[ruleKey+dnatSuffix]
_, hadSNAT := r.rules[ruleKey+snatSuffix]
_, hadFWD := r.rules[ruleKey+fwdSuffix]
if !hadDNAT && !hadSNAT && !hadFWD {
return nil
}
var merr *multierror.Error
if dnatRule, exists := r.rules[ruleKey+dnatSuffix]; exists {
if err := r.iptablesClient.Delete(tableNat, chainRTRDR, dnatRule...); err != nil {
merr = multierror.Append(merr, fmt.Errorf("delete DNAT rule: %w", err))
} else {
delete(r.rules, ruleKey+dnatSuffix)
}
delete(r.rules, ruleKey+dnatSuffix)
}
if snatRule, exists := r.rules[ruleKey+snatSuffix]; exists {
if err := r.iptablesClient.Delete(tableNat, chainRTNAT, snatRule...); err != nil {
merr = multierror.Append(merr, fmt.Errorf("delete SNAT rule: %w", err))
} else {
delete(r.rules, ruleKey+snatSuffix)
}
delete(r.rules, ruleKey+snatSuffix)
}
if fwdRule, exists := r.rules[ruleKey+fwdSuffix]; exists {
if err := r.iptablesClient.Delete(tableFilter, chainRTFWDOUT, fwdRule...); err != nil {
merr = multierror.Append(merr, fmt.Errorf("delete forward rule: %w", err))
} else {
delete(r.rules, ruleKey+fwdSuffix)
}
}
// Release the refcount only once all rules are gone from the kernel. On
// partial failure the failed entries stay in r.rules so a retry can remove
// them and release then.
if merr == nil {
if err := r.ipFwdState.ReleaseForwarding(r.v6); err != nil {
log.Errorf("%v", err)
}
delete(r.rules, ruleKey+fwdSuffix)
}
r.updateState()

View File

@@ -0,0 +1,249 @@
//go:build privileged
package nftables
import (
"net/netip"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
fw "github.com/netbirdio/netbird/client/firewall/manager"
"github.com/netbirdio/netbird/client/iface"
"github.com/netbirdio/netbird/client/iface/wgaddr"
)
func nftRefcountIfaceV4() *iFaceMock {
return &iFaceMock{
NameFunc: func() string { return "wt-refcount" },
AddressFunc: func() wgaddr.Address {
return wgaddr.Address{
IP: netip.MustParseAddr("100.96.0.1"),
Network: netip.MustParsePrefix("100.96.0.0/16"),
}
},
}
}
func nftRefcountIfaceDual() *iFaceMock {
return &iFaceMock{
NameFunc: func() string { return "wt-refcount" },
AddressFunc: func() wgaddr.Address {
return wgaddr.Address{
IP: netip.MustParseAddr("100.96.0.1"),
Network: netip.MustParsePrefix("100.96.0.0/16"),
IPv6: netip.MustParseAddr("fd00::1"),
IPv6Net: netip.MustParsePrefix("fd00::/64"),
}
},
}
}
func newNftRefcountManager(t *testing.T, dual bool) *Manager {
t.Helper()
if check() != NFTABLES {
t.Skip("nftables not supported on this system")
}
var ifMock *iFaceMock
if dual {
ifMock = nftRefcountIfaceDual()
} else {
ifMock = nftRefcountIfaceV4()
}
m, err := Create(ifMock, iface.DefaultMTU)
require.NoError(t, err, "create manager")
require.NoError(t, m.Init(nil), "init manager")
t.Cleanup(func() {
require.NoError(t, m.Close(nil), "close manager")
})
return m
}
func dnatV4(port uint16) fw.ForwardRule {
return fw.ForwardRule{
Protocol: fw.ProtocolTCP,
DestinationPort: fw.Port{Values: []uint16{port}},
TranslatedAddress: netip.MustParseAddr("100.96.0.2"),
TranslatedPort: fw.Port{Values: []uint16{80}},
}
}
func dnatV6(port uint16) fw.ForwardRule {
return fw.ForwardRule{
Protocol: fw.ProtocolTCP,
DestinationPort: fw.Port{Values: []uint16{port}},
TranslatedAddress: netip.MustParseAddr("fd00::2"),
TranslatedPort: fw.Port{Values: []uint16{80}},
}
}
// TestNftablesDNAT_RefcountBalancedV4 verifies that Add/Delete pairs leave the
// v4 refcount at zero.
func TestNftablesDNAT_RefcountBalancedV4(t *testing.T) {
m := newNftRefcountManager(t, false)
state := m.router.ipFwdState
r1, err := m.AddDNATRule(dnatV4(8081))
require.NoError(t, err, "add v4 dnat 1")
v4, v6 := state.Counts()
assert.Equal(t, 1, v4, "v4 refcount after first add")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
r2, err := m.AddDNATRule(dnatV4(8082))
require.NoError(t, err, "add v4 dnat 2")
v4, v6 = state.Counts()
assert.Equal(t, 2, v4, "v4 refcount after second add")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
require.NoError(t, m.DeleteDNATRule(r1), "delete v4 dnat 1")
v4, v6 = state.Counts()
assert.Equal(t, 1, v4, "v4 refcount after first delete")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
require.NoError(t, m.DeleteDNATRule(r2), "delete v4 dnat 2")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "v4 refcount after second delete")
assert.Equal(t, 0, v6, "v6 refcount unchanged")
}
// TestNftablesDNAT_RefcountBalancedV6 verifies the v6 path increments v6 only
// and decrements back to zero on Delete.
func TestNftablesDNAT_RefcountBalancedV6(t *testing.T) {
m := newNftRefcountManager(t, true)
require.NotNil(t, m.router6, "v6 router")
require.Same(t, m.router.ipFwdState, m.router6.ipFwdState, "shared state")
state := m.router.ipFwdState
r1, err := m.AddDNATRule(dnatV6(9091))
require.NoError(t, err, "add v6 dnat 1")
v4, v6 := state.Counts()
assert.Equal(t, 0, v4, "v4 refcount unchanged")
assert.Equal(t, 1, v6, "v6 refcount after first add")
r2, err := m.AddDNATRule(dnatV6(9092))
require.NoError(t, err, "add v6 dnat 2")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 2, v6, "v6 refcount after second add")
require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat 1")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "v4 refcount unchanged")
assert.Equal(t, 1, v6, "v6 refcount after first delete")
require.NoError(t, m.DeleteDNATRule(r2), "delete v6 dnat 2")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 0, v6, "v6 refcount after second delete")
}
// TestNftablesDNAT_DuplicateAddNoLeak verifies that a duplicate Add (same
// ForwardRule) does not double-increment the refcount.
func TestNftablesDNAT_DuplicateAddNoLeak(t *testing.T) {
m := newNftRefcountManager(t, true)
state := m.router.ipFwdState
rule := dnatV4(8083)
r1, err := m.AddDNATRule(rule)
require.NoError(t, err, "add v4 dnat")
v4, _ := state.Counts()
assert.Equal(t, 1, v4)
// duplicate add: same rule ID, must be a no-op for the refcount.
_, err = m.AddDNATRule(rule)
require.NoError(t, err, "duplicate add")
v4, _ = state.Counts()
assert.Equal(t, 1, v4, "duplicate add must not increment")
require.NoError(t, m.DeleteDNATRule(r1), "delete v4 dnat")
v4, _ = state.Counts()
assert.Equal(t, 0, v4, "single delete must drop to zero")
}
// TestNftablesDNAT_DeleteMissingNoUnderflow verifies deleting a rule that was
// never added does not underflow the refcount.
func TestNftablesDNAT_DeleteMissingNoUnderflow(t *testing.T) {
m := newNftRefcountManager(t, true)
state := m.router.ipFwdState
// Construct a Rule reference for something never added. The router stores
// rules by ID(), and DeleteDNATRule looks them up in r.rules; a missing
// entry must be a no-op rather than calling Release.
phantom := dnatV4(8099)
require.NoError(t, m.DeleteDNATRule(&phantom), "delete missing v4 dnat")
v4, v6 := state.Counts()
assert.Equal(t, 0, v4, "v4 refcount unaffected by missing delete")
assert.Equal(t, 0, v6, "v6 refcount unaffected")
phantom6 := dnatV6(9099)
require.NoError(t, m.DeleteDNATRule(&phantom6), "delete missing v6 dnat")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4)
assert.Equal(t, 0, v6, "v6 refcount unaffected by missing delete")
// And after a phantom delete, a real add still results in count=1.
r1, err := m.AddDNATRule(dnatV4(8100))
require.NoError(t, err, "add v4 dnat after phantom delete")
v4, _ = state.Counts()
assert.Equal(t, 1, v4, "real add still increments after phantom delete")
require.NoError(t, m.DeleteDNATRule(r1))
}
// TestNftablesRouting_RepeatedEnableSingleReference verifies that EnableRouting
// (called on every network-map update) holds at most one reference per family
// and a single DisableRouting drops both back to zero.
func TestNftablesRouting_RepeatedEnableSingleReference(t *testing.T) {
m := newNftRefcountManager(t, true)
state := m.router.ipFwdState
require.NoError(t, m.EnableRouting(), "first enable")
require.NoError(t, m.EnableRouting(), "second enable")
require.NoError(t, m.EnableRouting(), "third enable")
v4, v6 := state.Counts()
assert.Equal(t, 1, v4, "repeated enable holds a single v4 reference")
assert.Equal(t, 1, v6, "repeated enable holds a single v6 reference")
require.NoError(t, m.DisableRouting(), "disable")
v4, v6 = state.Counts()
assert.Equal(t, 0, v4, "single disable releases the v4 reference")
assert.Equal(t, 0, v6, "single disable releases the v6 reference")
}
// TestNftablesRouting_DisableKeepsDNATReference verifies that an unpaired
// DisableRouting does not release references held by active DNAT rules.
func TestNftablesRouting_DisableKeepsDNATReference(t *testing.T) {
m := newNftRefcountManager(t, true)
state := m.router.ipFwdState
r1, err := m.AddDNATRule(dnatV6(9095))
require.NoError(t, err, "add v6 dnat")
require.NoError(t, m.DisableRouting(), "unpaired disable")
_, v6 := state.Counts()
assert.Equal(t, 1, v6, "DNAT-held reference survives unpaired DisableRouting")
require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat")
_, v6 = state.Counts()
assert.Equal(t, 0, v6, "delete releases the DNAT reference")
}
// TestNftablesDNAT_DoubleDeleteNoUnderflow verifies that deleting the same rule
// twice does not underflow the refcount (the second delete is a no-op).
func TestNftablesDNAT_DoubleDeleteNoUnderflow(t *testing.T) {
m := newNftRefcountManager(t, true)
state := m.router.ipFwdState
r1, err := m.AddDNATRule(dnatV6(9093))
require.NoError(t, err)
_, v6 := state.Counts()
assert.Equal(t, 1, v6)
require.NoError(t, m.DeleteDNATRule(r1), "first delete")
_, v6 = state.Counts()
assert.Equal(t, 0, v6)
require.NoError(t, m.DeleteDNATRule(r1), "second delete must be no-op")
_, v6 = state.Counts()
assert.Equal(t, 0, v6, "double delete must not underflow")
}

View File

@@ -105,8 +105,8 @@ func (m *Manager) createIPv6Components(tableName string, wgIface iFaceMapper, mt
return fmt.Errorf("create v6 router: %w", err)
}
// Share the same IP forwarding state with the v4 router, since
// EnableIPForwarding controls both v4 and v6 sysctls.
// Share the per-family forwarding refcounter with the v4 router so a v4
// rule and a v6 rule against the same state machine cooperate cleanly.
m.router6.ipFwdState = m.router.ipFwdState
m.aclManager6, err = newAclManager(workTable6, wgIface, chainNameRoutingFw)
@@ -530,17 +530,12 @@ func (m *Manager) SetLogLevel(log.Level) {
}
func (m *Manager) EnableRouting() error {
if err := m.router.ipFwdState.RequestForwarding(); err != nil {
return fmt.Errorf("enable IP forwarding: %w", err)
}
return nil
// v6 only when the overlay actually has v6.
return m.router.ipFwdState.RequestRouting(m.router6 != nil)
}
func (m *Manager) DisableRouting() error {
if err := m.router.ipFwdState.ReleaseForwarding(); err != nil {
return fmt.Errorf("disable IP forwarding: %w", err)
}
return nil
return m.router.ipFwdState.ReleaseRouting()
}
// Flush rule/chain/set operations from the buffer

View File

@@ -93,7 +93,7 @@ func newRouter(workTable *nftables.Table, wgIface iFaceMapper, mtu uint16) (*rou
rules: make(map[string]*nftables.Rule),
af: familyForAddr(workTable.Family == nftables.TableFamilyIPv4),
wgIface: wgIface,
ipFwdState: ipfwdstate.NewIPForwardingState(),
ipFwdState: ipfwdstate.NewIPForwardingState(wgIface.Name()),
mtu: mtu,
}
@@ -1553,10 +1553,6 @@ func (r *router) refreshRulesMap() error {
}
func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
if err := r.ipFwdState.RequestForwarding(); err != nil {
return nil, err
}
ruleKey := rule.ID()
if _, exists := r.rules[ruleKey+dnatSuffix]; exists {
return rule, nil
@@ -1567,7 +1563,18 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
return nil, fmt.Errorf("convert protocol to number: %w", err)
}
// Request forwarding before queueing rules: addDnatRedirect/addDnatMasq
// buffer netlink messages on r.conn that the next caller's Flush would
// commit if we returned without flushing them ourselves.
v6 := r.af.tableFamily == nftables.TableFamilyIPv6
if err := r.ipFwdState.RequestForwarding(v6); err != nil {
return nil, fmt.Errorf("enable forwarding: %w", err)
}
if err := r.addDnatRedirect(rule, protoNum, ruleKey); err != nil {
if rerr := r.ipFwdState.ReleaseForwarding(v6); rerr != nil {
log.Warnf("rollback forwarding refcount: %v", rerr)
}
return nil, err
}
@@ -1579,6 +1586,11 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
// TODO: find chains with drop policies and add rules there
if err := r.conn.Flush(); err != nil {
if rerr := r.ipFwdState.ReleaseForwarding(v6); rerr != nil {
log.Warnf("rollback forwarding refcount: %v", rerr)
}
delete(r.rules, ruleKey+dnatSuffix)
delete(r.rules, ruleKey+snatSuffix)
return nil, fmt.Errorf("flush rules: %w", err)
}
@@ -1781,16 +1793,18 @@ func (r *router) addDnatMasq(rule firewall.ForwardRule, protoNum uint8, ruleKey
}
func (r *router) DeleteDNATRule(rule firewall.Rule) error {
if err := r.ipFwdState.ReleaseForwarding(); err != nil {
log.Errorf("%v", err)
}
ruleKey := rule.ID()
if err := r.refreshRulesMap(); err != nil {
return fmt.Errorf(refreshRulesMapError, err)
}
_, hadDNAT := r.rules[ruleKey+dnatSuffix]
_, hadSNAT := r.rules[ruleKey+snatSuffix]
if !hadDNAT && !hadSNAT {
return nil
}
var merr *multierror.Error
var needsFlush bool
@@ -1822,9 +1836,16 @@ func (r *router) DeleteDNATRule(rule firewall.Rule) error {
}
}
// Release the refcount only once the rules are gone from the kernel. On
// failure (including the refreshRulesMap error above) the rules and their
// map entries remain, keeping forwarding on until a retry removes them.
if merr == nil {
delete(r.rules, ruleKey+dnatSuffix)
delete(r.rules, ruleKey+snatSuffix)
if err := r.ipFwdState.ReleaseForwarding(r.af.tableFamily == nftables.TableFamilyIPv6); err != nil {
log.Errorf("%v", err)
}
}
return nberrors.FormatErrorOrNil(merr)

View File

@@ -22,6 +22,16 @@ import (
nbnet "github.com/netbirdio/netbird/client/net"
)
const (
// wgMsgTypeHandshakeInitiation is the lowest WireGuard message type.
wgMsgTypeHandshakeInitiation uint32 = 1
// wgMsgTypeTransport is the highest WireGuard message type.
wgMsgTypeTransport uint32 = 4
// wgMinMsgSize is the smallest WireGuard message: transport data with an empty
// payload, which is what a keepalive is.
wgMinMsgSize = 32
)
type receiverCreator struct {
iceBind *ICEBind
}
@@ -216,8 +226,15 @@ func (s *ICEBind) createReceiverFn(pc wgConn.BatchReader, conn *net.UDPConn, rxO
for i := 0; i < numMsgs; i++ {
msg := &(*msgs)[i]
// todo: handle err
if ok, _ := s.filterOutStunMessages(msg.Buffers, msg.N, msg.Addr); ok {
if ok, err := s.filterOutStunMessages(msg.Buffers, msg.N, msg.Addr); ok {
if err != nil {
log.Debugf("failed to handle STUN packet from %s: %v", msg.Addr, err)
}
// WireGuard reuses sizes and eps across reads and only skips a slot
// whose size is below the minimum message size. Leaving a consumed
// slot untouched makes it process this buffer again under the
// previous packet's length and endpoint.
sizes[i] = 0
continue
}
sizes[i] = msg.N
@@ -271,11 +288,16 @@ func (s *ICEBind) createOrUpdateMux() {
func (s *ICEBind) filterOutStunMessages(buffers [][]byte, n int, addr net.Addr) (bool, error) {
for i := range buffers {
if !stun.IsMessage(buffers[i]) {
if n > len(buffers[i]) {
continue
}
pkt := buffers[i][:n]
if isWireGuardMsg(pkt) || !stun.IsMessage(pkt) {
continue
}
msg, err := s.parseSTUNMessage(buffers[i][:n])
msg, err := s.parseSTUNMessage(pkt)
if err != nil {
buffers[i] = []byte{}
return true, err
@@ -347,18 +369,34 @@ func putMessages(msgs *[]ipv6.Message, msgsPool *sync.Pool) {
msgsPool.Put(msgs)
}
func isTransportPkg(buffers [][]byte, n int) bool {
// The first buffer should contain at least 4 bytes for type
if len(buffers[0]) < 4 {
return true
// isWireGuardMsg reports whether the packet carries a WireGuard message header: a
// little-endian uint32 message type in the range 1..4, which leaves the three bytes
// after the type byte zero, in a packet long enough to hold any WireGuard message.
//
// A well formed STUN message cannot take that shape. Its length field sits in the two
// bytes the type must leave zero, and for a message of at least wgMinMsgSize bytes that
// field holds at least 12, so the two framings do not overlap. The test has to be this
// tight because stun.IsMessage only looks at the magic cookie, which in a WireGuard
// message overlaps the receiver index: a session whose index happens to equal the cookie
// would otherwise have all of its inbound data misrouted to the STUN handler until the
// next rekey.
func isWireGuardMsg(pkt []byte) bool {
if len(pkt) < wgMinMsgSize {
return false
}
// WireGuard packet type is a little-endian uint32 at start
packetType := binary.LittleEndian.Uint32(buffers[0][:4])
// Check if packetType matches known WireGuard message types
if packetType == 4 && n > 32 {
return true
}
return false
msgType := binary.LittleEndian.Uint32(pkt[:4])
return msgType >= wgMsgTypeHandshakeInitiation && msgType <= wgMsgTypeTransport
}
// isTransportPkg reports whether the packet is WireGuard transport data carrying a
// payload, which is what counts as peer activity. A keepalive holds no payload and is
// exactly wgMinMsgSize bytes.
func isTransportPkg(buffers [][]byte, n int) bool {
if n < 4 || n > len(buffers[0]) {
return false
}
msgType := binary.LittleEndian.Uint32(buffers[0][:4])
return msgType == wgMsgTypeTransport && n > wgMinMsgSize
}

View File

@@ -0,0 +1,215 @@
//go:build !js
package bind
import (
"encoding/binary"
"net"
"testing"
"time"
"github.com/pion/stun/v3"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"golang.org/x/net/ipv4"
wgConn "golang.zx2c4.com/wireguard/conn"
)
// magicCookieBytes is the STUN magic cookie as it appears on the wire. In a
// WireGuard message the same offset holds the receiver (or sender) index, which is
// a random uint32, so a session can draw exactly this value.
var magicCookieBytes = []byte{0x21, 0x12, 0xA4, 0x42}
const testBufSize = 1500
// wgMsg builds a WireGuard message of the given type and size, with the index field
// at bytes 4:8 set to index.
func wgMsg(msgType uint32, size int, index []byte) []byte {
pkt := make([]byte, size)
binary.LittleEndian.PutUint32(pkt[:4], msgType)
copy(pkt[4:8], index)
return pkt
}
// intoBuffer copies pkt into a full-size receive buffer, the way the kernel read
// does, so tests see the same buffer/length split as the hot path.
func intoBuffer(pkt []byte) [][]byte {
buf := make([]byte, testBufSize)
copy(buf, pkt)
return [][]byte{buf}
}
func TestFilterOutStunMessages_PassesWireGuardWithCookieShapedIndex(t *testing.T) {
tests := []struct {
name string
msgType uint32
size int
}{
{"transport data", wgMsgTypeTransport, 128},
{"keepalive", wgMsgTypeTransport, wgMinMsgSize},
{"handshake initiation", wgMsgTypeHandshakeInitiation, 148},
{"handshake response", 2, 92},
{"cookie reply", 3, 64},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
pkt := wgMsg(tc.msgType, tc.size, magicCookieBytes)
require.True(t, stun.IsMessage(pkt), "precondition: pion sees this as STUN")
buffers := intoBuffer(pkt)
bind := &ICEBind{}
filtered, err := bind.filterOutStunMessages(buffers, tc.size, &net.UDPAddr{})
assert.NoError(t, err)
assert.False(t, filtered, "WireGuard message must be handed to WireGuard, not the STUN handler")
assert.Len(t, buffers[0], testBufSize, "buffer must be left intact for WireGuard")
})
}
}
func TestFilterOutStunMessages_FiltersRealSTUNMessage(t *testing.T) {
msg, err := stun.Build(stun.BindingRequest, stun.TransactionID, stun.Fingerprint)
require.NoError(t, err)
buffers := intoBuffer(msg.Raw)
bind := &ICEBind{}
filtered, err := bind.filterOutStunMessages(buffers, len(msg.Raw), &net.UDPAddr{})
assert.NoError(t, err)
assert.True(t, filtered, "STUN message must be consumed by the STUN handler")
assert.Empty(t, buffers[0], "consumed buffer must be emptied so WireGuard does not see it")
}
// TestIsWireGuardMsg_DisjointFromSTUN locks the invariant the filter relies on: a
// well formed STUN message long enough to be a WireGuard message always has a
// non-zero length field, so it cannot be mistaken for a WireGuard header.
func TestIsWireGuardMsg_DisjointFromSTUN(t *testing.T) {
types := []stun.MessageType{
stun.BindingRequest,
stun.BindingSuccess,
stun.BindingError,
{Method: stun.MethodBinding, Class: stun.ClassIndication},
}
for _, msgType := range types {
// Long enough that the length guard is not what makes this pass.
msg, err := stun.Build(msgType, stun.TransactionID,
stun.NewUsername("remoteUfrag:localUfrag"), stun.Fingerprint)
require.NoError(t, err)
require.GreaterOrEqual(t, len(msg.Raw), wgMinMsgSize, "precondition: %s", msgType)
assert.False(t, isWireGuardMsg(msg.Raw),
"%s must not look like a WireGuard message", msgType)
}
}
func TestIsWireGuardMsg(t *testing.T) {
tests := []struct {
name string
pkt []byte
want bool
}{
{"transport data", wgMsg(wgMsgTypeTransport, 128, nil), true},
{"handshake initiation", wgMsg(wgMsgTypeHandshakeInitiation, 148, nil), true},
{"unknown type 5", wgMsg(5, 128, nil), false},
{"type 0", wgMsg(0, 128, nil), false},
{"non-zero reserved byte", []byte{0x04, 0x00, 0x01, 0x00}, false},
{"too short", []byte{0x04, 0x00, 0x00}, false},
{"empty", nil, false},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
assert.Equal(t, tc.want, isWireGuardMsg(tc.pkt), "wrong classification for %s", tc.name)
})
}
}
// TestFilterOutStunMessages_IgnoresBytesBeyondPacket guards against classifying on
// buffer contents left over from an earlier, longer packet.
func TestFilterOutStunMessages_IgnoresBytesBeyondPacket(t *testing.T) {
buf := make([]byte, testBufSize)
copy(buf[4:8], magicCookieBytes)
buffers := [][]byte{buf}
bind := &ICEBind{}
filtered, err := bind.filterOutStunMessages(buffers, 2, &net.UDPAddr{})
assert.NoError(t, err)
assert.False(t, filtered, "a 2 byte packet must not be classified from stale buffer bytes")
}
// TestReceiveFn_ClearsSizeOfConsumedPacket covers the accounting WireGuard relies
// on: sizes is reused across reads, so a slot whose packet was consumed as STUN must
// be reported as empty. Otherwise WireGuard reprocesses the same buffer under the
// previous packet's length, which for a WireGuard-shaped packet means it is handled
// twice.
func TestReceiveFn_ClearsSizeOfConsumedPacket(t *testing.T) {
conn := listenUDP(t, "udp4", "127.0.0.1:0")
defer conn.Close()
recvFn := receiverCreator{setupICEBind(t)}.CreateReceiverFn(
ipv4.NewPacketConn(conn), conn, false, createMsgPool(),
)
msg, err := stun.Build(stun.BindingRequest, stun.TransactionID, stun.Fingerprint)
require.NoError(t, err)
sender := listenUDP(t, "udp4", "127.0.0.1:0")
defer sender.Close()
_, err = sender.WriteTo(msg.Raw, conn.LocalAddr())
require.NoError(t, err)
require.NoError(t, conn.SetReadDeadline(time.Now().Add(3*time.Second)))
bufs := [][]byte{make([]byte, 1500)}
// A leftover size from an earlier read, which is what makes the missing reset
// observable.
sizes := []int{148}
eps := make([]wgConn.Endpoint, 1)
n, err := recvFn(bufs, sizes, eps)
require.NoError(t, err)
require.Equal(t, 1, n)
assert.Zero(t, sizes[0], "consumed STUN packet must not leave a size behind for WireGuard")
}
func TestIsTransportPkg(t *testing.T) {
tests := []struct {
name string
pkt []byte
n int
want bool
}{
{"transport data with payload", wgMsg(wgMsgTypeTransport, 128, nil), 128, true},
{"keepalive", wgMsg(wgMsgTypeTransport, wgMinMsgSize, nil), wgMinMsgSize, false},
{"handshake initiation", wgMsg(wgMsgTypeHandshakeInitiation, 148, nil), 148, false},
{"stale type bytes beyond packet", wgMsg(wgMsgTypeTransport, 128, nil), 2, false},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
assert.Equal(t, tc.want, isTransportPkg(intoBuffer(tc.pkt), tc.n),
"wrong activity classification for %s", tc.name)
})
}
}
// TestFilterOutStunMessages_ConsumesSTUNWithWireGuardShapedType covers the one STUN
// encoding whose leading bytes collide with a WireGuard message type: method 0x080 as a
// request encodes to 0x0200, so the type byte reads as a handshake response and the byte
// after it is zero. Only the length check keeps such a message out of WireGuard's hands.
// pion implements no method in that range, so this is a synthetic worst case rather than
// traffic ICE produces.
func TestFilterOutStunMessages_ConsumesSTUNWithWireGuardShapedType(t *testing.T) {
msg, err := stun.Build(stun.NewType(stun.Method(0x080), stun.ClassRequest), stun.TransactionID)
require.NoError(t, err)
require.Equal(t, []byte{0x02, 0x00, 0x00, 0x00}, msg.Raw[:4],
"precondition: the leading bytes read as a WireGuard message type")
buffers := intoBuffer(msg.Raw)
bind := &ICEBind{}
filtered, err := bind.filterOutStunMessages(buffers, len(msg.Raw), &net.UDPAddr{})
assert.NoError(t, err)
assert.True(t, filtered, "STUN message must be consumed despite its WireGuard-shaped type")
}

View File

@@ -1,219 +0,0 @@
// Package approval brokers per-attempt user-accept prompts for inbound
// remote access (VNC today, SSH and others in the future). A caller pushes
// a Prompt; the broker emits a SystemEvent on the daemon→UI stream and
// blocks until the UI calls the daemon's RespondApproval RPC, the per-
// request timeout fires, or no subscriber is connected. The latter case
// fails closed so a backgrounded UI cannot silently bypass the gate.
package approval
import (
"context"
"errors"
"fmt"
"sync"
"time"
"github.com/google/uuid"
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/client/proto"
)
// Metadata keys the broker reserves on the emitted SystemEvent. Callers
// should not set these themselves; values in Prompt.Metadata that collide
// are overwritten by the broker.
const (
MetaRequestID = "request_id"
MetaKind = "kind"
MetaExpiresAt = "expires_at"
)
// ShortKeyFingerprint formats a hex-encoded Noise_IK static pubkey as a
// short, eyeball-able fingerprint to display in the approval dialog.
// The dashboard-supplied display name attached to a SessionPubKey isn't
// cryptographically asserted by the connecting client, so the prompt
// must also show something that IS: the key fingerprint, a hash of
// the static public key the client just proved possession of during the
// Noise handshake. Returns the empty string when the input is too short
// to plausibly be a hex pubkey, so the row is omitted rather than
// rendered as a misleading partial.
//
// Output format: 16 hex chars grouped as XXXX-XXXX-XXXX-XXXX (64 bits of
// fingerprint, resistant to random-prefix collisions and easy for a human
// to compare with an out-of-band reference).
func ShortKeyFingerprint(hexKey string) string {
if len(hexKey) < 8 {
return ""
}
src := hexKey
if len(src) > 16 {
src = src[:16]
}
var out []byte
for i, c := range src {
if i > 0 && i%4 == 0 {
out = append(out, '-')
}
out = append(out, byte(c))
}
return string(out)
}
// Kind values for the well-known prompt subjects. New subsystems should
// add a constant here so the UI can dispatch on a known string.
const (
KindVNC = "vnc"
KindSSH = "ssh"
)
// DefaultTimeout is the wall-clock window the user has to accept or deny a
// pending approval before the broker fails closed and returns ErrTimeout.
// Kept well under typical VNC client and dashboard connection timeouts so
// the RFB rejection actually reaches the browser instead of racing the
// browser's own "connection timed out" message.
const DefaultTimeout = 15 * time.Second
// timeoutValue returns the active timeout. It's a var so tests in this
// package can shorten the wait without exposing a setter on the public
// API. Production code always sees DefaultTimeout.
var timeoutValue = func() time.Duration { return DefaultTimeout }
// ErrNoSubscriber indicates no UI is connected to consume the prompt.
// The caller must reject the underlying connection (fail-closed).
var ErrNoSubscriber = errors.New("no UI subscriber connected for approval")
// ErrTimeout indicates the user did not respond within DefaultTimeout.
var ErrTimeout = errors.New("approval timed out")
// ErrDenied indicates the user explicitly denied the connection.
var ErrDenied = errors.New("approval denied")
// EventPublisher is the subset of peer.Status used to emit prompts.
type EventPublisher interface {
PublishEvent(
severity proto.SystemEvent_Severity,
category proto.SystemEvent_Category,
msg string,
userMsg string,
metadata map[string]string,
)
HasEventSubscribers() bool
}
// Prompt describes the pending request shown to the user. Kind selects
// the UI dispatch path (e.g. "vnc", "ssh"). Subject is the human-readable
// one-liner the UI may show as a title or notification body. Metadata is
// passed through verbatim and is the subsystem-specific payload (peer
// name, source IP, mode, etc.).
type Prompt struct {
Kind string
Subject string
Metadata map[string]string
}
// Decision carries the user's response to an approval prompt. ViewOnly is
// only meaningful when Accept is true; it lets the host grant the
// connection but signal the requester that input control is withheld.
type Decision struct {
Accept bool
ViewOnly bool
}
// Broker holds in-flight approval requests keyed by request ID.
type Broker struct {
pub EventPublisher
mu sync.Mutex
pending map[string]chan Decision
}
// New returns a broker that publishes prompts via pub.
func New(pub EventPublisher) *Broker {
return &Broker{
pub: pub,
pending: make(map[string]chan Decision),
}
}
// Request emits a SystemEvent for p and blocks until the UI calls Respond,
// ctx is cancelled, or DefaultTimeout elapses. Returns a Decision when
// the user replied; ErrDenied / ErrTimeout / ErrNoSubscriber / ctx.Err
// otherwise. Callers must treat any non-nil error as a deny.
func (b *Broker) Request(ctx context.Context, p Prompt) (Decision, error) {
var zero Decision
if b == nil || b.pub == nil {
return zero, fmt.Errorf("approval broker not configured")
}
if !b.pub.HasEventSubscribers() {
return zero, ErrNoSubscriber
}
id := uuid.NewString()
resp := make(chan Decision, 1)
b.mu.Lock()
b.pending[id] = resp
b.mu.Unlock()
defer b.dropPending(id)
timeout := timeoutValue()
expiresAt := time.Now().Add(timeout)
meta := make(map[string]string, len(p.Metadata)+3)
for k, v := range p.Metadata {
meta[k] = v
}
meta[MetaRequestID] = id
meta[MetaKind] = p.Kind
meta[MetaExpiresAt] = expiresAt.UTC().Format(time.RFC3339)
subject := p.Subject
if subject == "" {
subject = fmt.Sprintf("%s connection requires approval", p.Kind)
}
b.pub.PublishEvent(proto.SystemEvent_INFO, proto.SystemEvent_APPROVAL, subject, subject, meta)
log.Debugf("approval request %s (%s) emitted: %s", id, p.Kind, subject)
timer := time.NewTimer(timeout)
defer timer.Stop()
select {
case d := <-resp:
if !d.Accept {
return zero, ErrDenied
}
return d, nil
case <-timer.C:
return zero, ErrTimeout
case <-ctx.Done():
return zero, ctx.Err()
}
}
// Respond delivers the user's decision for id. Returns true when a pending
// request matched and was woken, false when id was unknown or already done.
func (b *Broker) Respond(id string, d Decision) bool {
if b == nil {
return false
}
b.mu.Lock()
ch, ok := b.pending[id]
if ok {
delete(b.pending, id)
}
b.mu.Unlock()
if !ok {
return false
}
select {
case ch <- d:
default:
}
return true
}
func (b *Broker) dropPending(id string) {
b.mu.Lock()
delete(b.pending, id)
b.mu.Unlock()
}

View File

@@ -1,434 +0,0 @@
package approval
import (
"context"
"sync"
"sync/atomic"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/netbirdio/netbird/client/proto"
)
// fakePublisher records published events and reports whether subscribers
// are connected. The subscribers flag is the security-critical signal:
// when false the broker must refuse to emit and the gate must fail closed.
type fakePublisher struct {
mu sync.Mutex
subscribers bool
events []*proto.SystemEvent
}
func (p *fakePublisher) PublishEvent(
severity proto.SystemEvent_Severity,
category proto.SystemEvent_Category,
msg string,
userMsg string,
metadata map[string]string,
) {
p.mu.Lock()
p.events = append(p.events, &proto.SystemEvent{
Severity: severity,
Category: category,
Message: msg,
UserMessage: userMsg,
Metadata: metadata,
})
p.mu.Unlock()
}
func (p *fakePublisher) HasEventSubscribers() bool {
p.mu.Lock()
defer p.mu.Unlock()
return p.subscribers
}
func (p *fakePublisher) lastEvent(t *testing.T) *proto.SystemEvent {
t.Helper()
p.mu.Lock()
defer p.mu.Unlock()
require.NotEmpty(t, p.events, "publisher saw no events")
return p.events[len(p.events)-1]
}
func (p *fakePublisher) eventCount() int {
p.mu.Lock()
defer p.mu.Unlock()
return len(p.events)
}
// TestRequestNoSubscriberFailsClosed is the core fail-closed invariant:
// when the UI is not subscribed, the broker must refuse without emitting
// an event or arming a waiter. A regression here is a silent bypass.
func TestRequestNoSubscriberFailsClosed(t *testing.T) {
pub := &fakePublisher{subscribers: false}
b := New(pub)
_, err := b.Request(context.Background(), Prompt{Kind: KindVNC, Subject: "test"})
assert.ErrorIs(t, err, ErrNoSubscriber)
assert.Equal(t, 0, pub.eventCount(), "no event must be emitted when fail-closed")
b.mu.Lock()
pending := len(b.pending)
b.mu.Unlock()
assert.Equal(t, 0, pending, "no waiter must be registered on fail-closed")
}
// TestRequestTimeoutDenies verifies that a request without a UI response
// returns ErrTimeout (deny) rather than nil (silent accept). Uses a short
// per-test broker timeout via Respond after the fact to keep the test fast.
func TestRequestTimeoutDenies(t *testing.T) {
// Replace DefaultTimeout for the lifetime of this test.
orig := DefaultTimeout
defaultTimeout(t, 60*time.Millisecond)
defer defaultTimeout(t, orig)
pub := &fakePublisher{subscribers: true}
b := New(pub)
start := time.Now()
_, err := b.Request(context.Background(), Prompt{Kind: KindVNC, Subject: "test"})
assert.ErrorIs(t, err, ErrTimeout, "missing user response must yield ErrTimeout, not nil")
assert.GreaterOrEqual(t, time.Since(start), 50*time.Millisecond, "timeout fired prematurely")
}
// TestRequestDenied returns ErrDenied when the UI responds with false.
func TestRequestDenied(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
var requestID string
done := make(chan error, 1)
go func() {
done <- requestErr(b, context.Background(), Prompt{Kind: KindVNC, Subject: "test"})
}()
requestID = waitForRequestID(t, pub)
require.True(t, b.Respond(requestID, Decision{Accept: false}))
select {
case err := <-done:
assert.ErrorIs(t, err, ErrDenied)
case <-time.After(time.Second):
t.Fatal("Request did not return after Respond(false)")
}
}
// TestRequestAccepted is the happy path. Failure here doesn't bypass the
// gate but breaks the feature.
func TestRequestAccepted(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
done := make(chan error, 1)
go func() {
done <- requestErr(b, context.Background(), Prompt{Kind: KindVNC, Subject: "test"})
}()
id := waitForRequestID(t, pub)
require.True(t, b.Respond(id, Decision{Accept: true}))
select {
case err := <-done:
assert.NoError(t, err)
case <-time.After(time.Second):
t.Fatal("Request did not return after Respond(true)")
}
}
// TestRequestCtxCancelDenies verifies that an upstream cancel (e.g. the
// engine shutting down mid-prompt) returns the cancel error rather than
// nil. A nil here would be a silent bypass on shutdown races.
func TestRequestCtxCancelDenies(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
ctx, cancel := context.WithCancel(context.Background())
done := make(chan error, 1)
go func() {
done <- requestErr(b, ctx, Prompt{Kind: KindVNC, Subject: "test"})
}()
// Wait until the prompt is in flight so cancel races a live waiter.
_ = waitForRequestID(t, pub)
cancel()
select {
case err := <-done:
assert.ErrorIs(t, err, context.Canceled)
case <-time.After(time.Second):
t.Fatal("Request did not return after ctx cancel")
}
}
// TestRespondUnknownIsNoop ensures a stray RespondApproval RPC cannot
// affect or accidentally accept any in-flight request whose id it doesn't
// match. Also confirms it doesn't panic.
func TestRespondUnknownIsNoop(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
// No in-flight prompts: Respond returns false.
assert.False(t, b.Respond("does-not-exist", Decision{Accept: true}))
// With an in-flight prompt, a wrong id still returns false and the
// prompt remains armed (eventually timing out as a deny).
defaultTimeout(t, 60*time.Millisecond)
defer defaultTimeout(t, DefaultTimeout)
done := make(chan error, 1)
go func() {
done <- requestErr(b, context.Background(), Prompt{Kind: KindVNC})
}()
realID := waitForRequestID(t, pub)
assert.False(t, b.Respond("totally-bogus", Decision{Accept: true}), "unknown id must not match")
assert.NotEqual(t, "totally-bogus", realID)
select {
case err := <-done:
assert.ErrorIs(t, err, ErrTimeout, "armed prompt must still time out, not accept")
case <-time.After(time.Second):
t.Fatal("prompt did not resolve")
}
}
// TestRespondAfterTimeoutNoop confirms a late accept response can't
// retroactively flip a denied (timed-out) request. The dropPending defer
// in Request must have removed the entry by the time Respond races in.
func TestRespondAfterTimeoutNoop(t *testing.T) {
defaultTimeout(t, 30*time.Millisecond)
defer defaultTimeout(t, DefaultTimeout)
pub := &fakePublisher{subscribers: true}
b := New(pub)
done := make(chan error, 1)
go func() {
done <- requestErr(b, context.Background(), Prompt{Kind: KindVNC})
}()
id := waitForRequestID(t, pub)
select {
case err := <-done:
require.ErrorIs(t, err, ErrTimeout)
case <-time.After(time.Second):
t.Fatal("prompt did not time out")
}
assert.False(t, b.Respond(id, Decision{Accept: true}), "late respond must be no-op")
}
// TestRespondDoubleNoop ensures a duplicate ack from the UI doesn't leak
// past the matched waiter or panic on a closed/full channel.
func TestRespondDoubleNoop(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
done := make(chan error, 1)
go func() {
done <- requestErr(b, context.Background(), Prompt{Kind: KindVNC})
}()
id := waitForRequestID(t, pub)
require.True(t, b.Respond(id, Decision{Accept: true}))
assert.False(t, b.Respond(id, Decision{Accept: false}), "second response must be no-op")
select {
case err := <-done:
assert.NoError(t, err)
case <-time.After(time.Second):
t.Fatal("prompt did not resolve")
}
}
// TestNilBrokerRequestErrors guards the engine pre-init path where the
// broker may not yet exist (or its publisher is nil): Request must
// error, never silently accept.
func TestNilBrokerRequestErrors(t *testing.T) {
var b *Broker
_, err := b.Request(context.Background(), Prompt{Kind: KindVNC})
assert.Error(t, err, "nil broker must error, never silently accept")
b2 := New(nil)
_, err = b2.Request(context.Background(), Prompt{Kind: KindVNC})
assert.Error(t, err, "broker with nil publisher must error, never silently accept")
}
// TestPromptMetadataInjected confirms the broker stamps request_id, kind,
// and expires_at on the emitted event. The UI relies on these keys; if
// they are dropped, the user cannot route the prompt and the response
// path breaks (which fails closed via timeout).
func TestPromptMetadataInjected(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
done := make(chan error, 1)
go func() {
done <- requestErr(b, context.Background(), Prompt{
Kind: KindVNC,
Subject: "VNC connection from peerA",
Metadata: map[string]string{"peer_name": "peerA"},
})
}()
id := waitForRequestID(t, pub)
ev := pub.lastEvent(t)
assert.Equal(t, proto.SystemEvent_APPROVAL, ev.Category)
assert.Equal(t, KindVNC, ev.Metadata[MetaKind])
assert.Equal(t, id, ev.Metadata[MetaRequestID])
assert.NotEmpty(t, ev.Metadata[MetaExpiresAt])
assert.Equal(t, "peerA", ev.Metadata["peer_name"], "caller metadata must pass through")
require.True(t, b.Respond(id, Decision{Accept: true}))
<-done
}
// TestConcurrentRequests verifies that two concurrent prompts are tracked
// independently. A bug that aliases ids would let one Respond unblock
// the wrong waiter (a silent accept across prompts).
func TestConcurrentRequests(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
const n = 20
results := make(chan error, n)
for i := 0; i < n; i++ {
go func() {
results <- requestErr(b, context.Background(), Prompt{Kind: KindVNC})
}()
}
ids := waitForNRequestIDs(t, pub, n)
require.Len(t, ids, n)
// Deny exactly half, accept the rest. Track outcome per id so we can
// match each Request's return value against the response we sent.
denySet := make(map[string]bool, n)
for i, id := range ids {
deny := i%2 == 0
denySet[id] = deny
require.True(t, b.Respond(id, Decision{Accept: !deny}))
}
// Collect all returns and check no nil errors slipped past a deny.
var accepted, denied atomic.Int32
for i := 0; i < n; i++ {
select {
case err := <-results:
if err == nil {
accepted.Add(1)
} else {
assert.ErrorIs(t, err, ErrDenied)
denied.Add(1)
}
case <-time.After(2 * time.Second):
t.Fatalf("only got %d/%d responses", i, n)
}
}
assert.Equal(t, int32(n/2), denied.Load())
assert.Equal(t, int32(n/2), accepted.Load())
}
// waitForRequestID blocks until the publisher sees its next event and
// returns the request_id stamped on it.
func waitForRequestID(t *testing.T, pub *fakePublisher) string {
t.Helper()
deadline := time.Now().Add(2 * time.Second)
for time.Now().Before(deadline) {
pub.mu.Lock()
count := len(pub.events)
var id string
if count > 0 {
id = pub.events[count-1].Metadata[MetaRequestID]
}
pub.mu.Unlock()
if id != "" {
return id
}
time.Sleep(2 * time.Millisecond)
}
t.Fatal("timeout waiting for emitted event")
return ""
}
func waitForNRequestIDs(t *testing.T, pub *fakePublisher, n int) []string {
t.Helper()
deadline := time.Now().Add(2 * time.Second)
for time.Now().Before(deadline) {
pub.mu.Lock()
count := len(pub.events)
pub.mu.Unlock()
if count >= n {
break
}
time.Sleep(2 * time.Millisecond)
}
pub.mu.Lock()
defer pub.mu.Unlock()
out := make([]string, 0, len(pub.events))
seen := make(map[string]struct{}, len(pub.events))
for _, ev := range pub.events {
id := ev.Metadata[MetaRequestID]
if id == "" {
continue
}
if _, dup := seen[id]; dup {
continue
}
seen[id] = struct{}{}
out = append(out, id)
}
if len(out) < n {
t.Fatalf("only got %d/%d request ids", len(out), n)
}
return out
}
// defaultTimeout swaps the broker's per-request wall-clock window so the
// timeout tests run quickly. Restores the prior value on the next call.
func defaultTimeout(t *testing.T, d time.Duration) {
t.Helper()
if d <= 0 {
t.Fatal("defaultTimeout must be > 0")
}
timeoutValue = func() time.Duration { return d }
}
// requestErr wraps Broker.Request to drop the Decision when tests only
// care about the error path. Keeps the goroutine bodies tight.
func requestErr(b *Broker, ctx context.Context, p Prompt) error {
_, err := b.Request(ctx, p)
return err
}
// TestRequestViewOnly checks the view-only outcome flows through Request's
// Decision return without being silently swallowed.
func TestRequestViewOnly(t *testing.T) {
pub := &fakePublisher{subscribers: true}
b := New(pub)
type result struct {
d Decision
err error
}
done := make(chan result, 1)
go func() {
d, err := b.Request(context.Background(), Prompt{Kind: KindVNC})
done <- result{d, err}
}()
id := waitForRequestID(t, pub)
require.True(t, b.Respond(id, Decision{Accept: true, ViewOnly: true}))
select {
case r := <-done:
assert.NoError(t, r.err)
assert.True(t, r.d.Accept)
assert.True(t, r.d.ViewOnly, "ViewOnly must survive the round-trip")
case <-time.After(time.Second):
t.Fatal("view-only request did not resolve")
}
}

View File

@@ -1,62 +0,0 @@
package approval
import "testing"
// TestShortKeyFingerprint locks in the format the VNC approval prompt
// shows to the user. The fingerprint is the user's only cryptographic
// anchor against a malicious management server that pushes a spoofed
// display name, so accidental changes to its format would silently
// undermine that defence.
func TestShortKeyFingerprint(t *testing.T) {
cases := []struct {
name string
in string
want string
}{
{
name: "full_32_byte_pubkey",
in: "0123456789abcdeffedcba9876543210ffeeddccbbaa99887766554433221100",
want: "0123-4567-89ab-cdef",
},
{
name: "exactly_16_chars",
in: "0123456789abcdef",
want: "0123-4567-89ab-cdef",
},
{
name: "borderline_8_chars",
in: "01234567",
want: "0123-4567",
},
{
name: "too_short_returns_empty",
in: "0123",
want: "",
},
{
name: "empty_returns_empty",
in: "",
want: "",
},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
got := ShortKeyFingerprint(tc.in)
if got != tc.want {
t.Fatalf("ShortKeyFingerprint(%q) = %q, want %q", tc.in, got, tc.want)
}
})
}
}
// TestShortKeyFingerprint_DistinctKeysDistinctOutputs guards against a
// formatting bug that would collapse different prefixes onto the same
// displayed fingerprint and let an attacker substitute their pubkey for
// a victim's while keeping the prompt visually identical.
func TestShortKeyFingerprint_DistinctKeysDistinctOutputs(t *testing.T) {
a := ShortKeyFingerprint("0123456789abcdef" + "rest_of_pubkey_ignored")
b := ShortKeyFingerprint("0123456789abcde0" + "rest_of_pubkey_ignored")
if a == b {
t.Fatalf("expected distinct outputs for distinct prefixes, both = %q", a)
}
}

View File

@@ -2,6 +2,7 @@ package auth
import (
"context"
"errors"
"net/url"
"strings"
"sync"
@@ -140,25 +141,21 @@ func (a *Auth) IsSSOSupported(ctx context.Context) (bool, error) {
// This avoids creating a new connection to the management server
func (a *Auth) GetOAuthFlow(ctx context.Context, forceDeviceAuth bool) (OAuthFlow, error) {
var flow OAuthFlow
var err error
err = a.withRetry(ctx, func(client *mgm.GrpcClient) error {
if forceDeviceAuth {
flow, err = a.getDeviceFlow(client)
return err
}
// the connection is owned by a and outlives this call, so a later fallback reuses it
newAuth := func(context.Context) (*Auth, func(), error) {
return a, func() {}, nil
}
// Try PKCE flow first
flow, err = a.getPKCEFlow(client)
if err != nil {
// If PKCE not supported, try Device flow
if s, ok := status.FromError(err); ok && (s.Code() == codes.NotFound || s.Code() == codes.Unimplemented) {
flow, err = a.getDeviceFlow(client)
return err
}
return err
err := a.withRetry(ctx, func(client *mgm.GrpcClient) error {
var err error
flow, err = oauthFlowWithFallback(a, client, flowOrder(forceDeviceAuth), "", newAuth)
var ssoUnavailable *ssoUnavailableError
if errors.As(err, &ssoUnavailable) {
return backoff.Permanent(err)
}
return nil
return err
})
return flow, err
@@ -344,7 +341,6 @@ func (a *Auth) setSystemInfoFlags(info *system.Info) {
a.config.RosenpassEnabled,
a.config.RosenpassPermissive,
a.config.ServerSSHAllowed,
a.config.ServerVNCAllowed,
a.config.DisableClientRoutes,
a.config.DisableServerRoutes,
a.config.DisableDNS,

View File

@@ -48,8 +48,17 @@ type DeviceAuthProviderConfig struct {
LoginHint string
}
// validateDeviceAuthConfig validates device authorization provider configuration
// validateDeviceAuthConfig validates device authorization provider configuration. A missing
// value means management does not have this flow configured, so the error wraps
// errFlowNotConfigured and the caller can fall back to the other flow.
func validateDeviceAuthConfig(config *DeviceAuthProviderConfig) error {
if err := checkDeviceAuthConfig(config); err != nil {
return fmt.Errorf("%w: %w", errFlowNotConfigured, err)
}
return nil
}
func checkDeviceAuthConfig(config *DeviceAuthProviderConfig) error {
errorMsgFormat := "invalid provider configuration received from management: %s value is empty. Contact your NetBird administrator"
if config.Audience == "" {
@@ -161,8 +170,12 @@ func (d *DeviceAuthorizationFlow) RequestAuthInfo(ctx context.Context) (AuthFlow
return AuthFlowInfo{}, fmt.Errorf("reading body failed with error: %v", err)
}
if res.StatusCode != 200 {
return AuthFlowInfo{}, fmt.Errorf("request device code returned status %d error: %s", res.StatusCode, string(body))
if res.StatusCode != http.StatusOK {
reqErr := fmt.Errorf("request device code returned status %d error: %s", res.StatusCode, string(body))
if deviceGrantUnsupported(res.StatusCode, body) {
return AuthFlowInfo{}, fmt.Errorf("%w: %w", errFlowNotConfigured, reqErr)
}
return AuthFlowInfo{}, reqErr
}
deviceCode := AuthFlowInfo{}
@@ -186,6 +199,34 @@ func (d *DeviceAuthorizationFlow) RequestAuthInfo(ctx context.Context) (AuthFlow
return deviceCode, err
}
// deviceGrantUnsupported reports whether the IdP's answer to a device code request means it does
// not serve the device authorization grant at all, rather than a transient or request-specific
// failure. An IdP that does not route the endpoint answers 404/405/501; one that knows the
// endpoint but has the grant disabled for this client answers with an OAuth 2.0 error code.
func deviceGrantUnsupported(statusCode int, body []byte) bool {
switch statusCode {
case http.StatusNotFound, http.StatusMethodNotAllowed, http.StatusNotImplemented:
return true
case http.StatusBadRequest, http.StatusUnauthorized, http.StatusForbidden:
default:
return false
}
var oauthErr struct {
Error string `json:"error"`
}
if err := json.Unmarshal(body, &oauthErr); err != nil {
return false
}
switch oauthErr.Error {
case "unsupported_grant_type", "unauthorized_client", "invalid_client":
return true
default:
return false
}
}
func appendLoginHint(uri, loginHint string) string {
if uri == "" || loginHint == "" {
return uri

View File

@@ -2,15 +2,19 @@ package auth
import (
"context"
"errors"
"fmt"
"net/http"
"net/url"
"runtime"
"sync"
log "github.com/sirupsen/logrus"
"google.golang.org/grpc/codes"
gstatus "google.golang.org/grpc/status"
"github.com/netbirdio/netbird/client/internal/profilemanager"
mgm "github.com/netbirdio/netbird/shared/management/client"
)
// OAuthFlow represents an interface for authorization using different OAuth 2.0 flows
@@ -59,77 +63,278 @@ func (t TokenInfo) GetTokenToUse() string {
return t.AccessToken
}
func shouldUseDeviceFlow(force bool, isUnixDesktopClient bool) bool {
return force || (runtime.GOOS == "linux" || runtime.GOOS == "freebsd") && !isUnixDesktopClient
// errFlowNotConfigured marks a flow this deployment does not offer: management returned no
// configuration for it, the configuration it returned is incomplete, or the IdP refuses to serve
// the grant. It is the only condition that makes the client try the other flow, so that a
// transient failure keeps failing on the flow the user actually wants.
var errFlowNotConfigured = errors.New("authorization flow is not configured")
// ssoUnavailableError reports that the management server offers no usable SSO flow at all.
// Retrying cannot help, so callers should surface it to the user instead of backing off.
type ssoUnavailableError struct {
msg string
}
// NewOAuthFlow initializes and returns the appropriate OAuth flow based on the management configuration
//
// It starts by initializing the PKCE.If this process fails, it resorts to the Device Code Flow,
// and if that also fails, the authentication process is deemed unsuccessful
//
// On Linux distros without desktop environment support, it only tries to initialize the Device Code Flow
// forceDeviceCodeFlow can be used to skip PKCE and go directly to Device Code Flow (e.g., for Android TV)
func NewOAuthFlow(ctx context.Context, config *profilemanager.Config, isUnixDesktopClient bool, forceDeviceCodeFlow bool, hint string) (OAuthFlow, error) {
if shouldUseDeviceFlow(forceDeviceCodeFlow, isUnixDesktopClient) {
return authenticateWithDeviceCodeFlow(ctx, config, hint)
}
pkceFlow, err := authenticateWithPKCEFlow(ctx, config, hint)
if err != nil {
log.Debugf("failed to initialize pkce authentication with error: %v\n", err)
log.Debug("falling back to device code flow")
return authenticateWithDeviceCodeFlow(ctx, config, hint)
}
return pkceFlow, nil
func (e *ssoUnavailableError) Error() string {
return e.msg
}
// authenticateWithPKCEFlow initializes the Proof Key for Code Exchange flow auth flow
func authenticateWithPKCEFlow(ctx context.Context, config *profilemanager.Config, hint string) (OAuthFlow, error) {
authClient, err := NewAuth(ctx, config.PrivateKey, config.ManagementURL, config)
if err != nil {
return nil, fmt.Errorf("failed to create auth client: %v", err)
}
defer authClient.Close()
// oauthFlowInit names one of the OAuth flows and builds it from the management configuration.
type oauthFlowInit struct {
name string
init func(a *Auth, client *mgm.GrpcClient, hint string) (OAuthFlow, error)
}
pkceFlowInfo, err := authClient.getPKCEFlow(authClient.client)
// authFactory hands out a management connection to build a flow with, plus the cleanup that
// releases it. Callers that own a long-lived connection return it with a no-op cleanup.
type authFactory func(ctx context.Context) (*Auth, func(), error)
// fallbackFlow wraps the flow that was picked at initialization time with the flows that were
// not tried. Whether the IdP actually serves a flow only shows up when the flow is run: an IdP
// with the device grant disabled answers the device code request with 404 even though
// management handed out a device flow configuration. When that happens the wrapper swaps in the
// next flow instead of failing the login.
type fallbackFlow struct {
mu sync.Mutex
active OAuthFlow
remaining []oauthFlowInit
hint string
newAuth authFactory
}
func (f *fallbackFlow) RequestAuthInfo(ctx context.Context) (AuthFlowInfo, error) {
info, err := f.current().RequestAuthInfo(ctx)
if err == nil || !isFlowUnavailable(err) {
return info, err
}
next, nextErr := f.initNext(ctx)
if nextErr != nil {
log.Debugf("failed to fall back to another authorization flow: %v", nextErr)
return AuthFlowInfo{}, err
}
return next.RequestAuthInfo(ctx)
}
func (f *fallbackFlow) WaitToken(ctx context.Context, info AuthFlowInfo) (TokenInfo, error) {
return f.current().WaitToken(ctx, info)
}
func (f *fallbackFlow) GetClientID(ctx context.Context) string {
return f.current().GetClientID(ctx)
}
func (f *fallbackFlow) current() OAuthFlow {
f.mu.Lock()
defer f.mu.Unlock()
return f.active
}
// initNext initializes the next flow this deployment offers and makes it the active one.
func (f *fallbackFlow) initNext(ctx context.Context) (OAuthFlow, error) {
f.mu.Lock()
defer f.mu.Unlock()
if len(f.remaining) == 0 {
return nil, errors.New("no authorization flow left to try")
}
a, cleanup, err := f.newAuth(ctx)
if err != nil {
return nil, fmt.Errorf("getting pkce authorization flow info failed with error: %v", err)
return nil, err
}
defer cleanup()
flow, remaining, err := initFirstAvailableFlow(a, a.client, f.remaining, f.hint)
if err != nil {
return nil, err
}
log.Infof("the identity provider does not serve the selected authorization flow, continuing with the next one")
f.active = flow
f.remaining = remaining
return flow, nil
}
// preferDeviceFlow reports whether the device code flow should be tried before PKCE. PKCE needs
// a browser on this host and a loopback listener to receive the redirect, neither of which
// exists on a Unix host without a graphical session. The GOOS guard keeps a caller that reports
// no graphical session on a platform that always has one from changing the preference.
func preferDeviceFlow(force bool, hasGraphicalSession bool) bool {
return force || (runtime.GOOS == "linux" || runtime.GOOS == "freebsd") && !hasGraphicalSession
}
// flowOrder returns both flows in the order they should be attempted.
func flowOrder(preferDevice bool) []oauthFlowInit {
pkce := oauthFlowInit{name: "pkce authorization flow", init: initPKCEFlow}
device := oauthFlowInit{name: "device code flow", init: initDeviceFlow}
if preferDevice {
return []oauthFlowInit{device, pkce}
}
return []oauthFlowInit{pkce, device}
}
func initPKCEFlow(a *Auth, client *mgm.GrpcClient, hint string) (OAuthFlow, error) {
flow, err := a.getPKCEFlow(client)
if err != nil {
return nil, err
}
if hint != "" {
pkceFlowInfo.SetLoginHint(hint)
flow.SetLoginHint(hint)
}
return pkceFlowInfo, nil
return flow, nil
}
// authenticateWithDeviceCodeFlow initializes the Device Code auth Flow
func authenticateWithDeviceCodeFlow(ctx context.Context, config *profilemanager.Config, hint string) (OAuthFlow, error) {
func initDeviceFlow(a *Auth, client *mgm.GrpcClient, hint string) (OAuthFlow, error) {
flow, err := a.getDeviceFlow(client)
if err != nil {
return nil, err
}
if hint != "" {
flow.SetLoginHint(hint)
}
return flow, nil
}
// NewOAuthFlow initializes and returns an OAuth flow based on the management configuration.
//
// Both flows are optional server side: management answers NotFound for a flow it has no
// configuration for. The preferred flow is tried first and the other one is used as a fallback,
// so a server that only offers one of them still works. forceDeviceCodeFlow prefers the device
// code flow regardless of platform (e.g. for Android TV).
func NewOAuthFlow(ctx context.Context, config *profilemanager.Config, hasGraphicalSession bool, forceDeviceCodeFlow bool, hint string) (OAuthFlow, error) {
authClient, err := NewAuth(ctx, config.PrivateKey, config.ManagementURL, config)
if err != nil {
return nil, fmt.Errorf("failed to create auth client: %v", err)
return nil, fmt.Errorf("create auth client: %w", err)
}
defer authClient.Close()
deviceFlowInfo, err := authClient.getDeviceFlow(authClient.client)
// the connection above is closed on return, so a later fallback opens its own
newAuth := func(ctx context.Context) (*Auth, func(), error) {
a, err := NewAuth(ctx, config.PrivateKey, config.ManagementURL, config)
if err != nil {
return nil, nil, fmt.Errorf("create auth client: %w", err)
}
return a, func() {
if err := a.Close(); err != nil {
log.Debugf("failed to close auth client: %v", err)
}
}, nil
}
flows := flowOrder(preferDeviceFlow(forceDeviceCodeFlow, hasGraphicalSession))
return oauthFlowWithFallback(authClient, authClient.client, flows, hint, newAuth)
}
// oauthFlowWithFallback initializes the first flow this deployment offers, moving on to the next
// one when a flow is not configured here. It only fails once every flow has been tried, and any
// flow left untried is handed to the returned flow so it can still fall back if the IdP rejects
// the flow that was picked.
func oauthFlowWithFallback(a *Auth, client *mgm.GrpcClient, flows []oauthFlowInit, hint string, newAuth authFactory) (OAuthFlow, error) {
flow, remaining, err := initFirstAvailableFlow(a, client, flows, hint)
if err != nil {
switch s, ok := gstatus.FromError(err); {
case ok && s.Code() == codes.NotFound:
return nil, fmt.Errorf("no SSO provider returned from management. " +
"Please proceed with setting up this device using setup keys " +
"https://docs.netbird.io/how-to/register-machines-using-setup-keys")
case ok && s.Code() == codes.Unimplemented:
return nil, fmt.Errorf("the management server, %s, does not support SSO providers, "+
"please update your server or use Setup Keys to login", config.ManagementURL)
default:
return nil, fmt.Errorf("getting device authorization flow info failed with error: %v", err)
return nil, err
}
if len(remaining) == 0 {
return flow, nil
}
return &fallbackFlow{
active: flow,
remaining: remaining,
hint: hint,
newAuth: newAuth,
}, nil
}
// initFirstAvailableFlow returns the first flow that could be initialized along with the flows
// after it, which are still untried.
func initFirstAvailableFlow(a *Auth, client *mgm.GrpcClient, flows []oauthFlowInit, hint string) (OAuthFlow, []oauthFlowInit, error) {
var errs []error
for i, f := range flows {
flow, err := f.init(a, client, hint)
if err == nil {
return flow, flows[i+1:], nil
}
errs = append(errs, fmt.Errorf("%s: %w", f.name, err))
// only a flow this deployment does not offer is worth replacing with another one
if !isFlowUnavailable(err) {
break
}
if i < len(flows)-1 {
log.Infof("%s is not configured (%v), falling back to %s", f.name, err, flows[i+1].name)
}
}
if hint != "" {
deviceFlowInfo.SetLoginHint(hint)
return nil, nil, flowInitError(a.mgmURL, errs)
}
// flowInitError turns the per-flow initialization errors into a single actionable error. The
// message stays neutral about what to do instead: SSO is also how a peer extends its session and
// authenticates SSH, where a setup key is no alternative. Callers that are enrolling a device add
// that advice themselves, see IsSSOUnavailable.
func flowInitError(mgmURL *url.URL, errs []error) error {
if allMatch(errs, isFlowUnimplemented) {
return &ssoUnavailableError{msg: fmt.Sprintf("the management server, %s, does not support SSO providers, "+
"please update your server", mgmURL)}
}
return deviceFlowInfo, nil
if allMatch(errs, isFlowUnavailable) {
return &ssoUnavailableError{msg: "the management server has no SSO provider configured: " +
"neither the pkce authorization flow nor the device code flow is available"}
}
return fmt.Errorf("initialize authorization flow: %w", errors.Join(errs...))
}
// IsSSOUnavailable reports whether err means the management server offers no usable SSO flow, so
// no retry and no other flow can help. Enrollment paths use it to point the user at setup keys.
func IsSSOUnavailable(err error) bool {
var ssoUnavailable *ssoUnavailableError
return errors.As(err, &ssoUnavailable)
}
func allMatch(errs []error, match func(error) bool) bool {
if len(errs) == 0 {
return false
}
for _, err := range errs {
if !match(err) {
return false
}
}
return true
}
// isFlowUnavailable reports whether the flow is not on offer here: management has no
// configuration for it (NotFound), predates the RPC entirely (Unimplemented), returned an
// incomplete configuration, or the IdP does not serve the grant.
func isFlowUnavailable(err error) bool {
return errors.Is(err, errFlowNotConfigured) ||
hasStatusCode(err, codes.NotFound) ||
hasStatusCode(err, codes.Unimplemented)
}
func isFlowUnimplemented(err error) bool {
return hasStatusCode(err, codes.Unimplemented)
}
func hasStatusCode(err error, code codes.Code) bool {
s, ok := gstatus.FromError(err)
if !ok {
return false
}
return s.Code() == code
}

View File

@@ -0,0 +1,221 @@
package auth
import (
"context"
"errors"
"fmt"
"net/url"
"runtime"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
mgm "github.com/netbirdio/netbird/shared/management/client"
)
// stubFlow is a minimal OAuthFlow returned by the fake initializers below. requestErr, when set,
// is what its RequestAuthInfo returns, standing in for an IdP that rejects the flow.
type stubFlow struct {
name string
hint string
requestErr error
}
func (s *stubFlow) RequestAuthInfo(context.Context) (AuthFlowInfo, error) {
if s.requestErr != nil {
return AuthFlowInfo{}, s.requestErr
}
return AuthFlowInfo{UserCode: s.name}, nil
}
func (s *stubFlow) WaitToken(context.Context, AuthFlowInfo) (TokenInfo, error) {
return TokenInfo{}, nil
}
func (s *stubFlow) GetClientID(context.Context) string {
return ""
}
// stubInit returns a flow initializer that yields a named stub flow, or err when err is non-nil.
func stubInit(name string, err error) oauthFlowInit {
return stubInitFlow(name, err, nil)
}
// stubInitFlow is stubInit with control over what the resulting flow's RequestAuthInfo returns.
func stubInitFlow(name string, initErr, requestErr error) oauthFlowInit {
return oauthFlowInit{
name: name,
init: func(_ *Auth, _ *mgm.GrpcClient, hint string) (OAuthFlow, error) {
if initErr != nil {
return nil, initErr
}
return &stubFlow{name: name, hint: hint, requestErr: requestErr}, nil
},
}
}
// stubAuthFactory hands out an Auth without a management connection, which the stub
// initializers above never touch.
func stubAuthFactory(a *Auth) authFactory {
return func(context.Context) (*Auth, func(), error) {
return a, func() {}, nil
}
}
func TestOAuthFlowWithFallback(t *testing.T) {
notFound := status.Error(codes.NotFound, "no device authorization flow information available")
unimplemented := status.Error(codes.Unimplemented, "unknown method")
incompleteConfig := fmt.Errorf("%w: Client ID value is empty", errFlowNotConfigured)
unreachable := status.Error(codes.Unavailable, "connection refused")
tests := []struct {
name string
flows []oauthFlowInit
expectedFlow string
expectedErr string
expectedNoSSO bool
}{
{
name: "preferred flow is used",
flows: []oauthFlowInit{stubInit("device", nil), stubInit("pkce", nil)},
expectedFlow: "device",
},
{
// the RedHat case: device code flow disabled on management, PKCE configured
name: "falls back when preferred flow is not configured",
flows: []oauthFlowInit{stubInit("device", notFound), stubInit("pkce", nil)},
expectedFlow: "pkce",
},
{
name: "falls back on an incomplete flow configuration",
flows: []oauthFlowInit{stubInit("pkce", incompleteConfig), stubInit("device", nil)},
expectedFlow: "device",
},
{
name: "does not fall back when the preferred flow fails for another reason",
flows: []oauthFlowInit{stubInit("pkce", unreachable), stubInit("device", nil)},
expectedErr: "connection refused",
},
{
// stays neutral about the remedy: --extend and SSH auth cannot use a setup key
name: "neither flow configured reports no SSO provider",
flows: []oauthFlowInit{stubInit("device", notFound), stubInit("pkce", notFound)},
expectedErr: "no SSO provider configured",
expectedNoSSO: true,
},
{
name: "old server without the flow RPCs asks for an update",
flows: []oauthFlowInit{stubInit("device", unimplemented), stubInit("pkce", unimplemented)},
expectedErr: "does not support SSO providers",
expectedNoSSO: true,
},
}
mgmURL, err := url.Parse("https://api.netbird.io:443")
require.NoError(t, err)
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
a := &Auth{mgmURL: mgmURL}
flow, err := oauthFlowWithFallback(a, nil, tt.flows, "user@example.com", stubAuthFactory(a))
if tt.expectedErr != "" {
require.Error(t, err)
assert.Contains(t, err.Error(), tt.expectedErr)
var ssoUnavailable *ssoUnavailableError
assert.Equal(t, tt.expectedNoSSO, errors.As(err, &ssoUnavailable),
"terminal SSO-unavailable classification mismatch for %v", err)
return
}
require.NoError(t, err)
stub := activeStub(t, flow)
assert.Equal(t, tt.expectedFlow, stub.name)
assert.Equal(t, "user@example.com", stub.hint, "login hint must be passed to the flow")
})
}
}
// activeStub unwraps the flow currently in use, which is behind a fallbackFlow whenever an
// untried flow is left.
func activeStub(t *testing.T, flow OAuthFlow) *stubFlow {
t.Helper()
if fallback, ok := flow.(*fallbackFlow); ok {
flow = fallback.current()
}
stub, ok := flow.(*stubFlow)
require.True(t, ok, "unexpected flow type %T", flow)
return stub
}
// TestFallbackFlowRequestAuthInfo covers the failure the RedHat report hit: management hands out
// a device flow configuration, but the IdP does not serve the grant and only says so when the
// device code is requested.
func TestFallbackFlowRequestAuthInfo(t *testing.T) {
mgmURL, err := url.Parse("https://api.netbird.io:443")
require.NoError(t, err)
a := &Auth{mgmURL: mgmURL}
idpRejects := fmt.Errorf("%w: request device code returned status 404", errFlowNotConfigured)
t.Run("swaps in the untried flow", func(t *testing.T) {
flows := []oauthFlowInit{stubInitFlow("device", nil, idpRejects), stubInit("pkce", nil)}
flow, err := oauthFlowWithFallback(a, nil, flows, "", stubAuthFactory(a))
require.NoError(t, err)
require.Equal(t, "device", activeStub(t, flow).name)
info, err := flow.RequestAuthInfo(context.Background())
require.NoError(t, err)
assert.Equal(t, "pkce", info.UserCode, "the request must be served by the fallback flow")
assert.Equal(t, "pkce", activeStub(t, flow).name, "the fallback flow must stay active for WaitToken")
})
t.Run("keeps the original error when nothing else is configured", func(t *testing.T) {
flows := []oauthFlowInit{
stubInitFlow("device", nil, idpRejects),
stubInit("pkce", status.Error(codes.NotFound, "no pkce authorization flow information available")),
}
flow, err := oauthFlowWithFallback(a, nil, flows, "", stubAuthFactory(a))
require.NoError(t, err)
_, err = flow.RequestAuthInfo(context.Background())
require.Error(t, err)
assert.Contains(t, err.Error(), "status 404")
})
t.Run("does not swap flows on an unrelated failure", func(t *testing.T) {
flows := []oauthFlowInit{
stubInitFlow("device", nil, errors.New("timeout talking to the IdP")),
stubInit("pkce", nil),
}
flow, err := oauthFlowWithFallback(a, nil, flows, "", stubAuthFactory(a))
require.NoError(t, err)
_, err = flow.RequestAuthInfo(context.Background())
require.Error(t, err)
assert.Equal(t, "device", activeStub(t, flow).name, "the preferred flow must stay active")
})
}
func TestFlowOrder(t *testing.T) {
assert.Equal(t, "pkce authorization flow", flowOrder(false)[0].name)
assert.Equal(t, "device code flow", flowOrder(true)[0].name)
assert.Len(t, flowOrder(false), 2, "both flows must always be attempted")
}
func TestPreferDeviceFlow(t *testing.T) {
isUnix := runtime.GOOS == "linux" || runtime.GOOS == "freebsd"
assert.True(t, preferDeviceFlow(true, true), "forced device flow wins over a desktop session")
assert.Equal(t, isUnix, preferDeviceFlow(false, false), "headless unix hosts prefer the device flow")
assert.False(t, preferDeviceFlow(false, true), "desktop clients prefer PKCE")
}

View File

@@ -62,8 +62,17 @@ type PKCEAuthProviderConfig struct {
LoginHint string
}
// validatePKCEConfig validates PKCE provider configuration
// validatePKCEConfig validates PKCE provider configuration. A missing value means management
// does not have this flow configured, so the error wraps errFlowNotConfigured and the caller can
// fall back to the other flow.
func validatePKCEConfig(config *PKCEAuthProviderConfig) error {
if err := checkPKCEConfig(config); err != nil {
return fmt.Errorf("%w: %w", errFlowNotConfigured, err)
}
return nil
}
func checkPKCEConfig(config *PKCEAuthProviderConfig) error {
errorMsgFormat := "invalid provider configuration received from management: %s value is empty. Contact your NetBird administrator"
if config.ClientID == "" {

View File

@@ -614,8 +614,6 @@ func createEngineConfig(key wgtypes.Key, config *profilemanager.Config, peerConf
RosenpassEnabled: config.RosenpassEnabled,
RosenpassPermissive: config.RosenpassPermissive,
ServerSSHAllowed: util.ReturnBoolWithDefaultTrue(config.ServerSSHAllowed),
ServerVNCAllowed: config.ServerVNCAllowed != nil && *config.ServerVNCAllowed,
DisableVNCApproval: config.DisableVNCApproval,
EnableSSHRoot: config.EnableSSHRoot,
EnableSSHSFTP: config.EnableSSHSFTP,
EnableSSHLocalPortForwarding: config.EnableSSHLocalPortForwarding,
@@ -699,7 +697,6 @@ func loginToManagement(ctx context.Context, client mgm.Client, pubSSHKey []byte,
config.RosenpassEnabled,
config.RosenpassPermissive,
config.ServerSSHAllowed,
config.ServerVNCAllowed,
config.DisableClientRoutes,
config.DisableServerRoutes,
config.DisableDNS,

View File

@@ -34,9 +34,8 @@ import (
"github.com/netbirdio/netbird/shared/netiputil"
)
const readmeContent = `Netbird debug bundle
This debug bundle contains the following files.
If the --anonymize flag is set, the files are anonymized to protect sensitive information.
const readmeContent = `This debug bundle contains the following files.
If anonymization is enabled (--anonymize / --anonymize-level), the files are anonymized to protect sensitive information.
status.txt: Anonymized status information of the NetBird client.
client.log: Most recent, anonymized client log file of the NetBird client.
@@ -70,21 +69,34 @@ capture.pcap: Packet capture in pcap format. Only present when capture was runni
Anonymization Process
The files in this bundle have been anonymized to protect sensitive information. Here's how the anonymization was applied:
The files in this bundle have been anonymized to protect sensitive information. The level applied to this bundle is recorded at the top of this file. Here's how the anonymization was applied:
IP Addresses
IPv4 addresses are replaced with addresses starting from 198.51.100.0
IPv6 addresses are replaced with addresses starting from 100::
Default level:
- Public IPv4 addresses are replaced with addresses starting from 198.51.100.0
- Public IPv6 addresses are replaced with addresses starting from 2001:db8:ffff::
- IPv6 unique local addresses (fc00::/7) are anonymized as well: their random global ID uniquely identifies the network.
- IP addresses from internal IPv4 ranges and well-known addresses are not anonymized (e.g. 8.8.8.8, 100.64.0.0/10, addresses starting with 192.168., 172.16., 10., 169.254., fe80::).
Strict level (--anonymize-level strict), in addition to the default level:
- Private (RFC 1918), CGNAT (100.64.0.0/10), and link-local (169.254.0.0/16, fe80::/10) addresses are anonymized too.
- Internal IPv4 addresses are replaced with addresses starting from 198.18.0.0 and internal IPv6 addresses with addresses starting from 2001:db8:1::, so internal addresses remain distinguishable from public ones.
- Addresses are mapped in order of first appearance: subnet structure, allocation scheme, and gateway conventions are not preserved. Prefix lengths of networks are preserved.
- Peer names in front of NetBird domains are replaced with numbered placeholders (e.g. peer-1.netbird.cloud), and subdomain labels of other domains with host-N placeholders.
- WireGuard public keys are replaced with consistent placeholder keys.
IP addresses from non public ranges and well known addresses are not anonymized (e.g. 8.8.8.8, 100.64.0.0/10, addresses starting with 192.168., 172.16., 10., etc.).
Reoccuring IP addresses are replaced with the same anonymized address.
Note: The anonymized IP addresses in the status file do not match those in the log and routes files. However, the anonymized IP addresses are consistent within the status file and across the routes and log files.
MAC Addresses
MAC addresses are replaced at every anonymization level with consistent placeholders counting up from 02:00:00:00:00:01. Broadcast, multicast, and all-zero addresses are kept. At the default level a preserved IPv6 link-local address may still embed a MAC address (EUI-64); the strict level anonymizes those addresses.
Domains
All domain names (except for the netbird domains) are replaced with randomly generated strings ending in ".domain". Anonymized domains are consistent across all files in the bundle.
Reoccuring domain names are replaced with the same anonymized domain.
At the strict level, the peer name labels in front of netbird domains are anonymized as well.
Sync Response
The network_map.json file contains the following anonymized information:
@@ -281,6 +293,7 @@ type BundleGenerator struct {
cliVersion string
anonymize bool
anonymizeLevel anonymize.Level
includeSystemInfo bool
logFileCount uint32
@@ -288,7 +301,10 @@ type BundleGenerator struct {
}
type BundleConfig struct {
Anonymize bool
Anonymize bool
// AnonymizeLevel selects how much the anonymizer redacts.
// anonymize.LevelStrict implies Anonymize.
AnonymizeLevel anonymize.Level
IncludeSystemInfo bool
LogFileCount uint32
}
@@ -327,8 +343,11 @@ func NewBundleGenerator(deps GeneratorDependencies, cfg BundleConfig) *BundleGen
uiLogOpener = openLogFile
}
anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
anonymizer.SetLevel(cfg.AnonymizeLevel)
return &BundleGenerator{
anonymizer: anonymize.NewAnonymizer(anonymize.DefaultAddresses()),
anonymizer: anonymizer,
internalConfig: deps.InternalConfig,
statusRecorder: deps.StatusRecorder,
@@ -345,7 +364,8 @@ func NewBundleGenerator(deps GeneratorDependencies, cfg BundleConfig) *BundleGen
daemonVersion: deps.DaemonVersion,
cliVersion: deps.CliVersion,
anonymize: cfg.Anonymize,
anonymize: cfg.Anonymize || cfg.AnonymizeLevel >= anonymize.LevelStrict,
anonymizeLevel: cfg.AnonymizeLevel,
includeSystemInfo: cfg.IncludeSystemInfo,
logFileCount: logFileCount,
}
@@ -485,7 +505,13 @@ func (g *BundleGenerator) addSystemInfo() {
}
func (g *BundleGenerator) addReadme() error {
readmeReader := strings.NewReader(readmeContent)
level := "none (anonymization disabled)"
if g.anonymize {
level = g.anonymizeLevel.String()
}
header := fmt.Sprintf("Netbird debug bundle\nAnonymization level applied to this bundle: %s\n", level)
readmeReader := strings.NewReader(header + readmeContent)
if err := g.addFileToZip(readmeReader, "README.txt"); err != nil {
return fmt.Errorf("add README file to zip: %w", err)
}
@@ -507,9 +533,10 @@ func (g *BundleGenerator) addStatus() error {
fullStatus := g.statusRecorder.GetFullStatus()
protoFullStatus := nbstatus.ToProtoFullStatus(fullStatus)
overview := nbstatus.ConvertToStatusOutputOverview(protoFullStatus, nbstatus.ConvertOptions{
Anonymize: g.anonymize,
ProfileName: profName,
DaemonVersion: g.daemonVersion,
Anonymize: g.anonymize,
AnonymizeLevel: g.anonymizeLevel,
ProfileName: profName,
DaemonVersion: g.daemonVersion,
})
overview.CliVersion = g.cliVersion
statusOutput := overview.FullDetailSummary()
@@ -662,7 +689,7 @@ func (g *BundleGenerator) addCommonConfigFields(configContent *strings.Builder)
configContent.WriteString("NetBird Client Configuration:\n\n")
if key, err := wgtypes.ParseKey(g.internalConfig.PrivateKey); err == nil {
configContent.WriteString(fmt.Sprintf("PublicKey: %s\n", key.PublicKey().String()))
configContent.WriteString(fmt.Sprintf("PublicKey: %s\n", g.anonymizer.AnonymizeWGKey(key.PublicKey().String())))
}
configContent.WriteString(fmt.Sprintf("WgIface: %s\n", g.internalConfig.WgIface))
configContent.WriteString(fmt.Sprintf("WgPort: %d\n", g.internalConfig.WgPort))
@@ -694,12 +721,6 @@ func (g *BundleGenerator) addCommonConfigFields(configContent *strings.Builder)
if g.internalConfig.SSHJWTCacheTTL != nil {
configContent.WriteString(fmt.Sprintf("SSHJWTCacheTTL: %d\n", *g.internalConfig.SSHJWTCacheTTL))
}
if g.internalConfig.ServerVNCAllowed != nil {
configContent.WriteString(fmt.Sprintf("ServerVNCAllowed: %v\n", *g.internalConfig.ServerVNCAllowed))
}
if g.internalConfig.DisableVNCApproval != nil {
configContent.WriteString(fmt.Sprintf("DisableVNCApproval: %v\n", *g.internalConfig.DisableVNCApproval))
}
configContent.WriteString(fmt.Sprintf("DisableClientRoutes: %v\n", g.internalConfig.DisableClientRoutes))
configContent.WriteString(fmt.Sprintf("DisableServerRoutes: %v\n", g.internalConfig.DisableServerRoutes))
@@ -958,6 +979,11 @@ func (g *BundleGenerator) addUpdateLogs() error {
}
baseName := filepath.Base(logFile)
data, err = g.anonymizeBytes(data)
if err != nil {
log.Warnf("skipping update log file %s: %v", baseName, err)
continue
}
if err := g.addFileToZip(bytes.NewReader(data), filepath.Join("update-logs", baseName)); err != nil {
return fmt.Errorf("add update log file %s to zip: %w", baseName, err)
}
@@ -985,6 +1011,13 @@ func (g *BundleGenerator) addCorruptedStateFiles() error {
}
fileName := filepath.Base(match)
// Corrupted state files usually fail structured JSON anonymization,
// so run them through the string anonymizer instead.
data, err = g.anonymizeBytes(data)
if err != nil {
log.Warnf("skipping corrupted state file %s: %v", fileName, err)
continue
}
if err := g.addFileToZip(bytes.NewReader(data), "corrupted_states/"+fileName); err != nil {
log.Warnf("Failed to add corrupted state file %s to zip: %v", fileName, err)
continue
@@ -996,6 +1029,27 @@ func (g *BundleGenerator) addCorruptedStateFiles() error {
return nil
}
// anonymizeBytes runs raw file content through the string anonymizer line by
// line when anonymization is enabled. It errors instead of returning partial
// content, so a caller never adds an unanonymized fallback to the bundle.
func (g *BundleGenerator) anonymizeBytes(data []byte) ([]byte, error) {
if !g.anonymize {
return data, nil
}
var buf bytes.Buffer
scanner := bufio.NewScanner(bytes.NewReader(data))
scanner.Buffer(make([]byte, 1024*1024), 1024*1024)
for scanner.Scan() {
buf.WriteString(g.anonymizer.AnonymizeString(scanner.Text()))
buf.WriteByte('\n')
}
if err := scanner.Err(); err != nil {
return nil, fmt.Errorf("anonymize content: %w", err)
}
return buf.Bytes(), nil
}
func (g *BundleGenerator) addMetrics() error {
if g.clientMetrics == nil {
log.Debugf("skipping metrics in debug bundle: no metrics collector")
@@ -1468,6 +1522,7 @@ func anonymizeRemotePeer(peer *mgmProto.RemotePeerConfig, anonymizer *anonymize.
}
peer.Fqdn = anonymizer.AnonymizeDomain(peer.Fqdn)
peer.WgPubKey = anonymizer.AnonymizeWGKey(peer.WgPubKey)
anonymizeSSHConfig(peer.SshConfig)
}

View File

@@ -844,6 +844,10 @@ func collectSysctls() string {
[]string{"net.ipv4.conf.all.src_valid_mark", "net.ipv4.conf.default.src_valid_mark"},
listInterfaceSysctls("ipv4", "src_valid_mark")...,
))
writeSysctlGroup(&builder, "accept_ra", append(
[]string{"net.ipv6.conf.all.accept_ra", "net.ipv6.conf.default.accept_ra"},
listInterfaceSysctls("ipv6", "accept_ra")...,
))
writeSysctlGroup(&builder, "conntrack", []string{
"net.netfilter.nf_conntrack_acct",
"net.netfilter.nf_conntrack_tcp_loose",

View File

@@ -864,8 +864,6 @@ func TestAddConfig_AllFieldsCovered(t *testing.T) {
RosenpassEnabled: true,
RosenpassPermissive: true,
ServerSSHAllowed: &bTrue,
ServerVNCAllowed: &bTrue,
DisableVNCApproval: &bTrue,
EnableSSHRoot: &bTrue,
EnableSSHSFTP: &bTrue,
EnableSSHLocalPortForwarding: &bTrue,

View File

@@ -35,14 +35,14 @@ func (g *BundleGenerator) toWGShowFormat(s *configurer.Stats) string {
var sb strings.Builder
sb.WriteString(fmt.Sprintf("interface: %s\n", s.DeviceName))
sb.WriteString(fmt.Sprintf(" public key: %s\n", s.PublicKey))
sb.WriteString(fmt.Sprintf(" public key: %s\n", g.anonymizer.AnonymizeWGKey(s.PublicKey)))
sb.WriteString(fmt.Sprintf(" listen port: %d\n", s.ListenPort))
if s.FWMark != 0 {
sb.WriteString(fmt.Sprintf(" fwmark: %#x\n", s.FWMark))
}
for _, peer := range s.Peers {
sb.WriteString(fmt.Sprintf("\npeer: %s\n", peer.PublicKey))
sb.WriteString(fmt.Sprintf("\npeer: %s\n", g.anonymizer.AnonymizeWGKey(peer.PublicKey)))
if peer.Endpoint.IP != nil {
if g.anonymize {
anonEndpoint := g.anonymizer.AnonymizeUDPAddr(peer.Endpoint)
@@ -54,7 +54,11 @@ func (g *BundleGenerator) toWGShowFormat(s *configurer.Stats) string {
if len(peer.AllowedIPs) > 0 {
var ipStrings []string
for _, ipnet := range peer.AllowedIPs {
ipStrings = append(ipStrings, ipnet.String())
ipStr := ipnet.String()
if g.anonymize {
ipStr = g.anonymizer.AnonymizeIPString(ipStr)
}
ipStrings = append(ipStrings, ipStr)
}
sb.WriteString(fmt.Sprintf(" allowed ips: %s\n", strings.Join(ipStrings, ", ")))
}

View File

@@ -267,18 +267,38 @@ func (s *systemConfigurator) getSystemDNSSettings() (SystemDNSSettings, error) {
return SystemDNSSettings{}, fmt.Errorf("sending the command: %w", err)
}
var dnsSettings SystemDNSSettings
dnsSettings, serverAddresses, err := parseSystemDNSSettings(b)
if err != nil {
return dnsSettings, err
}
s.mu.Lock()
s.origNameservers = serverAddresses
s.mu.Unlock()
return dnsSettings, nil
}
// parseSystemDNSSettings parses the output of `scutil show State:/Network/Service/<id>/DNS`.
// Lines that don't match the expected "index : value" shape are skipped: hosts with unusual
// network services (e.g. orphaned hardware ports) can produce entries without a value.
func parseSystemDNSSettings(out []byte) (SystemDNSSettings, []netip.Addr, error) {
// port is not exposed by scutil, default to 53
dnsSettings := SystemDNSSettings{ServerPort: DefaultPort}
var serverAddresses []netip.Addr
inSearchDomainsArray := false
inServerAddressesArray := false
scanner := bufio.NewScanner(bytes.NewReader(b))
scanner := bufio.NewScanner(bytes.NewReader(out))
for scanner.Scan() {
line := strings.TrimSpace(scanner.Text())
switch {
case strings.HasPrefix(line, "DomainName :"):
domainName := strings.TrimSpace(strings.Split(line, ":")[1])
dnsSettings.Domains = append(dnsSettings.Domains, domainName)
domainName := strings.TrimSpace(strings.TrimPrefix(line, "DomainName :"))
if domainName != "" {
dnsSettings.Domains = append(dnsSettings.Domains, domainName)
}
continue
case line == "SearchDomains : <array> {":
inSearchDomainsArray = true
continue
@@ -288,36 +308,45 @@ func (s *systemConfigurator) getSystemDNSSettings() (SystemDNSSettings, error) {
case line == "}":
inSearchDomainsArray = false
inServerAddressesArray = false
continue
}
if !inSearchDomainsArray && !inServerAddressesArray {
continue
}
parts := strings.SplitN(line, " : ", 2)
if len(parts) != 2 {
log.Debugf("skipping unexpected scutil DNS line %q", line)
continue
}
value := strings.TrimSpace(parts[1])
if value == "" {
continue
}
if inSearchDomainsArray {
searchDomain := strings.Split(line, " : ")[1]
dnsSettings.Domains = append(dnsSettings.Domains, searchDomain)
} else if inServerAddressesArray {
address := strings.Split(line, " : ")[1]
if ip, err := netip.ParseAddr(address); err == nil && !ip.IsUnspecified() {
ip = ip.Unmap()
serverAddresses = append(serverAddresses, ip)
// Prefer the first IPv4 server as ServerIP since our DNS listener is IPv4.
if !dnsSettings.ServerIP.IsValid() && ip.Is4() {
dnsSettings.ServerIP = ip
}
}
dnsSettings.Domains = append(dnsSettings.Domains, value)
continue
}
ip, err := netip.ParseAddr(value)
if err != nil || ip.IsUnspecified() {
continue
}
ip = ip.Unmap()
serverAddresses = append(serverAddresses, ip)
// Prefer the first IPv4 server as ServerIP since our DNS listener is IPv4.
if !dnsSettings.ServerIP.IsValid() && ip.Is4() {
dnsSettings.ServerIP = ip
}
}
if err := scanner.Err(); err != nil {
return dnsSettings, err
return dnsSettings, serverAddresses, err
}
// default to 53 port
dnsSettings.ServerPort = DefaultPort
s.mu.Lock()
s.origNameservers = serverAddresses
s.mu.Unlock()
return dnsSettings, nil
return dnsSettings, serverAddresses, nil
}
func (s *systemConfigurator) getOriginalNameservers() []netip.Addr {
@@ -435,11 +464,15 @@ func (s *systemConfigurator) getPrimaryService() (string, string, error) {
router := ""
for scanner.Scan() {
text := scanner.Text()
parts := strings.SplitN(text, ":", 2)
if len(parts) != 2 {
continue
}
if strings.Contains(text, "PrimaryService") {
primaryService = strings.TrimSpace(strings.Split(text, ":")[1])
primaryService = strings.TrimSpace(parts[1])
}
if strings.Contains(text, "Router") {
router = strings.TrimSpace(strings.Split(text, ":")[1])
router = strings.TrimSpace(parts[1])
}
}
if err := scanner.Err(); err != nil && err != io.EOF {

View File

@@ -328,6 +328,120 @@ func removeTestDNSKey(key string) error {
return err
}
func TestParseSystemDNSSettings(t *testing.T) {
tests := []struct {
name string
output string
expectedDomains []string
expectedServers []netip.Addr
expectedIP netip.Addr
}{
{
name: "well_formed",
output: `<dictionary> {
DomainName : example.com
SearchDomains : <array> {
0 : example.com
1 : corp.example.com
}
ServerAddresses : <array> {
0 : 192.168.1.1
1 : fd00::53
}
}
`,
expectedDomains: []string{"example.com", "example.com", "corp.example.com"},
expectedServers: []netip.Addr{netip.MustParseAddr("192.168.1.1"), netip.MustParseAddr("fd00::53")},
expectedIP: netip.MustParseAddr("192.168.1.1"),
},
{
// entries without a value after the separator used to panic with
// "index out of range [1] with length 1"
name: "malformed_array_entries_skipped",
output: `<dictionary> {
SearchDomains : <array> {
0 :
(null)
1 : corp.example.com
}
ServerAddresses : <array> {
0 :
1 : 192.168.1.1
}
}
`,
expectedDomains: []string{"corp.example.com"},
expectedServers: []netip.Addr{netip.MustParseAddr("192.168.1.1")},
expectedIP: netip.MustParseAddr("192.168.1.1"),
},
{
name: "domain_name_without_value_skipped",
output: `<dictionary> {
DomainName :
ServerAddresses : <array> {
0 : 192.168.1.1
}
}
`,
expectedServers: []netip.Addr{netip.MustParseAddr("192.168.1.1")},
expectedIP: netip.MustParseAddr("192.168.1.1"),
},
{
name: "ipv6_first_prefers_ipv4_server_ip",
output: `<dictionary> {
ServerAddresses : <array> {
0 : fd00::53
1 : 192.168.1.1
}
}
`,
expectedServers: []netip.Addr{netip.MustParseAddr("fd00::53"), netip.MustParseAddr("192.168.1.1")},
expectedIP: netip.MustParseAddr("192.168.1.1"),
},
{
name: "invalid_and_unspecified_addresses_skipped",
output: `<dictionary> {
ServerAddresses : <array> {
0 : (null)
1 : 0.0.0.0
2 : 192.168.1.1
}
}
`,
expectedServers: []netip.Addr{netip.MustParseAddr("192.168.1.1")},
expectedIP: netip.MustParseAddr("192.168.1.1"),
},
{
name: "v4_mapped_address_unmapped",
output: `<dictionary> {
ServerAddresses : <array> {
0 : ::ffff:192.168.1.1
}
}
`,
expectedServers: []netip.Addr{netip.MustParseAddr("192.168.1.1")},
expectedIP: netip.MustParseAddr("192.168.1.1"),
},
{
name: "empty_output",
output: "",
},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
settings, servers, err := parseSystemDNSSettings([]byte(tc.output))
require.NoError(t, err, "parsing should not fail")
assert.Equal(t, tc.expectedDomains, settings.Domains, "domains should match")
assert.Equal(t, tc.expectedServers, servers, "server addresses should match")
assert.Equal(t, tc.expectedIP, settings.ServerIP, "server IP should match")
assert.Equal(t, DefaultPort, settings.ServerPort, "server port should default to 53")
})
}
}
func TestGetOriginalNameservers(t *testing.T) {
configurator := &systemConfigurator{
createdKeys: make(map[string]struct{}),

View File

@@ -23,6 +23,7 @@ import (
"golang.zx2c4.com/wireguard/tun/netstack"
"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
"github.com/netbirdio/netbird/client/anonymize"
nberrors "github.com/netbirdio/netbird/client/errors"
"github.com/netbirdio/netbird/client/firewall"
"github.com/netbirdio/netbird/client/firewall/firewalld"
@@ -34,7 +35,6 @@ import (
"github.com/netbirdio/netbird/client/iface/udpmux"
"github.com/netbirdio/netbird/client/iface/wgaddr"
"github.com/netbirdio/netbird/client/internal/acl"
"github.com/netbirdio/netbird/client/internal/approval"
"github.com/netbirdio/netbird/client/internal/debug"
"github.com/netbirdio/netbird/client/internal/dns"
dnsconfig "github.com/netbirdio/netbird/client/internal/dns/config"
@@ -136,8 +136,6 @@ type EngineConfig struct {
RosenpassPermissive bool
ServerSSHAllowed bool
ServerVNCAllowed bool
DisableVNCApproval *bool
EnableSSHRoot *bool
EnableSSHSFTP *bool
EnableSSHLocalPortForwarding *bool
@@ -236,9 +234,7 @@ type Engine struct {
networkMonitor *networkmonitor.NetworkMonitor
sshServer sshServer
vncSrv vncServer
approvalBroker *approval.Broker
sshServer sshServer
statusRecorder *peer.Status
@@ -345,7 +341,6 @@ func NewEngine(
TURNs: []*stun.URI{},
networkSerial: 0,
statusRecorder: services.StatusRecorder,
approvalBroker: approval.New(services.StatusRecorder),
stateManager: services.StateManager,
portForwardManager: portforward.NewManager(),
checks: services.Checks,
@@ -420,10 +415,6 @@ func (e *Engine) stopLocked() {
log.Warnf("failed to stop SSH server: %v", err)
}
if err := e.stopVNCServer(); err != nil {
log.Warnf("failed to stop VNC server: %v", err)
}
e.cleanupSSHConfig()
if e.ingressGatewayMgr != nil {
@@ -1247,7 +1238,6 @@ func (e *Engine) applyInfoFlags(info *system.Info) {
e.config.RosenpassEnabled,
e.config.RosenpassPermissive,
&e.config.ServerSSHAllowed,
&e.config.ServerVNCAllowed,
e.config.DisableClientRoutes,
e.config.DisableServerRoutes,
e.config.DisableDNS,
@@ -1303,10 +1293,6 @@ func (e *Engine) updateConfig(conf *mgmProto.PeerConfig) error {
}
}
if err := e.updateVNC(); err != nil {
log.Warnf("failed handling VNC server setup: %v", err)
}
state := e.statusRecorder.GetLocalPeerState()
state.IP = e.wgInterface.Address().String()
state.IPv6 = e.wgInterface.Address().IPv6String()
@@ -1400,6 +1386,7 @@ func (e *Engine) handleBundle(params *mgmProto.BundleParameters) (*mgmProto.JobR
bundleJobParams := debug.BundleConfig{
Anonymize: params.Anonymize,
AnonymizeLevel: anonymize.ParseLevel(params.AnonymizeLevel),
IncludeSystemInfo: true,
LogFileCount: uint32(params.LogFileCount),
}
@@ -1607,11 +1594,6 @@ func (e *Engine) reconcilePeers(networkMap *mgmProto.NetworkMap) ([]*mgmProto.Re
}
}
// VNC auth: always sync, including nil so cleared auth on the management
// side is applied locally, and so it isn't skipped on the RemotePeersIsEmpty
// cleanup path.
e.updateVNCServerAuth(networkMap.GetVncAuth())
// cleanup request, most likely our peer has been deleted
if networkMap.GetRemotePeersIsEmpty() {
err := e.removeAllPeers()
@@ -2893,16 +2875,3 @@ func decodeRelayIP(b []byte) netip.Addr {
}
return ip.Unmap()
}
// RespondApproval relays the user's decision for a pending approval to
// the broker. viewOnly is honoured only when accept is true. Returns
// true when the request_id matched a live prompt.
func (e *Engine) RespondApproval(requestID string, accept, viewOnly bool) bool {
if e == nil || e.approvalBroker == nil {
return false
}
return e.approvalBroker.Respond(requestID, approval.Decision{
Accept: accept,
ViewOnly: accept && viewOnly,
})
}

View File

@@ -12,10 +12,10 @@ import (
firewallManager "github.com/netbirdio/netbird/client/firewall/manager"
"github.com/netbirdio/netbird/client/iface/netstack"
nftypes "github.com/netbirdio/netbird/client/internal/netflow/types"
sshauth "github.com/netbirdio/netbird/client/ssh/auth"
sshconfig "github.com/netbirdio/netbird/client/ssh/config"
sshserver "github.com/netbirdio/netbird/client/ssh/server"
mgmProto "github.com/netbirdio/netbird/shared/management/proto"
sshauth "github.com/netbirdio/netbird/shared/sessionauth"
sshuserhash "github.com/netbirdio/netbird/shared/sshauth"
)
@@ -237,18 +237,22 @@ func (e *Engine) startSSHServer(jwtConfig *sshserver.JWTConfig) error {
return errors.New("wg interface not initialized")
}
wgAddr := e.wgInterface.Address()
serverConfig := &sshserver.Config{
HostKeyPEM: e.config.SSHKey,
JWT: jwtConfig,
NetstackNet: e.wgInterface.GetNet(),
NetworkValidation: wgAddr,
HostKeyPEM: e.config.SSHKey,
JWT: jwtConfig,
}
server := sshserver.New(serverConfig)
wgAddr := e.wgInterface.Address()
server.SetNetworkValidation(wgAddr)
netbirdIP := wgAddr.IP
listenAddr := netip.AddrPortFrom(netbirdIP, sshserver.InternalSSHPort)
if netstackNet := e.wgInterface.GetNet(); netstackNet != nil {
server.SetNetstackNet(netstackNet)
}
e.configureSSHServer(server)
if err := server.Start(e.ctx, listenAddr); err != nil {

View File

@@ -1,329 +0,0 @@
//go:build !js && !ios && !android
package internal
import (
"context"
"errors"
"fmt"
"net/netip"
log "github.com/sirupsen/logrus"
firewallManager "github.com/netbirdio/netbird/client/firewall/manager"
"github.com/netbirdio/netbird/client/internal/approval"
"github.com/netbirdio/netbird/client/internal/metrics"
nftypes "github.com/netbirdio/netbird/client/internal/netflow/types"
"github.com/netbirdio/netbird/client/internal/peer"
"github.com/netbirdio/netbird/client/vnc"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
mgmProto "github.com/netbirdio/netbird/shared/management/proto"
sshauth "github.com/netbirdio/netbird/shared/sessionauth"
sshuserhash "github.com/netbirdio/netbird/shared/sshauth"
)
type vncServer interface {
Start(ctx context.Context, addr netip.AddrPort, network netip.Prefix) error
AddListener(ctx context.Context, addr netip.AddrPort, network netip.Prefix) error
Stop() error
ActiveSessions() []vncserver.ActiveSessionInfo
}
func (e *Engine) setupVNCPortRedirection() error {
if e.firewall == nil || e.wgInterface == nil {
return nil
}
localAddr := e.wgInterface.Address().IP
if !localAddr.IsValid() {
return errors.New("invalid local NetBird address")
}
if err := e.firewall.AddInboundDNAT(localAddr, firewallManager.ProtocolTCP, vnc.ExternalPort, vnc.InternalPort); err != nil {
return fmt.Errorf("add VNC port redirection: %w", err)
}
log.Infof("VNC port redirection: %s:%d -> %s:%d", localAddr, vnc.ExternalPort, localAddr, vnc.InternalPort)
if wgAddr := e.wgInterface.Address(); wgAddr.HasIPv6() {
v6 := wgAddr.IPv6
if err := e.firewall.AddInboundDNAT(v6, firewallManager.ProtocolTCP, vnc.ExternalPort, vnc.InternalPort); err != nil {
log.Warnf("failed to add IPv6 VNC port redirection: %v", err)
} else {
log.Infof("VNC port redirection: [%s]:%d -> [%s]:%d", v6, vnc.ExternalPort, v6, vnc.InternalPort)
}
}
return nil
}
func (e *Engine) cleanupVNCPortRedirection() error {
if e.firewall == nil || e.wgInterface == nil {
return nil
}
localAddr := e.wgInterface.Address().IP
if !localAddr.IsValid() {
return errors.New("invalid local NetBird address")
}
if err := e.firewall.RemoveInboundDNAT(localAddr, firewallManager.ProtocolTCP, vnc.ExternalPort, vnc.InternalPort); err != nil {
return fmt.Errorf("remove VNC port redirection: %w", err)
}
if wgAddr := e.wgInterface.Address(); wgAddr.HasIPv6() {
if err := e.firewall.RemoveInboundDNAT(wgAddr.IPv6, firewallManager.ProtocolTCP, vnc.ExternalPort, vnc.InternalPort); err != nil {
log.Debugf("failed to remove IPv6 VNC port redirection: %v", err)
}
}
return nil
}
// updateVNC handles starting/stopping the VNC server based on the config flag.
func (e *Engine) updateVNC() error {
if !e.config.ServerVNCAllowed {
if e.vncSrv != nil {
log.Info("VNC server disabled, stopping")
}
return e.stopVNCServer()
}
if e.config.BlockInbound {
log.Info("VNC server disabled because inbound connections are blocked")
return e.stopVNCServer()
}
if e.vncSrv != nil {
return nil
}
return e.startVNCServer()
}
func (e *Engine) startVNCServer() error {
if e.wgInterface == nil {
return errors.New("wg interface not initialized")
}
capturer, injector, ok := newPlatformVNC()
if !ok {
log.Debug("VNC server not supported on this platform")
return nil
}
netbirdIP := e.wgInterface.Address().IP
var sessionRecorder func(vncserver.SessionTick)
if e.clientMetrics != nil {
sessionRecorder = func(t vncserver.SessionTick) {
e.clientMetrics.RecordVNCSessionTick(e.ctx, metrics.VNCSessionTick{
Period: t.Period,
BytesOut: t.BytesOut,
Writes: t.Writes,
FBUs: t.FBUs,
MaxFBUBytes: t.MaxFBUBytes,
MaxFBURects: t.MaxFBURects,
MaxWriteBytes: t.MaxWriteBytes,
WriteNanos: t.WriteNanos,
})
}
}
serviceMode := vncNeedsServiceMode()
if serviceMode {
log.Info("VNC: running as system service, enabling service mode (per-session agent proxy)")
}
requireApproval := e.config.DisableVNCApproval == nil || !*e.config.DisableVNCApproval
srv := vncserver.New(vncserver.Config{
Capturer: capturer,
Injector: injector,
IdentityKey: e.config.WgPrivateKey[:],
ServiceMode: serviceMode,
SessionRecorder: sessionRecorder,
NetstackNet: e.wgInterface.GetNet(),
RequireApproval: requireApproval,
Approver: &vncApprover{broker: e.approvalBroker, statusRecorder: e.statusRecorder},
// Session start/stop is invisible to the peer status recorder, so push a
// snapshot ourselves; otherwise the UI's session list goes stale until an
// unrelated peer change happens to fire one.
OnSessionsChanged: e.statusRecorder.NotifyStateChange,
})
listenAddr := netip.AddrPortFrom(netbirdIP, vnc.InternalPort)
network := e.wgInterface.Address().Network
if err := srv.Start(e.ctx, listenAddr, network); err != nil {
return fmt.Errorf("start VNC server: %w", err)
}
if wgAddr := e.wgInterface.Address(); wgAddr.HasIPv6() {
v6Addr := netip.AddrPortFrom(wgAddr.IPv6, vnc.InternalPort)
if err := srv.AddListener(e.ctx, v6Addr, wgAddr.IPv6Net); err != nil {
log.Warnf("failed to add IPv6 VNC listener: %v", err)
}
}
e.vncSrv = srv
if netstackNet := e.wgInterface.GetNet(); netstackNet != nil {
if registrar, ok := e.firewall.(interface {
RegisterNetstackService(protocol nftypes.Protocol, port uint16)
}); ok {
registrar.RegisterNetstackService(nftypes.TCP, vnc.InternalPort)
log.Debugf("registered VNC service with netstack for TCP:%d", vnc.InternalPort)
}
}
if err := e.setupVNCPortRedirection(); err != nil {
log.Warnf("setup VNC port redirection: %v", err)
}
log.Info("VNC server enabled")
return nil
}
// updateVNCServerAuth updates VNC fine-grained access control from management.
// A nil vncAuth clears all authorized users and session pubkeys so management
// can revoke access by omitting the field on the next sync.
func (e *Engine) updateVNCServerAuth(vncAuth *mgmProto.VNCAuth) {
if e.vncSrv == nil {
return
}
vncSrv, ok := e.vncSrv.(*vncserver.Server)
if !ok {
return
}
if vncAuth == nil {
vncSrv.UpdateVNCAuth(&sshauth.Config{})
return
}
protoUsers := vncAuth.GetAuthorizedUsers()
authorizedUsers := make([]sshuserhash.UserIDHash, len(protoUsers))
for i, hash := range protoUsers {
if len(hash) != 16 {
log.Warnf("invalid VNC auth hash length %d, expected 16", len(hash))
return
}
authorizedUsers[i] = sshuserhash.UserIDHash(hash)
}
machineUsers := make(map[string][]uint32)
for osUser, indexes := range vncAuth.GetMachineUsers() {
machineUsers[osUser] = indexes.GetIndexes()
}
sessionPubKeys := make([]sshauth.SessionPubKey, 0, len(vncAuth.GetSessionPubKeys()))
for _, pk := range vncAuth.GetSessionPubKeys() {
pub := pk.GetPubKey()
if len(pub) != 32 {
log.Warnf("VNC session pubkey wrong length %d", len(pub))
continue
}
hash := pk.GetUserIdHash()
if len(hash) != 16 {
log.Warnf("VNC session user id hash wrong length %d", len(hash))
continue
}
sessionPubKeys = append(sessionPubKeys, sshauth.SessionPubKey{
PubKey: pub,
UserIDHash: sshuserhash.UserIDHash(hash),
DisplayName: pk.GetDisplayName(),
})
}
vncSrv.UpdateVNCAuth(&sshauth.Config{
AuthorizedUsers: authorizedUsers,
MachineUsers: machineUsers,
SessionPubKeys: sessionPubKeys,
})
}
// GetVNCServerStatus returns whether the VNC server is running and the list
// of active VNC sessions. The pointer is captured under syncMsgMux so a
// concurrent updateVNC/stopVNCServer cannot swap it out between the nil
// check and the ActiveSessions call.
func (e *Engine) GetVNCServerStatus() (enabled bool, sessions []vncserver.ActiveSessionInfo) {
e.syncMsgMux.Lock()
vncSrv := e.vncSrv
e.syncMsgMux.Unlock()
if vncSrv == nil {
return false, nil
}
return true, vncSrv.ActiveSessions()
}
func (e *Engine) stopVNCServer() error {
if e.vncSrv == nil {
return nil
}
if err := e.cleanupVNCPortRedirection(); err != nil {
log.Warnf("cleanup VNC port redirection: %v", err)
}
if e.wgInterface != nil && e.wgInterface.GetNet() != nil {
if registrar, ok := e.firewall.(interface {
UnregisterNetstackService(protocol nftypes.Protocol, port uint16)
}); ok {
registrar.UnregisterNetstackService(nftypes.TCP, vnc.InternalPort)
}
}
log.Info("stopping VNC server")
err := e.vncSrv.Stop()
e.vncSrv = nil
if err != nil {
return fmt.Errorf("stop VNC server: %w", err)
}
return nil
}
// vncApprover adapts the generic approval.Broker for the VNC server.
type vncApprover struct {
broker *approval.Broker
statusRecorder *peer.Status
}
func (a *vncApprover) Request(ctx context.Context, info vncserver.ApprovalInfo) (vncserver.ApprovalDecision, error) {
// Resolve the source overlay IP to a peer FQDN for the prompt label.
if info.PeerName == "" && info.SourceIP != "" && a.statusRecorder != nil {
if fqdn, ok := a.statusRecorder.PeerByIP(info.SourceIP); ok {
info.PeerName = fqdn
}
}
subject := fmt.Sprintf("VNC connection from %s", displayPeer(info))
meta := map[string]string{
"peer_name": info.PeerName,
"peer_pubkey": info.PeerPubKey,
"source_ip": info.SourceIP,
"mode": info.Mode,
"username": info.Username,
"initiator": info.Initiator,
}
d, err := a.broker.Request(ctx, approval.Prompt{
Kind: approval.KindVNC,
Subject: subject,
Metadata: meta,
})
if err != nil {
return vncserver.ApprovalDecision{}, err
}
return vncserver.ApprovalDecision{ViewOnly: d.ViewOnly}, nil
}
func displayPeer(info vncserver.ApprovalInfo) string {
if info.Initiator != "" {
return info.Initiator
}
if info.PeerName != "" {
return info.PeerName
}
if info.SourceIP != "" {
return info.SourceIP
}
if info.PeerPubKey != "" {
return info.PeerPubKey
}
return "unknown peer"
}

View File

@@ -1,31 +0,0 @@
//go:build freebsd
package internal
import (
"fmt"
log "github.com/sirupsen/logrus"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
// newConsoleVNC builds the FreeBSD console fallback: vt(4) framebuffer
// for capture, /dev/uinput for input. The uinput device requires the
// `uinput` kernel module (`kldload uinput`); without it, input init
// fails and we drop to a stub injector so the user still gets a
// view-only screen mirror.
func newConsoleVNC() (vncserver.ScreenCapturer, vncserver.InputInjector, error) {
poller := vncserver.NewFBPoller("")
w, h := poller.Width(), poller.Height()
if w == 0 || h == 0 {
poller.Close()
return nil, nil, fmt.Errorf("vt framebuffer init failed (vt may not allow mmap on this driver)")
}
if inj, err := vncserver.NewUInputInjector(w, h); err == nil {
return poller, inj, nil
} else {
log.Infof("VNC console: uinput unavailable (%v); view-only mode. Run `kldload uinput` to enable input.", err)
return poller, &vncserver.StubInputInjector{}, nil
}
}

View File

@@ -1,30 +0,0 @@
//go:build linux && !android
package internal
import (
"fmt"
log "github.com/sirupsen/logrus"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
// newConsoleVNC builds a framebuffer + uinput VNC backend for boxes
// without a running X server. Used as the auto-fallback when
// newPlatformVNC can't reach X. Returns an error when /dev/fb0 or
// /dev/uinput aren't usable so the caller can drop back to a stub.
func newConsoleVNC() (vncserver.ScreenCapturer, vncserver.InputInjector, error) {
poller := vncserver.NewFBPoller("")
w, h := poller.Width(), poller.Height()
if w == 0 || h == 0 {
poller.Close()
return nil, nil, fmt.Errorf("framebuffer capturer init failed (is /dev/fb0 readable?)")
}
inj, err := vncserver.NewUInputInjector(w, h)
if err != nil {
log.Debugf("uinput unavailable, falling back to view-only VNC: %v", err)
return poller, &vncserver.StubInputInjector{}, nil
}
return poller, inj, nil
}

View File

@@ -1,33 +0,0 @@
//go:build darwin && !ios
package internal
import (
"os"
log "github.com/sirupsen/logrus"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
func newPlatformVNC() (vncserver.ScreenCapturer, vncserver.InputInjector, bool) {
capturer := vncserver.NewMacPoller()
// No permission request here. Screen Recording is a user-scope TCC service,
// so a request from this process is dropped when it runs as a LaunchDaemon:
// no prompt appears and NetBird never even shows up in the Screen Recording
// list. The per-user agent asks instead, see newAgentResources.
injector, err := vncserver.NewMacInputInjector()
if err != nil {
log.Debugf("VNC: macOS input injector: %v", err)
return capturer, &vncserver.StubInputInjector{}, true
}
return capturer, injector, true
}
// vncNeedsServiceMode reports whether the running process is a system
// LaunchDaemon (root, parented by launchd). Daemons sit in the global
// bootstrap namespace and cannot talk to WindowServer; we route capture
// through a per-user agent in that case.
func vncNeedsServiceMode() bool {
return os.Geteuid() == 0 && os.Getppid() == 1
}

View File

@@ -1,23 +0,0 @@
//go:build js || ios || android
package internal
import (
log "github.com/sirupsen/logrus"
mgmProto "github.com/netbirdio/netbird/shared/management/proto"
)
type vncServer interface{}
func (e *Engine) updateVNC() error { return nil }
func (e *Engine) updateVNCServerAuth(auth *mgmProto.VNCAuth) {
if auth == nil {
return
}
log.Debugf("ignoring VNC auth push on platform without a VNC server: %d session pubkeys, %d authorized users",
len(auth.GetSessionPubKeys()), len(auth.GetAuthorizedUsers()))
}
func (e *Engine) stopVNCServer() error { return nil }

View File

@@ -1,13 +0,0 @@
//go:build windows
package internal
import vncserver "github.com/netbirdio/netbird/client/vnc/server"
func newPlatformVNC() (vncserver.ScreenCapturer, vncserver.InputInjector, bool) {
return vncserver.NewDesktopCapturer(), vncserver.NewWindowsInputInjector(), true
}
func vncNeedsServiceMode() bool {
return vncserver.GetCurrentSessionID() == 0
}

View File

@@ -1,35 +0,0 @@
//go:build (linux && !android) || freebsd
package internal
import (
log "github.com/sirupsen/logrus"
vncserver "github.com/netbirdio/netbird/client/vnc/server"
)
func newPlatformVNC() (vncserver.ScreenCapturer, vncserver.InputInjector, bool) {
// Prefer X11 when an X server is reachable. NewX11InputInjector probes
// DISPLAY (and /proc) eagerly, so a non-nil error here means no X.
injector, err := vncserver.NewX11InputInjector("", "", "")
if err == nil {
return vncserver.NewX11Poller("", ""), injector, true
}
log.Debugf("VNC: X11 not available: %v", err)
// Fallback for headless / pre-X states (kernel console, login manager
// without X, physical server in recovery): stream the framebuffer and
// inject input via /dev/uinput.
consoleCap, consoleInj, err := newConsoleVNC()
if err == nil {
log.Infof("VNC: using framebuffer console capture (%dx%d)", consoleCap.Width(), consoleCap.Height())
return consoleCap, consoleInj, true
}
log.Debugf("VNC: framebuffer console fallback unavailable: %v", err)
return &vncserver.StubCapturer{}, &vncserver.StubInputInjector{}, false
}
func vncNeedsServiceMode() bool {
return false
}

View File

@@ -4,11 +4,17 @@ package metrics
type ConnectionType string
const (
// ConnectionTypeICE represents a direct peer-to-peer connection using ICE
ConnectionTypeICE ConnectionType = "ice"
// ConnectionTypeICEP2P represents a direct peer-to-peer connection using ICE
ConnectionTypeICEP2P ConnectionType = "ice_p2p"
// ConnectionTypeICETurn represents an ICE connection through a TURN server
ConnectionTypeICETurn ConnectionType = "ice_turn"
// ConnectionTypeRelay represents a relayed connection
ConnectionTypeRelay ConnectionType = "relay"
// ConnectionTypeUnknown represents a connection with no active transport. It is not pushed.
ConnectionTypeUnknown ConnectionType = "unknown"
)
// String returns the string representation of the connection type

View File

@@ -120,36 +120,6 @@ func (m *influxDBMetrics) RecordSyncDuration(_ context.Context, agentInfo AgentI
m.trimLocked()
}
func (m *influxDBMetrics) RecordVNCSessionTick(_ context.Context, agentInfo AgentInfo, tick VNCSessionTick) {
tags := fmt.Sprintf("deployment_type=%s,version=%s,os=%s,arch=%s,peer_id=%s",
agentInfo.DeploymentType.String(),
agentInfo.Version,
agentInfo.OS,
agentInfo.Arch,
agentInfo.peerID,
)
m.mu.Lock()
defer m.mu.Unlock()
m.samples = append(m.samples, influxSample{
measurement: "netbird_vnc_traffic",
tags: tags,
fields: map[string]float64{
"period_seconds": tick.Period.Seconds(),
"bytes_out": float64(tick.BytesOut),
"writes": float64(tick.Writes),
"fbus": float64(tick.FBUs),
"max_fbu_bytes": float64(tick.MaxFBUBytes),
"max_fbu_rects": float64(tick.MaxFBURects),
"max_write_bytes": float64(tick.MaxWriteBytes),
"write_time_seconds": float64(tick.WriteNanos) / 1e9,
},
timestamp: time.Now(),
})
m.trimLocked()
}
func (m *influxDBMetrics) RecordSyncPhase(_ context.Context, agentInfo AgentInfo, phase string, duration time.Duration) {
tags := fmt.Sprintf("deployment_type=%s,version=%s,os=%s,arch=%s,peer_id=%s,phase=%s",
agentInfo.DeploymentType.String(),

View File

@@ -28,7 +28,7 @@ func TestInfluxDBMetrics_RecordAndExport(t *testing.T) {
WgHandshakeSuccess: time.Now().Add(-1 * time.Second),
}
m.RecordConnectionStages(context.Background(), agentInfo, "pair123", ConnectionTypeICE, false, ts)
m.RecordConnectionStages(context.Background(), agentInfo, "pair123", ConnectionTypeICEP2P, false, ts)
var buf bytes.Buffer
err := m.Export(&buf)
@@ -60,7 +60,7 @@ func TestInfluxDBMetrics_ExportDeterministicFieldOrder(t *testing.T) {
// Record multiple times and verify consistent field order
for i := 0; i < 10; i++ {
m.RecordConnectionStages(context.Background(), agentInfo, "pair123", ConnectionTypeICE, false, ts)
m.RecordConnectionStages(context.Background(), agentInfo, "pair123", ConnectionTypeICEP2P, false, ts)
}
var buf bytes.Buffer

View File

@@ -56,14 +56,33 @@ Measurement: `netbird_peer_connection`
Tags:
- `deployment_type`: "cloud" | "selfhosted" | "unknown"
- `connection_type`: "ice" | "relay"
- `connection_type`: "ice_p2p" | "ice_turn" | "relay" (see below)
- `attempt_type`: "initial" | "reconnection"
- `version`: NetBird version string
- `os`: Operating system (linux, darwin, windows, android, ios, etc.)
- `arch`: CPU architecture (amd64, arm64, etc.)
- `peer_id`: anonymised peer identifier (truncated SHA-256 of the WireGuard public key)
- `connection_pair_id`: deterministic identifier for the peer pair, identical on both sides
**Note:** `SignalingReceived` is set when the first offer or answer arrives from the remote peer (in both initial and reconnection paths). It excludes the potentially unbounded wait for the remote peer to come online.
#### `connection_type` values
Derived from the connection priority (`conntype.ConnPriority`) by `metricsConnType` in `client/internal/peer/conn.go`:
| Value | Priority | Traffic is |
|-------|----------|------------|
| `ice_p2p` | `ICEP2P` | direct peer-to-peer |
| `ice_turn` | `ICETurn` | relayed, through a TURN server |
| `relay` | `Relay` | relayed, through a NetBird relay |
| `unknown` | `None` or unrecognised | no active transport — **the sample is not pushed** |
**Direct traffic is `ice_p2p` only.** `ice_turn` is relayed despite being negotiated by ICE, matching `Conn.isRelayed`.
`None` means no transport is active: not established yet, or reset after a relay drop or a peer-state reset. Such a sample cannot be attributed to a transport, so `recordConnectionMetrics` drops it instead of pushing it — `unknown` therefore never appears in the bucket. Connection counts are counts of connections whose transport was known at sampling time.
**Samples recorded before 0.77 used a single `ice` value** which covered `ICEP2P`, `ICETurn` *and* `None`, so historical `ice` samples overstate direct connections by an unknown amount and must not be compared with `ice_p2p`.
### Sync Duration
Measurement: `netbird_sync`

View File

@@ -63,11 +63,6 @@ type metricsImplementation interface {
// RecordLoginDuration records how long the login to management took
RecordLoginDuration(ctx context.Context, agentInfo AgentInfo, duration time.Duration, success bool)
// RecordVNCSessionTick records a periodic snapshot of one VNC
// session's wire activity. Called once per metricsConn tick interval
// (and once at session close), only when the tick saw activity.
RecordVNCSessionTick(ctx context.Context, agentInfo AgentInfo, tick VNCSessionTick)
// Export exports metrics in InfluxDB line protocol format
Export(w io.Writer) error
@@ -87,21 +82,6 @@ type ClientMetrics struct {
pushCancel context.CancelFunc
}
// VNCSessionTick is one sampling slice of a VNC session's wire activity.
// BytesOut / Writes / FBUs / WriteNanos are deltas observed during this
// tick; Max* fields are the high-water marks observed during the tick.
// Period is the wall-clock duration the deltas cover.
type VNCSessionTick struct {
Period time.Duration
BytesOut uint64
Writes uint64
FBUs uint64
MaxFBUBytes uint64
MaxFBURects uint64
MaxWriteBytes uint64
WriteNanos uint64
}
// ConnectionStageTimestamps holds timestamps for each connection stage
type ConnectionStageTimestamps struct {
SignalingReceived time.Time // First signal received from remote peer (both initial and reconnection)
@@ -151,18 +131,6 @@ func (c *ClientMetrics) RecordSyncDuration(ctx context.Context, duration time.Du
c.impl.RecordSyncDuration(ctx, agentInfo, duration)
}
// RecordVNCSessionTick records a periodic snapshot of one VNC session.
func (c *ClientMetrics) RecordVNCSessionTick(ctx context.Context, tick VNCSessionTick) {
if c == nil {
return
}
c.mu.RLock()
agentInfo := c.agentInfo
c.mu.RUnlock()
c.impl.RecordVNCSessionTick(ctx, agentInfo, tick)
}
// RecordSyncPhase records the duration of a single sub-phase of sync processing
func (c *ClientMetrics) RecordSyncPhase(ctx context.Context, phase string, duration time.Duration) {
if c == nil {

View File

@@ -76,9 +76,6 @@ func (m *mockMetrics) RecordSyncPhase(_ context.Context, _ AgentInfo, _ string,
func (m *mockMetrics) RecordLoginDuration(_ context.Context, _ AgentInfo, _ time.Duration, _ bool) {
}
func (m *mockMetrics) RecordVNCSessionTick(_ context.Context, _ AgentInfo, _ VNCSessionTick) {
}
func (m *mockMetrics) Export(w io.Writer) error {
if m.exportData != "" {
_, err := w.Write([]byte(m.exportData))

View File

@@ -307,6 +307,8 @@ func (conn *Conn) Close(signalToRemote bool) {
if conn.wgWatcherCancel != nil {
conn.wgWatcherCancel()
conn.wgWatcher = nil
conn.wgWatcherCancel = nil
}
conn.workerRelay.CloseConn()
if conn.workerICE != nil {
@@ -959,12 +961,9 @@ func (conn *Conn) recordConnectionMetrics() {
priority := conn.currentConnPriority
conn.mu.Unlock()
var connType metrics.ConnectionType
switch priority {
case conntype.Relay:
connType = metrics.ConnectionTypeRelay
default:
connType = metrics.ConnectionTypeICE
connType := metricsConnType(priority)
if connType == metrics.ConnectionTypeUnknown {
return
}
// Record metrics with timestamps - duration calculation happens in metrics package
@@ -1065,3 +1064,16 @@ func boolToConnStatus(connected bool) guard.ConnStatus {
}
return guard.ConnStatusDisconnected
}
func metricsConnType(priority conntype.ConnPriority) metrics.ConnectionType {
switch priority {
case conntype.Relay:
return metrics.ConnectionTypeRelay
case conntype.ICETurn:
return metrics.ConnectionTypeICETurn
case conntype.ICEP2P:
return metrics.ConnectionTypeICEP2P
default:
return metrics.ConnectionTypeUnknown
}
}

View File

@@ -11,6 +11,8 @@ import (
"github.com/stretchr/testify/assert"
"github.com/netbirdio/netbird/client/iface"
"github.com/netbirdio/netbird/client/internal/metrics"
"github.com/netbirdio/netbird/client/internal/peer/conntype"
"github.com/netbirdio/netbird/client/internal/peer/dispatcher"
"github.com/netbirdio/netbird/client/internal/peer/guard"
"github.com/netbirdio/netbird/client/internal/peer/ice"
@@ -386,3 +388,33 @@ func TestConn_onWGDisconnected_NoEscalationWithoutRosenpass(t *testing.T) {
}
assert.Empty(t, disconnected, "escalation must be limited to rosenpass connections")
}
func TestMetricsConnType(t *testing.T) {
tests := []struct {
name string
priority conntype.ConnPriority
expected metrics.ConnectionType
}{
{"relay", conntype.Relay, metrics.ConnectionTypeRelay},
{"ice over turn is relayed, not p2p", conntype.ICETurn, metrics.ConnectionTypeICETurn},
{"direct p2p", conntype.ICEP2P, metrics.ConnectionTypeICEP2P},
{"unset priority is unknown, not p2p", conntype.None, metrics.ConnectionTypeUnknown},
{"unrecognised priority is unknown", conntype.ConnPriority(99), metrics.ConnectionTypeUnknown},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
assert.Equal(t, tc.expected, metricsConnType(tc.priority))
})
}
}
func TestMetricsConnType_RelayedMatchesIsRelayed(t *testing.T) {
for _, priority := range []conntype.ConnPriority{conntype.None, conntype.Relay, conntype.ICETurn, conntype.ICEP2P} {
conn := &Conn{currentConnPriority: priority}
tag := metricsConnType(priority)
relayedTag := tag == metrics.ConnectionTypeRelay || tag == metrics.ConnectionTypeICETurn
assert.Equal(t, conn.isRelayed(), relayedTag,
"priority %s: isRelayed and the %q metric tag must agree", priority, tag)
}
}

View File

@@ -1330,15 +1330,6 @@ func (d *Status) SubscribeToEvents() *EventSubscription {
}
}
// HasEventSubscribers reports whether any client is currently subscribed
// to the daemon's SystemEvent stream. Used by the VNC approval broker to
// fail closed when no UI is connected to prompt the user.
func (d *Status) HasEventSubscribers() bool {
d.eventMux.Lock()
defer d.eventMux.Unlock()
return len(d.eventStreams) > 0
}
// UnsubscribeFromEvents removes an event subscription
func (d *Status) UnsubscribeFromEvents(sub *EventSubscription) {
if sub == nil {

View File

@@ -70,8 +70,6 @@ type ConfigInput struct {
StateFilePath string
PreSharedKey *string
ServerSSHAllowed *bool
ServerVNCAllowed *bool
DisableVNCApproval *bool
EnableSSHRoot *bool
EnableSSHSFTP *bool
EnableSSHLocalPortForwarding *bool
@@ -126,8 +124,6 @@ type Config struct {
RosenpassEnabled bool
RosenpassPermissive bool
ServerSSHAllowed *bool
ServerVNCAllowed *bool
DisableVNCApproval *bool
EnableSSHRoot *bool
EnableSSHSFTP *bool
EnableSSHLocalPortForwarding *bool
@@ -460,33 +456,6 @@ func (config *Config) apply(input ConfigInput) (updated bool, err error) {
updated = true
}
if input.ServerVNCAllowed != nil {
if config.ServerVNCAllowed == nil || *input.ServerVNCAllowed != *config.ServerVNCAllowed {
if *input.ServerVNCAllowed {
log.Infof("enabling VNC server")
} else {
log.Infof("disabling VNC server")
}
config.ServerVNCAllowed = input.ServerVNCAllowed
updated = true
}
} else if config.ServerVNCAllowed == nil {
config.ServerVNCAllowed = util.False()
updated = true
}
if input.DisableVNCApproval != nil {
if config.DisableVNCApproval == nil || *input.DisableVNCApproval != *config.DisableVNCApproval {
if *input.DisableVNCApproval {
log.Infof("disabling VNC connection approval prompt")
} else {
log.Infof("enabling VNC connection approval prompt")
}
config.DisableVNCApproval = input.DisableVNCApproval
updated = true
}
}
if input.EnableSSHRoot != nil && (config.EnableSSHRoot == nil || *input.EnableSSHRoot != *config.EnableSSHRoot) {
if *input.EnableSSHRoot {
log.Infof("enabling SSH root login")
@@ -743,8 +712,6 @@ func (config *Config) applyMDMPolicy(policy *mdm.Policy) {
}
applyBool(mdm.KeyAllowServerSSH, func(v bool) { bv := v; config.ServerSSHAllowed = &bv })
applyBool(mdm.KeyAllowServerVNC, func(v bool) { bv := v; config.ServerVNCAllowed = &bv })
applyBool(mdm.KeyDisableVNCApproval, func(v bool) { bv := v; config.DisableVNCApproval = &bv })
applyBool(mdm.KeyDisableClientRoutes, func(v bool) { config.DisableClientRoutes = v })
applyBool(mdm.KeyDisableServerRoutes, func(v bool) { config.DisableServerRoutes = v })
applyBool(mdm.KeyBlockInbound, func(v bool) { config.BlockInbound = v })

View File

@@ -130,36 +130,6 @@ func TestApply_MDMBoolKeysOverrideOnDiskValue(t *testing.T) {
assert.True(t, cfg.Policy().HasKey(mdm.KeyRosenpassEnabled))
}
func TestApply_MDMVNCKeys(t *testing.T) {
tmp := filepath.Join(t.TempDir(), "config.json")
// Seed without MDM: VNC off, approval prompt on.
withMDMPolicy(t, mdm.NewPolicy(nil))
_, err := UpdateOrCreateConfig(ConfigInput{
ConfigPath: tmp,
ServerVNCAllowed: boolPtr(false),
DisableVNCApproval: boolPtr(false),
})
require.NoError(t, err)
// MDM enforces VNC on and disables the approval prompt.
withMDMPolicy(t, mdm.NewPolicy(map[string]any{
mdm.KeyAllowServerVNC: true,
mdm.KeyDisableVNCApproval: true,
}))
cfg, err := UpdateOrCreateConfig(ConfigInput{ConfigPath: tmp})
require.NoError(t, err)
require.NotNil(t, cfg)
require.NotNil(t, cfg.ServerVNCAllowed)
assert.True(t, *cfg.ServerVNCAllowed, "MDM override should flip on-disk false to true")
require.NotNil(t, cfg.DisableVNCApproval)
assert.True(t, *cfg.DisableVNCApproval)
assert.True(t, cfg.Policy().HasKey(mdm.KeyAllowServerVNC))
assert.True(t, cfg.Policy().HasKey(mdm.KeyDisableVNCApproval))
}
func TestApply_MDMLazyConnection(t *testing.T) {
cases := []struct {
name string

View File

@@ -2,54 +2,183 @@ package ipfwdstate
import (
"fmt"
"sync"
log "github.com/sirupsen/logrus"
"github.com/netbirdio/netbird/client/internal/routemanager/systemops"
)
// IPForwardingState is a struct that keeps track of the IP forwarding state.
// todo: read initial state of the IP forwarding from the system and reset the state based on it.
// todo: separate v4/v6 forwarding state, since the sysctls are independent
// (net.ipv4.ip_forward vs net.ipv6.conf.all.forwarding). Currently the nftables
// manager shares one instance between both routers, which works only because
// EnableIPForwarding enables both sysctls in a single call.
// IPForwardingState tracks v4 and v6 IP-forwarding sysctl enables with
// independent refcounts so a v4-only routing setup doesn't flip v6 sysctls.
type IPForwardingState struct {
enabledCounter int
mu sync.Mutex
v4Count int
v6Count int
// routingV4/routingV6 track whether the routing path currently holds a
// reference, so repeated EnableRouting calls (one per network-map update)
// hold at most one reference per family and an unpaired DisableRouting
// can't release references held by DNAT rules.
routingV4 bool
routingV6 bool
wgIfaceName string
v6Saved map[string]int
}
func NewIPForwardingState() *IPForwardingState {
return &IPForwardingState{}
// NewIPForwardingState returns a state tracker for the IP-forwarding sysctls.
// wgIfaceName is excluded from the per-interface accept_ra handling.
func NewIPForwardingState(wgIfaceName string) *IPForwardingState {
return &IPForwardingState{wgIfaceName: wgIfaceName}
}
func (f *IPForwardingState) RequestForwarding() error {
if f.enabledCounter != 0 {
f.enabledCounter++
// Counts returns the current v4 and v6 refcounts. Intended for diagnostics
// and tests.
func (f *IPForwardingState) Counts() (v4, v6 int) {
f.mu.Lock()
defer f.mu.Unlock()
return f.v4Count, f.v6Count
}
// RequestRouting takes the forwarding references for the routing path. It is
// idempotent: while routing already holds a reference, further calls don't
// increment the refcounts, and a v4-only request releases a previously held v6
// reference. A v6 sysctl failure is logged and not returned so it can't take
// down v4 routing (the sysctl may be unwritable, e.g. read-only /proc/sys or
// IPv6 disabled on the kernel command line); v6 is retried on the next call.
func (f *IPForwardingState) RequestRouting(v6 bool) error {
f.mu.Lock()
defer f.mu.Unlock()
if !f.routingV4 {
if err := f.requestV4(); err != nil {
return err
}
f.routingV4 = true
}
if !v6 {
if !f.routingV6 {
return nil
}
f.routingV6 = false
return f.releaseV6()
}
if f.routingV6 {
return nil
}
if err := systemops.EnableIPForwarding(); err != nil {
return fmt.Errorf("failed to enable IP forwarding with sysctl: %w", err)
if err := f.requestV6(); err != nil {
log.Warnf("enable IPv6 forwarding for routing: %v", err)
return nil
}
f.enabledCounter = 1
log.Info("IP forwarding enabled")
f.routingV6 = true
return nil
}
func (f *IPForwardingState) ReleaseForwarding() error {
if f.enabledCounter == 0 {
return nil
// ReleaseRouting releases the references RequestRouting holds. Calls without a
// held reference are no-ops.
func (f *IPForwardingState) ReleaseRouting() error {
f.mu.Lock()
defer f.mu.Unlock()
if f.routingV4 {
f.routingV4 = false
f.releaseV4()
}
if f.enabledCounter > 1 {
f.enabledCounter--
return nil
if f.routingV6 {
f.routingV6 = false
return f.releaseV6()
}
// if failed to disable IP forwarding we anyway decrement the counter
f.enabledCounter = 0
// todo call systemops.DisableIPForwarding()
return nil
}
// RequestForwarding enables the family's forwarding sysctl on first request.
func (f *IPForwardingState) RequestForwarding(v6 bool) error {
f.mu.Lock()
defer f.mu.Unlock()
if v6 {
return f.requestV6()
}
return f.requestV4()
}
// ReleaseForwarding decrements the family counter. The last v6 release restores
// what enable captured. v4 stays on: net.ipv4.ip_forward is co-owned by other
// tooling (docker, k8s, libvirt).
func (f *IPForwardingState) ReleaseForwarding(v6 bool) error {
f.mu.Lock()
defer f.mu.Unlock()
if v6 {
return f.releaseV6()
}
f.releaseV4()
return nil
}
func (f *IPForwardingState) requestV4() error {
if f.v4Count == 0 {
if err := systemops.EnableV4IPForwarding(); err != nil {
return fmt.Errorf("enable IPv4 forwarding: %w", err)
}
log.Info("IPv4 forwarding enabled")
}
f.v4Count++
return nil
}
func (f *IPForwardingState) releaseV4() {
if f.v4Count > 0 {
f.v4Count--
}
}
func (f *IPForwardingState) requestV6() error {
if f.v6Count == 0 {
saved, err := systemops.EnableV6IPForwarding(f.wgIfaceName)
if err != nil {
if rerr := systemops.DisableV6IPForwarding(saved); rerr != nil {
log.Warnf("rollback partial v6 sysctls: %v", rerr)
}
return fmt.Errorf("enable IPv6 forwarding: %w", err)
}
// A failed restore on a previous release keeps its saved values; those
// are the true originals, so keep them over what this enable captured.
if f.v6Saved == nil {
f.v6Saved = saved
} else {
for k, v := range saved {
if _, ok := f.v6Saved[k]; !ok {
f.v6Saved[k] = v
}
}
}
log.Info("IPv6 forwarding enabled")
}
f.v6Count++
return nil
}
func (f *IPForwardingState) releaseV6() error {
if f.v6Count == 0 {
return nil
}
f.v6Count--
if f.v6Count > 0 {
return nil
}
// Keep the saved values on failure so a later release or enable/release
// cycle can still restore them; re-restoring an already-restored key is a
// no-op since the sysctl already holds the desired value.
if err := systemops.DisableV6IPForwarding(f.v6Saved); err != nil {
return fmt.Errorf("disable IPv6 forwarding: %w", err)
}
f.v6Saved = nil
log.Info("IPv6 forwarding disabled")
return nil
}

View File

@@ -0,0 +1,39 @@
//go:build privileged
package ipfwdstate
import (
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// TestRequestRoutingV6ToV4Transition verifies that a v4-only routing request
// releases a previously held routing-owned v6 reference without touching
// references held by DNAT rules.
func TestRequestRoutingV6ToV4Transition(t *testing.T) {
f := NewIPForwardingState("wt-fwd-test")
require.NoError(t, f.RequestRouting(true), "request routing with v6")
v4, v6 := f.Counts()
assert.Equal(t, 1, v4, "v4 reference held")
assert.Equal(t, 1, v6, "v6 reference held")
require.NoError(t, f.RequestRouting(false), "request routing v4-only")
v4, v6 = f.Counts()
assert.Equal(t, 1, v4, "v4 reference kept")
assert.Equal(t, 0, v6, "routing-owned v6 reference released")
// A DNAT-held reference survives a v4-only routing request.
require.NoError(t, f.RequestForwarding(true), "dnat v6 reference")
require.NoError(t, f.RequestRouting(false), "repeat v4-only request")
_, v6 = f.Counts()
assert.Equal(t, 1, v6, "dnat-held v6 reference survives")
require.NoError(t, f.ReleaseForwarding(true), "release dnat v6 reference")
require.NoError(t, f.ReleaseRouting(), "release routing")
v4, v6 = f.Counts()
assert.Equal(t, 0, v4, "all v4 references released")
assert.Equal(t, 0, v6, "all v6 references released")
}

View File

@@ -58,11 +58,7 @@ func Setup(wgIface iface) (map[string]int, error) {
continue
}
// Escape '%' and '.' so they survive the dot-to-slash conversion in Set()
safeName := strings.ReplaceAll(intf.Name, "%", percentEscape)
safeName = strings.ReplaceAll(safeName, ".", dotEscape)
i := fmt.Sprintf(rpFilterInterfacePath, safeName)
i := fmt.Sprintf(rpFilterInterfacePath, EscapeInterfaceName(intf.Name))
oldVal, err := Set(i, 2, true)
if err != nil {
result = multierror.Append(result, err)
@@ -74,6 +70,13 @@ func Setup(wgIface iface) (map[string]int, error) {
return keys, nberrors.FormatErrorOrNil(result)
}
// EscapeInterfaceName escapes '%' and '.' in an interface name (e.g. VLANs
// like eth0.100) so the name survives the dot-to-slash conversion in Set.
func EscapeInterfaceName(name string) string {
safe := strings.ReplaceAll(name, "%", percentEscape)
return strings.ReplaceAll(safe, ".", dotEscape)
}
// Set sets a sysctl configuration, if onlyIfOne is true it will only set the new value if it's set to 1
func Set(key string, desiredValue int, onlyIfOne bool) (int, error) {
path := strings.ReplaceAll(key, ".", "/")

View File

@@ -32,8 +32,17 @@ func (r *SysOps) removeFromRouteTable(netip.Prefix, Nexthop) error {
return nil
}
func EnableIPForwarding() error {
log.Infof("Enable IP forwarding is not implemented on %s", runtime.GOOS)
func EnableV4IPForwarding() error {
log.Infof("Enable IPv4 forwarding is not implemented on %s", runtime.GOOS)
return nil
}
func EnableV6IPForwarding(string) (map[string]int, error) {
log.Infof("Enable IPv6 forwarding is not implemented on %s", runtime.GOOS)
return map[string]int{}, nil
}
func DisableV6IPForwarding(map[string]int) error {
return nil
}

View File

@@ -58,8 +58,17 @@ func (r *SysOps) removeFromRouteTable(netip.Prefix, Nexthop) error {
return nil
}
func EnableIPForwarding() error {
log.Infof("Enable IP forwarding is not implemented on %s", runtime.GOOS)
func EnableV4IPForwarding() error {
log.Infof("Enable IPv4 forwarding is not implemented on %s", runtime.GOOS)
return nil
}
func EnableV6IPForwarding(string) (map[string]int, error) {
log.Infof("Enable IPv6 forwarding is not implemented on %s", runtime.GOOS)
return map[string]int{}, nil
}
func DisableV6IPForwarding(map[string]int) error {
return nil
}

View File

@@ -763,13 +763,10 @@ func flushRoutes(tableID, family int) error {
return nberrors.FormatErrorOrNil(result)
}
func EnableIPForwarding() error {
func EnableV4IPForwarding() error {
if _, err := sysctl.Set(ipv4ForwardingPath, 1, false); err != nil {
return err
}
if _, err := sysctl.Set(ipv6ForwardingPath, 1, false); err != nil {
log.Warnf("failed to enable IPv6 forwarding: %v", err)
}
return nil
}

View File

@@ -43,8 +43,17 @@ func (r *SysOps) RemoveVPNRoute(prefix netip.Prefix, intf *net.Interface) error
return r.genericRemoveVPNRoute(prefix, intf)
}
func EnableIPForwarding() error {
log.Infof("Enable IP forwarding is not implemented on %s", runtime.GOOS)
func EnableV4IPForwarding() error {
log.Infof("Enable IPv4 forwarding is not implemented on %s", runtime.GOOS)
return nil
}
func EnableV6IPForwarding(string) (map[string]int, error) {
log.Infof("Enable IPv6 forwarding is not implemented on %s", runtime.GOOS)
return map[string]int{}, nil
}
func DisableV6IPForwarding(map[string]int) error {
return nil
}

View File

@@ -0,0 +1,92 @@
//go:build !android
package systemops
import (
"fmt"
"net"
"os"
"github.com/hashicorp/go-multierror"
log "github.com/sirupsen/logrus"
nberrors "github.com/netbirdio/netbird/client/errors"
"github.com/netbirdio/netbird/client/internal/routemanager/sysctl"
)
const (
// 1 (default) accepts RAs only while forwarding is off; 2 keeps RA
// acceptance on regardless, so RA-installed host defaults survive our
// v6 forwarding flip.
acceptRAInterfacePath = "net.ipv6.conf.%s.accept_ra"
acceptRADefaultPath = "net.ipv6.conf.default.accept_ra"
acceptRAProcPathFormat = "/proc/sys/net/ipv6/conf/%s/accept_ra"
)
// EnableV6IPForwarding bumps accept_ra=2 on host v6 interfaces before flipping
// forwarding=1, so RA-installed host defaults survive. Returns the prior values
// of sysctls we actually changed; entries already at the target are omitted.
func EnableV6IPForwarding(wgIfaceName string) (map[string]int, error) {
saved := map[string]int{}
bumpAcceptRA(saved, wgIfaceName)
oldVal, err := sysctl.Set(ipv6ForwardingPath, 1, false)
if err != nil {
return saved, err
}
if oldVal != 1 {
saved[ipv6ForwardingPath] = oldVal
}
return saved, nil
}
// DisableV6IPForwarding restores what EnableV6IPForwarding captured.
func DisableV6IPForwarding(saved map[string]int) error {
var result *multierror.Error
for key, value := range saved {
if _, err := sysctl.Set(key, value, false); err != nil {
result = multierror.Append(result, fmt.Errorf("restore %s: %w", key, err))
}
}
return nberrors.FormatErrorOrNil(result)
}
func bumpAcceptRA(saved map[string]int, wgIfaceName string) {
// Also bump conf.default so interfaces created while forwarding is on
// (hotplug, new Wi-Fi/dock) inherit accept_ra=2 and keep accepting RAs.
bumpAcceptRAKey(saved, acceptRADefaultPath)
interfaces, err := net.Interfaces()
if err != nil {
log.Warnf("list interfaces for accept_ra: %v", err)
return
}
for _, intf := range interfaces {
if intf.Name == "lo" || intf.Name == wgIfaceName {
continue
}
bumpAcceptRAForInterface(saved, intf.Name)
}
}
func bumpAcceptRAForInterface(saved map[string]int, name string) {
// Build procfs path from name, not the dotted key: VLAN names like eth0.100.
if _, err := os.Stat(fmt.Sprintf(acceptRAProcPathFormat, name)); err != nil {
return
}
bumpAcceptRAKey(saved, fmt.Sprintf(acceptRAInterfacePath, sysctl.EscapeInterfaceName(name)))
}
func bumpAcceptRAKey(saved map[string]int, key string) {
// onlyIfOne=true: leave admin overrides (0, 2) alone.
oldVal, err := sysctl.Set(key, 2, true)
if err != nil {
log.Warnf("bump %s: %v", key, err)
return
}
// With onlyIfOne, a write only happened when the old value was 1; values
// left untouched (0, 2) must not be recorded for restore.
if oldVal == 1 {
saved[key] = oldVal
}
}

View File

@@ -75,14 +75,6 @@ func New(filePath string) *Manager {
}
}
// FilePath returns the path of the underlying state file.
func (m *Manager) FilePath() string {
if m == nil {
return ""
}
return m.filePath
}
// Start starts the state manager periodic save routine
func (m *Manager) Start() {
if m == nil {

View File

@@ -14,6 +14,7 @@ import (
log "github.com/sirupsen/logrus"
nbAnonymize "github.com/netbirdio/netbird/client/anonymize"
"github.com/netbirdio/netbird/client/internal"
"github.com/netbirdio/netbird/client/internal/auth"
"github.com/netbirdio/netbird/client/internal/debug"
@@ -28,6 +29,13 @@ import (
types "github.com/netbirdio/netbird/upload-server/types"
)
// AnonymizeLevelDefault and AnonymizeLevelStrict are the accepted
// anonymizeLevel values for DebugBundle.
const (
AnonymizeLevelDefault = nbAnonymize.LevelDefaultString
AnonymizeLevelStrict = nbAnonymize.LevelStrictString
)
// ConnectionListener export internal Listener for mobile
type ConnectionListener interface {
peer.Listener
@@ -200,8 +208,10 @@ func (c *Client) Stop() {
// DebugBundle generates a debug bundle, uploads it and returns the upload key.
// It works with or without a running engine: when the engine is up it reuses
// the live config, sync response and client metrics; otherwise it loads the
// config from disk (or the preloaded tvOS config).
func (c *Client) DebugBundle(anonymize bool) (string, error) {
// config from disk (or the preloaded tvOS config). anonymizeLevel is "default"
// or "strict"; strict also anonymizes internal IP ranges, peer names, and
// WireGuard public keys, and implies anonymize.
func (c *Client) DebugBundle(anonymize bool, anonymizeLevel string) (string, error) {
cfg, cc := c.stateSnapshot()
// If the engine hasn't been started, load config so we can reach management.
@@ -251,6 +261,7 @@ func (c *Client) DebugBundle(anonymize bool) (string, error) {
deps,
debug.BundleConfig{
Anonymize: anonymize,
AnonymizeLevel: nbAnonymize.ParseLevel(anonymizeLevel),
IncludeSystemInfo: true,
},
)

View File

@@ -21,8 +21,6 @@ var allKeys = []string{
KeyBlockInbound,
KeyDisableMetricsCollection,
KeyAllowServerSSH,
KeyAllowServerVNC,
KeyDisableVNCApproval,
KeyDisableAutoConnect,
KeyDisableAutostart,
KeyPreSharedKey,

View File

@@ -36,8 +36,6 @@ const (
KeyBlockInbound = "blockInbound"
KeyDisableMetricsCollection = "disableMetricsCollection"
KeyAllowServerSSH = "allowServerSSH"
KeyAllowServerVNC = "allowServerVNC"
KeyDisableVNCApproval = "disableVNCApproval"
KeyDisableAutoConnect = "disableAutoConnect"
// KeyDisableAutostart suppresses the GUI's fresh-install
// launch-on-login default and marks the Settings toggle as

File diff suppressed because it is too large Load Diff

View File

@@ -1099,30 +1099,6 @@ func request_DaemonService_ExposeService_0(ctx context.Context, marshaler runtim
return stream, metadata, nil
}
func request_DaemonService_RespondApproval_0(ctx context.Context, marshaler runtime.Marshaler, client DaemonServiceClient, req *http.Request, pathParams map[string]string) (proto.Message, runtime.ServerMetadata, error) {
var (
protoReq RespondApprovalRequest
metadata runtime.ServerMetadata
)
if err := marshaler.NewDecoder(req.Body).Decode(&protoReq); err != nil && !errors.Is(err, io.EOF) {
return nil, metadata, status.Errorf(codes.InvalidArgument, "%v", err)
}
msg, err := client.RespondApproval(ctx, &protoReq, grpc.Header(&metadata.HeaderMD), grpc.Trailer(&metadata.TrailerMD))
return msg, metadata, err
}
func local_request_DaemonService_RespondApproval_0(ctx context.Context, marshaler runtime.Marshaler, server DaemonServiceServer, req *http.Request, pathParams map[string]string) (proto.Message, runtime.ServerMetadata, error) {
var (
protoReq RespondApprovalRequest
metadata runtime.ServerMetadata
)
if err := marshaler.NewDecoder(req.Body).Decode(&protoReq); err != nil && !errors.Is(err, io.EOF) {
return nil, metadata, status.Errorf(codes.InvalidArgument, "%v", err)
}
msg, err := server.RespondApproval(ctx, &protoReq)
return msg, metadata, err
}
func request_DaemonService_WailsUIReady_0(ctx context.Context, marshaler runtime.Marshaler, client DaemonServiceClient, req *http.Request, pathParams map[string]string) (proto.Message, runtime.ServerMetadata, error) {
var (
protoReq WailsUIReadyRequest
@@ -2001,26 +1977,6 @@ func RegisterDaemonServiceHandlerServer(ctx context.Context, mux *runtime.ServeM
runtime.HTTPError(ctx, mux, outboundMarshaler, w, req, err)
return
})
mux.Handle(http.MethodPost, pattern_DaemonService_RespondApproval_0, func(w http.ResponseWriter, req *http.Request, pathParams map[string]string) {
ctx, cancel := context.WithCancel(req.Context())
defer cancel()
var stream runtime.ServerTransportStream
ctx = grpc.NewContextWithServerTransportStream(ctx, &stream)
inboundMarshaler, outboundMarshaler := runtime.MarshalerForRequest(mux, req)
annotatedContext, err := runtime.AnnotateIncomingContext(ctx, mux, req, "/daemon.DaemonService/RespondApproval", runtime.WithHTTPPathPattern("/daemon.DaemonService/RespondApproval"))
if err != nil {
runtime.HTTPError(ctx, mux, outboundMarshaler, w, req, err)
return
}
resp, md, err := local_request_DaemonService_RespondApproval_0(annotatedContext, inboundMarshaler, server, req, pathParams)
md.HeaderMD, md.TrailerMD = metadata.Join(md.HeaderMD, stream.Header()), metadata.Join(md.TrailerMD, stream.Trailer())
annotatedContext = runtime.NewServerMetadataContext(annotatedContext, md)
if err != nil {
runtime.HTTPError(annotatedContext, mux, outboundMarshaler, w, req, err)
return
}
forward_DaemonService_RespondApproval_0(annotatedContext, mux, outboundMarshaler, w, req, resp, mux.GetForwardResponseOptions()...)
})
mux.Handle(http.MethodPost, pattern_DaemonService_WailsUIReady_0, func(w http.ResponseWriter, req *http.Request, pathParams map[string]string) {
ctx, cancel := context.WithCancel(req.Context())
defer cancel()
@@ -2846,23 +2802,6 @@ func RegisterDaemonServiceHandlerClient(ctx context.Context, mux *runtime.ServeM
}
forward_DaemonService_ExposeService_0(annotatedContext, mux, outboundMarshaler, w, req, func() (proto.Message, error) { return resp.Recv() }, mux.GetForwardResponseOptions()...)
})
mux.Handle(http.MethodPost, pattern_DaemonService_RespondApproval_0, func(w http.ResponseWriter, req *http.Request, pathParams map[string]string) {
ctx, cancel := context.WithCancel(req.Context())
defer cancel()
inboundMarshaler, outboundMarshaler := runtime.MarshalerForRequest(mux, req)
annotatedContext, err := runtime.AnnotateContext(ctx, mux, req, "/daemon.DaemonService/RespondApproval", runtime.WithHTTPPathPattern("/daemon.DaemonService/RespondApproval"))
if err != nil {
runtime.HTTPError(ctx, mux, outboundMarshaler, w, req, err)
return
}
resp, md, err := request_DaemonService_RespondApproval_0(annotatedContext, inboundMarshaler, client, req, pathParams)
annotatedContext = runtime.NewServerMetadataContext(annotatedContext, md)
if err != nil {
runtime.HTTPError(annotatedContext, mux, outboundMarshaler, w, req, err)
return
}
forward_DaemonService_RespondApproval_0(annotatedContext, mux, outboundMarshaler, w, req, resp, mux.GetForwardResponseOptions()...)
})
mux.Handle(http.MethodPost, pattern_DaemonService_WailsUIReady_0, func(w http.ResponseWriter, req *http.Request, pathParams map[string]string) {
ctx, cancel := context.WithCancel(req.Context())
defer cancel()
@@ -2929,7 +2868,6 @@ var (
pattern_DaemonService_StopCPUProfile_0 = runtime.MustPattern(runtime.NewPattern(1, []int{2, 0, 2, 1}, []string{"daemon.DaemonService", "StopCPUProfile"}, ""))
pattern_DaemonService_GetInstallerResult_0 = runtime.MustPattern(runtime.NewPattern(1, []int{2, 0, 2, 1}, []string{"daemon.DaemonService", "GetInstallerResult"}, ""))
pattern_DaemonService_ExposeService_0 = runtime.MustPattern(runtime.NewPattern(1, []int{2, 0, 2, 1}, []string{"daemon.DaemonService", "ExposeService"}, ""))
pattern_DaemonService_RespondApproval_0 = runtime.MustPattern(runtime.NewPattern(1, []int{2, 0, 2, 1}, []string{"daemon.DaemonService", "RespondApproval"}, ""))
pattern_DaemonService_WailsUIReady_0 = runtime.MustPattern(runtime.NewPattern(1, []int{2, 0, 2, 1}, []string{"daemon.DaemonService", "WailsUIReady"}, ""))
)
@@ -2979,6 +2917,5 @@ var (
forward_DaemonService_StopCPUProfile_0 = runtime.ForwardResponseMessage
forward_DaemonService_GetInstallerResult_0 = runtime.ForwardResponseMessage
forward_DaemonService_ExposeService_0 = runtime.ForwardResponseStream
forward_DaemonService_RespondApproval_0 = runtime.ForwardResponseMessage
forward_DaemonService_WailsUIReady_0 = runtime.ForwardResponseMessage
)

View File

@@ -152,14 +152,6 @@ service DaemonService {
// ExposeService exposes a local port via the NetBird reverse proxy
rpc ExposeService(ExposeServiceRequest) returns (stream ExposeServiceEvent) {}
// RespondApproval delivers the user's accept/deny decision for a
// pending user-approval prompt. The daemon pushes the prompt as a
// SystemEvent with category APPROVAL and metadata key "request_id";
// the UI calls this RPC with the same request_id to unblock whichever
// subsystem (VNC, SSH, ...) is waiting. The "kind" metadata key tells
// the UI which subsystem the prompt belongs to.
rpc RespondApproval(RespondApprovalRequest) returns (RespondApprovalResponse) {}
// WailsUIReady is a no-op probe the Wails UI calls once at startup. The UI
// only cares whether the daemon implements it: an Unimplemented response
// means the daemon predates this UI and is too old to drive it.
@@ -250,10 +242,6 @@ message LoginRequest {
optional bool disableSSHAuth = 38;
optional int32 sshJWTCacheTTL = 39;
optional bool disable_ipv6 = 40;
optional bool serverVNCAllowed = 41;
optional bool disableVNCApproval = 42;
}
message LoginResponse {
@@ -374,16 +362,12 @@ message GetConfigResponse {
bool disable_ipv6 = 27;
bool serverVNCAllowed = 28;
bool disableVNCApproval = 29;
// mDMManagedFields lists the names of configuration keys whose value is
// currently enforced by an MDM policy. Names match mdm.Key* constants
// (e.g. "managementURL", "disableClientRoutes"). UI/CLI clients should
// render the corresponding inputs as read-only and display a "managed
// by MDM" indicator.
repeated string mDMManagedFields = 30;
repeated string mDMManagedFields = 28;
}
// PeerState contains the latest state of a peer
@@ -468,25 +452,6 @@ message SSHServerState {
repeated SSHSessionInfo sessions = 2;
}
// VNCSessionInfo contains information about an active VNC session
message VNCSessionInfo {
string remoteAddress = 1;
string mode = 2;
string username = 3;
// userID is the Noise-verified session identity (hashed user ID from
// the ACL session-key entry), empty when auth is disabled.
string userID = 4;
// initiator is the human-readable display name of the dashboard user
// who minted the SessionPubKey, when known.
string initiator = 5;
}
// VNCServerState contains the latest state of the VNC server
message VNCServerState {
bool enabled = 1;
repeated VNCSessionInfo sessions = 2;
}
// FullStatus contains the full state held by the Status instance
message FullStatus {
ManagementState managementState = 1;
@@ -507,7 +472,6 @@ message FullStatus {
// on it to know when to re-fetch ListNetworks via the push stream, instead
// of polling on every status snapshot.
uint64 networksRevision = 11;
VNCServerState vncServerState = 12;
}
// Networks
@@ -576,6 +540,11 @@ message DebugBundleRequest {
// untrusted TLS certificate. Restricted to privileged callers; for
// self-hosted upload servers.
bool uploadInsecure = 7;
// anonymizeLevel selects how much the anonymizer redacts: "default"
// (or empty) keeps internal IP ranges, "strict" also anonymizes them.
// Unknown values are treated as "strict". Only meaningful with anonymize;
// "strict" implies it.
string anonymizeLevel = 8;
}
message DebugBundleResponse {
@@ -707,7 +676,6 @@ message SystemEvent {
AUTHENTICATION = 2;
CONNECTIVITY = 3;
SYSTEM = 4;
APPROVAL = 5;
}
string id = 1;
@@ -798,10 +766,6 @@ message SetConfigRequest {
optional bool disableSSHAuth = 33;
optional int32 sshJWTCacheTTL = 34;
optional bool disable_ipv6 = 35;
optional bool serverVNCAllowed = 36;
optional bool disableVNCApproval = 37;
}
message SetConfigResponse{}
@@ -930,6 +894,10 @@ message GetPeerSSHHostKeyResponse {
message RequestJWTAuthRequest {
// hint for OIDC login_hint parameter (typically email address)
optional string hint = 1;
// hasGraphicalSession tells the daemon that the caller has a graphical session,
// which decides whether PKCE or the device code flow is preferred. The daemon
// cannot detect this itself: it does not inherit the session environment.
bool hasGraphicalSession = 2;
}
// RequestJWTAuthResponse contains authentication flow information
@@ -973,6 +941,10 @@ message RequestExtendAuthSessionRequest {
// Optional OIDC login_hint (typically the user's email) to pre-fill the
// IdP login form.
optional string hint = 1;
// hasGraphicalSession tells the daemon that the caller has a graphical session,
// which decides whether PKCE or the device code flow is preferred. The daemon
// cannot detect this itself: it does not inherit the session environment.
bool hasGraphicalSession = 2;
}
// RequestExtendAuthSessionResponse carries the verification URI the UI
@@ -1091,18 +1063,3 @@ message StartBundleCaptureRequest {
message StartBundleCaptureResponse {}
message StopBundleCaptureRequest {}
message StopBundleCaptureResponse {}
message RespondApprovalRequest {
// request_id matches the SystemEvent metadata key emitted by the daemon
// when a subsystem awaits user approval for an inbound connection.
string request_id = 1;
// accept is true if the user approved the request, false if they
// denied it. A missing or unknown request_id is treated as a no-op.
bool accept = 2;
// view_only signals that the user granted the connection but withheld
// input control. Only meaningful when accept is true; ignored when
// accept is false.
bool view_only = 3;
}
message RespondApprovalResponse {}

View File

@@ -64,7 +64,6 @@ const (
DaemonService_StopCPUProfile_FullMethodName = "/daemon.DaemonService/StopCPUProfile"
DaemonService_GetInstallerResult_FullMethodName = "/daemon.DaemonService/GetInstallerResult"
DaemonService_ExposeService_FullMethodName = "/daemon.DaemonService/ExposeService"
DaemonService_RespondApproval_FullMethodName = "/daemon.DaemonService/RespondApproval"
DaemonService_WailsUIReady_FullMethodName = "/daemon.DaemonService/WailsUIReady"
)
@@ -168,13 +167,6 @@ type DaemonServiceClient interface {
GetInstallerResult(ctx context.Context, in *InstallerResultRequest, opts ...grpc.CallOption) (*InstallerResultResponse, error)
// ExposeService exposes a local port via the NetBird reverse proxy
ExposeService(ctx context.Context, in *ExposeServiceRequest, opts ...grpc.CallOption) (grpc.ServerStreamingClient[ExposeServiceEvent], error)
// RespondApproval delivers the user's accept/deny decision for a
// pending user-approval prompt. The daemon pushes the prompt as a
// SystemEvent with category APPROVAL and metadata key "request_id";
// the UI calls this RPC with the same request_id to unblock whichever
// subsystem (VNC, SSH, ...) is waiting. The "kind" metadata key tells
// the UI which subsystem the prompt belongs to.
RespondApproval(ctx context.Context, in *RespondApprovalRequest, opts ...grpc.CallOption) (*RespondApprovalResponse, error)
// WailsUIReady is a no-op probe the Wails UI calls once at startup. The UI
// only cares whether the daemon implements it: an Unimplemented response
// means the daemon predates this UI and is too old to drive it.
@@ -675,16 +667,6 @@ func (c *daemonServiceClient) ExposeService(ctx context.Context, in *ExposeServi
// This type alias is provided for backwards compatibility with existing code that references the prior non-generic stream type by name.
type DaemonService_ExposeServiceClient = grpc.ServerStreamingClient[ExposeServiceEvent]
func (c *daemonServiceClient) RespondApproval(ctx context.Context, in *RespondApprovalRequest, opts ...grpc.CallOption) (*RespondApprovalResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(RespondApprovalResponse)
err := c.cc.Invoke(ctx, DaemonService_RespondApproval_FullMethodName, in, out, cOpts...)
if err != nil {
return nil, err
}
return out, nil
}
func (c *daemonServiceClient) WailsUIReady(ctx context.Context, in *WailsUIReadyRequest, opts ...grpc.CallOption) (*WailsUIReadyResponse, error) {
cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
out := new(WailsUIReadyResponse)
@@ -795,13 +777,6 @@ type DaemonServiceServer interface {
GetInstallerResult(context.Context, *InstallerResultRequest) (*InstallerResultResponse, error)
// ExposeService exposes a local port via the NetBird reverse proxy
ExposeService(*ExposeServiceRequest, grpc.ServerStreamingServer[ExposeServiceEvent]) error
// RespondApproval delivers the user's accept/deny decision for a
// pending user-approval prompt. The daemon pushes the prompt as a
// SystemEvent with category APPROVAL and metadata key "request_id";
// the UI calls this RPC with the same request_id to unblock whichever
// subsystem (VNC, SSH, ...) is waiting. The "kind" metadata key tells
// the UI which subsystem the prompt belongs to.
RespondApproval(context.Context, *RespondApprovalRequest) (*RespondApprovalResponse, error)
// WailsUIReady is a no-op probe the Wails UI calls once at startup. The UI
// only cares whether the daemon implements it: an Unimplemented response
// means the daemon predates this UI and is too old to drive it.
@@ -951,9 +926,6 @@ func (UnimplementedDaemonServiceServer) GetInstallerResult(context.Context, *Ins
func (UnimplementedDaemonServiceServer) ExposeService(*ExposeServiceRequest, grpc.ServerStreamingServer[ExposeServiceEvent]) error {
return status.Error(codes.Unimplemented, "method ExposeService not implemented")
}
func (UnimplementedDaemonServiceServer) RespondApproval(context.Context, *RespondApprovalRequest) (*RespondApprovalResponse, error) {
return nil, status.Error(codes.Unimplemented, "method RespondApproval not implemented")
}
func (UnimplementedDaemonServiceServer) WailsUIReady(context.Context, *WailsUIReadyRequest) (*WailsUIReadyResponse, error) {
return nil, status.Error(codes.Unimplemented, "method WailsUIReady not implemented")
}
@@ -1760,24 +1732,6 @@ func _DaemonService_ExposeService_Handler(srv interface{}, stream grpc.ServerStr
// This type alias is provided for backwards compatibility with existing code that references the prior non-generic stream type by name.
type DaemonService_ExposeServiceServer = grpc.ServerStreamingServer[ExposeServiceEvent]
func _DaemonService_RespondApproval_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(RespondApprovalRequest)
if err := dec(in); err != nil {
return nil, err
}
if interceptor == nil {
return srv.(DaemonServiceServer).RespondApproval(ctx, in)
}
info := &grpc.UnaryServerInfo{
Server: srv,
FullMethod: DaemonService_RespondApproval_FullMethodName,
}
handler := func(ctx context.Context, req interface{}) (interface{}, error) {
return srv.(DaemonServiceServer).RespondApproval(ctx, req.(*RespondApprovalRequest))
}
return interceptor(ctx, in, info, handler)
}
func _DaemonService_WailsUIReady_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
in := new(WailsUIReadyRequest)
if err := dec(in); err != nil {
@@ -1967,10 +1921,6 @@ var DaemonService_ServiceDesc = grpc.ServiceDesc{
MethodName: "GetInstallerResult",
Handler: _DaemonService_GetInstallerResult_Handler,
},
{
MethodName: "RespondApproval",
Handler: _DaemonService_RespondApproval_Handler,
},
{
MethodName: "WailsUIReady",
Handler: _DaemonService_WailsUIReady_Handler,

View File

@@ -111,7 +111,7 @@ func (s *Server) StartCapture(req *proto.StartCaptureRequest, stream proto.Daemo
return status.Errorf(codes.Internal, "create capture session: %v", err)
}
engine, err := s.claimCapture(sess, func() { pw.Close() })
engine, err := s.claimCapture(sess)
if err != nil {
sess.Stop()
pw.Close()
@@ -190,7 +190,10 @@ func (s *Server) StartBundleCapture(_ context.Context, req *proto.StartBundleCap
s.stopBundleCaptureLocked()
s.cleanupBundleCapture()
s.evictActiveCaptureLocked()
if s.activeCapture != nil {
return nil, status.Error(codes.FailedPrecondition, "another capture is already running")
}
engine, err := s.getCaptureEngineLocked()
if err != nil {
@@ -301,58 +304,29 @@ func (s *Server) cleanupBundleCapture() {
s.bundleCapture = nil
}
// claimCapture reserves the engine's capture slot for sess. If another
// capture is already running it is evicted: a previous streaming session
// whose gRPC client died and never freed the slot stays stuck otherwise,
// and a bundle capture is just informational state.
func (s *Server) claimCapture(sess *capture.Session, cancel func()) (*internal.Engine, error) {
// claimCapture reserves the engine's capture slot for sess. Returns
// FailedPrecondition if another capture is already active.
func (s *Server) claimCapture(sess *capture.Session) (*internal.Engine, error) {
s.mutex.Lock()
defer s.mutex.Unlock()
s.evictActiveCaptureLocked()
if s.activeCapture != nil {
return nil, status.Error(codes.FailedPrecondition, "another capture is already running")
}
engine, err := s.getCaptureEngineLocked()
if err != nil {
return nil, err
}
s.activeCapture = sess
s.activeCaptureCancel = cancel
return engine, nil
}
// evictActiveCaptureLocked tears down whatever capture currently owns
// the engine slot so a fresh claim can succeed. Caller must hold mutex.
func (s *Server) evictActiveCaptureLocked() {
if s.activeCapture == nil {
return
}
if s.bundleCapture != nil && s.bundleCapture.sess == s.activeCapture {
log.Infof("evicting running bundle capture to start a new capture")
s.stopBundleCaptureLocked()
return
}
log.Infof("evicting previous streaming capture to start a new one")
prev := s.activeCapture
cancel := s.activeCaptureCancel
if engine, err := s.getCaptureEngineLocked(); err == nil {
if err := engine.SetCapture(nil); err != nil {
log.Debugf("clear previous capture: %v", err)
}
}
s.activeCapture = nil
s.activeCaptureCancel = nil
prev.Stop()
if cancel != nil {
cancel()
}
}
// releaseCapture clears the active-capture owner if it still matches sess.
func (s *Server) releaseCapture(sess *capture.Session) {
s.mutex.Lock()
defer s.mutex.Unlock()
if s.activeCapture == sess {
s.activeCapture = nil
s.activeCaptureCancel = nil
}
}
@@ -367,7 +341,6 @@ func (s *Server) clearCaptureIfOwner(sess *capture.Session, engine *internal.Eng
log.Debugf("clear capture: %v", err)
}
s.activeCapture = nil
s.activeCaptureCancel = nil
}
func (s *Server) getCaptureEngineLocked() (*internal.Engine, error) {

View File

@@ -16,6 +16,7 @@ import (
"google.golang.org/grpc/codes"
gstatus "google.golang.org/grpc/status"
"github.com/netbirdio/netbird/client/anonymize"
"github.com/netbirdio/netbird/client/internal/debug"
"github.com/netbirdio/netbird/client/internal/ipcauth"
"github.com/netbirdio/netbird/client/proto"
@@ -122,6 +123,7 @@ func (s *Server) generateDebugBundle(req *proto.DebugBundleRequest, uiOpener deb
},
debug.BundleConfig{
Anonymize: req.GetAnonymize(),
AnonymizeLevel: anonymize.ParseLevel(req.GetAnonymizeLevel()),
IncludeSystemInfo: req.GetSystemInfo(),
LogFileCount: req.GetLogFileCount(),
},

View File

@@ -297,8 +297,6 @@ func mdmManagedFieldConflicts(msg *proto.SetConfigRequest, policy *mdm.Policy) [
conflictBool(mdm.KeyRosenpassPermissive, msg.RosenpassPermissive),
conflictBool(mdm.KeyDisableAutoConnect, msg.DisableAutoConnect),
conflictBool(mdm.KeyAllowServerSSH, msg.ServerSSHAllowed),
conflictBool(mdm.KeyAllowServerVNC, msg.ServerVNCAllowed),
conflictBool(mdm.KeyDisableVNCApproval, msg.DisableVNCApproval),
conflictBool(mdm.KeyDisableClientRoutes, msg.DisableClientRoutes),
conflictBool(mdm.KeyDisableServerRoutes, msg.DisableServerRoutes),
conflictBool(mdm.KeyBlockInbound, msg.BlockInbound),
@@ -334,8 +332,6 @@ func setConfigRequestHasConfigOverrides(msg *proto.SetConfigRequest) bool {
msg.Mtu != nil ||
msg.DisableAutoConnect != nil ||
msg.ServerSSHAllowed != nil ||
msg.ServerVNCAllowed != nil ||
msg.DisableVNCApproval != nil ||
msg.NetworkMonitor != nil ||
msg.DisableClientRoutes != nil ||
msg.DisableServerRoutes != nil ||
@@ -374,8 +370,6 @@ func loginRequestHasConfigOverrides(msg *proto.LoginRequest) bool {
msg.WireguardPort != nil ||
msg.DisableAutoConnect != nil ||
msg.ServerSSHAllowed != nil ||
msg.ServerVNCAllowed != nil ||
msg.DisableVNCApproval != nil ||
msg.RosenpassPermissive != nil ||
len(msg.ExtraIFaceBlacklist) > 0 ||
msg.NetworkMonitor != nil ||
@@ -424,8 +418,6 @@ func loginRequestMDMConflicts(msg *proto.LoginRequest, policy *mdm.Policy) []str
conflictBool(mdm.KeyRosenpassPermissive, msg.RosenpassPermissive),
conflictBool(mdm.KeyDisableAutoConnect, msg.DisableAutoConnect),
conflictBool(mdm.KeyAllowServerSSH, msg.ServerSSHAllowed),
conflictBool(mdm.KeyAllowServerVNC, msg.ServerVNCAllowed),
conflictBool(mdm.KeyDisableVNCApproval, msg.DisableVNCApproval),
conflictBool(mdm.KeyDisableClientRoutes, msg.DisableClientRoutes),
conflictBool(mdm.KeyDisableServerRoutes, msg.DisableServerRoutes),
conflictBool(mdm.KeyBlockInbound, msg.BlockInbound),

View File

@@ -122,12 +122,8 @@ type Server struct {
captureEnabled bool
bundleCapture *bundleCapture
// activeCapture is the session currently installed on the engine; guarded by s.mutex.
activeCapture *capture.Session
// activeCaptureCancel tears down the streaming pipe/cancel for the
// active streaming capture so eviction unblocks the StartCapture RPC
// handler. Nil for bundle captures (they own their own context).
activeCaptureCancel func()
networksDisabled bool
activeCapture *capture.Session
networksDisabled bool
sleepHandler *sleephandler.SleepHandler
@@ -557,8 +553,6 @@ func (s *Server) setConfigInputFromRequest(msg *proto.SetConfigRequest) (profile
config.RosenpassPermissive = msg.RosenpassPermissive
config.DisableAutoConnect = msg.DisableAutoConnect
config.ServerSSHAllowed = msg.ServerSSHAllowed
config.ServerVNCAllowed = msg.ServerVNCAllowed
config.DisableVNCApproval = msg.DisableVNCApproval
config.NetworkMonitor = msg.NetworkMonitor
config.DisableClientRoutes = msg.DisableClientRoutes
config.DisableServerRoutes = msg.DisableServerRoutes
@@ -688,6 +682,11 @@ func (s *Server) Login(callerCtx context.Context, msg *proto.LoginRequest) (*pro
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, msg.IsUnixDesktopClient, false, hint)
if err != nil {
state.Set(internal.StatusLoginFailed)
// enrolling a device is the one flow a setup key can replace
if auth.IsSSOUnavailable(err) {
return nil, fmt.Errorf("%w. Set this device up with a setup key instead: "+
"https://docs.netbird.io/how-to/register-machines-using-setup-keys", err)
}
return nil, err
}
@@ -1578,7 +1577,6 @@ func (s *Server) buildStatusResponse(ctx context.Context, msg *proto.StatusReque
pbFullStatus := fullStatus.ToProto()
pbFullStatus.Events = s.statusRecorder.GetEventHistory()
pbFullStatus.SshServerState = s.getSSHServerState()
pbFullStatus.VncServerState = s.getVNCServerState()
pbFullStatus.NetworksRevision = s.statusRecorder.GetNetworksRevision()
statusResponse.FullStatus = pbFullStatus
}
@@ -1619,38 +1617,6 @@ func (s *Server) getSSHServerState() *proto.SSHServerState {
return sshServerState
}
// getVNCServerState retrieves the current VNC server state.
func (s *Server) getVNCServerState() *proto.VNCServerState {
s.mutex.Lock()
connectClient := s.connectClient
s.mutex.Unlock()
if connectClient == nil {
return nil
}
engine := connectClient.Engine()
if engine == nil {
return nil
}
enabled, sessions := engine.GetVNCServerStatus()
pbSessions := make([]*proto.VNCSessionInfo, 0, len(sessions))
for _, sess := range sessions {
pbSessions = append(pbSessions, &proto.VNCSessionInfo{
RemoteAddress: sess.RemoteAddress,
Mode: sess.Mode,
Username: sess.Username,
UserID: sess.UserID,
Initiator: sess.Initiator,
})
}
return &proto.VNCServerState{
Enabled: enabled,
Sessions: pbSessions,
}
}
// GetPeerSSHHostKey retrieves SSH host key for a specific peer
func (s *Server) GetPeerSSHHostKey(
ctx context.Context,
@@ -1762,8 +1728,8 @@ func (s *Server) RequestJWTAuth(
hint = profilemanager.GetLoginHint()
}
isDesktop := isUnixRunningDesktop()
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, isDesktop, false, hint)
// the daemon has no graphical session of its own, only the caller can answer this
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, msg.GetHasGraphicalSession(), false, hint)
if err != nil {
return nil, gstatus.Errorf(codes.Internal, "failed to create OAuth flow: %v", err)
}
@@ -1866,8 +1832,8 @@ func (s *Server) RequestExtendAuthSession(
hint = profilemanager.GetLoginHint()
}
isDesktop := isUnixRunningDesktop()
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, isDesktop, false, hint)
// the daemon has no graphical session of its own, only the caller can answer this
oAuthFlow, err := auth.NewOAuthFlow(ctx, config, msg.GetHasGraphicalSession(), false, hint)
if err != nil {
return nil, gstatus.Errorf(codes.Internal, "failed to create OAuth flow: %v", err)
}
@@ -2039,37 +2005,6 @@ func (s *Server) ExposeService(req *proto.ExposeServiceRequest, srv proto.Daemon
return nil
}
// RespondApproval relays the user's accept/deny decision for a pending
// approval prompt to the engine's broker. Unknown or already-resolved
// request_ids are silently no-op'd so a slow UI cannot deny a prompt the
// user already handled (or that already timed out).
func (s *Server) RespondApproval(_ context.Context, msg *proto.RespondApprovalRequest) (*proto.RespondApprovalResponse, error) {
if msg.GetRequestId() == "" {
return nil, gstatus.Errorf(codes.InvalidArgument, "request_id is required")
}
s.mutex.Lock()
connectClient := s.connectClient
s.mutex.Unlock()
if connectClient == nil {
return nil, gstatus.Errorf(codes.FailedPrecondition, "client not initialized")
}
engine := connectClient.Engine()
if engine == nil {
return nil, gstatus.Errorf(codes.FailedPrecondition, "engine not running")
}
if !engine.RespondApproval(msg.GetRequestId(), msg.GetAccept(), msg.GetViewOnly()) {
log.Debugf("approval response for unknown request_id %s", msg.GetRequestId())
}
return &proto.RespondApprovalResponse{}, nil
}
func isUnixRunningDesktop() bool {
if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
return false
}
return os.Getenv("DESKTOP_SESSION") != "" || os.Getenv("XDG_CURRENT_DESKTOP") != ""
}
func (s *Server) runProbes(ctx context.Context, waitForProbeResult bool) {
if s.connectClient == nil {
return
@@ -2170,8 +2105,6 @@ func (s *Server) GetConfig(ctx context.Context, req *proto.GetConfigRequest) (*p
Mtu: int64(cfg.MTU),
DisableAutoConnect: cfg.DisableAutoConnect,
ServerSSHAllowed: *cfg.ServerSSHAllowed,
ServerVNCAllowed: cfg.ServerVNCAllowed != nil && *cfg.ServerVNCAllowed,
DisableVNCApproval: cfg.DisableVNCApproval != nil && *cfg.DisableVNCApproval,
RosenpassEnabled: cfg.RosenpassEnabled,
RosenpassPermissive: cfg.RosenpassPermissive,
BlockInbound: cfg.BlockInbound,

View File

@@ -109,30 +109,6 @@ func TestSetConfig_MDMReject_SingleField(t *testing.T) {
assert.Equal(t, []string{mdm.KeyManagementURL}, v.GetFields())
}
func TestSetConfig_MDMReject_VNCFields(t *testing.T) {
withMDMPolicy(t, mdm.NewPolicy(map[string]any{
mdm.KeyAllowServerVNC: true,
mdm.KeyDisableVNCApproval: false,
}))
s, ctx, profName, username, _ := setupServerWithProfile(t)
vncAllowed := false
disableApproval := true
_, err := s.SetConfig(ctx, &proto.SetConfigRequest{
ProfileName: profName,
Username: username,
ServerVNCAllowed: &vncAllowed,
DisableVNCApproval: &disableApproval,
})
v := extractViolation(t, err)
assert.ElementsMatch(t, []string{
mdm.KeyAllowServerVNC,
mdm.KeyDisableVNCApproval,
}, v.GetFields())
}
func TestSetConfig_MDMReject_MultipleFields(t *testing.T) {
withMDMPolicy(t, mdm.NewPolicy(map[string]any{
mdm.KeyManagementURL: "https://mdm.example.com:443",

View File

@@ -61,8 +61,6 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
rosenpassEnabled := true
rosenpassPermissive := true
serverSSHAllowed := true
serverVNCAllowed := true
disableVNCApproval := true
interfaceName := "utun100"
wireguardPort := int64(51820)
preSharedKey := "test-psk"
@@ -87,8 +85,6 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
RosenpassEnabled: &rosenpassEnabled,
RosenpassPermissive: &rosenpassPermissive,
ServerSSHAllowed: &serverSSHAllowed,
ServerVNCAllowed: &serverVNCAllowed,
DisableVNCApproval: &disableVNCApproval,
InterfaceName: &interfaceName,
WireguardPort: &wireguardPort,
OptionalPreSharedKey: &preSharedKey,
@@ -132,10 +128,6 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
require.Equal(t, rosenpassPermissive, cfg.RosenpassPermissive)
require.NotNil(t, cfg.ServerSSHAllowed)
require.Equal(t, serverSSHAllowed, *cfg.ServerSSHAllowed)
require.NotNil(t, cfg.ServerVNCAllowed)
require.Equal(t, serverVNCAllowed, *cfg.ServerVNCAllowed)
require.NotNil(t, cfg.DisableVNCApproval)
require.Equal(t, disableVNCApproval, *cfg.DisableVNCApproval)
require.Equal(t, interfaceName, cfg.WgIface)
require.Equal(t, int(wireguardPort), cfg.WgPort)
require.Equal(t, preSharedKey, cfg.PreSharedKey)
@@ -188,8 +180,6 @@ func verifyAllFieldsCovered(t *testing.T, req *proto.SetConfigRequest) {
"RosenpassEnabled": true,
"RosenpassPermissive": true,
"ServerSSHAllowed": true,
"ServerVNCAllowed": true,
"DisableVNCApproval": true,
"InterfaceName": true,
"WireguardPort": true,
"OptionalPreSharedKey": true,
@@ -250,8 +240,6 @@ func TestCLIFlags_MappedToSetConfig(t *testing.T) {
"enable-rosenpass": "RosenpassEnabled",
"rosenpass-permissive": "RosenpassPermissive",
"allow-server-ssh": "ServerSSHAllowed",
"allow-server-vnc": "ServerVNCAllowed",
"disable-vnc-approval": "DisableVNCApproval",
"interface-name": "InterfaceName",
"wireguard-port": "WireguardPort",
"preshared-key": "OptionalPreSharedKey",

View File

@@ -26,51 +26,40 @@ import (
// daemon's SSH server into a root (or unauthenticated) shell.
// - Enabling the SSH server at all is what makes the above reachable, and a
// profile the caller owns is not a privilege they hold.
// - Enabling the VNC server exposes the console session, which on a
// multi-user host belongs to another user, and disabling its approval
// prompt removes that user's only say in it.
// - While a remote-access server (SSH or VNC) is enabled, repointing the
// profile at another management identity hands authorization decisions,
// including which keys and users are accepted, to whoever controls that
// identity. Changing the management URL and deregistering the peer are both
// ways to do that.
// - While the SSH server is enabled, repointing the profile at another
// management identity hands SSH authorization decisions, including which
// keys and users are accepted, to whoever controls that identity. Changing
// the management URL and deregistering the peer are both ways to do that.
//
// Everything else stays unauthenticated, so this is not an authorization model:
// it only refuses the changes that would let a local user become root, or reach
// another user's desktop. A caller whose identity cannot be established is
// refused as well.
// it only refuses the changes that would let a local user become root. A caller
// whose identity cannot be established is refused as well.
// privilegedConfigChange is the subset of a config request that crosses the
// user-to-root boundary. Fields are nil or empty when the request leaves them
// untouched.
type privilegedConfigChange struct {
managementURL string
serverSSHAllowed *bool
enableSSHRoot *bool
disableSSHAuth *bool
serverVNCAllowed *bool
disableVNCApproval *bool
managementURL string
serverSSHAllowed *bool
enableSSHRoot *bool
disableSSHAuth *bool
}
func privilegedChangeFromSetConfig(msg *proto.SetConfigRequest) privilegedConfigChange {
return privilegedConfigChange{
managementURL: msg.GetManagementUrl(),
serverSSHAllowed: msg.ServerSSHAllowed,
enableSSHRoot: msg.EnableSSHRoot,
disableSSHAuth: msg.DisableSSHAuth,
serverVNCAllowed: msg.ServerVNCAllowed,
disableVNCApproval: msg.DisableVNCApproval,
managementURL: msg.GetManagementUrl(),
serverSSHAllowed: msg.ServerSSHAllowed,
enableSSHRoot: msg.EnableSSHRoot,
disableSSHAuth: msg.DisableSSHAuth,
}
}
func privilegedChangeFromLogin(msg *proto.LoginRequest) privilegedConfigChange {
return privilegedConfigChange{
managementURL: msg.GetManagementUrl(),
serverSSHAllowed: msg.ServerSSHAllowed,
enableSSHRoot: msg.EnableSSHRoot,
disableSSHAuth: msg.DisableSSHAuth,
serverVNCAllowed: msg.ServerVNCAllowed,
disableVNCApproval: msg.DisableVNCApproval,
managementURL: msg.GetManagementUrl(),
serverSSHAllowed: msg.ServerSSHAllowed,
enableSSHRoot: msg.EnableSSHRoot,
disableSSHAuth: msg.DisableSSHAuth,
}
}
@@ -94,21 +83,15 @@ func requirePrivilegeForConfigChange(ctx context.Context, stored *profilemanager
return denyPrivileged(ctx, "enabling the NetBird SSH server", ipcauth.UpCommand("--allow-server-ssh"))
}
if err := requirePrivilegeForVNCChange(ctx, stored, change); err != nil {
return err
}
// Only guard the management binding while a remote-access server is enabled:
// that is when the management identity decides who may open a shell or reach
// the desktop here.
server, enabled := enabledRemoteAccessServer(stored)
if !enabled {
// Only guard the management binding while the SSH server is enabled: that is
// when the management identity decides who may open a shell here.
if !sshServerEnabled(stored) {
return nil
}
if change.managementURL != "" && !sameManagementURL(stored.ManagementURL, change.managementURL) {
return denyPrivileged(ctx,
fmt.Sprintf("changing the management URL while the NetBird %s server is enabled", server),
"changing the management URL while the NetBird SSH server is enabled",
ipcauth.UpCommand("-m "+change.managementURL))
}
@@ -116,21 +99,20 @@ func requirePrivilegeForConfigChange(ctx context.Context, stored *profilemanager
}
// requirePrivilegeForDeregistration refuses to deregister the peer from the
// management server when the caller is not privileged and the profile has a
// remote-access server enabled. Deregistering frees the peer's key to be
// registered against another management identity, which is the same handover the
// management server when the caller is not privileged and the profile has the
// SSH server enabled. Deregistering frees the peer's key to be registered
// against another management identity, which is the same handover the
// management URL check refuses.
//
// Callers that treat deregistration as best-effort (profile removal) continue
// without it; callers that were asked to deregister surface the error.
func requirePrivilegeForDeregistration(ctx context.Context, cfg *profilemanager.Config) error {
server, enabled := enabledRemoteAccessServer(cfg)
if !enabled {
if !sshServerEnabled(cfg) {
return nil
}
return denyPrivileged(ctx,
fmt.Sprintf("deregistering this peer while the NetBird %s server is enabled", server),
"deregistering this peer while the NetBird SSH server is enabled",
ipcauth.ElevatedCommand("netbird logout"))
}

View File

@@ -1,58 +0,0 @@
package server
import (
"context"
"github.com/netbirdio/netbird/client/internal/ipcauth"
"github.com/netbirdio/netbird/client/internal/profilemanager"
)
// The VNC half of the privilege gate described in ssh_gate.go. The daemon runs
// as root/LocalSystem and the VNC server it hosts attaches to the console
// session, so both of these cross the user-to-root boundary:
//
// - Enabling the VNC server publishes the console desktop, keyboard and
// mouse to whoever the account authorizes. On a multi-user host that
// desktop belongs to another user, and the profile the caller owns is not a
// privilege they hold over it.
// - Disabling the approval prompt removes the console user's per-connection
// consent, turning an authorized VNC session into a silent one.
func requirePrivilegeForVNCChange(ctx context.Context, stored *profilemanager.Config, change privilegedConfigChange) error {
if enables(storedFlag(stored, func(c *profilemanager.Config) *bool { return c.DisableVNCApproval }), change.disableVNCApproval) {
return denyPrivileged(ctx, "disabling VNC connection approval", ipcauth.UpCommand("--disable-vnc-approval"))
}
if enables(storedFlag(stored, func(c *profilemanager.Config) *bool { return c.ServerVNCAllowed }), change.serverVNCAllowed) {
return denyPrivileged(ctx, "enabling the NetBird VNC server", ipcauth.UpCommand("--allow-server-vnc"))
}
return nil
}
// vncServerEnabled reports whether the profile currently runs the VNC server.
//
// Unlike the SSH server, VNC defaults to off: the flag was introduced with the
// server itself, so a nil value is a config written before VNC existed and
// means the server is not running.
func vncServerEnabled(cfg *profilemanager.Config) bool {
if cfg == nil || cfg.ServerVNCAllowed == nil {
return false
}
return *cfg.ServerVNCAllowed
}
// enabledRemoteAccessServer names a remote-access server the profile currently
// runs, and whether it runs any. It decides whether the management-binding and
// deregistration guards apply, since either server hands authorization
// decisions to the management identity the profile points at. SSH is reported
// first when both are on: it is the more privileged of the two.
func enabledRemoteAccessServer(cfg *profilemanager.Config) (string, bool) {
switch {
case sshServerEnabled(cfg):
return "SSH", true
case vncServerEnabled(cfg):
return "VNC", true
default:
return "", false
}
}

View File

@@ -1,178 +0,0 @@
package server
import (
"testing"
"github.com/netbirdio/netbird/client/internal/profilemanager"
)
func TestRequirePrivilegeForConfigChange_VNCFlags(t *testing.T) {
tests := []struct {
name string
stored *profilemanager.Config
change privilegedConfigChange
privileged bool
wantDeny bool
}{
{
name: "enabling the vnc server unprivileged is refused",
stored: &profilemanager.Config{ServerVNCAllowed: boolPtr(false)},
change: privilegedConfigChange{serverVNCAllowed: boolPtr(true)},
wantDeny: true,
},
{
name: "enabling the vnc server as root is allowed",
stored: &profilemanager.Config{ServerVNCAllowed: boolPtr(false)},
change: privilegedConfigChange{serverVNCAllowed: boolPtr(true)},
privileged: true,
},
{
name: "restating an already enabled vnc server is not a change",
stored: &profilemanager.Config{ServerVNCAllowed: boolPtr(true)},
change: privilegedConfigChange{serverVNCAllowed: boolPtr(true)},
},
{
name: "turning the vnc server off is not guarded",
stored: &profilemanager.Config{ServerVNCAllowed: boolPtr(true)},
change: privilegedConfigChange{serverVNCAllowed: boolPtr(false)},
},
{
// Unlike SSH, a nil flag means off: the flag shipped with the server.
name: "a config written before vnc existed counts as off, so enabling is refused",
stored: &profilemanager.Config{},
change: privilegedConfigChange{serverVNCAllowed: boolPtr(true)},
wantDeny: true,
},
{
name: "a profile with no config yet counts as off, so enabling is refused",
stored: nil,
change: privilegedConfigChange{serverVNCAllowed: boolPtr(true)},
wantDeny: true,
},
{
name: "disabling the vnc approval prompt unprivileged is refused",
stored: &profilemanager.Config{DisableVNCApproval: boolPtr(false)},
change: privilegedConfigChange{disableVNCApproval: boolPtr(true)},
wantDeny: true,
},
{
name: "disabling the vnc approval prompt as root is allowed",
stored: &profilemanager.Config{DisableVNCApproval: boolPtr(false)},
change: privilegedConfigChange{disableVNCApproval: boolPtr(true)},
privileged: true,
},
{
name: "re-enabling the vnc approval prompt is not guarded",
stored: &profilemanager.Config{DisableVNCApproval: boolPtr(true)},
change: privilegedConfigChange{disableVNCApproval: boolPtr(false)},
},
{
name: "restating a disabled approval prompt is not a change",
stored: &profilemanager.Config{DisableVNCApproval: boolPtr(true)},
change: privilegedConfigChange{disableVNCApproval: boolPtr(true)},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
ctx := userCtx()
if tt.privileged {
ctx = rootCtx()
}
err := requirePrivilegeForConfigChange(ctx, tt.stored, tt.change)
if tt.wantDeny {
assertDenied(t, err)
return
}
assertAllowed(t, err)
})
}
}
// The management binding and deregistration guards protect either remote-access
// server, so the VNC server alone must arm them even with SSH off.
func TestRequirePrivilegeForConfigChange_ManagementURLWithVNCOnly(t *testing.T) {
vncOnly := &profilemanager.Config{
ServerSSHAllowed: boolPtr(false),
ServerVNCAllowed: boolPtr(true),
ManagementURL: mustURL(t, "https://api.netbird.io:443"),
DisableVNCApproval: boolPtr(false),
}
err := requirePrivilegeForConfigChange(userCtx(), vncOnly,
privilegedConfigChange{managementURL: "https://attacker.example.com:443"})
assertDenied(t, err)
// The same move is the administrator's to make.
assertAllowed(t, requirePrivilegeForConfigChange(rootCtx(), vncOnly,
privilegedConfigChange{managementURL: "https://selfhosted.example.com:443"}))
// Restating the stored binding is not a change, so it is never refused.
assertAllowed(t, requirePrivilegeForConfigChange(userCtx(), vncOnly,
privilegedConfigChange{managementURL: "https://api.netbird.io"}))
}
func TestRequirePrivilegeForDeregistration_VNCOnly(t *testing.T) {
vncOnly := &profilemanager.Config{ServerSSHAllowed: boolPtr(false), ServerVNCAllowed: boolPtr(true)}
assertDenied(t, requirePrivilegeForDeregistration(userCtx(), vncOnly))
assertAllowed(t, requirePrivilegeForDeregistration(rootCtx(), vncOnly))
bothOff := &profilemanager.Config{ServerSSHAllowed: boolPtr(false), ServerVNCAllowed: boolPtr(false)}
assertAllowed(t, requirePrivilegeForDeregistration(userCtx(), bothOff))
}
// enabledRemoteAccessServer names the server in the refusal, so the user is told
// which one is holding the binding down. SSH wins when both are on: it is the
// more privileged of the two.
func TestEnabledRemoteAccessServer(t *testing.T) {
tests := []struct {
name string
cfg *profilemanager.Config
wantServer string
wantOn bool
}{
{
name: "ssh only",
cfg: &profilemanager.Config{ServerSSHAllowed: boolPtr(true), ServerVNCAllowed: boolPtr(false)},
wantServer: "SSH",
wantOn: true,
},
{
name: "vnc only",
cfg: &profilemanager.Config{ServerSSHAllowed: boolPtr(false), ServerVNCAllowed: boolPtr(true)},
wantServer: "VNC",
wantOn: true,
},
{
name: "both on reports ssh",
cfg: &profilemanager.Config{ServerSSHAllowed: boolPtr(true), ServerVNCAllowed: boolPtr(true)},
wantServer: "SSH",
wantOn: true,
},
{
name: "both off",
cfg: &profilemanager.Config{ServerSSHAllowed: boolPtr(false), ServerVNCAllowed: boolPtr(false)},
},
{
// A nil SSH flag means on (legacy configs), so it still arms the guard.
name: "legacy config with no flags at all reports ssh",
cfg: &profilemanager.Config{},
wantServer: "SSH",
wantOn: true,
},
{
name: "no config",
cfg: nil,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
server, on := enabledRemoteAccessServer(tt.cfg)
if on != tt.wantOn || server != tt.wantServer {
t.Fatalf("enabledRemoteAccessServer() = (%q, %v), want (%q, %v)", server, on, tt.wantServer, tt.wantOn)
}
})
}
}

View File

@@ -1,4 +1,4 @@
package sessionauth
package auth
import (
"errors"
@@ -15,8 +15,6 @@ const (
DefaultUserIDClaim = "sub"
// Wildcard is a special user ID that matches all users
Wildcard = "*"
// sessionPubKeyLen is the size of an X25519 static public key in bytes.
sessionPubKeyLen = 32
)
var (
@@ -24,7 +22,6 @@ var (
ErrUserNotAuthorized = errors.New("user is not authorized to access this peer")
ErrNoMachineUserMapping = errors.New("no authorization mapping for OS user")
ErrUserNotMappedToOSUser = errors.New("user is not authorized to login as OS user")
ErrSessionKeyNotKnown = errors.New("session pubkey not registered")
)
// Authorizer handles SSH fine-grained access control authorization
@@ -38,17 +35,6 @@ type Authorizer struct {
// machineUsers maps OS login usernames to lists of authorized user indexes
machineUsers map[string][]uint32
// sessionPubKeys maps an X25519 static public key (as map-safe
// array) to the hashed user identity that key authenticates as.
// Populated from management's temporary-access flow; used by VNC to
// authenticate via the Noise_IK handshake.
sessionPubKeys map[[sessionPubKeyLen]byte]sshuserhash.UserIDHash
// sessionDisplayNames mirrors sessionPubKeys with the optional
// human-readable display name management associated with each
// session key. Used by the per-connection UI approval prompt; not
// consulted by any authorization decision.
sessionDisplayNames map[[sessionPubKeyLen]byte]string
// mu protects the list of users
mu sync.RWMutex
}
@@ -64,29 +50,13 @@ type Config struct {
// MachineUsers maps OS login usernames to indexes in AuthorizedUsers
// If a user wants to login as a specific OS user, their index must be in the corresponding list
MachineUsers map[string][]uint32
// SessionPubKeys binds ephemeral X25519 static public keys to hashed
// user identities. Populated for VNC; ignored on the SSH side.
SessionPubKeys []SessionPubKey
}
// SessionPubKey is a single ephemeral-key entry: the 32-byte X25519
// static public key plus the hashed user identity it authenticates as,
// optionally plus a human-readable display name for the UI approval
// prompt to identify the requester.
type SessionPubKey struct {
PubKey []byte
UserIDHash sshuserhash.UserIDHash
DisplayName string
}
// NewAuthorizer creates a new SSH authorizer with empty configuration
func NewAuthorizer() *Authorizer {
a := &Authorizer{
userIDClaim: DefaultUserIDClaim,
machineUsers: make(map[string][]uint32),
sessionPubKeys: make(map[[sessionPubKeyLen]byte]sshuserhash.UserIDHash),
sessionDisplayNames: make(map[[sessionPubKeyLen]byte]string),
userIDClaim: DefaultUserIDClaim,
machineUsers: make(map[string][]uint32),
}
return a
@@ -102,8 +72,6 @@ func (a *Authorizer) Update(config *Config) {
a.userIDClaim = DefaultUserIDClaim
a.authorizedUsers = []sshuserhash.UserIDHash{}
a.machineUsers = make(map[string][]uint32)
a.sessionPubKeys = make(map[[sessionPubKeyLen]byte]sshuserhash.UserIDHash)
a.sessionDisplayNames = make(map[[sessionPubKeyLen]byte]string)
log.Info("SSH authorization cleared")
return
}
@@ -126,35 +94,8 @@ func (a *Authorizer) Update(config *Config) {
}
a.machineUsers = machineUsers
sessionPubKeys := make(map[[sessionPubKeyLen]byte]sshuserhash.UserIDHash, len(config.SessionPubKeys))
sessionDisplayNames := make(map[[sessionPubKeyLen]byte]string, len(config.SessionPubKeys))
conflicted := make(map[[sessionPubKeyLen]byte]struct{})
for _, e := range config.SessionPubKeys {
if len(e.PubKey) != sessionPubKeyLen {
continue
}
var key [sessionPubKeyLen]byte
copy(key[:], e.PubKey)
if _, bad := conflicted[key]; bad {
continue
}
if existing, ok := sessionPubKeys[key]; ok && existing != e.UserIDHash {
log.Warnf("SSH auth: session pubkey bound to conflicting user hashes; dropping binding")
delete(sessionPubKeys, key)
delete(sessionDisplayNames, key)
conflicted[key] = struct{}{}
continue
}
sessionPubKeys[key] = e.UserIDHash
if e.DisplayName != "" {
sessionDisplayNames[key] = e.DisplayName
}
}
a.sessionPubKeys = sessionPubKeys
a.sessionDisplayNames = sessionDisplayNames
log.Debugf("SSH auth: updated with %d authorized users, %d machine user mappings, %d session pubkeys",
len(config.AuthorizedUsers), len(machineUsers), len(sessionPubKeys))
log.Debugf("SSH auth: updated with %d authorized users, %d machine user mappings",
len(config.AuthorizedUsers), len(machineUsers))
}
// Authorize validates if a user is authorized to login as the specified OS user.
@@ -214,54 +155,6 @@ func (a *Authorizer) GetUserIDClaim() string {
return a.userIDClaim
}
// LookupSessionKey resolves a Noise-verified static public key to the
// hashed user identity registered with it. Fails closed when the key is
// unknown.
func (a *Authorizer) LookupSessionKey(pubKey []byte) (sshuserhash.UserIDHash, error) {
var zero sshuserhash.UserIDHash
if len(pubKey) != sessionPubKeyLen {
return zero, fmt.Errorf("session pubkey wrong length: %d", len(pubKey))
}
var key [sessionPubKeyLen]byte
copy(key[:], pubKey)
a.mu.RLock()
hash, ok := a.sessionPubKeys[key]
a.mu.RUnlock()
if !ok {
return zero, ErrSessionKeyNotKnown
}
return hash, nil
}
// LookupSessionDisplayName returns the human-readable display name
// management associated with a session pubkey, or empty string when none
// is recorded. Never returns an error: a missing/unknown key reports as
// "" and the caller falls back to other identifiers.
func (a *Authorizer) LookupSessionDisplayName(pubKey []byte) string {
if len(pubKey) != sessionPubKeyLen {
return ""
}
var key [sessionPubKeyLen]byte
copy(key[:], pubKey)
a.mu.RLock()
name := a.sessionDisplayNames[key]
a.mu.RUnlock()
return name
}
// AuthorizeOSUserBySessionKey resolves the OS-user mapping for a session
// key. Mirrors Authorize but skips the JWT-hash step since the key has
// already been verified and the user identity hash is in hand.
func (a *Authorizer) AuthorizeOSUserBySessionKey(userIDHash sshuserhash.UserIDHash, osUsername string) (string, error) {
a.mu.RLock()
defer a.mu.RUnlock()
userIndex, found := a.findUserIndex(userIDHash)
if !found {
return "", fmt.Errorf("session user (hash: %s) not in authorized list for OS user %q: %w", userIDHash, osUsername, ErrUserNotAuthorized)
}
return a.checkMachineUserMapping("session", osUsername, userIndex)
}
// findUserIndex finds the index of a hashed user ID in the authorized users list
// Returns the index and true if found, 0 and false if not found
func (a *Authorizer) findUserIndex(hashedUserID sshuserhash.UserIDHash) (int, bool) {

View File

@@ -1,7 +1,6 @@
package sessionauth
package auth
import (
"errors"
"testing"
"github.com/stretchr/testify/assert"
@@ -611,61 +610,3 @@ func TestAuthorizer_Wildcard_WithPartialIndexes_AllowsAllUsers(t *testing.T) {
assert.Error(t, err)
assert.ErrorIs(t, err, ErrUserNotAuthorized, "unauthorized user should be denied")
}
func TestAuthorizer_LookupSessionKey_Valid(t *testing.T) {
pub := bytesRepeat(0x11, sessionPubKeyLen)
userHash, err := sshauth.HashUserID("alice")
require.NoError(t, err)
a := NewAuthorizer()
a.Update(&Config{
AuthorizedUsers: []sshauth.UserIDHash{userHash},
MachineUsers: map[string][]uint32{Wildcard: {0}},
SessionPubKeys: []SessionPubKey{{PubKey: pub, UserIDHash: userHash}},
})
got, err := a.LookupSessionKey(pub)
require.NoError(t, err)
assert.Equal(t, userHash, got)
if _, err := a.AuthorizeOSUserBySessionKey(got, "alice"); err != nil {
t.Fatalf("AuthorizeOSUserBySessionKey: %v", err)
}
}
func TestAuthorizer_LookupSessionKey_UnknownPub(t *testing.T) {
a := NewAuthorizer()
a.Update(&Config{})
_, err := a.LookupSessionKey(bytesRepeat(0x22, sessionPubKeyLen))
require.ErrorIs(t, err, ErrSessionKeyNotKnown)
}
func TestAuthorizer_LookupSessionKey_WrongLength(t *testing.T) {
a := NewAuthorizer()
_, err := a.LookupSessionKey([]byte("short"))
require.Error(t, err)
}
func TestAuthorizer_LookupSessionKey_UpdateClears(t *testing.T) {
pub := bytesRepeat(0x33, sessionPubKeyLen)
userHash, err := sshauth.HashUserID("alice")
require.NoError(t, err)
a := NewAuthorizer()
a.Update(&Config{SessionPubKeys: []SessionPubKey{{PubKey: pub, UserIDHash: userHash}}})
if _, err := a.LookupSessionKey(pub); err != nil {
t.Fatalf("setup lookup: %v", err)
}
a.Update(&Config{})
if _, err := a.LookupSessionKey(pub); !errors.Is(err, ErrSessionKeyNotKnown) {
t.Fatalf("expected ErrSessionKeyNotKnown, got %v", err)
}
}
func bytesRepeat(b byte, n int) []byte {
out := make([]byte, n)
for i := range out {
out[i] = b
}
return out
}

View File

@@ -13,6 +13,7 @@ import (
"golang.org/x/crypto/ssh"
"github.com/netbirdio/netbird/client/proto"
"github.com/netbirdio/netbird/util"
)
const (
@@ -92,7 +93,8 @@ func printAuthInstructions(stderr io.Writer, authResponse *proto.RequestJWTAuthR
// RequestJWTToken requests or retrieves a JWT token for SSH authentication
func RequestJWTToken(ctx context.Context, client proto.DaemonServiceClient, stdout, stderr io.Writer, useCache bool, hint string, openBrowser func(string) error) (string, error) {
req := &proto.RequestJWTAuthRequest{}
// the ssh client runs in the user's session, the daemon does not: tell it what we can see
req := &proto.RequestJWTAuthRequest{HasGraphicalSession: util.HasGraphicalSession()}
if hint != "" {
req.Hint = &hint
}
@@ -193,4 +195,3 @@ func buildAddressList(hostname string, remote net.Addr) []string {
}
return addresses
}

View File

@@ -26,10 +26,10 @@ import (
cryptossh "golang.org/x/crypto/ssh"
nbssh "github.com/netbirdio/netbird/client/ssh"
sshauth "github.com/netbirdio/netbird/client/ssh/auth"
"github.com/netbirdio/netbird/client/ssh/server"
"github.com/netbirdio/netbird/client/ssh/testutil"
nbjwt "github.com/netbirdio/netbird/shared/auth/jwt"
sshauth "github.com/netbirdio/netbird/shared/sessionauth"
sshuserhash "github.com/netbirdio/netbird/shared/sshauth"
)

View File

@@ -243,7 +243,7 @@ func (s *Server) setUserEnvironmentVariables(envMap map[string]string, userProfi
// prepareCommandEnv prepares environment variables for command execution on Windows
func (s *Server) prepareCommandEnv(logger *log.Entry, localUser *user.User, session ssh.Session) []string {
username, domain := s.parseUsername(localUser.Username)
username, domain := parseUsername(localUser.Username)
userEnv, err := s.getUserEnvironment(logger, username, domain)
if err != nil {
log.Debugf("failed to get user environment for %s\\%s, using fallback: %v", domain, username, err)
@@ -383,7 +383,7 @@ func (s *Server) executeCommandWithPty(logger *log.Entry, session ssh.Session, _
return false
}
username, domain := s.parseUsername(localUser.Username)
username, domain := parseUsername(localUser.Username)
shell := getUserShell(localUser.Uid)
req := PtyExecutionRequest{

View File

@@ -23,11 +23,11 @@ import (
"github.com/stretchr/testify/require"
nbssh "github.com/netbirdio/netbird/client/ssh"
sshauth "github.com/netbirdio/netbird/client/ssh/auth"
"github.com/netbirdio/netbird/client/ssh/client"
"github.com/netbirdio/netbird/client/ssh/detection"
"github.com/netbirdio/netbird/client/ssh/testutil"
nbjwt "github.com/netbirdio/netbird/shared/auth/jwt"
sshauth "github.com/netbirdio/netbird/shared/sessionauth"
sshuserhash "github.com/netbirdio/netbird/shared/sshauth"
)

Some files were not shown because too many files have changed in this diff Show More