Bump github.com/jackc/pgx/v5 from 5.5.5 to 5.9.2

Bumps [github.com/jackc/pgx/v5](https://github.com/jackc/pgx) from 5.5.5 to 5.9.2.
- [Changelog](https://github.com/jackc/pgx/blob/master/CHANGELOG.md)
- [Commits](https://github.com/jackc/pgx/compare/v5.5.5...v5.9.2)

---
updated-dependencies:
- dependency-name: github.com/jackc/pgx/v5
  dependency-version: 5.9.2
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>

.github/workflows/release.yml

@@ -293,11 +293,8 @@ jobs:
           ${{ steps.goreleaser.outputs.artifacts }}
           JSON
 
-          # dockers_v2 artifacts have no top-level goarch field, so match the
-          # per-platform -amd64 tag suffix instead; it works for both the old
-          # dockers and the new dockers_v2 image naming.
           mapfile -t src_images < <(
-            jq -r '.[] | select(.type == "Docker Image") | .name | select(startswith("ghcr.io/") and endswith("-amd64"))' /tmp/goreleaser-artifacts.json
+            jq -r '.[] | select(.type == "Docker Image") | select(.goarch == "amd64") | .name | select(startswith("ghcr.io/"))' /tmp/goreleaser-artifacts.json
           )
 
           for src in "${src_images[@]}"; do

.gitignore

@@ -1,4 +1,3 @@
-.claude
 .idea
 .run
 *.iml

client/iface/configurer/kernel_unix.go

@@ -17,15 +17,12 @@ import (
 
 type KernelConfigurer struct {
 	deviceName string
-	statsCache *statsCache
 }
 
 func NewKernelConfigurer(deviceName string) *KernelConfigurer {
-	c := &KernelConfigurer{
+	return &KernelConfigurer{
 		deviceName: deviceName,
 	}
-	c.statsCache = newStatsCache(statsCacheTTL, c.fetchStats)
-	return c
 }
 
 func (c *KernelConfigurer) ConfigureInterface(privateKey string, port int) error {
@@ -249,6 +246,12 @@ func (c *KernelConfigurer) configure(config wgtypes.Config) error {
 		}
 	}()
 
+	// validate if device with name exists
+	_, err = wg.Device(c.deviceName)
+	if err != nil {
+		return err
+	}
+
 	return wg.ConfigureDevice(c.deviceName, config)
 }
 
@@ -297,14 +300,6 @@ func (c *KernelConfigurer) FullStats() (*Stats, error) {
 }
 
 func (c *KernelConfigurer) GetStats() (map[string]WGStats, error) {
-	return c.statsCache.get()
-}
-
-func (c *KernelConfigurer) LastActivities() map[string]monotime.Time {
-	return nil
-}
-
-func (c *KernelConfigurer) fetchStats() (map[string]WGStats, error) {
 	stats := make(map[string]WGStats)
 	wg, err := wgctrl.New()
 	if err != nil {
@@ -331,3 +326,7 @@ func (c *KernelConfigurer) fetchStats() (map[string]WGStats, error) {
 	}
 	return stats, nil
 }
+
+func (c *KernelConfigurer) LastActivities() map[string]monotime.Time {
+	return nil
+}

client/iface/configurer/stats_cache.go

@@ -1,52 +0,0 @@
-package configurer
-
-import (
-	"sync"
-	"time"
-
-	"golang.org/x/sync/singleflight"
-)
-
-const statsCacheTTL = 1 * time.Second
-
-type statsCache struct {
-	ttl   time.Duration
-	fetch func() (map[string]WGStats, error)
-
-	mu       sync.RWMutex
-	value    map[string]WGStats
-	expireAt time.Time
-
-	sf singleflight.Group
-}
-
-func newStatsCache(ttl time.Duration, fetch func() (map[string]WGStats, error)) *statsCache {
-	return &statsCache{ttl: ttl, fetch: fetch}
-}
-
-func (c *statsCache) get() (map[string]WGStats, error) {
-	c.mu.RLock()
-	if c.value != nil && time.Now().Before(c.expireAt) {
-		value := c.value
-		c.mu.RUnlock()
-		return value, nil
-	}
-	c.mu.RUnlock()
-
-	value, err, _ := c.sf.Do("stats", func() (interface{}, error) {
-		res, err := c.fetch()
-		if err != nil {
-			return nil, err
-		}
-
-		c.mu.Lock()
-		c.value = res
-		c.expireAt = time.Now().Add(c.ttl)
-		c.mu.Unlock()
-		return res, nil
-	})
-	if err != nil {
-		return nil, err
-	}
-	return value.(map[string]WGStats), nil
-}

client/iface/configurer/stats_cache_test.go

@@ -1,70 +0,0 @@
-package configurer
-
-import (
-	"errors"
-	"sync"
-	"sync/atomic"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/require"
-)
-
-func TestStatsCache_CachesWithinTTL(t *testing.T) {
-	var calls atomic.Int64
-	c := newStatsCache(50*time.Millisecond, func() (map[string]WGStats, error) {
-		calls.Add(1)
-		return map[string]WGStats{"p": {}}, nil
-	})
-
-	for i := 0; i < 10; i++ {
-		_, err := c.get()
-		require.NoError(t, err)
-	}
-	require.Equal(t, int64(1), calls.Load(), "within TTL only one underlying fetch")
-
-	time.Sleep(60 * time.Millisecond)
-	_, err := c.get()
-	require.NoError(t, err)
-	require.Equal(t, int64(2), calls.Load(), "after TTL expiry a fresh fetch happens")
-}
-
-func TestStatsCache_SingleFlight(t *testing.T) {
-	var calls atomic.Int64
-	release := make(chan struct{})
-	c := newStatsCache(time.Minute, func() (map[string]WGStats, error) {
-		calls.Add(1)
-		<-release
-		return map[string]WGStats{}, nil
-	})
-
-	const n = 50
-	var wg sync.WaitGroup
-	wg.Add(n)
-	for i := 0; i < n; i++ {
-		go func() {
-			defer wg.Done()
-			_, _ = c.get()
-		}()
-	}
-	time.Sleep(20 * time.Millisecond)
-	close(release)
-	wg.Wait()
-
-	require.Equal(t, int64(1), calls.Load(), "concurrent misses collapse into one fetch")
-}
-
-func TestStatsCache_ErrorNotCached(t *testing.T) {
-	var calls atomic.Int64
-	wantErr := errors.New("dump failed")
-	c := newStatsCache(time.Minute, func() (map[string]WGStats, error) {
-		calls.Add(1)
-		return nil, wantErr
-	})
-
-	_, err := c.get()
-	require.ErrorIs(t, err, wantErr)
-	_, err = c.get()
-	require.ErrorIs(t, err, wantErr)
-	require.Equal(t, int64(2), calls.Load(), "errors are not cached; each call retries")
-}

client/iface/configurer/usp.go

@@ -40,7 +40,6 @@ type WGUSPConfigurer struct {
 	device           *device.Device
 	deviceName       string
 	activityRecorder *bind.ActivityRecorder
-	statsCache       *statsCache
 
 	uapiListener net.Listener
 }
@@ -51,19 +50,16 @@ func NewUSPConfigurer(device *device.Device, deviceName string, activityRecorder
 		deviceName:       deviceName,
 		activityRecorder: activityRecorder,
 	}
-	wgCfg.statsCache = newStatsCache(statsCacheTTL, wgCfg.fetchStats)
 	wgCfg.startUAPI()
 	return wgCfg
 }
 
 func NewUSPConfigurerNoUAPI(device *device.Device, deviceName string, activityRecorder *bind.ActivityRecorder) *WGUSPConfigurer {
-	wgCfg := &WGUSPConfigurer{
+	return &WGUSPConfigurer{
 		device:           device,
 		deviceName:       deviceName,
 		activityRecorder: activityRecorder,
 	}
-	wgCfg.statsCache = newStatsCache(statsCacheTTL, wgCfg.fetchStats)
-	return wgCfg
 }
 
 func (c *WGUSPConfigurer) ConfigureInterface(privateKey string, port int) error {
@@ -352,10 +348,6 @@ func (t *WGUSPConfigurer) Close() {
 }
 
 func (t *WGUSPConfigurer) GetStats() (map[string]WGStats, error) {
-	return t.statsCache.get()
-}
-
-func (t *WGUSPConfigurer) fetchStats() (map[string]WGStats, error) {
 	ipc, err := t.device.IpcGet()
 	if err != nil {
 		return nil, fmt.Errorf("ipc get: %w", err)

client/internal/engine.go

@@ -522,7 +522,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 		} else {
 			log.Infof("running rosenpass in strict mode")
 		}
-		e.rpManager, err = rosenpass.NewManager(e.config.PreSharedKey, e.config.WgIfaceName, publicKey)
+		e.rpManager, err = rosenpass.NewManager(e.config.PreSharedKey, e.config.WgIfaceName)
 		if err != nil {
 			return fmt.Errorf("create rosenpass manager: %w", err)
 		}

client/internal/peer/conn.go

@@ -30,11 +30,6 @@ import (
 	relayClient "github.com/netbirdio/netbird/shared/relay/client"
 )
 
-// wgTimeoutEscalationThreshold is the number of consecutive WireGuard
-// handshake timeouts after which the rosenpass state for the peer is
-// considered desynced and gets reset.
-const wgTimeoutEscalationThreshold = 3
-
 // MetricsRecorder is an interface for recording peer connection metrics
 type MetricsRecorder interface {
 	RecordConnectionStages(
@@ -123,9 +118,6 @@ type Conn struct {
 	wgWatcher       *WGWatcher
 	wgWatcherWg     sync.WaitGroup
 	wgWatcherCancel context.CancelFunc
-	// wgTimeouts counts consecutive WireGuard handshake timeouts without a
-	// successful handshake in between. Guarded by mu.
-	wgTimeouts int
 
 	// used to store the remote Rosenpass key for Relayed connection in case of connection update from ice
 	rosenpassRemoteKey []byte
@@ -691,29 +683,6 @@ func (conn *Conn) onWGDisconnected() {
 	default:
 		conn.Log.Debugf("No active connection to close on WG timeout")
 	}
-
-	conn.escalateWGTimeoutLocked()
-}
-
-// escalateWGTimeoutLocked resets the peer's rosenpass state after repeated
-// handshake timeouts. With rosenpass enabled, persistent timeouts mean the
-// preshared keys have desynced; the renewal exchange runs over the dead
-// tunnel and cannot resync them. Reporting the peer disconnected drops its
-// rosenpass state, so the next connection configuration programs the
-// rendezvous key and the tunnel can bootstrap again. Callers must hold mu.
-func (conn *Conn) escalateWGTimeoutLocked() {
-	if conn.config.RosenpassConfig.PubKey == nil {
-		return
-	}
-
-	conn.wgTimeouts++
-	if conn.wgTimeouts < wgTimeoutEscalationThreshold || conn.onDisconnected == nil {
-		return
-	}
-	conn.wgTimeouts = 0
-
-	conn.Log.Warnf("%d consecutive WireGuard handshake timeouts, resetting rosenpass state for peer", wgTimeoutEscalationThreshold)
-	conn.onDisconnected(conn.config.WgConfig.RemoteKey)
 }
 
 func (conn *Conn) updateRelayStatus(relayServerAddr string, rosenpassPubKey []byte, updateTime time.Time) {
@@ -843,7 +812,7 @@ func (conn *Conn) enableWgWatcherIfNeeded(enabledTime time.Time) {
 	conn.wgWatcherWg.Add(1)
 	go func() {
 		defer conn.wgWatcherWg.Done()
-		conn.wgWatcher.EnableWgWatcher(wgWatcherCtx, enabledTime, conn.onWGDisconnected, conn.onWGHandshakeSuccess, conn.onWGCheckSuccess)
+		conn.wgWatcher.EnableWgWatcher(wgWatcherCtx, enabledTime, conn.onWGDisconnected, conn.onWGHandshakeSuccess)
 	}()
 }
 
@@ -923,15 +892,6 @@ func (conn *Conn) onWGHandshakeSuccess(when time.Time) {
 	conn.recordConnectionMetrics()
 }
 
-// onWGCheckSuccess is called for every watcher check that observed a fresh
-// handshake, including handshakes of connections that were already up when
-// the watcher started.
-func (conn *Conn) onWGCheckSuccess() {
-	conn.mu.Lock()
-	conn.wgTimeouts = 0
-	conn.mu.Unlock()
-}
-
 // recordConnectionMetrics records connection stage timestamps as metrics
 func (conn *Conn) recordConnectionMetrics() {
 	if conn.metricsRecorder == nil {

client/internal/peer/conn_test.go

@@ -7,7 +7,6 @@ import (
 	"testing"
 	"time"
 
-	log "github.com/sirupsen/logrus"
 	"github.com/stretchr/testify/assert"
 
 	"github.com/netbirdio/netbird/client/iface"
@@ -305,84 +304,3 @@ func TestConn_presharedKey_RosenpassManaged(t *testing.T) {
 		t.Fatalf("expected non-nil presharedKey before Rosenpass manages PSK")
 	}
 }
-
-func newWGTimeoutTestConn(rosenpassEnabled bool, disconnected *[]string) *Conn {
-	cfg := ConnConfig{
-		Key:      "LLHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
-		LocalKey: "RRHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU=",
-		WgConfig: WgConfig{RemoteKey: "LLHf3Ma6z6mdLbriAJbqhX7+nM/B71lgw2+91q3LfhU="},
-	}
-	if rosenpassEnabled {
-		cfg.RosenpassConfig = RosenpassConfig{PubKey: []byte("dummykey")}
-	}
-
-	conn := &Conn{
-		ctx:           context.Background(),
-		config:        cfg,
-		Log:           log.WithField("peer", cfg.Key),
-		metricsStages: &MetricsStages{},
-	}
-	conn.SetOnDisconnected(func(remotePeer string) {
-		*disconnected = append(*disconnected, remotePeer)
-	})
-	return conn
-}
-
-// TestConn_onWGDisconnected_EscalatesToRosenpassReset: repeated handshake
-// timeouts with rosenpass enabled mean the preshared keys have desynced. The
-// renewal exchange runs over the dead tunnel and cannot resync them, so after
-// wgTimeoutEscalationThreshold consecutive timeouts the conn must report the
-// peer disconnected, dropping its rosenpass state so the next configuration
-// programs the rendezvous key.
-func TestConn_onWGDisconnected_EscalatesToRosenpassReset(t *testing.T) {
-	var disconnected []string
-	conn := newWGTimeoutTestConn(true, &disconnected)
-
-	for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
-		conn.onWGDisconnected()
-	}
-	assert.Empty(t, disconnected, "escalation must not fire below the threshold")
-
-	conn.onWGDisconnected()
-	assert.Equal(t, []string{conn.config.WgConfig.RemoteKey}, disconnected,
-		"reaching the threshold must report the peer disconnected once")
-
-	for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
-		conn.onWGDisconnected()
-	}
-	assert.Len(t, disconnected, 1, "escalation must restart counting after firing")
-
-	conn.onWGDisconnected()
-	assert.Len(t, disconnected, 2, "continued timeouts must escalate again")
-}
-
-// TestConn_onWGDisconnected_CheckSuccessResetsEscalation: a successful
-// handshake between timeouts means the tunnel recovered; the counter must
-// start over.
-func TestConn_onWGDisconnected_CheckSuccessResetsEscalation(t *testing.T) {
-	var disconnected []string
-	conn := newWGTimeoutTestConn(true, &disconnected)
-
-	for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
-		conn.onWGDisconnected()
-	}
-	conn.onWGCheckSuccess()
-
-	for i := 0; i < wgTimeoutEscalationThreshold-1; i++ {
-		conn.onWGDisconnected()
-	}
-	assert.Empty(t, disconnected, "handshake success must reset the timeout count")
-}
-
-// TestConn_onWGDisconnected_NoEscalationWithoutRosenpass: without rosenpass
-// there is no per-peer key state to reset; repeated timeouts must not report
-// disconnects.
-func TestConn_onWGDisconnected_NoEscalationWithoutRosenpass(t *testing.T) {
-	var disconnected []string
-	conn := newWGTimeoutTestConn(false, &disconnected)
-
-	for i := 0; i < wgTimeoutEscalationThreshold*3; i++ {
-		conn.onWGDisconnected()
-	}
-	assert.Empty(t, disconnected, "escalation must be limited to rosenpass connections")
-}

client/internal/peer/guard/guard.go

@@ -85,11 +85,7 @@ func (g *Guard) reconnectLoopWithRetry(ctx context.Context, callback func()) {
 	defer g.srWatcher.RemoveListener(srReconnectedChan)
 
 	ticker := g.initialTicker(ctx)
-	defer func() {
-		// If backoff.Ticker.send is blocked, context.Done will not close the Ticker goroutine.
-		// We have to explicitly call Stop, even if we use backoff.WithContext.
-		ticker.Stop()
-	}()
+	defer ticker.Stop()
 
 	tickerChannel := ticker.C
 

client/internal/peer/guard/guard_leak_test.go

@@ -1,92 +0,0 @@
-package guard
-
-import (
-	"context"
-	"runtime"
-	"strings"
-	"sync"
-	"testing"
-	"time"
-
-	log "github.com/sirupsen/logrus"
-
-	"github.com/netbirdio/netbird/client/internal/peer/ice"
-)
-
-func newTestGuard(status connStatusFunc) *Guard {
-	srw := NewSRWatcher(nil, nil, nil, ice.Config{})
-	return NewGuard(log.WithField("test", "guard"), status, 50*time.Millisecond, srw)
-}
-
-// countBackoffTickerGoroutines returns how many goroutines are currently sitting
-// in backoff/v4.(*Ticker).run (a ticker goroutine that has not exited).
-func countBackoffTickerGoroutines() int {
-	buf := make([]byte, 1<<25) // 32MB
-	n := runtime.Stack(buf, true)
-	return strings.Count(string(buf[:n]), "backoff/v4.(*Ticker).run")
-}
-
-// TestGuard_ReconnectTicker_NoGoroutineLeakOnShutdown reproduces a observed
-// leak: after a shutdown burst, ticker run/send goroutines stay parked
-// forever even though every reconnect loop has exited.
-func TestGuard_ReconnectTicker_NoGoroutineLeakOnShutdown(t *testing.T) {
-	before := countBackoffTickerGoroutines()
-
-	const peers = 6000
-	cancels := make([]context.CancelFunc, 0, peers)
-	var wg sync.WaitGroup
-
-	// A status check slower than the tick cadence. This models the real
-	// isConnectedOnAllWay/callback doing work: while the loop is busy in the
-	// handler, the ticker fires the next tick and parks in send(), because
-	// send() never selects on ctx.
-	slowStatus := func() ConnStatus {
-		time.Sleep(70 * time.Millisecond)
-		return ConnStatusConnected
-	}
-
-	for range peers {
-		g := newTestGuard(slowStatus)
-		ctx, cancel := context.WithCancel(context.Background())
-		cancels = append(cancels, cancel)
-		wg.Add(1)
-		go func() {
-			defer wg.Done()
-			g.Start(ctx, func() {})
-		}()
-		// Force the live ticker to be a newReconnectTicker.
-		g.SetRelayedConnDisconnected()
-	}
-
-	// Let the replacement tickers get past their 800ms initial interval, so
-	// many are parked in send() waiting on the (slow) consumer when we tear
-	// everything down.
-	time.Sleep(1500 * time.Millisecond)
-
-	// Shutdown burst: cancel every peer at once, like engine teardown.
-	for _, c := range cancels {
-		c()
-	}
-
-	// Every reconnect loop must return
-	waitCh := make(chan struct{})
-	go func() { wg.Wait(); close(waitCh) }()
-	select {
-	case <-waitCh:
-	case <-time.After(30 * time.Second):
-		t.Fatal("not all reconnect loops returned after ctx cancel")
-	}
-
-	// Give any correctly-stopped ticker goroutines time to unwind.
-	for range 50 {
-		runtime.Gosched()
-		time.Sleep(10 * time.Millisecond)
-	}
-
-	leaked := countBackoffTickerGoroutines() - before
-	t.Logf("backoff Ticker.run goroutines still parked after teardown of %d peers: %d", peers, leaked)
-	if leaked > 0 {
-		t.Errorf("LEAK: %d backoff ticker goroutines parked after all reconnect loops exited "+
-			"(defer ticker.Stop() stops the initial ticker, not the live replacement)", leaked)
-	}
-}

client/internal/peer/wg_watcher.go

@@ -71,11 +71,9 @@ func (w *WGWatcher) PrepareInitialHandshake() (ok bool) {
 
 // EnableWgWatcher runs the WireGuard watcher loop using the handshake baseline captured by
 // PrepareInitialHandshake. The watcher runs until ctx is cancelled. Caller is responsible
-// for context lifecycle management. onHandshakeSuccessFn is called only for the first
-// handshake observed by this run, onCheckSuccessFn for every check that observed a fresh
-// handshake, including the first.
-func (w *WGWatcher) EnableWgWatcher(ctx context.Context, enabledTime time.Time, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time), onCheckSuccessFn func()) {
-	w.periodicHandshakeCheck(ctx, onDisconnectedFn, onHandshakeSuccessFn, onCheckSuccessFn, enabledTime, w.initialHandshake)
+// for context lifecycle management.
+func (w *WGWatcher) EnableWgWatcher(ctx context.Context, enabledTime time.Time, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time)) {
+	w.periodicHandshakeCheck(ctx, onDisconnectedFn, onHandshakeSuccessFn, enabledTime, w.initialHandshake)
 
 	w.muEnabled.Lock()
 	w.enabled = false
@@ -92,7 +90,7 @@ func (w *WGWatcher) Reset() {
 }
 
 // wgStateCheck help to check the state of the WireGuard handshake and relay connection
-func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time), onCheckSuccessFn func(), enabledTime time.Time, initialHandshake time.Time) {
+func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time), enabledTime time.Time, initialHandshake time.Time) {
 	w.log.Infof("WireGuard watcher started")
 
 	timer := time.NewTimer(wgHandshakeOvertime)
@@ -119,10 +117,6 @@ func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn
 				}
 			}
 
-			if onCheckSuccessFn != nil && ctx.Err() == nil {
-				onCheckSuccessFn()
-			}
-
 			lastHandshake = *handshake
 
 			resetTime := time.Until(handshake.Add(checkPeriod))

client/internal/peer/wg_watcher_test.go

@@ -24,72 +24,6 @@ func (m *MocWgIface) disconnect() {
 	m.stop = true
 }
 
-type mockHandshakeStats struct {
-	mu        sync.Mutex
-	handshake time.Time
-}
-
-func (m *mockHandshakeStats) GetStats() (map[string]configurer.WGStats, error) {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	return map[string]configurer.WGStats{"": {LastHandshake: m.handshake}}, nil
-}
-
-func (m *mockHandshakeStats) advance() {
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	m.handshake = time.Now()
-}
-
-// TestWGWatcher_CheckSuccessCallback: onCheckSuccessFn must fire for a fresh
-// handshake even when the watcher started with an existing handshake baseline,
-// the case where onHandshakeSuccessFn stays silent.
-func TestWGWatcher_CheckSuccessCallback(t *testing.T) {
-	// checkPeriod bounds how stale a handshake may be before the watcher treats it
-	// as a suspended-machine timeout. The first check fires after wgHandshakeOvertime,
-	// so keep checkPeriod well above any scheduling jitter to avoid a false timeout
-	// converting the expected success into a disconnect on a loaded runner.
-	checkPeriod = 1 * time.Minute
-	wgHandshakeOvertime = 1 * time.Second
-
-	mlog := log.WithField("peer", "tet")
-	// Use an old baseline so advance() yields a strictly newer handshake even on
-	// platforms with coarse clock resolution (Windows), where two time.Now() calls
-	// microseconds apart can return the same instant and read as a timed-out handshake.
-	stats := &mockHandshakeStats{handshake: time.Now().Add(-time.Hour)}
-	watcher := NewWGWatcher(mlog, stats, "", newStateDump("peer", mlog, &Status{}))
-
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	require.True(t, watcher.PrepareInitialHandshake())
-
-	firstHandshake := make(chan struct{}, 1)
-	checkSuccess := make(chan struct{}, 1)
-	go watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {
-		firstHandshake <- struct{}{}
-	}, func() {
-		select {
-		case checkSuccess <- struct{}{}:
-		default:
-		}
-	})
-
-	stats.advance()
-
-	select {
-	case <-checkSuccess:
-	case <-time.After(10 * time.Second):
-		t.Errorf("timeout waiting for check success callback")
-	}
-
-	select {
-	case <-firstHandshake:
-		t.Errorf("first-handshake callback must not fire for a non-zero baseline")
-	default:
-	}
-}
-
 func TestWGWatcher_EnableWgWatcher(t *testing.T) {
 	checkPeriod = 5 * time.Second
 	wgHandshakeOvertime = 1 * time.Second
@@ -110,7 +44,7 @@ func TestWGWatcher_EnableWgWatcher(t *testing.T) {
 		onDisconnected <- struct{}{}
 	}, func(when time.Time) {
 		mlog.Infof("onHandshakeSuccess: %v", when)
-	}, nil)
+	})
 
 	// wait for initial reading
 	time.Sleep(2 * time.Second)
@@ -139,7 +73,7 @@ func TestWGWatcher_ReEnable(t *testing.T) {
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
-		watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {}, nil)
+		watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {})
 	}()
 	cancel()
 
@@ -155,7 +89,7 @@ func TestWGWatcher_ReEnable(t *testing.T) {
 	onDisconnected := make(chan struct{}, 1)
 	go watcher.EnableWgWatcher(ctx, time.Now(), func() {
 		onDisconnected <- struct{}{}
-	}, func(when time.Time) {}, nil)
+	}, func(when time.Time) {})
 
 	time.Sleep(2 * time.Second)
 	mocWgIface.disconnect()

client/internal/rosenpass/manager.go

@@ -39,7 +39,6 @@ type rpServer interface {
 
 type Manager struct {
 	ifaceName    string
-	localWgKey   wgtypes.Key
 	spk          []byte
 	ssk          []byte
 	rpKeyHash    string
@@ -52,9 +51,8 @@ type Manager struct {
 	wgIface      PresharedKeySetter
 }
 
-// NewManager creates a new Rosenpass manager. localWgKey is the local
-// WireGuard public key, used to derive the per-peer rendezvous key.
-func NewManager(preSharedKey *wgtypes.Key, wgIfaceName string, localWgKey wgtypes.Key) (*Manager, error) {
+// NewManager creates a new Rosenpass manager
+func NewManager(preSharedKey *wgtypes.Key, wgIfaceName string) (*Manager, error) {
 	public, secret, err := rp.GenerateKeyPair()
 	if err != nil {
 		return nil, err
@@ -64,7 +62,6 @@ func NewManager(preSharedKey *wgtypes.Key, wgIfaceName string, localWgKey wgtype
 	log.Tracef("generated new rosenpass key pair with public key %s", rpKeyHash)
 	return &Manager{
 		ifaceName:    wgIfaceName,
-		localWgKey:   localWgKey,
 		rpKeyHash:    rpKeyHash,
 		spk:          public,
 		ssk:          secret,
@@ -76,7 +73,7 @@ func NewManager(preSharedKey *wgtypes.Key, wgIfaceName string, localWgKey wgtype
 		// nil receiver in addPeer -> m.rpWgHandler.AddPeer. generateConfig will
 		// replace it with a fresh handler on each Run() to clear stale peer
 		// state from previous engine sessions.
-		rpWgHandler: NewNetbirdHandler((*[32]byte)(preSharedKey), localWgKey),
+		rpWgHandler: NewNetbirdHandler(),
 		lock:        sync.Mutex{},
 	}, nil
 }
@@ -164,7 +161,7 @@ func (m *Manager) generateConfig() (rp.Config, error) {
 	cfg.Peers = []rp.PeerConfig{}
 
 	m.lock.Lock()
-	m.rpWgHandler = NewNetbirdHandler(m.preSharedKey, m.localWgKey)
+	m.rpWgHandler = NewNetbirdHandler()
 	if m.wgIface != nil {
 		m.rpWgHandler.SetInterface(m.wgIface)
 	}

client/internal/rosenpass/manager_test.go

@@ -85,7 +85,7 @@ func newTestManager(spkFirstByte byte, mock *mockServer) *Manager {
 		ssk:         make([]byte, 32),
 		rpKeyHash:   "test-hash",
 		rpPeerIDs:   make(map[string]*rp.PeerID),
-		rpWgHandler: NewNetbirdHandler(nil, wgtypes.Key{0x01}),
+		rpWgHandler: NewNetbirdHandler(),
 		server:      mock,
 	}
 }
@@ -255,7 +255,7 @@ func TestAddPeer_NilServer_ReturnsErrorNoCrash(t *testing.T) {
 // issue #4341 cannot occur in the window between NewManager and Run().
 func TestNewManager_PreInitializesHandler(t *testing.T) {
 	psk := wgtypes.Key{}
-	m, err := NewManager(&psk, "wt0", wgtypes.Key{0x01})
+	m, err := NewManager(&psk, "wt0")
 	require.NoError(t, err)
 	require.NotNil(t, m.rpWgHandler, "rpWgHandler must be initialized in NewManager")
 }
@@ -329,10 +329,10 @@ func TestIsPresharedKeyInitialized_AddedButNotHandshaken_ReturnsFalse(t *testing
 	require.False(t, m.IsPresharedKeyInitialized(wgKey))
 }
 
-// --- NetbirdHandler.applyKey ----------------------------------------------
+// --- NetbirdHandler.outputKey ----------------------------------------------
 
-func TestHandler_ApplyKey_FirstCallUsesUpdateOnlyFalse(t *testing.T) {
-	h := NewNetbirdHandler(nil, wgtypes.Key{0x01})
+func TestHandler_OutputKey_FirstCallUsesUpdateOnlyFalse(t *testing.T) {
+	h := NewNetbirdHandler()
 	iface := &mockIface{}
 	h.SetInterface(iface)
 
@@ -348,8 +348,8 @@ func TestHandler_ApplyKey_FirstCallUsesUpdateOnlyFalse(t *testing.T) {
 	require.Equal(t, wgKey.String(), iface.calls[0].peerKey)
 }
 
-func TestHandler_ApplyKey_SubsequentCallsUseUpdateOnlyTrue(t *testing.T) {
-	h := NewNetbirdHandler(nil, wgtypes.Key{0x01})
+func TestHandler_OutputKey_SubsequentCallsUseUpdateOnlyTrue(t *testing.T) {
+	h := NewNetbirdHandler()
 	iface := &mockIface{}
 	h.SetInterface(iface)
 
@@ -364,8 +364,8 @@ func TestHandler_ApplyKey_SubsequentCallsUseUpdateOnlyTrue(t *testing.T) {
 	require.True(t, iface.calls[1].updateOnly, "subsequent rotations must use updateOnly=true")
 }
 
-func TestHandler_ApplyKey_NilInterface_NoCrashNoCall(t *testing.T) {
-	h := NewNetbirdHandler(nil, wgtypes.Key{0x01})
+func TestHandler_OutputKey_NilInterface_NoCrashNoCall(t *testing.T) {
+	h := NewNetbirdHandler()
 	// no SetInterface — iface remains nil
 	pid := rp.PeerID{0x03}
 	h.AddPeer(pid, "wt0", rp.Key(wgtypes.Key{}))
@@ -374,8 +374,8 @@ func TestHandler_ApplyKey_NilInterface_NoCrashNoCall(t *testing.T) {
 	h.HandshakeCompleted(pid, rp.Key{})
 }
 
-func TestHandler_ApplyKey_UnknownPeer_NoCall(t *testing.T) {
-	h := NewNetbirdHandler(nil, wgtypes.Key{0x01})
+func TestHandler_OutputKey_UnknownPeer_NoCall(t *testing.T) {
+	h := NewNetbirdHandler()
 	iface := &mockIface{}
 	h.SetInterface(iface)
 
@@ -384,7 +384,7 @@ func TestHandler_ApplyKey_UnknownPeer_NoCall(t *testing.T) {
 }
 
 func TestHandler_RemovePeer_ClearsInitializedState(t *testing.T) {
-	h := NewNetbirdHandler(nil, wgtypes.Key{0x01})
+	h := NewNetbirdHandler()
 	iface := &mockIface{}
 	h.SetInterface(iface)
 
@@ -398,7 +398,7 @@ func TestHandler_RemovePeer_ClearsInitializedState(t *testing.T) {
 }
 
 func TestHandler_SetInterfaceAfterAddPeer_StillReceivesKey(t *testing.T) {
-	h := NewNetbirdHandler(nil, wgtypes.Key{0x01})
+	h := NewNetbirdHandler()
 	pid := rp.PeerID{0x05}
 	wgKey := wgtypes.Key{0xEE}
 	h.AddPeer(pid, "wt0", rp.Key(wgKey))

client/internal/rosenpass/netbird_handler.go

@@ -18,34 +18,19 @@ type PresharedKeySetter interface {
 type wireGuardPeer struct {
 	Interface string
 	PublicKey rp.Key
-	// initialized is true once a completed exchange has set a
-	// Rosenpass-managed PSK for this peer.
-	initialized bool
-	// chainKey is the key output by the last completed exchange, advanced by
-	// one ratchet step on expiry. Nil until the first exchange completes and
-	// after the peer has fallen back to the rendezvous key.
-	chainKey *wgtypes.Key
-	// expiries counts failed renewals since the last completed exchange.
-	expiries int
 }
 
 type NetbirdHandler struct {
-	mu    sync.Mutex
-	iface PresharedKeySetter
-	// preSharedKey is the account-level preshared key, used as the rendezvous
-	// key when set. Nil means the deterministic seed key is used instead.
-	preSharedKey *[32]byte
-	// localWgKey is the local WireGuard public key, one of the two inputs to
-	// the deterministic seed key.
-	localWgKey wgtypes.Key
-	peers      map[rp.PeerID]*wireGuardPeer
+	mu               sync.Mutex
+	iface            PresharedKeySetter
+	peers            map[rp.PeerID]wireGuardPeer
+	initializedPeers map[rp.PeerID]bool
 }
 
-func NewNetbirdHandler(preSharedKey *[32]byte, localWgKey wgtypes.Key) *NetbirdHandler {
+func NewNetbirdHandler() *NetbirdHandler {
 	return &NetbirdHandler{
-		preSharedKey: preSharedKey,
-		localWgKey:   localWgKey,
-		peers:        map[rp.PeerID]*wireGuardPeer{},
+		peers:            map[rp.PeerID]wireGuardPeer{},
+		initializedPeers: map[rp.PeerID]bool{},
 	}
 }
 
@@ -57,16 +42,10 @@ func (h *NetbirdHandler) SetInterface(iface PresharedKeySetter) {
 	h.iface = iface
 }
 
-// AddPeer registers a peer with the handler. Re-adding a known peer (every
-// reconnection does) keeps its key recovery state.
 func (h *NetbirdHandler) AddPeer(pid rp.PeerID, intf string, pk rp.Key) {
 	h.mu.Lock()
 	defer h.mu.Unlock()
-	if existing, ok := h.peers[pid]; ok && existing.PublicKey == pk {
-		existing.Interface = intf
-		return
-	}
-	h.peers[pid] = &wireGuardPeer{
+	h.peers[pid] = wireGuardPeer{
 		Interface: intf,
 		PublicKey: pk,
 	}
@@ -76,6 +55,7 @@ func (h *NetbirdHandler) RemovePeer(pid rp.PeerID) {
 	h.mu.Lock()
 	defer h.mu.Unlock()
 	delete(h.peers, pid)
+	delete(h.initializedPeers, pid)
 }
 
 // IsPeerInitialized returns true if Rosenpass has completed a handshake
@@ -83,120 +63,50 @@ func (h *NetbirdHandler) RemovePeer(pid rp.PeerID) {
 func (h *NetbirdHandler) IsPeerInitialized(pid rp.PeerID) bool {
 	h.mu.Lock()
 	defer h.mu.Unlock()
-	peer, ok := h.peers[pid]
-	return ok && peer.initialized
+	return h.initializedPeers[pid]
 }
 
-// HandshakeCompleted programs the freshly exchanged output key and resets the
-// peer's key recovery state.
 func (h *NetbirdHandler) HandshakeCompleted(pid rp.PeerID, key rp.Key) {
-	psk := wgtypes.Key(key)
-
-	h.mu.Lock()
-	defer h.mu.Unlock()
-
-	peer, ok := h.peers[pid]
-	if !ok {
-		return
-	}
-	if peer.expiries > 0 {
-		log.Infof("rosenpass exchange completed for peer %s after %d expired renewals", wgtypes.Key(peer.PublicKey), peer.expiries)
-	}
-	// chainKey tracks the shared exchange output regardless of the local write
-	// outcome, so both ends still converge on the next expiry.
-	peer.chainKey = &psk
-	peer.expiries = 0
-	if !h.applyKeyLocked(pid, psk, peer.initialized) {
-		return
-	}
-	peer.initialized = true
+	h.outputKey(rp.KeyOutputReasonStale, pid, key)
 }
 
-// HandshakeExpired replaces the expired key. The renewal exchange runs over
-// the tunnel keyed by the PSK itself, so the replacement must be derivable on
-// both ends without communication: the first expiry ratchets the last shared
-// key forward, repeated expiries (and expiries without a completed exchange)
-// fall back to the rendezvous key and drop the peer out of the initialized
-// state so connection reconfigurations reprogram the rendezvous key as well.
 func (h *NetbirdHandler) HandshakeExpired(pid rp.PeerID) {
+	key, _ := rp.GeneratePresharedKey()
+	h.outputKey(rp.KeyOutputReasonStale, pid, key)
+}
+
+func (h *NetbirdHandler) outputKey(_ rp.KeyOutputReason, pid rp.PeerID, psk rp.Key) {
 	h.mu.Lock()
-	defer h.mu.Unlock()
+	iface := h.iface
+	wg, ok := h.peers[pid]
+	isInitialized := h.initializedPeers[pid]
+	h.mu.Unlock()
 
-	peer, ok := h.peers[pid]
-	if !ok {
-		return
-	}
-
-	peer.expiries++
-
-	var psk wgtypes.Key
-	if peer.chainKey != nil && peer.expiries == 1 {
-		log.Infof("rosenpass key for peer %s expired without renewal, advancing to ratcheted key", wgtypes.Key(peer.PublicKey))
-		psk = RatchetKey(*peer.chainKey)
-		peer.chainKey = &psk
-	} else {
-		rendezvous, err := h.rendezvousKey(peer)
-		if err != nil {
-			// Fail closed: without a rendezvous key the expired key must
-			// still be rotated out, even if the replacement is unusable.
-			log.Errorf("failed to derive rendezvous key, replacing expired key with a random one: %v", err)
-			h.applyRandomKeyLocked(pid)
-			return
-		}
-		log.Warnf("rosenpass key for peer %s expired %d times without renewal, falling back to the rendezvous key", wgtypes.Key(peer.PublicKey), peer.expiries)
-		psk = rendezvous
-		peer.chainKey = nil
-		peer.initialized = false
-	}
-
-	h.applyKeyLocked(pid, psk, true)
-}
-
-// rendezvousKey returns the key both ends converge on without communication:
-// the account-level preshared key when configured, the deterministic seed key
-// otherwise. It mirrors the key that peer connections program when Rosenpass
-// does not manage the peer yet.
-func (h *NetbirdHandler) rendezvousKey(peer *wireGuardPeer) (wgtypes.Key, error) {
-	if h.preSharedKey != nil {
-		return *h.preSharedKey, nil
-	}
-
-	seed, err := DeterministicSeedKey(h.localWgKey.String(), wgtypes.Key(peer.PublicKey).String())
-	if err != nil {
-		return wgtypes.Key{}, err
-	}
-	return *seed, nil
-}
-
-// applyKeyLocked writes the preshared key for the peer to the WireGuard
-// interface and reports whether the write succeeded. Callers must hold h.mu
-// for the whole state-mutation-plus-write so that a concurrent completion and
-// expiry cannot reorder their writes relative to the in-memory chain key.
-func (h *NetbirdHandler) applyKeyLocked(pid rp.PeerID, psk wgtypes.Key, updateOnly bool) bool {
-	peer, ok := h.peers[pid]
-	if !ok {
-		return false
-	}
-
-	if h.iface == nil {
+	if iface == nil {
 		log.Warn("rosenpass: interface not set, cannot update preshared key")
-		return false
-	}
-
-	peerKey := wgtypes.Key(peer.PublicKey).String()
-	if err := h.iface.SetPresharedKey(peerKey, psk, updateOnly); err != nil {
-		log.Errorf("Failed to apply rosenpass key: %v", err)
-		return false
-	}
-
-	return true
-}
-
-func (h *NetbirdHandler) applyRandomKeyLocked(pid rp.PeerID) {
-	key, err := rp.GeneratePresharedKey()
-	if err != nil {
-		log.Errorf("failed to generate random preshared key: %v", err)
 		return
 	}
-	h.applyKeyLocked(pid, wgtypes.Key(key), true)
+
+	if !ok {
+		return
+	}
+
+	peerKey := wgtypes.Key(wg.PublicKey).String()
+	pskKey := wgtypes.Key(psk)
+
+	// Use updateOnly=true for later rotations (peer already has Rosenpass PSK)
+	// Use updateOnly=false for first rotation (peer has original/empty PSK)
+	if err := iface.SetPresharedKey(peerKey, pskKey, isInitialized); err != nil {
+		log.Errorf("Failed to apply rosenpass key: %v", err)
+		return
+	}
+
+	// Mark peer as isInitialized after the successful first rotation
+	if !isInitialized {
+		h.mu.Lock()
+		if _, exists := h.peers[pid]; exists {
+			h.initializedPeers[pid] = true
+		}
+		h.mu.Unlock()
+	}
 }

client/internal/rosenpass/netbird_handler_test.go

@@ -1,250 +0,0 @@
-package rosenpass
-
-import (
-	"testing"
-
-	rp "cunicu.li/go-rosenpass"
-	"github.com/stretchr/testify/require"
-	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
-)
-
-// handlerTestLink wires two NetbirdHandlers as the two ends of a single
-// tunnel: handler A manages the rosenpass peer B and vice versa, the way two
-// NetBird clients see each other.
-type handlerTestLink struct {
-	handlerA, handlerB *NetbirdHandler
-	ifaceA, ifaceB     *mockIface
-	pidA, pidB         rp.PeerID
-	wgKeyA, wgKeyB     wgtypes.Key
-}
-
-func newHandlerTestLink(t *testing.T, preSharedKey *[32]byte) *handlerTestLink {
-	t.Helper()
-
-	link := &handlerTestLink{
-		ifaceA: &mockIface{},
-		ifaceB: &mockIface{},
-	}
-	link.pidA[0] = 0xaa
-	link.pidB[0] = 0xbb
-	link.wgKeyA[31] = 1
-	link.wgKeyB[31] = 2
-
-	link.handlerA = NewNetbirdHandler(preSharedKey, link.wgKeyA)
-	link.handlerB = NewNetbirdHandler(preSharedKey, link.wgKeyB)
-
-	link.handlerA.SetInterface(link.ifaceA)
-	link.handlerB.SetInterface(link.ifaceB)
-
-	link.handlerA.AddPeer(link.pidB, "wt0", rp.Key(link.wgKeyB))
-	link.handlerB.AddPeer(link.pidA, "wt0", rp.Key(link.wgKeyA))
-
-	return link
-}
-
-// complete simulates a completed rosenpass exchange: both ends derive the
-// same output key.
-func (l *handlerTestLink) complete(osk rp.Key) {
-	l.handlerA.HandshakeCompleted(l.pidB, osk)
-	l.handlerB.HandshakeCompleted(l.pidA, osk)
-}
-
-// expire simulates a failed key renewal on both ends.
-func (l *handlerTestLink) expire() {
-	l.handlerA.HandshakeExpired(l.pidB)
-	l.handlerB.HandshakeExpired(l.pidA)
-}
-
-func lastPSK(t *testing.T, m *mockIface) wgtypes.Key {
-	t.Helper()
-	m.mu.Lock()
-	defer m.mu.Unlock()
-	require.NotEmpty(t, m.calls, "expected at least one SetPresharedKey call")
-	return m.calls[len(m.calls)-1].psk
-}
-
-func TestHandshakeCompleted_SetsKeyAndInitializes(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	require.Equal(t, wgtypes.Key(osk), lastPSK(t, link.ifaceA), "completed exchange must program the osk")
-	require.False(t, link.ifaceA.calls[0].updateOnly, "first rotation must not be update-only")
-	require.True(t, link.handlerA.IsPeerInitialized(link.pidB), "peer must be initialized after first completed exchange")
-
-	link.complete(osk)
-	require.True(t, link.ifaceA.calls[1].updateOnly, "later rotations must be update-only")
-}
-
-// TestHandshakeExpired_BothSidesConverge encodes the core recovery invariant:
-// rosenpass renewals run over the tunnel that the PSK itself keys, so when a
-// renewal fails on both ends, both ends must fall back to the same key or the
-// tunnel can never handshake again.
-func TestHandshakeExpired_BothSidesConverge(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	link.expire()
-	keyA := lastPSK(t, link.ifaceA)
-	keyB := lastPSK(t, link.ifaceB)
-	require.NotEqual(t, wgtypes.Key(osk), keyA, "expired key must be rotated out")
-	require.Equal(t, keyA, keyB, "both ends must converge on the same key after expiry")
-
-	link.expire()
-	require.Equal(t, lastPSK(t, link.ifaceA), lastPSK(t, link.ifaceB),
-		"both ends must still converge after repeated expiries")
-}
-
-// TestHandshakeExpired_ExpiryWithoutCompletionConverges covers the bootstrap
-// case: the initial exchange never completed (the tunnel ran on the rendezvous
-// key), so an expiry must not replace the working key with an unrecoverable
-// one on either end.
-func TestHandshakeExpired_ExpiryWithoutCompletionConverges(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	link.expire()
-	require.Equal(t, lastPSK(t, link.ifaceA), lastPSK(t, link.ifaceB),
-		"both ends must converge when the exchange never completed")
-}
-
-// TestHandshakeExpired_RepeatedExpiryClearsInitialized: once renewals keep
-// failing, the peer must drop out of the initialized state so the next
-// connection reconfiguration reprograms the rendezvous key instead of
-// preserving a poisoned rosenpass-managed key.
-func TestHandshakeExpired_RepeatedExpiryClearsInitialized(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	link.expire()
-	link.expire()
-
-	require.False(t, link.handlerA.IsPeerInitialized(link.pidB),
-		"repeated expiries must clear the initialized state")
-	require.False(t, link.handlerB.IsPeerInitialized(link.pidA),
-		"repeated expiries must clear the initialized state")
-}
-
-// TestHandshakeCompleted_AfterExpiryRecovers: a completed exchange after a
-// desync must fully reset the recovery state.
-func TestHandshakeCompleted_AfterExpiryRecovers(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk1, osk2 rp.Key
-	osk1[0] = 1
-	osk2[0] = 2
-
-	link.complete(osk1)
-	link.expire()
-	link.expire()
-
-	link.complete(osk2)
-	require.Equal(t, wgtypes.Key(osk2), lastPSK(t, link.ifaceA), "new exchange must program the fresh osk")
-	require.True(t, link.handlerA.IsPeerInitialized(link.pidB), "peer must be initialized again after recovery")
-
-	link.expire()
-	require.Equal(t, lastPSK(t, link.ifaceA), lastPSK(t, link.ifaceB),
-		"recovered link must converge again on the next expiry")
-	require.NotEqual(t, wgtypes.Key(osk2), lastPSK(t, link.ifaceA), "expired key must be rotated out")
-}
-
-// TestHandshakeExpired_FirstExpiryRatchetsLastKey: the first expiry must
-// derive the replacement from the last shared key, so an attacker who only
-// blocks the renewal exchange gains nothing over the previous key.
-func TestHandshakeExpired_FirstExpiryRatchetsLastKey(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	link.expire()
-	require.Equal(t, RatchetKey(wgtypes.Key(osk)), lastPSK(t, link.ifaceA),
-		"first expiry must program the ratcheted key")
-	require.True(t, link.handlerA.IsPeerInitialized(link.pidB),
-		"ratchet step must keep the peer initialized so reconfigurations preserve the key")
-}
-
-// TestHandshakeExpired_RepeatedExpiryFallsBackToSeed: once the ratchet key
-// also fails, both ends must land on the same key that peer connections
-// program for uninitialized peers, so a reconnect completes the recovery.
-func TestHandshakeExpired_RepeatedExpiryFallsBackToSeed(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	link.expire()
-	link.expire()
-
-	seed, err := DeterministicSeedKey(link.wgKeyA.String(), link.wgKeyB.String())
-	require.NoError(t, err)
-	require.Equal(t, *seed, lastPSK(t, link.ifaceA), "repeated expiry must fall back to the seed key")
-	require.Equal(t, *seed, lastPSK(t, link.ifaceB), "repeated expiry must fall back to the seed key")
-}
-
-// TestHandshakeExpired_ConfiguredPSKUsedAsRendezvous: with an account-level
-// preshared key configured, the fallback must be that key, matching what peer
-// connections program for uninitialized peers.
-func TestHandshakeExpired_ConfiguredPSKUsedAsRendezvous(t *testing.T) {
-	psk := &[32]byte{0x77}
-	link := newHandlerTestLink(t, psk)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	link.expire()
-	link.expire()
-
-	require.Equal(t, wgtypes.Key(*psk), lastPSK(t, link.ifaceA),
-		"fallback must be the configured preshared key")
-	require.Equal(t, wgtypes.Key(*psk), lastPSK(t, link.ifaceB),
-		"fallback must be the configured preshared key on both ends")
-}
-
-// TestHandshakeExpired_ExpiryWritesAreUpdateOnly: expiry replacements must
-// never create a WireGuard peer that connection management has removed.
-func TestHandshakeExpired_ExpiryWritesAreUpdateOnly(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-
-	link.expire()
-	link.expire()
-
-	for _, call := range link.ifaceA.calls[1:] {
-		require.True(t, call.updateOnly, "expiry writes must be update-only")
-	}
-}
-
-// TestAddPeer_ReAddKeepsRecoveryState: reconnections re-add the peer on every
-// OnConnected; that must not reset the expiry chain state.
-func TestAddPeer_ReAddKeepsRecoveryState(t *testing.T) {
-	link := newHandlerTestLink(t, nil)
-
-	var osk rp.Key
-	osk[0] = 0x42
-	link.complete(osk)
-	link.expire()
-
-	link.handlerA.AddPeer(link.pidB, "wt0", rp.Key(link.wgKeyB))
-	require.True(t, link.handlerA.IsPeerInitialized(link.pidB),
-		"re-adding a known peer must keep its state")
-
-	link.expire()
-	seed, err := DeterministicSeedKey(link.wgKeyA.String(), link.wgKeyB.String())
-	require.NoError(t, err)
-	require.Equal(t, *seed, lastPSK(t, link.ifaceA),
-		"second expiry after re-add must continue to the seed fallback")
-}

client/internal/rosenpass/seed.go

@@ -1,28 +1,11 @@
 package rosenpass
 
 import (
-	"crypto/sha256"
 	"fmt"
 
 	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
 )
 
-// ratchetLabel domain-separates the expiry ratchet from other uses of the
-// rosenpass output key.
-const ratchetLabel = "netbird-rosenpass-expiry-ratchet"
-
-// RatchetKey derives the successor preshared key from the previous Rosenpass
-// output key. When a key expires without a completed renewal, both peers
-// advance their last shared key by one ratchet step: the expired key is
-// rotated out while both ends still converge on an identical, non-public
-// replacement without communicating.
-func RatchetKey(prev wgtypes.Key) wgtypes.Key {
-	input := make([]byte, 0, len(ratchetLabel)+len(prev))
-	input = append(input, ratchetLabel...)
-	input = append(input, prev[:]...)
-	return sha256.Sum256(input)
-}
-
 // DeterministicSeedKey derives a 32-byte WireGuard preshared key from a pair
 // of peer public keys. Both peers, given the same key pair, produce the same
 // output regardless of which side runs the function: the inputs are ordered

client/internal/routemanager/exit_node_selection_test.go

@@ -1,191 +0,0 @@
-package routemanager
-
-import (
-	"net/netip"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/netbirdio/netbird/client/internal/routeselector"
-	"github.com/netbirdio/netbird/route"
-)
-
-func newExitNodeTestManager() *DefaultManager {
-	return &DefaultManager{routeSelector: routeselector.NewRouteSelector()}
-}
-
-func exitRoute(netID, peer string, skipAutoApply bool) *route.Route {
-	return &route.Route{
-		NetID:         route.NetID(netID),
-		Network:       netip.MustParsePrefix("0.0.0.0/0"),
-		Peer:          peer,
-		SkipAutoApply: skipAutoApply,
-	}
-}
-
-func TestPickPreferredExitNode(t *testing.T) {
-	tests := []struct {
-		name string
-		info exitNodeInfo
-		want route.NetID
-	}{
-		{
-			name: "persisted user selection wins over management",
-			info: exitNodeInfo{
-				allIDs:               []route.NetID{"a", "b", "c"},
-				userSelected:         []route.NetID{"b"},
-				selectedByManagement: []route.NetID{"a"},
-			},
-			want: "b",
-		},
-		{
-			name: "multiple user-selected self-heal to deterministic min",
-			info: exitNodeInfo{
-				allIDs:       []route.NetID{"a", "b", "c"},
-				userSelected: []route.NetID{"c", "a"},
-			},
-			want: "a",
-		},
-		{
-			name: "explicit opt-out keeps none",
-			info: exitNodeInfo{
-				allIDs:         []route.NetID{"a", "b"},
-				userDeselected: []route.NetID{"a", "b"},
-			},
-			want: "",
-		},
-		{
-			name: "fresh defaults to management auto-apply pick",
-			info: exitNodeInfo{
-				allIDs:               []route.NetID{"a", "b", "c"},
-				selectedByManagement: []route.NetID{"b"},
-			},
-			want: "b",
-		},
-		{
-			name: "no user pick and no management auto-apply selects none",
-			info: exitNodeInfo{
-				allIDs: []route.NetID{"c", "a", "b"},
-			},
-			want: "",
-		},
-		{
-			name: "user-deselect does not block a management auto-apply sibling",
-			info: exitNodeInfo{
-				allIDs:               []route.NetID{"a", "b"},
-				userDeselected:       []route.NetID{"a"},
-				selectedByManagement: []route.NetID{"b"},
-			},
-			want: "b",
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			assert.Equal(t, tt.want, pickPreferredExitNode(tt.info), "preferred exit node")
-		})
-	}
-}
-
-func TestEnforceSingleExitNode(t *testing.T) {
-	m := newExitNodeTestManager()
-	all := []route.NetID{"a", "b", "c"}
-
-	m.enforceSingleExitNode("b", all)
-	assert.False(t, m.routeSelector.IsSelected("a"), "a should be deselected")
-	assert.True(t, m.routeSelector.IsSelected("b"), "b should be the only selected exit node")
-	assert.False(t, m.routeSelector.IsSelected("c"), "c should be deselected")
-
-	// Switching the preferred node moves the single selection.
-	m.enforceSingleExitNode("c", all)
-	assert.False(t, m.routeSelector.IsSelected("a"), "a stays deselected")
-	assert.False(t, m.routeSelector.IsSelected("b"), "b should now be deselected")
-	assert.True(t, m.routeSelector.IsSelected("c"), "c should now be selected")
-
-	// Empty preferred turns every exit node off.
-	m.enforceSingleExitNode("", all)
-	for _, id := range all {
-		assert.False(t, m.routeSelector.IsSelected(id), "no exit node should be selected")
-	}
-}
-
-func TestEnforceSingleExitNode_RespectsDeselectAll(t *testing.T) {
-	m := newExitNodeTestManager()
-	m.routeSelector.DeselectAllRoutes()
-
-	m.enforceSingleExitNode("b", []route.NetID{"a", "b"})
-
-	assert.True(t, m.routeSelector.IsDeselectAll(), "global deselect-all must stay in effect")
-	assert.False(t, m.routeSelector.IsSelected("b"), "no exit node should be forced on while deselect-all is set")
-}
-
-func TestUpdateRouteSelectorFromManagement_FreshSelectsOne(t *testing.T) {
-	m := newExitNodeTestManager()
-	routes := route.HAMap{
-		"exitA|0.0.0.0/0":    {exitRoute("exitA", "p1", false)},
-		"exitB|0.0.0.0/0":    {exitRoute("exitB", "p2", false)},
-		"lan|192.168.1.0/24": {{NetID: "lan", Network: netip.MustParsePrefix("192.168.1.0/24"), Peer: "p3"}},
-		"exitC|0.0.0.0/0":    {exitRoute("exitC", "p4", false)},
-	}
-
-	m.updateRouteSelectorFromManagement(routes)
-
-	// Exactly one exit node (the deterministic first) is selected.
-	assert.True(t, m.routeSelector.IsSelected("exitA"), "exitA is the deterministic default")
-	assert.False(t, m.routeSelector.IsSelected("exitB"), "exitB must not also be selected")
-	assert.False(t, m.routeSelector.IsSelected("exitC"), "exitC must not also be selected")
-	// Non-exit routes are left at their default-on state.
-	assert.True(t, m.routeSelector.IsSelected("lan"), "non-exit route selection is untouched")
-}
-
-func TestUpdateRouteSelectorFromManagement_HonorsPersistedPick(t *testing.T) {
-	m := newExitNodeTestManager()
-	routes := route.HAMap{
-		"exitA|0.0.0.0/0": {exitRoute("exitA", "p1", false)},
-		"exitB|0.0.0.0/0": {exitRoute("exitB", "p2", false)},
-	}
-	all := []route.NetID{"exitA", "exitB"}
-
-	// Simulate the state the runtime select path leaves behind: exactly one
-	// exit node explicitly selected, its sibling deselected.
-	require.NoError(t, m.routeSelector.SelectRoutes([]route.NetID{"exitB"}, true, all))
-	require.NoError(t, m.routeSelector.DeselectRoutes([]route.NetID{"exitA"}, all))
-
-	m.updateRouteSelectorFromManagement(routes)
-
-	assert.True(t, m.routeSelector.IsSelected("exitB"), "persisted pick must stay selected")
-	assert.False(t, m.routeSelector.IsSelected("exitA"), "the other exit node stays deselected")
-}
-
-func TestUpdateRouteSelectorFromManagement_OptOutKeepsNone(t *testing.T) {
-	m := newExitNodeTestManager()
-	routes := route.HAMap{
-		"exitA|0.0.0.0/0": {exitRoute("exitA", "p1", false)},
-		"exitB|0.0.0.0/0": {exitRoute("exitB", "p2", false)},
-	}
-	all := []route.NetID{"exitA", "exitB"}
-
-	// User deselected exit nodes and selected none.
-	require.NoError(t, m.routeSelector.DeselectRoutes(all, all))
-
-	m.updateRouteSelectorFromManagement(routes)
-
-	assert.False(t, m.routeSelector.IsSelected("exitA"), "opt-out keeps exitA off")
-	assert.False(t, m.routeSelector.IsSelected("exitB"), "opt-out keeps exitB off")
-}
-
-func TestUpdateRouteSelectorFromManagement_NoAutoApplySelectsNone(t *testing.T) {
-	m := newExitNodeTestManager()
-	// SkipAutoApply=true: management offers the exit nodes but doesn't request
-	// auto-activation, so none should be selected until the user picks one.
-	routes := route.HAMap{
-		"exitA|0.0.0.0/0": {exitRoute("exitA", "p1", true)},
-		"exitB|0.0.0.0/0": {exitRoute("exitB", "p2", true)},
-	}
-
-	m.updateRouteSelectorFromManagement(routes)
-
-	assert.False(t, m.routeSelector.IsSelected("exitA"), "no auto-apply keeps exitA off")
-	assert.False(t, m.routeSelector.IsSelected("exitB"), "no auto-apply keeps exitB off")
-}

client/internal/routemanager/manager.go

@@ -701,13 +701,7 @@ func resolveURLsToIPs(urls []string) []net.IP {
 	return ips
 }
 
-// updateRouteSelectorFromManagement reconciles exit-node selection on every
-// network map: it keeps at most one exit node selected — the user's persisted
-// pick, else whatever management marks for auto-apply (SkipAutoApply=false),
-// else none. We never auto-activate an exit node the map doesn't request; it
-// stays off until the user picks it. Exit nodes are mutually exclusive, but the
-// RouteSelector stores routes with default-on semantics, so without this every
-// available exit node would report selected at once.
+// updateRouteSelectorFromManagement updates the route selector based on the isSelected status from the management server
 func (m *DefaultManager) updateRouteSelectorFromManagement(clientRoutes route.HAMap) {
 	m.mirrorV6ExitPairSelections(clientRoutes)
 
@@ -718,14 +712,13 @@ func (m *DefaultManager) updateRouteSelectorFromManagement(clientRoutes route.HA
 		return
 	}
 
-	info := m.collectExitNodeInfo(clientRoutes)
-	if len(info.allIDs) == 0 {
+	exitNodeInfo := m.collectExitNodeInfo(clientRoutes)
+	if len(exitNodeInfo.allIDs) == 0 {
 		return
 	}
 
-	preferred := pickPreferredExitNode(info)
-	m.enforceSingleExitNode(preferred, info.allIDs)
-	m.logExitNodeUpdate(info, preferred)
+	m.updateExitNodeSelections(exitNodeInfo)
+	m.logExitNodeUpdate(exitNodeInfo)
 }
 
 // mirrorV6ExitPairSelections keeps every synthesized "-v6" exit route's selection
@@ -753,10 +746,6 @@ type exitNodeInfo struct {
 	userDeselected       []route.NetID
 }
 
-// collectExitNodeInfo categorises the available exit nodes by their persisted
-// selection state. It keys on the base (v4) NetID and skips the synthesized
-// "-v6" partner, which inherits its base's selection through the RouteSelector
-// — counting it separately would double-count the pair.
 func (m *DefaultManager) collectExitNodeInfo(clientRoutes route.HAMap) exitNodeInfo {
 	var info exitNodeInfo
 
@@ -766,9 +755,6 @@ func (m *DefaultManager) collectExitNodeInfo(clientRoutes route.HAMap) exitNodeI
 		}
 
 		netID := haID.NetID()
-		if strings.HasSuffix(string(netID), route.V6ExitSuffix) {
-			continue
-		}
 		info.allIDs = append(info.allIDs, netID)
 
 		if m.routeSelector.HasUserSelectionForRoute(netID) {
@@ -805,52 +791,45 @@ func (m *DefaultManager) checkManagementSelection(routes []*route.Route, netID r
 	}
 }
 
-// pickPreferredExitNode chooses the single exit node to keep selected. In order:
-//   - a persisted user selection wins (deterministic if several survive from
-//     legacy state, so the set self-heals down to one);
-//   - otherwise activate only what management marks for auto-apply
-//     (SkipAutoApply=false); the lexicographically first if it marks several.
-//
-// Returns "" when neither holds — we never force an arbitrary exit node on. A
-// route the map doesn't auto-apply stays off until the user selects it.
-// info.userDeselected is informational only: an explicit deselect simply keeps
-// that route out of both lists above, so it can't be picked.
-func pickPreferredExitNode(info exitNodeInfo) route.NetID {
-	if len(info.userSelected) > 0 {
-		return minNetID(info.userSelected)
-	}
-	if len(info.selectedByManagement) > 0 {
-		return minNetID(info.selectedByManagement)
-	}
-	return ""
+func (m *DefaultManager) updateExitNodeSelections(info exitNodeInfo) {
+	routesToDeselect := m.getRoutesToDeselect(info.allIDs)
+	m.deselectExitNodes(routesToDeselect)
+	m.selectExitNodesByManagement(info.selectedByManagement, info.allIDs)
 }
 
-// enforceSingleExitNode makes preferred the only selected exit node: every other
-// available exit node is deselected and preferred (if any) is selected, without
-// disturbing non-exit route selections. The whole reconciliation runs under a
-// single RouteSelector lock (SetExclusiveExitNode) so a concurrent deselect-all
-// cannot interleave and get undone; a global deselect-all is left untouched so
-// the user's "all off" stays in effect.
-func (m *DefaultManager) enforceSingleExitNode(preferred route.NetID, allIDs []route.NetID) {
-	m.routeSelector.SetExclusiveExitNode(preferred, allIDs)
-}
-
-func (m *DefaultManager) logExitNodeUpdate(info exitNodeInfo, preferred route.NetID) {
-	log.Debugf("Exit node selection: %d available, preferred=%q (%d user-selected, %d user-deselected, %d management-selected)",
-		len(info.allIDs), preferred, len(info.userSelected), len(info.userDeselected), len(info.selectedByManagement))
-}
-
-// minNetID returns the lexicographically smallest NetID, for a deterministic
-// default pick that stays stable across restarts.
-func minNetID(ids []route.NetID) route.NetID {
-	if len(ids) == 0 {
-		return ""
-	}
-	best := ids[0]
-	for _, id := range ids[1:] {
-		if id < best {
-			best = id
+func (m *DefaultManager) getRoutesToDeselect(allIDs []route.NetID) []route.NetID {
+	var routesToDeselect []route.NetID
+	for _, netID := range allIDs {
+		if !m.routeSelector.HasUserSelectionForRoute(netID) {
+			routesToDeselect = append(routesToDeselect, netID)
 		}
 	}
-	return best
+	return routesToDeselect
+}
+
+func (m *DefaultManager) deselectExitNodes(routesToDeselect []route.NetID) {
+	if len(routesToDeselect) == 0 {
+		return
+	}
+
+	err := m.routeSelector.DeselectRoutes(routesToDeselect, routesToDeselect)
+	if err != nil {
+		log.Warnf("Failed to deselect exit nodes: %v", err)
+	}
+}
+
+func (m *DefaultManager) selectExitNodesByManagement(selectedByManagement []route.NetID, allIDs []route.NetID) {
+	if len(selectedByManagement) == 0 {
+		return
+	}
+
+	err := m.routeSelector.SelectRoutes(selectedByManagement, true, allIDs)
+	if err != nil {
+		log.Warnf("Failed to select exit nodes: %v", err)
+	}
+}
+
+func (m *DefaultManager) logExitNodeUpdate(info exitNodeInfo) {
+	log.Debugf("Updated route selector: %d exit nodes available, %d selected by management, %d user-selected, %d user-deselected",
+		len(info.allIDs), len(info.selectedByManagement), len(info.userSelected), len(info.userDeselected))
 }

client/internal/routeselector/routeselector.go

@@ -115,38 +115,7 @@ func (rs *RouteSelector) DeselectAllRoutes() {
 	clear(rs.selectedRoutes)
 }
 
-// SetExclusiveExitNode atomically makes preferred the only selected exit node
-// among exitIDs: every other ID in exitIDs is deselected and preferred (when
-// non-empty) is selected, all under a single lock. Holding the lock across the
-// whole reconciliation prevents a concurrent DeselectAllRoutes from interleaving
-// between the deselect and select steps and being silently undone. A global
-// deselect-all is left untouched so the user's "all off" stays in effect;
-// non-exit routes are never referenced, so their selection is preserved.
-func (rs *RouteSelector) SetExclusiveExitNode(preferred route.NetID, exitIDs []route.NetID) {
-	rs.mu.Lock()
-	defer rs.mu.Unlock()
-
-	if rs.deselectAll {
-		return
-	}
-
-	for _, id := range exitIDs {
-		if id == preferred {
-			continue
-		}
-		rs.deselectedRoutes[id] = struct{}{}
-		delete(rs.selectedRoutes, id)
-	}
-
-	if preferred != "" {
-		delete(rs.deselectedRoutes, preferred)
-		rs.selectedRoutes[preferred] = struct{}{}
-	}
-}
-
-// IsDeselectAll reports whether the global "deselect all" flag is set, i.e. the
-// user explicitly disabled every route. Callers enforcing per-route invariants
-// (e.g. single exit node) should leave the selection untouched when it is.
+// IsDeselectAll reports whether the user has explicitly deselected all routes.
 func (rs *RouteSelector) IsDeselectAll() bool {
 	rs.mu.RLock()
 	defer rs.mu.RUnlock()

client/internal/statemanager/manager.go

@@ -1,7 +1,6 @@
 package statemanager
 
 import (
-	"bytes"
 	"context"
 	"encoding/json"
 	"errors"
@@ -306,11 +305,6 @@ func (m *Manager) loadStateFile(deleteCorrupt bool) (map[string]json.RawMessage,
 
 	var rawStates map[string]json.RawMessage
 	if err := json.Unmarshal(data, &rawStates); err != nil {
-		if len(bytes.TrimSpace(data)) == 0 {
-			log.Warnf("state file %s is empty (%d bytes)", m.filePath, len(data))
-		} else {
-			log.Warnf("state file %s has malformed content (%d bytes)", m.filePath, len(data))
-		}
 		m.handleCorruptedState(deleteCorrupt)
 		return nil, fmt.Errorf("unmarshal states: %w", err)
 	}

client/ios/NetBirdSDK/login.go

@@ -44,25 +44,10 @@ type Auth struct {
 // NewAuth instantiate Auth struct and validate the management URL
 func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
 	inputCfg := profilemanager.ConfigInput{
-		ConfigPath:    cfgPath,
 		ManagementURL: mgmURL,
 	}
 
-	// Load the existing config when a config file is already present so an
-	// interactive re-login reuses the peer's persisted WireGuard private key
-	// (and thus its identity) instead of generating a fresh one. Generating a
-	// new key registers a brand-new peer on the management server on every
-	// re-auth (named after the fallback hostname). Only fall back to a fresh
-	// in-memory config for the first-time login when no config file exists yet.
-	// DirectUpdateOrCreateConfig uses non-atomic writes so it also works inside
-	// the tvOS App Group sandbox where atomic temp-file+rename is blocked.
-	var cfg *profilemanager.Config
-	var err error
-	if cfgPath != "" {
-		cfg, err = profilemanager.DirectUpdateOrCreateConfig(inputCfg)
-	} else {
-		cfg, err = profilemanager.CreateInMemoryConfig(inputCfg)
-	}
+	cfg, err := profilemanager.CreateInMemoryConfig(inputCfg)
 	if err != nil {
 		return nil, err
 	}

client/server/mdm.go

@@ -3,7 +3,6 @@ package server
 import (
 	"context"
 	"fmt"
-	"net/url"
 	"time"
 
 	log "github.com/sirupsen/logrus"
@@ -182,37 +181,6 @@ func conflictBool(key string, p *bool) conflictCheck {
 	}
 }
 
-func canonicalURL(s string) string {
-	u, err := url.ParseRequestURI(s)
-	if err != nil {
-		return s
-	}
-	if u.Port() == "" {
-		switch u.Scheme {
-		case "https":
-			u.Host += ":443"
-		case "http":
-			u.Host += ":80"
-		}
-	}
-	return u.String()
-}
-
-// conflictURL is conflictString for URL-typed keys: both sides are
-// normalized via canonicalURL before comparison.
-func conflictURL(key, got string) conflictCheck {
-	return conflictCheck{
-		key: key,
-		check: func(pol *mdm.Policy) bool {
-			if got == "" {
-				return true
-			}
-			want, ok := pol.GetString(key)
-			return ok && canonicalURL(want) == canonicalURL(got)
-		},
-	}
-}
-
 // conflictString builds a conflictCheck for a string MDM key. An empty
 // `got` is treated as "field not set" (no override requested); otherwise
 // the check returns true only when the policy contains the key and its
@@ -288,7 +256,7 @@ func mdmManagedFieldConflicts(msg *proto.SetConfigRequest, policy *mdm.Policy) [
 	}
 
 	return resolveConflicts(policy, []conflictCheck{
-		conflictURL(mdm.KeyManagementURL, msg.ManagementUrl),
+		conflictString(mdm.KeyManagementURL, msg.ManagementUrl),
 		conflictString(mdm.KeyPreSharedKey, pskGot),
 		conflictBool(mdm.KeyRosenpassEnabled, msg.RosenpassEnabled),
 		conflictBool(mdm.KeyRosenpassPermissive, msg.RosenpassPermissive),
@@ -409,7 +377,7 @@ func loginRequestMDMConflicts(msg *proto.LoginRequest, policy *mdm.Policy) []str
 	}
 
 	return resolveConflicts(policy, []conflictCheck{
-		conflictURL(mdm.KeyManagementURL, msg.ManagementUrl),
+		conflictString(mdm.KeyManagementURL, msg.ManagementUrl),
 		conflictString(mdm.KeyPreSharedKey, pskGot),
 		conflictBool(mdm.KeyRosenpassEnabled, msg.RosenpassEnabled),
 		conflictBool(mdm.KeyRosenpassPermissive, msg.RosenpassPermissive),

client/server/setconfig_mdm_test.go

@@ -181,43 +181,6 @@ func TestSetConfig_MDMAllow_NonManagedFields(t *testing.T) {
 	require.NotNil(t, resp)
 }
 
-// TestSetConfig_MDMAllow_ManagementURLPortNormalized covers the
-// regression from discussion #6483: MDM URL without explicit port vs
-// UI echo with the parseURL-appended default port must be treated as
-// a no-op echo, not a conflict.
-func TestSetConfig_MDMAllow_ManagementURLPortNormalized(t *testing.T) {
-	tests := []struct {
-		name      string
-		mdmURL    string
-		submitURL string
-	}{
-		{"policy_no_port_submit_with_443", "https://netbird.corp.example", "https://netbird.corp.example:443"},
-		{"policy_with_443_submit_no_port", "https://netbird.corp.example:443", "https://netbird.corp.example"},
-		{"http_policy_no_port_submit_with_80", "http://netbird.corp.example", "http://netbird.corp.example:80"},
-	}
-
-	for _, tc := range tests {
-		t.Run(tc.name, func(t *testing.T) {
-			withMDMPolicy(t, mdm.NewPolicy(map[string]any{
-				mdm.KeyManagementURL: tc.mdmURL,
-			}))
-
-			s, ctx, profName, username, _ := setupServerWithProfile(t)
-
-			rosenpassEnabled := true
-			resp, err := s.SetConfig(ctx, &proto.SetConfigRequest{
-				ProfileName:      profName,
-				Username:         username,
-				ManagementUrl:    tc.submitURL,
-				RosenpassEnabled: &rosenpassEnabled,
-			})
-
-			require.NoError(t, err, "port-normalized URL echo must not trip MDM conflict gate")
-			require.NotNil(t, resp)
-		})
-	}
-}
-
 func TestSetConfig_MDMEmpty_NoEnforcement(t *testing.T) {
 	// No MDM policy active: any field can be written.
 	withMDMPolicy(t, mdm.NewPolicy(nil))

client/system/process.go

@@ -7,7 +7,7 @@ import (
 	"os"
 	"slices"
 
-	"github.com/shirou/gopsutil/v4/process"
+	"github.com/shirou/gopsutil/v3/process"
 )
 
 // getRunningProcesses returns a list of running process paths. The context bounds the work:

client/system/process_test.go

@@ -4,7 +4,7 @@ import (
 	"context"
 	"testing"
 
-	"github.com/shirou/gopsutil/v4/process"
+	"github.com/shirou/gopsutil/v3/process"
 )
 
 func Benchmark_getRunningProcesses(b *testing.B) {

e2e/agentnetwork/chat_test.go

@@ -91,7 +91,7 @@ func availableProviders() []providerCase {
 		if region == "" {
 			region = "us-east-1"
 		}
-		ps = append(ps, providerCase{name: "bedrock", catalogID: "bedrock_api", upstream: "https://bedrock-runtime." + region + ".amazonaws.com", apiKey: k, model: "us.anthropic.claude-haiku-4-5", kind: harness.WireBedrock})
+		ps = append(ps, providerCase{name: "bedrock", catalogID: "bedrock_api", upstream: "https://bedrock-runtime." + region + ".amazonaws.com", apiKey: k, model: "us.anthropic.claude-haiku-4-5", kind: harness.WireMessages})
 	}
 	return ps
 }
@@ -224,12 +224,9 @@ func TestProvidersMatrix(t *testing.T) {
 				var c int
 				var b string
 				var cerr error
-				switch pc.kind {
-				case harness.WireVertex:
+				if pc.kind == harness.WireVertex {
 					c, b, cerr = cl.Vertex(ctx, settings.Endpoint, proxyIP, pc.project, pc.region, pc.model, "Reply with exactly: pong", sessionID)
-				case harness.WireBedrock:
-					c, b, cerr = cl.Bedrock(ctx, settings.Endpoint, proxyIP, pc.model, "Reply with exactly: pong", sessionID)
-				default:
+				} else {
 					c, b, cerr = cl.Chat(ctx, settings.Endpoint, proxyIP, pc.kind, pc.model, "Reply with exactly: pong", sessionID)
 				}
 				if cerr == nil {

e2e/agentnetwork/guardrail_test.go

@@ -1,168 +0,0 @@
-//go:build e2e
-
-package agentnetwork
-
-import (
-	"context"
-	"strings"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/netbirdio/netbird/e2e/harness"
-	"github.com/netbirdio/netbird/shared/management/http/api"
-)
-
-// catalogModel returns the normalized catalog id the proxy stamps for a
-// path-routed provider's configured model — the form the guardrail allowlist is
-// compared against (region prefix / @version stripped).
-func catalogModel(pc providerCase) string {
-	switch pc.kind {
-	case harness.WireBedrock:
-		return strings.TrimPrefix(pc.model, "us.")
-	case harness.WireVertex:
-		return strings.SplitN(pc.model, "@", 2)[0]
-	default:
-		return pc.model
-	}
-}
-
-// disallowedModel returns a valid-shaped model id for the provider that is NOT
-// the configured/allowed one, so the guardrail must reject it before the
-// request ever reaches the upstream.
-func disallowedModel(pc providerCase) string {
-	switch pc.kind {
-	case harness.WireBedrock:
-		return "us.anthropic.claude-opus-4-8"
-	case harness.WireVertex:
-		return "claude-opus-4-8@20250101"
-	default:
-		return "unlisted-model"
-	}
-}
-
-// sendModel drives one request for the given model through the provider's native
-// wire shape and returns the HTTP status.
-func sendModel(ctx context.Context, t *testing.T, cl *harness.Client, endpoint, proxyIP string, pc providerCase, model string) int {
-	t.Helper()
-	var code int
-	var err error
-	switch pc.kind {
-	case harness.WireBedrock:
-		code, _, err = cl.Bedrock(ctx, endpoint, proxyIP, model, "Reply with exactly: pong", "")
-	case harness.WireVertex:
-		code, _, err = cl.Vertex(ctx, endpoint, proxyIP, pc.project, pc.region, model, "Reply with exactly: pong", "")
-	default:
-		code, _, err = cl.Chat(ctx, endpoint, proxyIP, pc.kind, model, "Reply with exactly: pong", "")
-	}
-	require.NoError(t, err, "request must reach the proxy for %s", pc.name)
-	return code
-}
-
-// TestModelAllowlistEnforced provisions a Model Allowlist guardrail limiting each
-// path-routed provider (Bedrock, Vertex) to its configured model, then drives
-// requests over the tunnel: the allowed model returns 200 while a model outside
-// the allowlist is denied 403 by the guardrail before it reaches the upstream.
-// This is the coverage missing for #6751 — the model for these providers travels
-// in the URL path, and the allowlist must be enforced there.
-func TestModelAllowlistEnforced(t *testing.T) {
-	var providers []providerCase
-	for _, pc := range availableProviders() {
-		if pc.kind == harness.WireBedrock || pc.kind == harness.WireVertex {
-			providers = append(providers, pc)
-		}
-	}
-	if len(providers) == 0 {
-		t.Skip("no path-routed provider keys set (AWS_BEARER_TOKEN_BEDROCK / GOOGLE_VERTEX_*); source ~/.llm-keys")
-	}
-
-	ctx, cancel := context.WithTimeout(context.Background(), 20*time.Minute)
-	defer cancel()
-
-	grp, err := srv.API().Groups.Create(ctx, api.PostApiGroupsJSONRequestBody{Name: "e2e-allowlist"})
-	require.NoError(t, err, "create group")
-	t.Cleanup(func() { _ = srv.API().Groups.Delete(context.Background(), grp.Id) })
-
-	ephemeral := false
-	sk, err := srv.API().SetupKeys.Create(ctx, api.PostApiSetupKeysJSONRequestBody{
-		Name:       "e2e-allowlist-client",
-		Type:       "reusable",
-		ExpiresIn:  86400,
-		UsageLimit: 0,
-		AutoGroups: []string{grp.Id},
-		Ephemeral:  &ephemeral,
-	})
-	require.NoError(t, err, "mint setup key")
-
-	// Providers with their configured (allowed) models; the first bootstraps the cluster.
-	ids := make([]string, 0, len(providers))
-	allowed := make([]string, 0, len(providers))
-	for i, pc := range providers {
-		req := providerRequest(pc)
-		if i == 0 {
-			req.BootstrapCluster = ptr(harness.AgentNetworkCluster)
-		}
-		prov, perr := srv.CreateProvider(ctx, req)
-		require.NoError(t, perr, "create provider %s", pc.name)
-		id := prov.Id
-		ids = append(ids, id)
-		allowed = append(allowed, catalogModel(pc))
-		t.Cleanup(func() { _ = srv.DeleteProvider(context.Background(), id) })
-	}
-
-	// Guardrail allowlisting exactly the configured models.
-	var gr api.AgentNetworkGuardrailRequest
-	gr.Name = "e2e-allowlist"
-	gr.Checks.ModelAllowlist.Enabled = true
-	gr.Checks.ModelAllowlist.Models = allowed
-	guard, err := srv.CreateGuardrail(ctx, gr)
-	require.NoError(t, err, "create guardrail")
-	t.Cleanup(func() { _ = srv.DeleteGuardrail(context.Background(), guard.Id) })
-
-	enabled := true
-	pol, err := srv.CreatePolicy(ctx, api.AgentNetworkPolicyRequest{
-		Name:                   "e2e-allowlist",
-		Enabled:                &enabled,
-		SourceGroups:           []string{grp.Id},
-		DestinationProviderIds: ids,
-		GuardrailIds:           &[]string{guard.Id},
-	})
-	require.NoError(t, err, "create policy")
-	t.Cleanup(func() { _ = srv.DeletePolicy(context.Background(), pol.Id) })
-
-	settings, err := srv.GetSettings(ctx)
-	require.NoError(t, err, "read settings for endpoint")
-	require.NotEmpty(t, settings.Endpoint, "agent-network endpoint must be assigned")
-
-	proxyToken, err := srv.CreateProxyTokenCLI(ctx, "e2e-proxy-allowlist")
-	require.NoError(t, err, "mint proxy token via CLI")
-	px, err := harness.StartProxy(ctx, srv, proxyToken)
-	require.NoError(t, err, "start proxy")
-	t.Cleanup(func() { _ = px.Terminate(context.Background()) })
-
-	cl, err := harness.StartClient(ctx, srv, sk.Key)
-	require.NoError(t, err, "start client")
-	t.Cleanup(func() { _ = cl.Terminate(context.Background()) })
-
-	require.NoError(t, cl.WaitConnected(ctx, 90*time.Second), "client must connect to management")
-	if err := cl.WaitProxyPeer(ctx, 180*time.Second); err != nil {
-		t.Fatalf("client did not see the proxy peer: %v\n=== proxy logs ===\n%s", err, px.Logs(context.Background()))
-	}
-	proxyIP, err := cl.ResolveProxyIP(ctx, settings.Endpoint)
-	require.NoError(t, err, "resolve agent-network endpoint to proxy IP")
-
-	for _, pc := range providers {
-		pc := pc
-		t.Run(pc.name, func(t *testing.T) {
-			// The admin's allowlisted model is served end to end.
-			assert.Equal(t, 200, sendModel(ctx, t, cl, settings.Endpoint, proxyIP, pc, pc.model),
-				"allowlisted model must be permitted for %s", pc.name)
-			// A model outside the allowlist is rejected by the guardrail (before
-			// the upstream), regardless of whether it is a real catalog model.
-			assert.Equal(t, 403, sendModel(ctx, t, cl, settings.Endpoint, proxyIP, pc, disallowedModel(pc)),
-				"model outside the allowlist must be denied for %s", pc.name)
-		})
-	}
-}

e2e/agentnetwork/vllm_test.go

@@ -1,171 +0,0 @@
-//go:build e2e
-
-package agentnetwork
-
-import (
-	"context"
-	"strings"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/require"
-
-	"github.com/netbirdio/netbird/e2e/harness"
-	"github.com/netbirdio/netbird/shared/management/http/api"
-)
-
-// TestVLLMProvider proves the proxy supports a self-hosted vLLM backend. vLLM is
-// OpenAI-compatible, so it uses the "vllm" catalog entry (KindCustom) and is
-// reached over plain HTTP — no TLS anywhere on the path:
-//
-//	client --tunnel--> netbird proxy --http--> vllm (:8000, OpenAI-compatible)
-//
-// The mock vLLM server answers /v1/chat/completions with an OpenAI-shaped
-// completion carrying a non-zero usage block. The test asserts the chat returns
-// 200 with the completion, that the request is recorded in the access log by its
-// session id, and that vLLM's usage block is metered into a consumption row —
-// which together prove request routing, response parsing, and token accounting
-// all work for a self-hosted OpenAI-compatible provider.
-//
-// It needs no external credentials (the mock ignores auth), so it always runs.
-func TestVLLMProvider(t *testing.T) {
-	ctx, cancel := context.WithTimeout(context.Background(), 15*time.Minute)
-	defer cancel()
-
-	vllm, err := harness.StartVLLM(ctx, srv)
-	require.NoError(t, err, "start mock vLLM server")
-	t.Cleanup(func() { _ = vllm.Terminate(context.Background()) })
-
-	grp, err := srv.API().Groups.Create(ctx, api.PostApiGroupsJSONRequestBody{Name: "e2e-vllm"})
-	require.NoError(t, err, "create group")
-	t.Cleanup(func() { _ = srv.API().Groups.Delete(context.Background(), grp.Id) })
-
-	ephemeral := false
-	sk, err := srv.API().SetupKeys.Create(ctx, api.PostApiSetupKeysJSONRequestBody{
-		Name:       "e2e-vllm-client",
-		Type:       "reusable",
-		ExpiresIn:  86400,
-		UsageLimit: 0,
-		AutoGroups: []string{grp.Id},
-		Ephemeral:  &ephemeral,
-	})
-	require.NoError(t, err, "mint setup key")
-	require.NotEmpty(t, sk.Key, "setup key plaintext")
-
-	// vLLM provider pointed at the mock over plain HTTP. The mock ignores auth,
-	// so a dummy key satisfies the "Bearer ${API_KEY}" template. The served model
-	// is enumerated so the router dispatches this model string to this provider.
-	dummyKey := "sk-vllm-e2e"
-	prov, err := srv.CreateProvider(ctx, api.AgentNetworkProviderRequest{
-		Name:             "vllm",
-		ProviderId:       "vllm",
-		UpstreamUrl:      vllm.URL,
-		ApiKey:           &dummyKey,
-		Enabled:          ptr(true),
-		BootstrapCluster: ptr(harness.AgentNetworkCluster),
-		Models: &[]api.AgentNetworkProviderModel{
-			{Id: harness.VLLMModel, InputPer1k: 0.001, OutputPer1k: 0.002},
-		},
-	})
-	require.NoError(t, err, "create vllm provider")
-	t.Cleanup(func() { _ = srv.DeleteProvider(context.Background(), prov.Id) })
-
-	// Token limit far above the handful of tokens this test drives, so it never
-	// blocks but switches on usage metering — the switch that makes consumption
-	// rows get recorded.
-	enabled := true
-	pol, err := srv.CreatePolicy(ctx, api.AgentNetworkPolicyRequest{
-		Name:                   "e2e-vllm-allow",
-		Enabled:                &enabled,
-		SourceGroups:           []string{grp.Id},
-		DestinationProviderIds: []string{prov.Id},
-		Limits: &api.AgentNetworkPolicyLimits{
-			TokenLimit: api.AgentNetworkPolicyTokenLimit{
-				Enabled:       true,
-				GroupCap:      10_000_000,
-				UserCap:       10_000_000,
-				WindowSeconds: 60,
-			},
-		},
-	})
-	require.NoError(t, err, "create policy")
-	t.Cleanup(func() { _ = srv.DeletePolicy(context.Background(), pol.Id) })
-
-	settings, err := srv.GetSettings(ctx)
-	require.NoError(t, err, "read settings")
-	require.NotEmpty(t, settings.Endpoint, "endpoint must be assigned")
-
-	proxyToken, err := srv.CreateProxyTokenCLI(ctx, "e2e-vllm-proxy")
-	require.NoError(t, err, "mint proxy token")
-	px, err := harness.StartProxy(ctx, srv, proxyToken)
-	require.NoError(t, err, "start proxy")
-	t.Cleanup(func() { _ = px.Terminate(context.Background()) })
-
-	cl, err := harness.StartClient(ctx, srv, sk.Key)
-	require.NoError(t, err, "start client")
-	t.Cleanup(func() { _ = cl.Terminate(context.Background()) })
-
-	require.NoError(t, cl.WaitConnected(ctx, 90*time.Second), "client must connect to management")
-	if err := cl.WaitProxyPeer(ctx, 180*time.Second); err != nil {
-		t.Fatalf("client did not see the proxy peer: %v\n=== proxy logs ===\n%s", err, px.Logs(context.Background()))
-	}
-	proxyIP, err := cl.ResolveProxyIP(ctx, settings.Endpoint)
-	require.NoError(t, err, "resolve endpoint to proxy IP")
-
-	before, _ := srv.ListAccessLogs(ctx)
-	sessionID := "e2e-session-vllm"
-
-	// Retry to absorb tunnel/DNS jitter on the first call.
-	var code int
-	var body string
-	deadline := time.Now().Add(90 * time.Second)
-	for time.Now().Before(deadline) {
-		c, b, cerr := cl.Chat(ctx, settings.Endpoint, proxyIP, harness.WireChat, harness.VLLMModel, "Reply with exactly: pong", sessionID)
-		if cerr == nil {
-			code, body = c, b
-			if code == 200 {
-				break
-			}
-		}
-		time.Sleep(5 * time.Second)
-	}
-	require.Equal(t, 200, code,
-		"chat through the vLLM provider must return 200; body: %s\n=== vllm logs ===\n%s\n=== proxy logs ===\n%s",
-		body, vllm.Logs(context.Background()), px.Logs(context.Background()))
-	require.True(t, strings.Contains(body, "chat.completion"),
-		"body should be an OpenAI-compatible chat completion; got: %s", body)
-
-	// The request must surface as an access-log row carrying our session id.
-	require.Eventually(t, func() bool {
-		logs, lerr := srv.ListAccessLogs(ctx)
-		return lerr == nil && logs.TotalRecords > before.TotalRecords
-	}, 30*time.Second, 2*time.Second, "an access-log row should be ingested for the vLLM provider")
-
-	require.Eventually(t, func() bool {
-		logs, lerr := srv.ListAccessLogs(ctx)
-		if lerr != nil {
-			return false
-		}
-		for _, r := range logs.Data {
-			if r.SessionId != nil && *r.SessionId == sessionID {
-				return true
-			}
-		}
-		return false
-	}, 30*time.Second, 2*time.Second, "session id %q must be recorded in an access-log row", sessionID)
-
-	// vLLM's usage block (prompt_tokens=11, completion_tokens=2) must be parsed
-	// and metered into a consumption row with positive token counts.
-	require.Eventually(t, func() bool {
-		rows, lerr := srv.ListConsumption(ctx)
-		if lerr != nil {
-			return false
-		}
-		for _, r := range rows {
-			if r.TokensInput > 0 && r.TokensOutput > 0 {
-				return true
-			}
-		}
-		return false
-	}, 60*time.Second, 3*time.Second, "vLLM usage must be metered into a consumption row")
-}

e2e/harness/agentnetwork.go

@@ -107,17 +107,6 @@ func (c *Combined) DeletePolicy(ctx context.Context, id string) error {
 	return anDelete(ctx, c, "/api/agent-network/policies/"+id)
 }
 
-// CreateGuardrail creates an agent-network guardrail (e.g. a model allowlist)
-// that can then be attached to a policy via its GuardrailIds.
-func (c *Combined) CreateGuardrail(ctx context.Context, req api.AgentNetworkGuardrailRequest) (api.AgentNetworkGuardrail, error) {
-	return anRequest[api.AgentNetworkGuardrail](ctx, c, http.MethodPost, "/api/agent-network/guardrails", req)
-}
-
-// DeleteGuardrail removes a guardrail by id.
-func (c *Combined) DeleteGuardrail(ctx context.Context, id string) error {
-	return anDelete(ctx, c, "/api/agent-network/guardrails/"+id)
-}
-
 // GetSettings returns the account's agent-network settings row. It exists only
 // after the first provider create bootstraps it.
 func (c *Combined) GetSettings(ctx context.Context) (api.AgentNetworkSettings, error) {

e2e/harness/client.go

@@ -194,11 +194,6 @@ const (
 	// WireVertex is the Anthropic-on-Vertex rawPredict shape: the client posts
 	// the full Vertex model path and the proxy mints the SA OAuth token.
 	WireVertex = "vertex"
-	// WireBedrock is the native AWS Bedrock InvokeModel shape: the model id
-	// travels in the URL path (/model/{id}/invoke), not the body, so the proxy
-	// routes by path. This is what a Bedrock SDK client sends and the shape the
-	// model-allowlist guardrail must enforce.
-	WireBedrock = "bedrock"
 )
 
 // Chat issues a chat-completion POST to the agent-network endpoint over the
@@ -231,17 +226,6 @@ func (cl *Client) Vertex(ctx context.Context, endpoint, proxyIP, project, region
 	return cl.post(ctx, endpoint, proxyIP, path, body, withSessionID(nil, sessionID))
 }
 
-// Bedrock issues a native AWS Bedrock InvokeModel POST over the tunnel. The
-// model id is carried in the request path (/model/{id}/invoke), so the proxy
-// routes by path; the body uses the bedrock anthropic_version rather than a
-// model field. A non-empty sessionID is sent as the universal x-session-id
-// header the proxy records.
-func (cl *Client) Bedrock(ctx context.Context, endpoint, proxyIP, model, prompt, sessionID string) (int, string, error) {
-	path := "/model/" + model + "/invoke"
-	body := fmt.Sprintf(`{"anthropic_version":"bedrock-2023-05-31","max_tokens":64,"messages":[{"role":"user","content":%q}]}`, prompt)
-	return cl.post(ctx, endpoint, proxyIP, path, body, withSessionID(nil, sessionID))
-}
-
 // withSessionID appends the x-session-id header when sessionID is non-empty.
 func withSessionID(headers []string, sessionID string) []string {
 	if sessionID == "" {

e2e/harness/vllm.go

@@ -1,113 +0,0 @@
-//go:build e2e
-
-package harness
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"path/filepath"
-	"time"
-
-	"github.com/docker/docker/api/types/container"
-	"github.com/testcontainers/testcontainers-go"
-	"github.com/testcontainers/testcontainers-go/wait"
-)
-
-const (
-	vllmImage = "nginx:alpine"
-	vllmAlias = "vllm"
-	vllmPort  = "8000/tcp"
-	// VLLMModel is the served model id the mock advertises and echoes back. It
-	// matches a real small model commonly served by vLLM so the provider's
-	// enumerated model and the client's request line up.
-	VLLMModel = "Qwen/Qwen2.5-0.5B-Instruct"
-)
-
-// vllmNginxConf emulates a vLLM OpenAI-compatible server over plain HTTP (vLLM's
-// default: no TLS, port 8000). It answers /v1/models with a one-model list and
-// any chat/completions path with a canned OpenAI-shaped chat completion carrying
-// a non-zero usage block, so the proxy's OpenAI parser records real token
-// consumption. Running actual vLLM in CI is infeasible (GPU + multi-GB model
-// download), so this stands in for the wire contract the proxy depends on.
-const vllmNginxConf = `pid /tmp/nginx.pid;
-events {}
-http {
-  server {
-    listen 8000;
-    location = /v1/models {
-      default_type application/json;
-      return 200 '{"object":"list","data":[{"id":"Qwen/Qwen2.5-0.5B-Instruct","object":"model","owned_by":"vllm"}]}';
-    }
-    location / {
-      default_type application/json;
-      return 200 '{"id":"chatcmpl-e2e-vllm","object":"chat.completion","created":1700000000,"model":"Qwen/Qwen2.5-0.5B-Instruct","choices":[{"index":0,"message":{"role":"assistant","content":"pong"},"finish_reason":"stop"}],"usage":{"prompt_tokens":11,"completion_tokens":2,"total_tokens":13}}';
-    }
-  }
-}
-`
-
-// VLLM is a mock vLLM OpenAI-compatible server on the combined server's network,
-// reachable at http://vllm:8000. A "vllm" provider points at it to exercise the
-// proxy's support for self-hosted OpenAI-compatible backends.
-type VLLM struct {
-	container testcontainers.Container
-	workDir   string
-	// URL is the upstream URL the vllm provider points at (http://<alias>:8000).
-	URL string
-}
-
-// StartVLLM runs the mock vLLM server on the shared network over plain HTTP.
-func StartVLLM(ctx context.Context, c *Combined) (*VLLM, error) {
-	workDir, err := os.MkdirTemp("/tmp", "nb-e2e-vllm-*")
-	if err != nil {
-		return nil, fmt.Errorf("create vllm work dir: %w", err)
-	}
-	// Widen so the (non-root worker) nginx container can traverse the bind mount.
-	if err := os.Chmod(workDir, 0o755); err != nil { //nolint:gosec // throwaway e2e config dir
-		return nil, fmt.Errorf("chmod vllm dir: %w", err)
-	}
-	if err := os.WriteFile(filepath.Join(workDir, "nginx.conf"), []byte(vllmNginxConf), 0o644); err != nil { //nolint:gosec // non-secret e2e config
-		return nil, fmt.Errorf("write nginx conf: %w", err)
-	}
-
-	req := testcontainers.ContainerRequest{
-		Image:          vllmImage,
-		ExposedPorts:   []string{vllmPort},
-		Networks:       []string{c.network.Name},
-		NetworkAliases: map[string][]string{c.network.Name: {vllmAlias}},
-		Cmd:            []string{"nginx", "-c", "/conf/nginx.conf", "-g", "daemon off;"},
-		HostConfigModifier: func(hc *container.HostConfig) {
-			hc.Binds = append(hc.Binds, workDir+":/conf:ro")
-		},
-		WaitingFor: wait.ForListeningPort(vllmPort).WithStartupTimeout(60 * time.Second),
-	}
-
-	ctr, err := testcontainers.GenericContainer(ctx, testcontainers.GenericContainerRequest{
-		ContainerRequest: req,
-		Started:          true,
-	})
-	if err != nil {
-		_ = os.RemoveAll(workDir)
-		return nil, fmt.Errorf("start vllm container: %w", err)
-	}
-
-	return &VLLM{container: ctr, workDir: workDir, URL: "http://" + vllmAlias + ":8000"}, nil
-}
-
-// Logs returns the vLLM container logs, for diagnostics on failure.
-func (v *VLLM) Logs(ctx context.Context) string {
-	return containerLogs(ctx, v.container)
-}
-
-// Terminate stops the vLLM container and cleans its work dir.
-func (v *VLLM) Terminate(ctx context.Context) error {
-	var err error
-	if v.container != nil {
-		err = v.container.Terminate(ctx)
-	}
-	if v.workDir != "" {
-		_ = os.RemoveAll(v.workDir)
-	}
-	return err
-}

go.mod

@@ -2,7 +2,7 @@ module github.com/netbirdio/netbird
 
 go 1.25.5
 
-toolchain go1.25.12
+toolchain go1.25.11
 
 require (
 	cunicu.li/go-rosenpass v0.5.42
@@ -19,8 +19,8 @@ require (
 	github.com/spf13/cobra v1.10.2
 	github.com/spf13/pflag v1.0.9
 	github.com/vishvananda/netlink v1.3.1
-	golang.org/x/crypto v0.54.0
-	golang.org/x/sys v0.47.0
+	golang.org/x/crypto v0.50.0
+	golang.org/x/sys v0.43.0
 	golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173
 	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10
 	golang.zx2c4.com/wireguard/windows v0.5.3
@@ -72,7 +72,7 @@ require (
 	github.com/hashicorp/go-multierror v1.1.1
 	github.com/hashicorp/go-secure-stdlib/base62 v0.1.2
 	github.com/hashicorp/go-version v1.7.0
-	github.com/jackc/pgx/v5 v5.5.5
+	github.com/jackc/pgx/v5 v5.9.2
 	github.com/libdns/route53 v1.5.0
 	github.com/libp2p/go-nat v0.2.0
 	github.com/libp2p/go-netroute v0.4.0
@@ -104,7 +104,6 @@ require (
 	github.com/redis/go-redis/v9 v9.7.3
 	github.com/rs/xid v1.3.0
 	github.com/shirou/gopsutil/v3 v3.24.4
-	github.com/shirou/gopsutil/v4 v4.25.8
 	github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966
 	github.com/songgao/water v0.0.0-20200317203138-2b4b6d7c09d8
 	github.com/stretchr/testify v1.11.1
@@ -128,11 +127,11 @@ require (
 	goauthentik.io/api/v3 v3.2023051.3
 	golang.org/x/exp v0.0.0-20250620022241-b7579e27df2b
 	golang.org/x/mobile v0.0.0-20251113184115-a159579294ab
-	golang.org/x/mod v0.37.0
-	golang.org/x/net v0.56.0
+	golang.org/x/mod v0.34.0
+	golang.org/x/net v0.53.0
 	golang.org/x/oauth2 v0.36.0
-	golang.org/x/sync v0.22.0
-	golang.org/x/term v0.45.0
+	golang.org/x/sync v0.20.0
+	golang.org/x/term v0.42.0
 	golang.org/x/time v0.15.0
 	google.golang.org/api v0.276.0
 	gopkg.in/yaml.v3 v3.0.1
@@ -246,8 +245,8 @@ require (
 	github.com/huin/goupnp v1.2.0 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/jackc/pgpassfile v1.0.0 // indirect
-	github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a // indirect
-	github.com/jackc/puddle/v2 v2.2.1 // indirect
+	github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 // indirect
+	github.com/jackc/puddle/v2 v2.2.2 // indirect
 	github.com/jackpal/go-nat-pmp v1.0.2 // indirect
 	github.com/jeandeaual/go-locale v0.0.0-20250612000132-0ef82f21eade // indirect
 	github.com/jinzhu/inflection v1.0.0 // indirect
@@ -309,6 +308,7 @@ require (
 	github.com/russellhaering/goxmldsig v1.6.0 // indirect
 	github.com/ryanuber/go-glob v1.0.0 // indirect
 	github.com/rymdport/portal v0.4.2 // indirect
+	github.com/shirou/gopsutil/v4 v4.25.8 // indirect
 	github.com/shoenig/go-m1cpu v0.2.1 // indirect
 	github.com/shopspring/decimal v1.4.0 // indirect
 	github.com/spf13/cast v1.7.0 // indirect
@@ -332,8 +332,8 @@ require (
 	go.uber.org/multierr v1.11.0 // indirect
 	go.yaml.in/yaml/v2 v2.4.3 // indirect
 	golang.org/x/image v0.33.0 // indirect
-	golang.org/x/text v0.40.0 // indirect
-	golang.org/x/tools v0.47.0 // indirect
+	golang.org/x/text v0.36.0 // indirect
+	golang.org/x/tools v0.43.0 // indirect
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
 	google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 // indirect
 	gopkg.in/square/go-jose.v2 v2.6.0 // indirect

go.sum

@@ -368,12 +368,12 @@ github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
 github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM=
 github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg=
-github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a h1:bbPeKD0xmW/Y25WS6cokEszi5g+S0QxI/d45PkRi7Nk=
-github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
-github.com/jackc/pgx/v5 v5.5.5 h1:amBjrZVmksIdNjxGW/IiIMzxMKZFelXbUoPNb+8sjQw=
-github.com/jackc/pgx/v5 v5.5.5/go.mod h1:ez9gk+OAat140fv9ErkZDYFWmXLfV+++K0uAOiwgm1A=
-github.com/jackc/puddle/v2 v2.2.1 h1:RhxXJtFG022u4ibrCSMSiu5aOq1i77R3OHKNJj77OAk=
-github.com/jackc/puddle/v2 v2.2.1/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7UlwOQYpKFhBabPMi4aNAfoODPEFNiAnClxo=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
+github.com/jackc/pgx/v5 v5.9.2 h1:3ZhOzMWnR4yJ+RW1XImIPsD1aNSz4T4fyP7zlQb56hw=
+github.com/jackc/pgx/v5 v5.9.2/go.mod h1:mal1tBGAFfLHvZzaYh77YS/eC6IX9OWbRV1QIIM0Jn4=
+github.com/jackc/puddle/v2 v2.2.2 h1:PR8nw+E/1w0GLuRFSmiioY6UooMp6KJv0/61nB7icHo=
+github.com/jackc/puddle/v2 v2.2.2/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
 github.com/jackpal/go-nat-pmp v1.0.2 h1:KzKSgb7qkJvOUTqYl9/Hg/me3pWgBmERKrTGD7BdWus=
 github.com/jackpal/go-nat-pmp v1.0.2/go.mod h1:QPH045xvCAeXUZOxsnwmrtiCoxIr9eob+4orBN1SBKc=
 github.com/jcmturner/aescts/v2 v2.0.0 h1:9YKLH6ey7H4eDBXW8khjYslgyqG2xZikXP0EQFKrle8=
@@ -781,8 +781,8 @@ golang.org/x/crypto v0.18.0/go.mod h1:R0j02AL6hcrfOiy9T4ZYp/rcWeMxM3L6QYxlOuEG1m
 golang.org/x/crypto v0.19.0/go.mod h1:Iy9bg/ha4yyC70EfRS8jz+B6ybOBKMaSxLj6P6oBDfU=
 golang.org/x/crypto v0.23.0/go.mod h1:CKFgDieR+mRhux2Lsu27y0fO304Db0wZe70UKqHu0v8=
 golang.org/x/crypto v0.31.0/go.mod h1:kDsLvtWBEx7MV9tJOj9bnXsPbxwJQ6csT/x4KIN4Ssk=
-golang.org/x/crypto v0.54.0 h1:YLIA59K4fiNzHzjnZt2tUJQjQtUWfWbeHBqKtk3eScw=
-golang.org/x/crypto v0.54.0/go.mod h1:KWL8ny2AZdGR2cWmzeHrp2azQPGogOv+HeQaVEXC2dk=
+golang.org/x/crypto v0.50.0 h1:zO47/JPrL6vsNkINmLoo/PH1gcxpls50DNogFvB5ZGI=
+golang.org/x/crypto v0.50.0/go.mod h1:3muZ7vA7PBCE6xgPX7nkzzjiUq87kRItoJQM1Yo8S+Q=
 golang.org/x/exp v0.0.0-20250620022241-b7579e27df2b h1:M2rDM6z3Fhozi9O7NWsxAkg/yqS/lQJ6PmkyIV3YP+o=
 golang.org/x/exp v0.0.0-20250620022241-b7579e27df2b/go.mod h1:3//PLf8L/X+8b4vuAfHzxeRUl04Adcb341+IGKfnqS8=
 golang.org/x/image v0.33.0 h1:LXRZRnv1+zGd5XBUVRFmYEphyyKJjQjCRiOuAP3sZfQ=
@@ -799,8 +799,8 @@ golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
 golang.org/x/mod v0.12.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
 golang.org/x/mod v0.15.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c=
 golang.org/x/mod v0.17.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c=
-golang.org/x/mod v0.37.0 h1:vF1DjpVEshcIqoEaauuHebaLk1O1forxjxBaVn884JQ=
-golang.org/x/mod v0.37.0/go.mod h1:m8S8VeM9r4dzDwjrKO0a1sZP3YjeMamRRlD+fmR2Q/0=
+golang.org/x/mod v0.34.0 h1:xIHgNUUnW6sYkcM5Jleh05DvLOtwc6RitGHbDk4akRI=
+golang.org/x/mod v0.34.0/go.mod h1:ykgH52iCZe79kzLLMhyCUzhMci+nQj+0XkbXpNYtVjY=
 golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190603091049-60506f45cf65/go.mod h1:HSz+uSET+XFnRR8LxR5pz3Of3rY3CfYBVs4xY44aLks=
@@ -819,8 +819,8 @@ golang.org/x/net v0.15.0/go.mod h1:idbUs1IY1+zTqbi8yxTbhexhEEk5ur9LInksu6HrEpk=
 golang.org/x/net v0.20.0/go.mod h1:z8BVo6PvndSri0LbOE3hAn0apkU+1YvI6E70E9jsnvY=
 golang.org/x/net v0.21.0/go.mod h1:bIjVDfnllIU7BJ2DNgfnXvpSvtn8VRwhlsaeUTyUS44=
 golang.org/x/net v0.25.0/go.mod h1:JkAGAh7GEvH74S6FOH42FLoXpXbE/aqXSrIQjXgsiwM=
-golang.org/x/net v0.56.0 h1:Rw8j/hFzGvJUZwNBXnAtf5sVDVt+65SK2C7IxCxZt5o=
-golang.org/x/net v0.56.0/go.mod h1:D3Ku6r+V6JROoZK144D2XfMHFcMq/0zSfLelVTCFKec=
+golang.org/x/net v0.53.0 h1:d+qAbo5L0orcWAr0a9JweQpjXF19LMXJE8Ey7hwOdUA=
+golang.org/x/net v0.53.0/go.mod h1:JvMuJH7rrdiCfbeHoo3fCQU24Lf5JJwT9W3sJFulfgs=
 golang.org/x/oauth2 v0.8.0/go.mod h1:yr7u4HXZRm1R1kBWqr/xKNqewf0plRYoB7sla+BCIXE=
 golang.org/x/oauth2 v0.36.0 h1:peZ/1z27fi9hUOFCAZaHyrpWG5lwe0RJEEEeH0ThlIs=
 golang.org/x/oauth2 v0.36.0/go.mod h1:YDBUJMTkDnJS+A4BP4eZBjCqtokkg1hODuPjwiGPO7Q=
@@ -835,8 +835,8 @@ golang.org/x/sync v0.3.0/go.mod h1:FU7BRWz2tNW+3quACPkgCx/L+uEAv1htQ0V83Z9Rj+Y=
 golang.org/x/sync v0.6.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
 golang.org/x/sync v0.7.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
 golang.org/x/sync v0.10.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
-golang.org/x/sync v0.22.0 h1:SZjpbeLmrCk4xhRSZFNZW5gFUeCeFgjekvI/+gfScek=
-golang.org/x/sync v0.22.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
+golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20180909124046-d0be0721c37e/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
@@ -872,8 +872,8 @@ golang.org/x/sys v0.17.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/sys v0.19.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/sys v0.20.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/sys v0.28.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/sys v0.47.0 h1:o7XGOvZQCADBQQ4Y7VNq2dRWQR7JmOUW8Kxx4ZsNgWs=
-golang.org/x/sys v0.47.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
+golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/telemetry v0.0.0-20240228155512-f48c80bd79b2/go.mod h1:TeRTkGYfJXctD9OcfyVLyj2J3IxLnKwHJR8f4D8a3YE=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
@@ -886,8 +886,8 @@ golang.org/x/term v0.16.0/go.mod h1:yn7UURbUtPyrVJPGPq404EukNFxcm/foM+bV/bfcDsY=
 golang.org/x/term v0.17.0/go.mod h1:lLRBjIVuehSbZlaOtGMbcMncT+aqLLLmKrsjNrUguwk=
 golang.org/x/term v0.20.0/go.mod h1:8UkIAJTvZgivsXaD6/pH6U9ecQzZ45awqEOzuCvwpFY=
 golang.org/x/term v0.27.0/go.mod h1:iMsnZpn0cago0GOrHO2+Y7u7JPn5AylBrcoWkElMTSM=
-golang.org/x/term v0.45.0 h1:NwWyBmoJCbfTHpxrWoZ9C6/VxOf7ic219I8xZZFdrf0=
-golang.org/x/term v0.45.0/go.mod h1:9aqxs0blBcrm/n0L9QW0aRVD+ktan8ssZromtqJC43w=
+golang.org/x/term v0.42.0 h1:UiKe+zDFmJobeJ5ggPwOshJIVt6/Ft0rcfrXZDLWAWY=
+golang.org/x/term v0.42.0/go.mod h1:Dq/D+snpsbazcBG5+F9Q1n2rXV8Ma+71xEjTRufARgY=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
@@ -899,8 +899,8 @@ golang.org/x/text v0.13.0/go.mod h1:TvPlkZtksWOMsz7fbANvkp4WM8x/WCo/om8BMLbz+aE=
 golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 golang.org/x/text v0.15.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 golang.org/x/text v0.21.0/go.mod h1:4IBbMaMmOPCJ8SecivzSH54+73PCFmPWxNTLm+vZkEQ=
-golang.org/x/text v0.40.0 h1:Ub2Z6/xjgF1WrYQz2nuITOEegKFtiIy+rieRJ5lHZKs=
-golang.org/x/text v0.40.0/go.mod h1:hpnzDAfGV753zIKo+wk3u1bVKCGPbrnF7+7LBF/UHVY=
+golang.org/x/text v0.36.0 h1:JfKh3XmcRPqZPKevfXVpI1wXPTqbkE5f7JA92a55Yxg=
+golang.org/x/text v0.36.0/go.mod h1:NIdBknypM8iqVmPiuco0Dh6P5Jcdk8lJL0CUebqK164=
 golang.org/x/time v0.15.0 h1:bbrp8t3bGUeFOx08pvsMYRTCVSMk89u4tKbNOZbp88U=
 golang.org/x/time v0.15.0/go.mod h1:Y4YMaQmXwGQZoFaVFk4YpCt4FLQMYKZe9oeV/f4MSno=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
@@ -914,8 +914,8 @@ golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc
 golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU=
 golang.org/x/tools v0.13.0/go.mod h1:HvlwmtVNQAhOuCjW7xxvovg8wbNq7LwfXh/k7wXUl58=
 golang.org/x/tools v0.21.1-0.20240508182429-e35e4ccd0d2d/go.mod h1:aiJjzUbINMkxbQROHiO6hDPo2LHcIPhhQsa9DLh0yGk=
-golang.org/x/tools v0.47.0 h1:7Kn5x/d1svx/PzryTsqeoZN4TZwqeH5pGWjefhLi/1Q=
-golang.org/x/tools v0.47.0/go.mod h1:dFHnyTvFWY212G+h7ZY4Vsp/K3U4/7W9TyVaAul8uCA=
+golang.org/x/tools v0.43.0 h1:12BdW9CeB3Z+J/I/wj34VMl8X+fEXBxVR90JeMX5E7s=
+golang.org/x/tools v0.43.0/go.mod h1:uHkMso649BX2cZK6+RpuIPXS3ho2hZo4FVwfoy1vIk0=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=

infrastructure_files/getting-started.sh

@@ -351,6 +351,11 @@ initialize_default_values() {
   NETBIRD_STUN_PORT=3478
 
   # Docker images
+  # Record whether the operator explicitly pinned the server/proxy images via
+  # env vars, so the agent-network preset can pick its own defaults without
+  # clobbering an explicit override.
+  NETBIRD_SERVER_IMAGE_EXPLICIT=${NETBIRD_SERVER_IMAGE:+true}
+  NETBIRD_PROXY_IMAGE_EXPLICIT=${NETBIRD_PROXY_IMAGE:+true}
   DASHBOARD_IMAGE=${DASHBOARD_IMAGE:-"netbirdio/dashboard:latest"}
   # Combined server replaces separate signal, relay, and management containers
   NETBIRD_SERVER_IMAGE=${NETBIRD_SERVER_IMAGE:-"netbirdio/netbird-server:latest"}
@@ -410,6 +415,15 @@ apply_agent_network_preset() {
   ENABLE_PROXY="true"
   ENABLE_CROWDSEC="false"
 
+  # Agent-network ships dedicated server/proxy images. Honor an explicit
+  # env override; otherwise pin the agent-network builds.
+  if [[ "${NETBIRD_SERVER_IMAGE_EXPLICIT}" != "true" ]]; then
+    NETBIRD_SERVER_IMAGE="netbirdio/netbird-server:0.74.0-rc.2"
+  fi
+  if [[ "${NETBIRD_PROXY_IMAGE_EXPLICIT}" != "true" ]]; then
+    NETBIRD_PROXY_IMAGE="netbirdio/reverse-proxy:0.74.0-rc.2"
+  fi
+
   if [[ -n "${NETBIRD_LETSENCRYPT_EMAIL}" ]]; then
     TRAEFIK_ACME_EMAIL="${NETBIRD_LETSENCRYPT_EMAIL}"
   else

management/internals/modules/agentnetwork/catalog/catalog.go

@@ -627,21 +627,6 @@ var providers = []Provider{
 		},
 		Models: []Model{},
 	},
-	{
-		// vLLM is an OpenAI-compatible self-hosted server. It behaves like
-		// the generic custom entry; it gets its own catalog id purely so it
-		// surfaces as a named "vLLM" choice in the provider picker.
-		ID:                 "vllm",
-		Kind:               KindCustom,
-		Name:               "vLLM",
-		Description:        "Self-hosted vLLM (OpenAI-compatible)",
-		DefaultHost:        "",
-		AuthHeaderName:     "Authorization",
-		AuthHeaderTemplate: "Bearer ${API_KEY}",
-		DefaultContentType: "application/json",
-		BrandColor:         "#30A2FF",
-		Models:             []Model{},
-	},
 	{
 		ID:                 "custom",
 		Kind:               KindCustom,

management/internals/modules/peers/manager.go

@@ -224,6 +224,30 @@ func (m *managerImpl) CreateProxyPeer(ctx context.Context, accountID string, pee
 		return nil
 	}
 
+	// Dedupe stale embedded peer records for the same (account, cluster).
+	// The proxy generates a fresh WireGuard keypair on every startup
+	// (proxy/internal/roundtrip/netbird.go), so without this sweep the
+	// prior embedded peer would linger forever — holding its CGNAT IP
+	// allocation, polluting other peers' rosters, and (most visibly)
+	// leaving the synth DNS pointing at the dead address. The
+	// (account, cluster) tuple identifies "the embedded peer for this
+	// proxy instance at this cluster"; any record matching that tuple
+	// with a different pubkey is by definition stale and must go.
+	staleIDs, err := m.findStaleEmbeddedProxyPeers(ctx, accountID, cluster, peerKey)
+	if err != nil {
+		return fmt.Errorf("scan for stale embedded proxy peers: %w", err)
+	}
+	if len(staleIDs) > 0 {
+		// userID="" + checkConnected=false: the deletion is initiated
+		// by management itself on behalf of the freshly-registering
+		// proxy, not by an end user; the stale peer may still be
+		// marked Connected from its prior session, but its session is
+		// dead by definition (its key no longer exists).
+		if err := m.DeletePeers(ctx, accountID, staleIDs, "", false); err != nil {
+			return fmt.Errorf("delete stale embedded proxy peers %v: %w", staleIDs, err)
+		}
+	}
+
 	name := fmt.Sprintf("proxy-%s", xid.New().String())
 	newPeer := &peer.Peer{
 		Ephemeral: true,
@@ -249,3 +273,29 @@ func (m *managerImpl) CreateProxyPeer(ctx context.Context, accountID string, pee
 
 	return nil
 }
+
+// findStaleEmbeddedProxyPeers returns the peer IDs of embedded proxy peer
+// records in accountID that target the same cluster but carry a different
+// WireGuard pubkey than the freshly-registering one. Used by CreateProxyPeer
+// to garbage-collect stale records left behind when the proxy restarts with a
+// regenerated keypair.
+func (m *managerImpl) findStaleEmbeddedProxyPeers(ctx context.Context, accountID, cluster, newKey string) ([]string, error) {
+	account, err := m.store.GetAccount(ctx, accountID)
+	if err != nil {
+		return nil, err
+	}
+	var stale []string
+	for _, p := range account.Peers {
+		if p == nil || !p.ProxyMeta.Embedded {
+			continue
+		}
+		if p.ProxyMeta.Cluster != cluster {
+			continue
+		}
+		if p.Key == newKey {
+			continue
+		}
+		stale = append(stale, p.ID)
+	}
+	return stale, nil
+}

management/internals/server/server.go

@@ -15,7 +15,7 @@ import (
 	"go.opentelemetry.io/otel/metric"
 	"golang.org/x/crypto/acme/autocert"
 	"golang.org/x/net/http2"
-	"golang.org/x/net/http2/h2c" //nolint:staticcheck
+	"golang.org/x/net/http2/h2c"
 	"google.golang.org/grpc"
 
 	"github.com/netbirdio/netbird/encryption"
@@ -382,7 +382,6 @@ func (s *BaseServer) serveGRPCWithHTTP(ctx context.Context, listener net.Listene
 			// the following magic is needed to support HTTP2 without TLS
 			// and still share a single port between gRPC and HTTP APIs
 			h1s := &http.Server{
-				//nolint:staticcheck // h2c also handles the HTTP/1 Upgrade mechanism, which http.Server's UnencryptedHTTP2 does not
 				Handler: h2c.NewHandler(handler, &http2.Server{}),
 			}
 			err = h1s.Serve(listener)

management/internals/shared/grpc/conversion.go

@@ -47,13 +47,16 @@ func init() {
 	precomputedDeprecatedRemotePeersConstraint = constraint
 }
 
-// toNetbirdConfig converts the server configuration to the wire representation. It returns
-// nil when no server config is set (the fan-out network-map path) because clients treat any
-// non-nil config as authoritative: a config without a relay section is interpreted as relay
-// disabled and wipes the clients' relay URLs.
 func toNetbirdConfig(config *nbconfig.Config, turnCredentials *Token, relayToken *Token, extraSettings *types.ExtraSettings, settings *types.Settings) *proto.NetbirdConfig {
 	if config == nil {
-		return nil
+		if settings == nil {
+			return nil
+		}
+		return &proto.NetbirdConfig{
+			Metrics: &proto.MetricsConfig{
+				Enabled: settings.MetricsPushEnabled,
+			},
+		}
 	}
 
 	var stuns []*proto.HostConfig

management/internals/shared/grpc/conversion_test.go

@@ -8,13 +8,11 @@ import (
 	"time"
 
 	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
 
 	nbdns "github.com/netbirdio/netbird/dns"
 	"github.com/netbirdio/netbird/management/internals/controllers/network_map"
 	"github.com/netbirdio/netbird/management/internals/controllers/network_map/controller/cache"
 	nbconfig "github.com/netbirdio/netbird/management/internals/server/config"
-	"github.com/netbirdio/netbird/management/server/types"
 )
 
 func TestToProtocolDNSConfigWithCache(t *testing.T) {
@@ -265,39 +263,3 @@ func TestEncodeSessionExpiresAt(t *testing.T) {
 		assert.True(t, got.AsTime().Equal(deadline))
 	})
 }
-
-// TestToNetbirdConfig_RelayInvariant guards against the v0.74.0 relay-wipe regression.
-// Clients treat any non-nil NetbirdConfig as authoritative and interpret a missing relay
-// section as relay disabled, wiping their relay URLs. toNetbirdConfig must therefore
-// return nil when no server config is set (the fan-out network-map path) instead of a
-// partial config, and a result built from a relay-enabled config must carry the relay
-// section.
-func TestToNetbirdConfig_RelayInvariant(t *testing.T) {
-	settings := &types.Settings{MetricsPushEnabled: true}
-
-	t.Run("nil server config returns nil config", func(t *testing.T) {
-		nbCfg := toNetbirdConfig(nil, nil, nil, nil, settings)
-		assert.Nil(t, nbCfg, "fan-out updates must not carry a partial NetbirdConfig even when settings are present")
-	})
-
-	t.Run("relay-enabled config carries relay section", func(t *testing.T) {
-		cfg := &nbconfig.Config{
-			Stuns: []*nbconfig.Host{{Proto: nbconfig.UDP, URI: "stun:stun.example.com:3478"}},
-			TURNConfig: &nbconfig.TURNConfig{
-				Turns: []*nbconfig.Host{{Proto: nbconfig.UDP, URI: "turn:turn.example.com:3478", Username: "user", Password: "pass"}},
-			},
-			Relay:  &nbconfig.Relay{Addresses: []string{"rels://relay.example.com:443"}},
-			Signal: &nbconfig.Host{Proto: nbconfig.HTTP, URI: "signal.example.com:10000"},
-		}
-		relayToken := &Token{Payload: "token-payload", Signature: "token-signature"}
-
-		nbCfg := toNetbirdConfig(cfg, nil, relayToken, nil, settings)
-		require.NotNil(t, nbCfg)
-		require.NotNil(t, nbCfg.Relay, "non-nil NetbirdConfig must include the relay section")
-		assert.Equal(t, cfg.Relay.Addresses, nbCfg.Relay.Urls, "relay URLs should match the server config")
-		assert.Equal(t, relayToken.Payload, nbCfg.Relay.TokenPayload, "relay token payload should be set")
-		assert.Equal(t, relayToken.Signature, nbCfg.Relay.TokenSignature, "relay token signature should be set")
-		require.NotNil(t, nbCfg.Metrics)
-		assert.True(t, nbCfg.Metrics.Enabled, "metrics flag should carry the settings value")
-	})
-}

management/server/agentnetwork_proxypeer_restart_test.go

@@ -0,0 +1,199 @@
+package server
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/netbirdio/netbird/management/internals/modules/peers"
+	"github.com/netbirdio/netbird/management/internals/modules/agentnetwork"
+	agenttypes "github.com/netbirdio/netbird/management/internals/modules/agentnetwork/types"
+	nbpeer "github.com/netbirdio/netbird/management/server/peer"
+	"github.com/netbirdio/netbird/management/server/permissions"
+	"github.com/netbirdio/netbird/management/server/store"
+	"github.com/netbirdio/netbird/management/server/types"
+)
+
+// TestAgentNetwork_ProxyRestart_PropagatesNewPeerAndDropsStale is the no-mock
+// regression guard for the bug the user reported: restarting the proxy creates
+// a fresh embedded peer with a NEW WireGuard public key (the proxy generates
+// the keypair on every startup at proxy/internal/roundtrip/netbird.go:312).
+// The PRIOR embedded peer record is never deleted on management, so the
+// account accumulates a stale peer holding a stale CGNAT IP. Other peers
+// in the account either keep routing to the dead IP, or — if synth DNS
+// picks the wrong record — never see the new IP at all.
+//
+// What this test exercises (no mocks):
+//   - real SQLite test store
+//   - real DefaultAccountManager, network-map controller, peer-update channels
+//   - real peers.Manager.CreateProxyPeer path (the very method the proxy
+//     invokes over gRPC on every startup)
+//   - real agentnetwork.Manager + synth chain so the client receives a
+//     concrete DNS record that must point at the LATEST proxy peer.
+//
+// Pre-fix expected behavior (red): two embedded peers exist after the
+// "restart"; the synth DNS record points at the stale one; the client
+// receives an update reflecting the new peer but the old one lingers.
+// Post-fix expected behavior (green): exactly one embedded peer exists
+// after restart (with the new key) AND the client's network map carries
+// the synth DNS pointing at that new peer's CGNAT IP.
+func TestAgentNetwork_ProxyRestart_PropagatesNewPeerAndDropsStale(t *testing.T) {
+	am, updateManager, err := createManager(t)
+	require.NoError(t, err, "createManager must succeed")
+	ctx := context.Background()
+
+	const (
+		accountID   = "an-restart-acct"
+		adminUserID = "an-restart-admin"
+		groupAID    = "an-restart-grp-A"
+		clusterAddr = "eu.proxy.netbird.io"
+		clientKey   = "BhRPtynAAYRDy08+q4HTMsos8fs4plTP4NOSh7C1ry8="
+		// Two different proxy pubkeys — the "before" and "after" of a
+		// proxy-process restart with fresh-keypair generation.
+		proxyKey1 = "Aaaaa1aaaaYRDy08+q4HTMsos8fs4plTP4NOSh7C1ry8="
+		proxyKey2 = "Bbbbb2bbbbYRDy08+q4HTMsos8fs4plTP4NOSh7C1ry8="
+	)
+
+	// --- Account scaffold ---
+	account := newAccountWithId(ctx, accountID, adminUserID, "an-restart.test", "", "", false)
+	require.NoError(t, am.Store.SaveAccount(ctx, account))
+
+	clientPeer := &nbpeer.Peer{
+		Key:      clientKey,
+		Name:     "an-restart-client",
+		DNSLabel: "an-restart-client",
+		Meta:     nbpeer.PeerSystemMeta{Hostname: "an-restart-client", GoOS: "linux", WtVersion: "development"},
+	}
+	addedClient, _, _, _, err := am.AddPeer(ctx, "", "", adminUserID, clientPeer, false)
+	require.NoError(t, err, "AddPeer for client must succeed")
+	require.NoError(t, am.MarkPeerConnected(ctx, clientKey, accountID, time.Now().UnixNano(), &types.NetworkMap{}),
+		"MarkPeerConnected for the client peer must succeed (affected-peer fan-out skips disconnected peers)")
+
+	// Place the client in group A so the synth policy reaches it.
+	account, err = am.Store.GetAccount(ctx, accountID)
+	require.NoError(t, err)
+	account.Groups[groupAID] = &types.Group{ID: groupAID, Name: "groupA", Peers: []string{addedClient.ID}}
+	require.NoError(t, am.Store.SaveAccount(ctx, account), "SaveAccount must persist group A")
+
+	// --- Real peers + agent-network managers ---
+	permMgr := permissions.NewManager(am.Store)
+	peersMgr := peers.NewManager(am.Store, permMgr)
+	peersMgr.SetAccountManager(am)
+	peersMgr.SetNetworkMapController(am.networkMapController)
+	agentMgr := agentnetwork.NewManager(am.Store, permMgr, am, nil)
+
+	// Subscribe BEFORE any state-mutating call so we don't lose the update
+	// that contains the synth DNS record.
+	clientCh := updateManager.CreateChannel(ctx, addedClient.ID)
+	t.Cleanup(func() { updateManager.CloseChannel(ctx, addedClient.ID) })
+	drain(clientCh)
+
+	// --- First proxy startup: register peer key K1, then mark it
+	// connected. In production the proxy follows CreateProxyPeer with the
+	// regular sync stream which lands on MarkPeerConnected; the synth DNS
+	// path filters out peers that aren't Connected (types/account.go:323),
+	// so without this step no DNS record would be emitted.
+	require.NoError(t, peersMgr.CreateProxyPeer(ctx, accountID, proxyKey1, clusterAddr),
+		"first CreateProxyPeer (proxy startup) must succeed")
+
+	peer1ID, err := am.Store.GetPeerIDByKey(ctx, store.LockingStrengthNone, proxyKey1)
+	require.NoError(t, err, "proxy peer for K1 must be persisted after CreateProxyPeer")
+	require.NotEmpty(t, peer1ID)
+
+	require.NoError(t, am.MarkPeerConnected(ctx, proxyKey1, accountID, time.Now().UnixNano(), &types.NetworkMap{}),
+		"MarkPeerConnected for K1 must succeed")
+
+	account, err = am.Store.GetAccount(ctx, accountID)
+	require.NoError(t, err)
+	proxyIP1 := account.Peers[peer1ID].IP.String()
+	require.NotEmpty(t, proxyIP1, "K1 must have an assigned overlay IP")
+
+	// --- Provider + policy. CreateProvider / CreatePolicy trigger the
+	// agentnetwork reconcile which runs UpdateAccountPeers; the resulting
+	// NetworkMap delivered to the client carries the synth DNS record
+	// pointing at K1's IP. ---
+	provider, err := agentMgr.CreateProvider(ctx, adminUserID, &agenttypes.Provider{
+		AccountID:   accountID,
+		ProviderID:  "openai_api",
+		Name:        "openai-test",
+		UpstreamURL: "https://api.openai.com",
+		APIKey:      "sk-test-key",
+		Enabled:     true,
+		Models:      []agenttypes.ProviderModel{{ID: "gpt-5.4"}},
+	}, clusterAddr)
+	require.NoError(t, err, "CreateProvider must succeed")
+
+	_, err = agentMgr.CreatePolicy(ctx, adminUserID, &agenttypes.Policy{
+		AccountID:              accountID,
+		Name:                   "p1",
+		Enabled:                true,
+		SourceGroups:           []string{groupAID},
+		DestinationProviderIDs: []string{provider.ID},
+	})
+	require.NoError(t, err, "CreatePolicy must succeed")
+
+	settings, err := am.Store.GetAgentNetworkSettings(ctx, store.LockingStrengthNone, accountID)
+	require.NoError(t, err)
+	fqdn := settings.Endpoint()
+
+	rdata1 := awaitZoneRData(clientCh, clusterAddr, fqdn, true)
+	require.Equal(t, proxyIP1, rdata1,
+		"client must receive a synth DNS record pointing at K1's overlay IP after the synth path runs")
+	drain(clientCh)
+
+	// --- Proxy restart: NEW keypair K2, same account, same cluster ---
+	require.NoError(t, peersMgr.CreateProxyPeer(ctx, accountID, proxyKey2, clusterAddr),
+		"second CreateProxyPeer (proxy restart with fresh keypair) must succeed")
+
+	peer2ID, err := am.Store.GetPeerIDByKey(ctx, store.LockingStrengthNone, proxyKey2)
+	require.NoError(t, err, "proxy peer for K2 must be persisted after restart")
+	require.NotEmpty(t, peer2ID)
+
+	require.NoError(t, am.MarkPeerConnected(ctx, proxyKey2, accountID, time.Now().UnixNano(), &types.NetworkMap{}),
+		"MarkPeerConnected for K2 must succeed")
+
+	// In production the agent's sync stream pulls a fresh NetworkMap as
+	// part of its normal reconcile cadence; in this isolated test
+	// MarkPeerConnected's affected-peer fan-out can race the channel-side
+	// buffer in a way that swallows the synth-DNS-bearing update before
+	// our await reads it. Trigger an explicit account-wide fan-out so the
+	// assertion below tests what production actually delivers, not the
+	// in-test buffer race.
+	am.UpdateAccountPeers(ctx, accountID, types.UpdateReason{Resource: types.UpdateResourcePeer, Operation: types.UpdateOperationUpdate})
+
+	account, err = am.Store.GetAccount(ctx, accountID)
+	require.NoError(t, err)
+	proxyIP2 := account.Peers[peer2ID].IP.String()
+	require.NotEmpty(t, proxyIP2, "K2 must have an assigned overlay IP")
+	require.NotEqual(t, proxyIP1, proxyIP2, "K2 must get a different overlay IP than K1 (sanity)")
+
+	// CRITICAL ASSERTION 1: K1 must no longer be in the store. The SqlStore
+	// returns ("", nil) for a missing key rather than NotFound, so assert
+	// on the returned ID being empty.
+	staleID, err := am.Store.GetPeerIDByKey(ctx, store.LockingStrengthNone, proxyKey1)
+	require.NoError(t, err, "GetPeerIDByKey for a missing peer must not error")
+	assert.Empty(t, staleID,
+		"stale embedded proxy peer K1 must be removed when a new embedded peer registers for the same (account, cluster); pre-fix this assertion fails because management never cleans up the prior peer record")
+
+	// CRITICAL ASSERTION 2: exactly one embedded proxy peer remains, and it
+	// is K2.
+	account, err = am.Store.GetAccount(ctx, accountID)
+	require.NoError(t, err)
+	embeddedKeys := []string{}
+	for _, p := range account.Peers {
+		if p.ProxyMeta.Embedded {
+			embeddedKeys = append(embeddedKeys, p.Key)
+		}
+	}
+	assert.Equal(t, []string{proxyKey2}, embeddedKeys,
+		"after a proxy restart exactly one embedded proxy peer should remain — the one with the new key K2")
+
+	// CRITICAL ASSERTION 3: the synth DNS record the client receives now
+	// points at K2's IP, not K1's.
+	rdata2 := awaitZoneRData(clientCh, clusterAddr, fqdn, true)
+	assert.Equal(t, proxyIP2, rdata2,
+		"after proxy restart, the client's synth DNS record must point at the NEW embedded peer's IP, not the stale K1 IP")
+}

management/server/peer_test.go

@@ -1048,7 +1048,11 @@ func testUpdateAccountPeers(t *testing.T) {
 
 			for _, channel := range peerChannels {
 				update := <-channel
-				assert.Nil(t, update.Update.NetbirdConfig, "fan-out updates must not carry a NetbirdConfig; clients treat a config without relay as relay disabled and wipe their relay URLs")
+				assert.NotNil(t, update.Update.NetbirdConfig)
+				assert.Nil(t, update.Update.NetbirdConfig.Stuns)
+				assert.Nil(t, update.Update.NetbirdConfig.Turns)
+				assert.Nil(t, update.Update.NetbirdConfig.Signal)
+				assert.Nil(t, update.Update.NetbirdConfig.Relay)
 				assert.Equal(t, tc.peers, len(update.Update.NetworkMap.RemotePeers))
 				assert.Equal(t, tc.peers*2, len(update.Update.NetworkMap.FirewallRules))
 			}

management/server/types/account.go

@@ -305,8 +305,7 @@ func (a *Account) SynthesizePrivateServiceZones(peerID string) []nbdns.CustomZon
 			zone = &nbdns.CustomZone{
 				Domain:           dns.Fqdn(serviceDomainZone),
 				Records:          []nbdns.SimpleRecord{},
-				NonAuthoritative:     true,
-				SearchDomainDisabled: true,
+				NonAuthoritative: true,
 			}
 			zonesByApex[serviceDomainZone] = zone
 		}

proxy/internal/middleware/builtin/llm_guardrail/middleware.go

@@ -25,14 +25,6 @@ const (
 	denyCodeModel    = "llm_policy.model_blocked"
 	denyReasonModel  = "model_blocked"
 	denyMessageModel = "model is not in the policy allowlist"
-	// Deny reason used when an allowlist is configured but the request model
-	// could not be determined. URL/path-routed providers (AWS Bedrock, Google
-	// Vertex, ...) carry the model outside the JSON body, so a request shape the
-	// parser does not recognise reaches the guardrail with no model. Such a
-	// request must be denied (fail closed), never waved through.
-	denyCodeModelUnknown    = "llm_policy.model_unknown"
-	denyReasonModelUnknown  = "model_unknown"
-	denyMessageModelUnknown = "request model could not be determined for the policy allowlist"
 )
 
 // Middleware enforces the model allowlist and optionally captures the
@@ -116,37 +108,23 @@ func (m *Middleware) evaluateAllowlist(model string, modelPresent bool) *middlew
 	if len(m.cfg.ModelAllowlist) == 0 {
 		return nil
 	}
-	// Fail closed: with an allowlist configured, a request whose model the
-	// upstream parser could not extract (absent or empty) must be denied rather
-	// than allowed. This is what enforces the allowlist for URL/path-routed
-	// providers (Bedrock, Vertex, ...) whose model lives outside the JSON body.
-	if !modelPresent || normaliseModel(model) == "" {
-		return denyModel("", denyCodeModelUnknown, denyMessageModelUnknown, denyReasonModelUnknown)
+	if !modelPresent {
+		return nil
 	}
 	if m.modelInAllowlist(model) {
 		return nil
 	}
-	return denyModel(model, denyCodeModel, denyMessageModel, denyReasonModel)
-}
-
-// denyModel builds a 403 deny Output for a model-allowlist rejection. model is
-// included in the details only when non-empty.
-func denyModel(model, code, message, reason string) *middleware.Output {
-	details := map[string]string{}
-	if model != "" {
-		details["model"] = model
-	}
 	return &middleware.Output{
 		Decision:   middleware.DecisionDeny,
 		DenyStatus: 403,
 		DenyReason: &middleware.DenyReason{
-			Code:    code,
-			Message: message,
-			Details: details,
+			Code:    denyCodeModel,
+			Message: denyMessageModel,
+			Details: map[string]string{"model": model},
 		},
 		Metadata: []middleware.KV{
 			{Key: middleware.KeyLLMPolicyDecision, Value: "deny"},
-			{Key: middleware.KeyLLMPolicyReason, Value: reason},
+			{Key: middleware.KeyLLMPolicyReason, Value: denyReasonModel},
 		},
 	}
 }

proxy/internal/middleware/builtin/llm_guardrail/middleware_test.go

@@ -102,44 +102,13 @@ func TestAllowlistCaseInsensitive(t *testing.T) {
 	}
 }
 
-func TestAllowlistMissingModelKeyDenies(t *testing.T) {
-	// Fail closed: with an allowlist configured, a request whose model the
-	// parser could not extract (URL/path-routed providers such as Bedrock or
-	// Vertex whose shape wasn't recognised) must be denied, not allowed.
+func TestAllowlistMissingModelKeyAllows(t *testing.T) {
 	mw := New(Config{ModelAllowlist: []string{"gpt-4o"}})
 	out, err := mw.Invoke(context.Background(), newInput())
 	require.NoError(t, err)
-	require.NotNil(t, out)
-	assert.Equal(t, middleware.DecisionDeny, out.Decision, "absent model must be denied when an allowlist is set")
-	assert.Equal(t, 403, out.DenyStatus, "deny status must be 403")
-	require.NotNil(t, out.DenyReason, "deny reason must be populated")
-	assert.Equal(t, "llm_policy.model_unknown", out.DenyReason.Code, "deny code must be model_unknown")
+	assert.Equal(t, middleware.DecisionAllow, out.Decision, "missing model key must allow even with non-empty allowlist")
 	dec, _ := metaValue(t, out.Metadata, middleware.KeyLLMPolicyDecision)
-	assert.Equal(t, "deny", dec, "decision must be deny when model key is absent")
-	reason, _ := metaValue(t, out.Metadata, middleware.KeyLLMPolicyReason)
-	assert.Equal(t, "model_unknown", reason, "reason metadata must be model_unknown")
-}
-
-func TestAllowlistEmptyModelValueDenies(t *testing.T) {
-	// A present-but-empty model is as undeterminable as an absent one.
-	mw := New(Config{ModelAllowlist: []string{"gpt-4o"}})
-	out, err := mw.Invoke(context.Background(), newInput(
-		middleware.KV{Key: middleware.KeyLLMModel, Value: "   "},
-	))
-	require.NoError(t, err)
-	require.NotNil(t, out)
-	assert.Equal(t, middleware.DecisionDeny, out.Decision, "empty model must be denied when an allowlist is set")
-	require.NotNil(t, out.DenyReason, "deny reason must be populated")
-	assert.Equal(t, "llm_policy.model_unknown", out.DenyReason.Code, "deny code must be model_unknown")
-}
-
-func TestAllowlistEmptyListAllowsMissingModel(t *testing.T) {
-	// Without an allowlist there is nothing to enforce, so a missing model is
-	// still allowed — the fail-closed rule only applies when a list is set.
-	mw := New(Config{})
-	out, err := mw.Invoke(context.Background(), newInput())
-	require.NoError(t, err)
-	assert.Equal(t, middleware.DecisionAllow, out.Decision, "no allowlist must allow even without a model")
+	assert.Equal(t, "allow", dec, "decision must be allow when model key is absent")
 }
 
 func TestPromptCaptureDisabledEmitsNoPrompt(t *testing.T) {

proxy/internal/middleware/builtin/llm_request_parser/guardrail_allowlist_test.go

@@ -1,106 +0,0 @@
-package llm_request_parser
-
-import (
-	"context"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/netbirdio/netbird/proxy/internal/middleware"
-	"github.com/netbirdio/netbird/proxy/internal/middleware/builtin/llm_guardrail"
-)
-
-// runParserGuardrail runs the request parser then the model-allowlist guardrail
-// in SlotOnRequest order, threading the parser's metadata into the guardrail the
-// same way the real chain does. It returns the guardrail decision so tests can
-// assert allowlist enforcement for URL/path-routed providers end to end.
-func runParserGuardrail(t *testing.T, url string, body []byte, allowlist []string) *middleware.Output {
-	t.Helper()
-	parser := newMiddleware(t)
-	parsed, err := parser.Invoke(context.Background(), &middleware.Input{
-		Slot: middleware.SlotOnRequest,
-		URL:  url,
-		Body: body,
-	})
-	require.NoError(t, err, "parser must not error")
-
-	guard := llm_guardrail.New(llm_guardrail.Config{ModelAllowlist: allowlist})
-	out, err := guard.Invoke(context.Background(), &middleware.Input{
-		Slot:     middleware.SlotOnRequest,
-		Metadata: parsed.Metadata,
-	})
-	require.NoError(t, err, "guardrail must not error")
-	require.NotNil(t, out, "guardrail must return an output")
-	return out
-}
-
-// TestModelAllowlist_URLRoutedProviders validates that the model allowlist is
-// enforced for providers whose model travels in the URL path (AWS Bedrock,
-// Google Vertex) rather than the JSON body. The "unknown action" case is the
-// regression guard for #6751: a Bedrock request shape the parser cannot map to a
-// model must fail closed under an allowlist instead of bypassing it.
-func TestModelAllowlist_URLRoutedProviders(t *testing.T) {
-	const bedrockBody = `{"anthropic_version":"bedrock-2023-05-31","messages":[{"role":"user","content":"hi"}]}`
-	const vertexBody = `{"anthropic_version":"vertex-2023-10-16","messages":[{"role":"user","content":"hi"}]}`
-
-	tests := []struct {
-		name      string
-		url       string
-		body      string
-		allowlist []string
-		decision  middleware.Decision
-		denyCode  string
-	}{
-		{
-			name:      "bedrock allowed model passes",
-			url:       "https://bedrock-runtime.us-east-1.amazonaws.com/model/us.anthropic.claude-haiku-4-5-v1:0/invoke",
-			body:      bedrockBody,
-			allowlist: []string{"anthropic.claude-haiku-4-5"},
-			decision:  middleware.DecisionAllow,
-		},
-		{
-			name:      "bedrock disallowed model denied",
-			url:       "https://bedrock-runtime.us-east-1.amazonaws.com/model/us.anthropic.claude-opus-4-8-v1:0/invoke",
-			body:      bedrockBody,
-			allowlist: []string{"anthropic.claude-haiku-4-5"},
-			decision:  middleware.DecisionDeny,
-			denyCode:  "llm_policy.model_blocked",
-		},
-		{
-			name:      "bedrock unknown action fails closed",
-			url:       "https://bedrock-runtime.us-east-1.amazonaws.com/model/us.anthropic.claude-opus-4-8-v1:0/some-future-action",
-			body:      bedrockBody,
-			allowlist: []string{"anthropic.claude-haiku-4-5"},
-			decision:  middleware.DecisionDeny,
-			denyCode:  "llm_policy.model_unknown",
-		},
-		{
-			name:      "vertex disallowed model denied",
-			url:       "/v1/projects/p/locations/global/publishers/anthropic/models/claude-opus-4-8@20250101:rawPredict",
-			body:      vertexBody,
-			allowlist: []string{"claude-haiku-4-5"},
-			decision:  middleware.DecisionDeny,
-			denyCode:  "llm_policy.model_blocked",
-		},
-		{
-			name:      "vertex allowed model passes",
-			url:       "/v1/projects/p/locations/global/publishers/anthropic/models/claude-haiku-4-5@20250101:rawPredict",
-			body:      vertexBody,
-			allowlist: []string{"claude-haiku-4-5"},
-			decision:  middleware.DecisionAllow,
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			out := runParserGuardrail(t, tt.url, []byte(tt.body), tt.allowlist)
-			assert.Equal(t, tt.decision, out.Decision, "unexpected decision for %s", tt.name)
-			if tt.decision == middleware.DecisionDeny {
-				require.NotNil(t, out.DenyReason, "deny reason must be set for %s", tt.name)
-				assert.Equal(t, 403, out.DenyStatus, "deny status must be 403 for %s", tt.name)
-				assert.Equal(t, tt.denyCode, out.DenyReason.Code, "deny code for %s", tt.name)
-			}
-		})
-	}
-}

shared/management/client/grpc.go

@@ -33,15 +33,10 @@ const ConnectTimeout = 10 * time.Second
 const healthCheckTimeout = 5 * time.Second
 
 const (
-	// EnvMaxRecvMsgSize overrides the default gRPC max receive message size
+	// EnvMaxRecvMsgSize overrides the default gRPC max receive message size (4 MB)
 	// for the management client connection. Value is in bytes.
 	EnvMaxRecvMsgSize = "NB_MANAGEMENT_GRPC_MAX_MSG_SIZE"
 
-	// defaultMaxRecvMsgSize is the max gRPC receive message size used for the
-	// management client connection when EnvMaxRecvMsgSize is unset or invalid.
-	// It overrides the gRPC library default of 4 MB.
-	defaultMaxRecvMsgSize = 1024 * 1024 * 16
-
 	errMsgMgmtPublicKey    = "failed getting Management Service public key: %s"
 	errMsgNoMgmtConnection = "no connection to management"
 )
@@ -89,22 +84,22 @@ type ExposeResponse struct {
 }
 
 // MaxRecvMsgSize returns the configured max gRPC receive message size from
-// the environment, or defaultMaxRecvMsgSize (16 MB) if unset or invalid.
+// the environment, or 0 if unset (which uses the gRPC default of 4 MB).
 func MaxRecvMsgSize() int {
 	val := os.Getenv(EnvMaxRecvMsgSize)
 	if val == "" {
-		return defaultMaxRecvMsgSize
+		return 0
 	}
 
 	size, err := strconv.Atoi(val)
 	if err != nil {
 		log.Warnf("invalid %s value %q, using default: %v", EnvMaxRecvMsgSize, val, err)
-		return defaultMaxRecvMsgSize
+		return 0
 	}
 
 	if size <= 0 {
 		log.Warnf("invalid %s value %d, must be positive, using default", EnvMaxRecvMsgSize, size)
-		return defaultMaxRecvMsgSize
+		return 0
 	}
 
 	return size

shared/management/client/grpc_test.go

@@ -21,11 +21,11 @@ func TestMaxRecvMsgSize(t *testing.T) {
 		envValue string
 		expected int
 	}{
-		{name: "unset returns default", envValue: "", expected: defaultMaxRecvMsgSize},
+		{name: "unset returns 0", envValue: "", expected: 0},
 		{name: "valid value", envValue: "10485760", expected: 10485760},
-		{name: "non-numeric returns default", envValue: "abc", expected: defaultMaxRecvMsgSize},
-		{name: "negative returns default", envValue: "-1", expected: defaultMaxRecvMsgSize},
-		{name: "zero returns default", envValue: "0", expected: defaultMaxRecvMsgSize},
+		{name: "non-numeric returns 0", envValue: "abc", expected: 0},
+		{name: "negative returns 0", envValue: "-1", expected: 0},
+		{name: "zero returns 0", envValue: "0", expected: 0},
 	}
 
 	for _, tt := range tests {

shared/relay/client/manager.go

@@ -30,16 +30,11 @@ type RelayTrack struct {
 	relayClient *Client
 	err         error
 	created     time.Time
-	// ready is closed once the dial started by openConnVia finishes (relayClient
-	// or err is set). Callers reusing a track wait on this instead of the track
-	// lock, so the dial never runs under rt.Lock.
-	ready chan struct{}
 }
 
 func NewRelayTrack() *RelayTrack {
 	return &RelayTrack{
 		created: time.Now(),
-		ready:   make(chan struct{}),
 	}
 }
 
@@ -331,24 +326,34 @@ func (m *Manager) openConnVia(ctx context.Context, serverAddress, peerKey string
 	// check if already has a connection to the desired relay server
 	m.relayClientsMutex.RLock()
 	rt, ok := m.relayClients[serverAddress]
-	m.relayClientsMutex.RUnlock()
 	if ok {
-		return m.openConnOnTrack(ctx, rt, peerKey)
+		rt.RLock()
+		m.relayClientsMutex.RUnlock()
+		defer rt.RUnlock()
+		if rt.err != nil {
+			return nil, rt.err
+		}
+		return rt.relayClient.OpenConn(ctx, peerKey)
 	}
+	m.relayClientsMutex.RUnlock()
 
 	// if not, establish a new connection but check it again (because changed the lock type) before starting the
 	// connection
 	m.relayClientsMutex.Lock()
 	rt, ok = m.relayClients[serverAddress]
 	if ok {
+		rt.RLock()
 		m.relayClientsMutex.Unlock()
-		return m.openConnOnTrack(ctx, rt, peerKey)
+		defer rt.RUnlock()
+		if rt.err != nil {
+			return nil, rt.err
+		}
+		return rt.relayClient.OpenConn(ctx, peerKey)
 	}
 
-	// Publish the track and release the map lock BEFORE dialing, so the dial does
-	// not run under rt.Lock (which would block RelayStates and the cleanup loop
-	// for the full dial). Concurrent callers find this track and wait on rt.ready.
+	// create a new relay client and store it in the relayClients map
 	rt = NewRelayTrack()
+	rt.Lock()
 	m.relayClients[serverAddress] = rt
 	m.relayClientsMutex.Unlock()
 
@@ -356,10 +361,8 @@ func (m *Manager) openConnVia(ctx context.Context, serverAddress, peerKey string
 	relayClient.SetTransportFallback(m.transportFallback)
 	err := relayClient.Connect(m.ctx)
 	if err != nil {
-		rt.Lock()
 		rt.err = err
 		rt.Unlock()
-		close(rt.ready)
 		m.relayClientsMutex.Lock()
 		delete(m.relayClients, serverAddress)
 		m.relayClientsMutex.Unlock()
@@ -367,34 +370,14 @@ func (m *Manager) openConnVia(ctx context.Context, serverAddress, peerKey string
 	}
 	// if connection closed then delete the relay client from the list
 	relayClient.SetOnDisconnectListener(m.onServerDisconnected)
-	rt.Lock()
 	rt.relayClient = relayClient
 	rt.Unlock()
-	close(rt.ready)
 
-	return relayClient.OpenConn(ctx, peerKey)
-}
-
-// openConnOnTrack opens a peer connection through an existing relay track,
-// waiting for the dial started by another openConnVia call to finish. It waits
-// on rt.ready rather than the track lock, so it neither holds nor contends the
-// track lock across the dial.
-func (m *Manager) openConnOnTrack(ctx context.Context, rt *RelayTrack, peerKey string) (net.Conn, error) {
-	select {
-	case <-rt.ready:
-	case <-ctx.Done():
-		return nil, ctx.Err()
+	conn, err := relayClient.OpenConn(ctx, peerKey)
+	if err != nil {
+		return nil, err
 	}
-
-	rt.RLock()
-	defer rt.RUnlock()
-	if rt.err != nil {
-		return nil, rt.err
-	}
-	if rt.relayClient == nil {
-		return nil, ErrRelayClientNotConnected
-	}
-	return rt.relayClient.OpenConn(ctx, peerKey)
+	return conn, nil
 }
 
 func (m *Manager) onServerConnected() {
@@ -493,13 +476,6 @@ func (m *Manager) cleanUpUnusedRelays() {
 			continue
 		}
 
-		// dial still in progress (openConnVia publishes the track before Connect
-		// completes and no longer holds rt.Lock during it), nothing to clean up.
-		if rt.relayClient == nil {
-			rt.Unlock()
-			continue
-		}
-
 		if time.Since(rt.created) <= m.keepUnusedServerTime {
 			rt.Unlock()
 			continue

shared/relay/client/manager_cleanup_test.go

@@ -1,60 +0,0 @@
-package client
-
-import (
-	"context"
-	"net/netip"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/require"
-)
-
-// TestCleanUpUnusedRelays_DoesNotBlockOnRealHangingDial drives a real, hanging foreign
-// relay dial and asserts cleanUpUnusedRelays does not stall behind it.
-func TestCleanUpUnusedRelays_DoesNotBlockOnRealHangingDial(t *testing.T) {
-	serverAddr := stallingRelayListener(t)
-
-	mCtx, mCancel := context.WithCancel(context.Background())
-	t.Cleanup(mCancel)
-
-	m := NewManager(mCtx, nil, "alice", 1280)
-
-	dialDone := make(chan struct{})
-	go func() {
-		defer close(dialDone)
-		_, _ = m.openConnVia(mCtx, serverAddr, "peerKey", netip.Addr{})
-	}()
-
-	// The track appears in the map once the dial is in flight.
-	require.Eventually(t, func() bool {
-		m.relayClientsMutex.RLock()
-		defer m.relayClientsMutex.RUnlock()
-		_, ok := m.relayClients[serverAddr]
-		return ok
-	}, 5*time.Second, 5*time.Millisecond, "relay dial did not start")
-
-	cleanupDone := make(chan struct{})
-	go func() {
-		defer close(cleanupDone)
-		m.cleanUpUnusedRelays()
-	}()
-
-	select {
-	case <-cleanupDone:
-	case <-time.After(2 * time.Second):
-		t.Fatal("cleanUpUnusedRelays blocked on an in-progress relay dial while holding the relay map lock")
-	}
-
-	m.relayClientsMutex.RLock()
-	_, stillTracked := m.relayClients[serverAddr]
-	m.relayClientsMutex.RUnlock()
-	require.True(t, stillTracked, "an in-progress relay dial must not be evicted by cleanup")
-
-	// Release the hanging dial so the goroutine can exit cleanly.
-	mCancel()
-	select {
-	case <-dialDone:
-	case <-time.After(5 * time.Second):
-		t.Fatal("openConnVia did not return after context cancellation")
-	}
-}

shared/relay/client/manager_relaystates_test.go

@@ -1,91 +0,0 @@
-package client
-
-import (
-	"context"
-	"net"
-	"net/netip"
-	"sync"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/require"
-)
-
-// stallingRelayListener accepts TCP connections and holds them open without ever
-// responding, so a relay handshake dialed against it blocks until its context is
-// cancelled. It returns the "rel://host:port" URL to dial.
-func stallingRelayListener(t *testing.T) string {
-	t.Helper()
-
-	ln, err := net.Listen("tcp", "127.0.0.1:0")
-	require.NoError(t, err)
-
-	var mu sync.Mutex
-	var conns []net.Conn
-	go func() {
-		for {
-			c, err := ln.Accept()
-			if err != nil {
-				return
-			}
-			mu.Lock()
-			conns = append(conns, c)
-			mu.Unlock()
-		}
-	}()
-	t.Cleanup(func() {
-		_ = ln.Close()
-		mu.Lock()
-		for _, c := range conns {
-			_ = c.Close()
-		}
-		mu.Unlock()
-	})
-
-	return "rel://" + ln.Addr().String()
-}
-
-// TestRelayStates_DoesNotBlockOnRealHangingDial is a regression test for
-// RelayStates() called by a "status -d command" hanging behind an in-progress
-// relay dial.
-func TestRelayStates_DoesNotBlockOnRealHangingDial(t *testing.T) {
-	serverAddr := stallingRelayListener(t)
-
-	mCtx, mCancel := context.WithCancel(context.Background())
-	t.Cleanup(mCancel)
-
-	m := NewManager(mCtx, nil, "alice", 1280)
-
-	dialDone := make(chan struct{})
-	go func() {
-		defer close(dialDone)
-		_, _ = m.openConnVia(mCtx, serverAddr, "peerKey", netip.Addr{})
-	}()
-
-	require.Eventually(t, func() bool {
-		m.relayClientsMutex.RLock()
-		defer m.relayClientsMutex.RUnlock()
-		_, ok := m.relayClients[serverAddr]
-		return ok
-	}, 5*time.Second, 5*time.Millisecond, "relay dial did not start")
-
-	done := make(chan []RelayConnState, 1)
-	go func() {
-		done <- m.RelayStates()
-	}()
-
-	select {
-	case states := <-done:
-		require.Empty(t, states, "a relay still being dialed carries no state and must be omitted")
-	case <-time.After(2 * time.Second):
-		t.Fatal("RelayStates blocked on a foreign relay whose Connect() is in progress")
-	}
-
-	// Release the hanging dial so the goroutine can exit cleanly.
-	mCancel()
-	select {
-	case <-dialDone:
-	case <-time.After(5 * time.Second):
-		t.Fatal("openConnVia did not return after context cancellation")
-	}
-}

signal/cmd/run.go

@@ -16,7 +16,7 @@ import (
 	"go.opentelemetry.io/otel/metric"
 	"golang.org/x/crypto/acme/autocert"
 	"golang.org/x/net/http2"
-	"golang.org/x/net/http2/h2c" //nolint:staticcheck
+	"golang.org/x/net/http2/h2c"
 
 	"github.com/netbirdio/netbird/shared/metrics"
 
@@ -281,7 +281,6 @@ func serveHTTP(httpListener net.Listener, handler http.Handler) {
 	go func() {
 		// Use h2c to support HTTP/2 without TLS (needed for gRPC)
 		h1s := &http.Server{
-			//nolint:staticcheck // h2c also handles the HTTP/1 Upgrade mechanism, which http.Server's UnencryptedHTTP2 does not
 			Handler: h2c.NewHandler(handler, &http2.Server{}),
 		}
 		err := h1s.Serve(httpListener)