Merge remote-tracking branch 'origin/main' into refactor/permissions-manager

client/Dockerfile

@@ -4,7 +4,7 @@
 #   sudo podman build -t localhost/netbird:latest -f client/Dockerfile --ignorefile .dockerignore-client .
 #   sudo podman run --rm -it --cap-add={BPF,NET_ADMIN,NET_RAW} localhost/netbird:latest
 
-FROM alpine:3.23.3
+FROM alpine:3.24
 # iproute2: busybox doesn't display ip rules properly
 RUN apk add --no-cache \
     bash \
@@ -17,10 +17,11 @@ ENV \
     NETBIRD_BIN="/usr/local/bin/netbird" \
     NB_LOG_FILE="console,/var/log/netbird/client.log" \
     NB_DAEMON_ADDR="unix:///var/run/netbird.sock" \
+    NB_ENABLE_CAPTURE="false" \
     NB_ENTRYPOINT_SERVICE_TIMEOUT="30"
 
 ENTRYPOINT [ "/usr/local/bin/netbird-entrypoint.sh" ]
-
-ARG NETBIRD_BINARY=netbird
+ARG TARGETPLATFORM
+ARG NETBIRD_BINARY=$TARGETPLATFORM/netbird
 COPY client/netbird-entrypoint.sh /usr/local/bin/netbird-entrypoint.sh
 COPY "${NETBIRD_BINARY}"  /usr/local/bin/netbird

client/Dockerfile-rootless

@@ -4,7 +4,7 @@
 #   podman build -t localhost/netbird:latest -f client/Dockerfile --ignorefile .dockerignore-client .
 #   podman run --rm -it --cap-add={BPF,NET_ADMIN,NET_RAW} localhost/netbird:latest
 
-FROM alpine:3.22.0
+FROM alpine:3.24
 
 RUN apk add --no-cache \
       bash \
@@ -23,10 +23,11 @@ ENV \
     NB_DAEMON_ADDR="unix:///var/lib/netbird/netbird.sock" \
     NB_LOG_FILE="console,/var/lib/netbird/client.log" \
     NB_DISABLE_DNS="true" \
+    NB_ENABLE_CAPTURE="false" \
     NB_ENTRYPOINT_SERVICE_TIMEOUT="30"
 
 ENTRYPOINT [ "/usr/local/bin/netbird-entrypoint.sh" ]
-
-ARG NETBIRD_BINARY=netbird
+ARG TARGETPLATFORM
+ARG NETBIRD_BINARY=$TARGETPLATFORM/netbird
 COPY client/netbird-entrypoint.sh /usr/local/bin/netbird-entrypoint.sh
 COPY "${NETBIRD_BINARY}"  /usr/local/bin/netbird

client/android/client.go

@@ -7,13 +7,16 @@ import (
 	"fmt"
 	"os"
 	"slices"
+	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"golang.org/x/exp/maps"
 
 	log "github.com/sirupsen/logrus"
 
+	nbAnonymize "github.com/netbirdio/netbird/client/anonymize"
 	"github.com/netbirdio/netbird/client/iface/device"
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/debug"
@@ -24,6 +27,7 @@ import (
 	"github.com/netbirdio/netbird/client/internal/routemanager"
 	"github.com/netbirdio/netbird/client/internal/stdnet"
 	"github.com/netbirdio/netbird/client/net"
+	"github.com/netbirdio/netbird/client/netevents"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/formatter"
 	"github.com/netbirdio/netbird/route"
@@ -31,10 +35,12 @@ import (
 	types "github.com/netbirdio/netbird/upload-server/types"
 )
 
-// ConnectionListener export internal Listener for mobile
-type ConnectionListener interface {
-	peer.Listener
-}
+// AnonymizeLevelDefault and AnonymizeLevelStrict are the accepted
+// anonymizeLevel values for DebugBundle.
+const (
+	AnonymizeLevelDefault = nbAnonymize.LevelDefaultString
+	AnonymizeLevelStrict  = nbAnonymize.LevelStrictString
+)
 
 // TunAdapter export internal TunAdapter for mobile
 type TunAdapter interface {
@@ -56,6 +62,12 @@ type DnsReadyListener interface {
 	dns.ReadyListener
 }
 
+// TunSettings is a snapshot of the settings the TUN device is rebuilt with
+type TunSettings struct {
+	Routes        string
+	SearchDomains string
+}
+
 func init() {
 	formatter.SetLogcatFormatter(log.StandardLogger())
 }
@@ -70,18 +82,51 @@ type Client struct {
 	deviceName            string
 	uiVersion             string
 	networkChangeListener listener.NetworkChangeListener
+	// netMgr outlives engine restarts: it mirrors the OS connectivity, not
+	// the engine lifecycle. Run and RunWithoutLogin inject its state and
+	// sweeper into each new ConnectClient.
+	netMgr *netevents.Manager
 
 	stateMu       sync.RWMutex
 	connectClient *internal.ConnectClient
 	config        *profilemanager.Config
 	cacheDir      string
+
+	// mdmSource holds the per-Client MDM policy source and its change
+	// detector as one unit. Set by SetMDMPolicyFetcher (called from the
+	// Kotlin side). Each Run passes the loader to the resolved Config so
+	// applyMDMPolicy picks up the active overlay. Nil means "MDM
+	// enforcement off for this Client".
+	mdmSource atomic.Pointer[mdmSource]
+
+	// Identifies the running profile for the SSO login hint; see profile_state.go.
+	cfgPath string
+
+	stateChangeMu    sync.Mutex
+	stateChangeSubID string
+	eventSub         *peer.EventSubscription
+	// Closed to stop the watch goroutines from delivering buffered items to a
+	// listener that has been removed or replaced. See stopStateChangeWatchLocked.
+	stateChangeDone chan struct{}
+
+	// Latched "the server wants an interactive login": survives the engine
+	// restarts that replace the run loop's context state. See Client.Status.
+	// Guarded by loginRequiredMu together with loginCleared, which counts
+	// clears so a stale observation cannot re-latch over one.
+	loginRequiredMu sync.Mutex
+	loginRequired   bool
+	loginCleared    uint64
+
+	extendMu     sync.Mutex
+	extendCancel context.CancelFunc
 }
 
-func (c *Client) setState(cfg *profilemanager.Config, cacheDir string, cc *internal.ConnectClient) {
+func (c *Client) setState(cfg *profilemanager.Config, cacheDir string, cfgPath string, cc *internal.ConnectClient) {
 	c.stateMu.Lock()
 	defer c.stateMu.Unlock()
 	c.config = cfg
 	c.cacheDir = cacheDir
+	c.cfgPath = cfgPath
 	c.connectClient = cc
 }
 
@@ -91,6 +136,16 @@ func (c *Client) stateSnapshot() (*profilemanager.Config, string, *internal.Conn
 	return c.config, c.cacheDir, c.connectClient
 }
 
+// authSnapshot returns the config together with the path it was loaded from, in
+// one lock: the path identifies the profile whose account email backs the login
+// hint, so reading it separately could pair one profile's config with another's
+// hint when a profile switch lands in between.
+func (c *Client) authSnapshot() (*profilemanager.Config, string, *internal.ConnectClient) {
+	c.stateMu.RLock()
+	defer c.stateMu.RUnlock()
+	return c.config, c.cfgPath, c.connectClient
+}
+
 func (c *Client) getConnectClient() *internal.ConnectClient {
 	c.stateMu.RLock()
 	defer c.stateMu.RUnlock()
@@ -102,14 +157,17 @@ func NewClient(androidSDKVersion int, deviceName string, uiVersion string, tunAd
 	execWorkaround(androidSDKVersion)
 
 	net.SetAndroidProtectSocketFn(tunAdapter.ProtectSocket)
+	system.SetIFaceDiscover(iFaceDiscover)
+	recorder := peer.NewRecorder("")
 	return &Client{
 		deviceName:            deviceName,
 		uiVersion:             uiVersion,
 		tunAdapter:            tunAdapter,
 		iFaceDiscover:         iFaceDiscover,
-		recorder:              peer.NewRecorder(""),
+		recorder:              recorder,
 		ctxCancelLock:         &sync.Mutex{},
 		networkChangeListener: networkChangeListener,
+		netMgr:                netevents.NewManager(recorder),
 	}
 }
 
@@ -129,6 +187,7 @@ func (c *Client) Run(platformFiles PlatformFiles, urlOpener URLOpener, isAndroid
 	if err != nil {
 		return err
 	}
+	c.applyMDMOverlay(cfg)
 	c.recorder.UpdateManagementAddress(cfg.ManagementURL.String())
 	c.recorder.UpdateRosenpass(cfg.RosenpassEnabled, cfg.RosenpassPermissive)
 
@@ -143,16 +202,23 @@ func (c *Client) Run(platformFiles PlatformFiles, urlOpener URLOpener, isAndroid
 	defer c.ctxCancel()
 	c.ctxCancelLock.Unlock()
 
-	auth := NewAuthWithConfig(ctx, cfg)
+	auth := NewAuthWithConfig(ctx, cfg, cfgFile)
 	err = auth.login(urlOpener, isAndroidTV)
 	if err != nil {
 		return err
 	}
-
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder)
-	c.setState(cfg, cacheDir, connectClient)
+
+	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder,
+		internal.WithNetEvents(c.netMgr))
+	c.setState(cfg, cacheDir, cfgFile, connectClient)
+	// This path runs the interactive SSO flow, so reaching here means the peer
+	// is authenticated again — release the latch Status() reports from. Clear
+	// only once the fresh connect client is installed: until then Status()
+	// still reads the previous run's context state, which holds the NeedsLogin
+	// that prompted this login, and would re-latch what was just cleared.
+	c.clearLoginRequired()
 	return connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile, cacheDir)
 }
 
@@ -173,6 +239,7 @@ func (c *Client) RunWithoutLogin(platformFiles PlatformFiles, dns *DNSList, dnsR
 	if err != nil {
 		return err
 	}
+	c.applyMDMOverlay(cfg)
 	c.recorder.UpdateManagementAddress(cfg.ManagementURL.String())
 	c.recorder.UpdateRosenpass(cfg.RosenpassEnabled, cfg.RosenpassPermissive)
 
@@ -186,8 +253,9 @@ func (c *Client) RunWithoutLogin(platformFiles PlatformFiles, dns *DNSList, dnsR
 
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder)
-	c.setState(cfg, cacheDir, connectClient)
+	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder,
+		internal.WithNetEvents(c.netMgr))
+	c.setState(cfg, cacheDir, cfgFile, connectClient)
 	return connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile, cacheDir)
 }
 
@@ -216,9 +284,49 @@ func (c *Client) RenewTun(fd int) error {
 	return e.RenewTun(fd)
 }
 
+func (c *Client) GetTunSettings() (*TunSettings, error) {
+	cc := c.getConnectClient()
+	if cc == nil {
+		return nil, fmt.Errorf("engine not running")
+	}
+
+	e := cc.Engine()
+	if e == nil {
+		return nil, fmt.Errorf("engine not initialized")
+	}
+
+	routes, searchDomains := e.TunSettings()
+	return &TunSettings{
+		Routes:        strings.Join(routes, ";"),
+		SearchDomains: strings.Join(searchDomains, ";"),
+	}, nil
+}
+
+// SetNetworkAvailable feeds OS-reported network availability into the client.
+// While unavailable, the internal reconnect loops suspend their attempts and
+// the connection listener reports NoNetwork instead of Connecting; when
+// availability returns, the loops resume immediately with a fresh backoff.
+// Losing the last network also sweeps the registered connections: nothing can
+// redial while offline, so the stale sockets would otherwise stay silently
+// "connected" until their own timeouts and the client would keep reporting
+// Connected with no network at all.
+func (c *Client) SetNetworkAvailable(available bool) {
+	c.netMgr.SetNetworkAvailable(available)
+}
+
+// NotifyNetworkChange marks the management, signal and relay connections
+// stale after the OS switched networks and schedules a sweep that cuts
+// whatever has not redialed on the new network by then. The engine and the
+// TUN device stay untouched.
+func (c *Client) NotifyNetworkChange() {
+	c.netMgr.NotifyNetworkChange()
+}
+
 // DebugBundle generates a debug bundle, uploads it, and returns the upload key.
-// It works both with and without a running engine.
-func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (string, error) {
+// It works both with and without a running engine. anonymizeLevel is "default"
+// or "strict"; strict also anonymizes internal IP ranges, peer names, and
+// WireGuard public keys, and implies anonymize.
+func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool, anonymizeLevel string) (string, error) {
 	cfg, cacheDir, cc := c.stateSnapshot()
 
 	// If the engine hasn't been started, load config from disk
@@ -230,6 +338,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		if err != nil {
 			return "", fmt.Errorf("load config: %w", err)
 		}
+		c.applyMDMOverlay(cfg)
 		cacheDir = platformFiles.CacheDir()
 	}
 
@@ -237,6 +346,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		InternalConfig: cfg,
 		StatusRecorder: c.recorder,
 		TempDir:        cacheDir,
+		StatePath:      platformFiles.StateFilePath(),
 	}
 
 	if cc != nil {
@@ -247,6 +357,9 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		deps.SyncResponse = resp
 
 		if e := cc.Engine(); e != nil {
+			deps.RefreshStatus = func() {
+				e.RunHealthProbes(context.Background(), true)
+			}
 			if cm := e.GetClientMetrics(); cm != nil {
 				deps.ClientMetrics = cm
 			}
@@ -257,6 +370,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		deps,
 		debug.BundleConfig{
 			Anonymize:         anonymize,
+			AnonymizeLevel:    nbAnonymize.ParseLevel(anonymizeLevel),
 			IncludeSystemInfo: true,
 		},
 	)
@@ -274,7 +388,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 	uploadCtx, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
 	defer cancel()
 
-	key, err := debug.UploadDebugBundle(uploadCtx, types.DefaultBundleURL, cfg.ManagementURL.String(), path)
+	key, err := debug.UploadDebugBundle(uploadCtx, types.DefaultBundleURL, cfg.ManagementURL.String(), path, false)
 	if err != nil {
 		return "", fmt.Errorf("upload debug bundle: %w", err)
 	}
@@ -296,15 +410,37 @@ func (c *Client) SetInfoLogLevel() {
 // PeersList return with the list of the PeerInfos
 func (c *Client) PeersList() *PeerInfoArray {
 
+	// The recorder only caches transfer counters and handshake times; nothing
+	// refreshes them on its own, so without this they read as zero. The desktop
+	// daemon does the same before serving a full peer status.
+	if err := c.recorder.RefreshWireGuardStats(); err != nil {
+		log.Debugf("failed to refresh WireGuard stats: %v", err)
+	}
+
 	fullStatus := c.recorder.GetFullStatus()
 
 	peerInfos := make([]PeerInfo, len(fullStatus.Peers))
 	for n, p := range fullStatus.Peers {
 		pi := PeerInfo{
-			p.IP,
-			p.FQDN,
-			int(p.ConnStatus),
-			PeerRoutes{routes: maps.Keys(p.GetRoutes())},
+			IP:         p.IP,
+			IPv6:       p.IPv6,
+			FQDN:       p.FQDN,
+			ConnStatus: int(p.ConnStatus),
+			Routes:     PeerRoutes{routes: maps.Keys(p.GetRoutes())},
+
+			PubKey:                     p.PubKey,
+			Latency:                    formatDuration(p.Latency),
+			LatencyMs:                  p.Latency.Milliseconds(),
+			BytesRx:                    p.BytesRx,
+			BytesTx:                    p.BytesTx,
+			ConnStatusUpdate:           formatTime(p.ConnStatusUpdate),
+			Relayed:                    p.Relayed,
+			RosenpassEnabled:           p.RosenpassEnabled,
+			LastWireguardHandshake:     formatTime(p.LastWireguardHandshake),
+			LocalIceCandidateType:      p.LocalIceCandidateType,
+			RemoteIceCandidateType:     p.RemoteIceCandidateType,
+			LocalIceCandidateEndpoint:  p.LocalIceCandidateEndpoint,
+			RemoteIceCandidateEndpoint: p.RemoteIceCandidateEndpoint,
 		}
 		peerInfos[n] = pi
 	}
@@ -336,43 +472,84 @@ func (c *Client) Networks() *NetworkArray {
 		return nil
 	}
 
+	routesMap := routeManager.GetClientRoutesWithNetID()
+	v6Merged := route.V6ExitMergeSet(routesMap)
+	resolvedDomains := c.recorder.GetResolvedDomainsStates()
+
 	networkArray := &NetworkArray{
 		items: make([]Network, 0),
 	}
 
-	resolvedDomains := c.recorder.GetResolvedDomainsStates()
-
-	for id, routes := range routeManager.GetClientRoutesWithNetID() {
+	for id, routes := range routesMap {
 		if len(routes) == 0 {
 			continue
 		}
-
-		r := routes[0]
-		domains := c.getNetworkDomainsFromRoute(r, resolvedDomains)
-		netStr := r.Network.String()
-
-		if r.IsDynamic() {
-			netStr = r.Domains.SafeString()
-		}
-
-		routePeer, err := c.recorder.GetPeer(routes[0].Peer)
-		if err != nil {
-			log.Errorf("could not get peer info for %s: %v", routes[0].Peer, err)
+		if _, skip := v6Merged[id]; skip {
 			continue
 		}
-		network := Network{
-			Name:       string(id),
-			Network:    netStr,
-			Peer:       routePeer.FQDN,
-			Status:     routePeer.ConnStatus.String(),
-			IsSelected: routeSelector.IsSelected(id),
-			Domains:    domains,
+
+		network := c.buildNetwork(id, routes, routeSelector.IsSelected(id), resolvedDomains, v6Merged)
+		if network == nil {
+			continue
 		}
-		networkArray.Add(network)
+		networkArray.Add(*network)
 	}
 	return networkArray
 }
 
+func (c *Client) buildNetwork(id route.NetID, routes []*route.Route, selected bool, resolvedDomains map[domain.Domain]peer.ResolvedDomainInfo, v6Merged map[route.NetID]struct{}) *Network {
+	r := routes[0]
+	netStr := r.Network.String()
+	if r.IsDynamic() {
+		netStr = r.Domains.SafeString()
+	}
+
+	routePeer, err := c.findBestRoutePeer(routes)
+	if err != nil {
+		log.Errorf("could not get peer info for route %s: %v", id, err)
+		return nil
+	}
+
+	network := &Network{
+		Name:       string(id),
+		Network:    netStr,
+		Peer:       routePeer.FQDN,
+		Status:     routePeer.ConnStatus.String(),
+		IsSelected: selected,
+		Domains:    c.getNetworkDomainsFromRoute(r, resolvedDomains),
+	}
+
+	if route.IsV4DefaultRoute(r.Network) && route.HasV6ExitPair(id, v6Merged) {
+		network.Network = "0.0.0.0/0, ::/0"
+	}
+
+	return network
+}
+
+// findBestRoutePeer returns the peer actively routing traffic for the given
+// HA route group. Falls back to the first connected peer, then the first peer.
+func (c *Client) findBestRoutePeer(routes []*route.Route) (peer.State, error) {
+	netStr := routes[0].Network.String()
+
+	fullStatus := c.recorder.GetFullStatus()
+	for _, p := range fullStatus.Peers {
+		if _, ok := p.GetRoutes()[netStr]; ok {
+			return p, nil
+		}
+	}
+
+	for _, r := range routes {
+		p, err := c.recorder.GetPeer(r.Peer)
+		if err != nil {
+			continue
+		}
+		if p.ConnStatus == peer.StatusConnected {
+			return p, nil
+		}
+	}
+	return c.recorder.GetPeer(routes[0].Peer)
+}
+
 // OnUpdatedHostDNS update the DNS servers addresses for root zones
 func (c *Client) OnUpdatedHostDNS(list *DNSList) error {
 	dnsServer, err := dns.GetServerDns()
@@ -386,7 +563,11 @@ func (c *Client) OnUpdatedHostDNS(list *DNSList) error {
 
 // SetConnectionListener set the network connection listener
 func (c *Client) SetConnectionListener(listener ConnectionListener) {
-	c.recorder.SetConnectionListener(listener)
+	if listener == nil {
+		c.recorder.RemoveConnectionListener()
+		return
+	}
+	c.recorder.SetConnectionListener(connectionListenerAdapter{listener})
 }
 
 // RemoveConnectionListener remove connection listener
@@ -394,10 +575,6 @@ func (c *Client) RemoveConnectionListener() {
 	c.recorder.RemoveConnectionListener()
 }
 
-func (c *Client) toggleRoute(command routeCommand) error {
-	return command.toggleRoute()
-}
-
 func (c *Client) getRouteManager() (routemanager.Manager, error) {
 	client := c.getConnectClient()
 	if client == nil {
@@ -417,22 +594,22 @@ func (c *Client) getRouteManager() (routemanager.Manager, error) {
 	return manager, nil
 }
 
-func (c *Client) SelectRoute(route string) error {
+func (c *Client) SelectRoute(id string) error {
 	manager, err := c.getRouteManager()
 	if err != nil {
 		return err
 	}
 
-	return c.toggleRoute(selectRouteCommand{route: route, manager: manager})
+	return manager.SelectRoutes([]route.NetID{route.NetID(id)}, true)
 }
 
-func (c *Client) DeselectRoute(route string) error {
+func (c *Client) DeselectRoute(id string) error {
 	manager, err := c.getRouteManager()
 	if err != nil {
 		return err
 	}
 
-	return c.toggleRoute(deselectRouteCommand{route: route, manager: manager})
+	return manager.DeselectRoutes([]route.NetID{route.NetID(id)})
 }
 
 // getNetworkDomainsFromRoute extracts domains from a route and enriches each domain
@@ -467,3 +644,28 @@ func exportEnvList(list *EnvList) {
 		}
 	}
 }
+
+// formatDuration renders a duration for display, trimming the fractional part
+// to two digits so latencies read as "12.34ms" rather than "12.345678ms".
+func formatDuration(d time.Duration) string {
+	ds := d.String()
+	dotIndex := strings.Index(ds, ".")
+	if dotIndex == -1 {
+		return ds
+	}
+
+	endIndex := min(dotIndex+3, len(ds))
+
+	// Skip the remaining digits so only the unit suffix is appended back.
+	unitStart := endIndex
+	for unitStart < len(ds) && ds[unitStart] >= '0' && ds[unitStart] <= '9' {
+		unitStart++
+	}
+	return ds[:endIndex] + ds[unitStart:]
+}
+
+// formatTime renders a timestamp in UTC using a fixed layout. The zero time is
+// passed through as-is so the UI can recognise it and show "never" instead.
+func formatTime(t time.Time) string {
+	return t.UTC().Format("2006-01-02 15:04:05")
+}

client/android/client_mdm.go

@@ -0,0 +1,52 @@
+//go:build android
+
+package android
+
+import (
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mdm"
+)
+
+type mdmSource struct {
+	loader   *mdm.Loader
+	detector *mdm.ChangeDetector
+}
+
+// SetMDMPolicyFetcher registers the native-provided MDM policy fetcher on
+// this Client; passing nil disables MDM enforcement.
+func (c *Client) SetMDMPolicyFetcher(p PolicyFetcher) {
+	loader := loaderFor(p)
+	c.mdmSource.Store(&mdmSource{loader: loader, detector: mdm.NewChangeDetector(loader)})
+}
+
+// HasMDMPolicyChanged re-reads the managed configuration and reports whether
+// it changed since the last observation; call it from the native OS-change
+// notification and restart the engine only on true.
+func (c *Client) HasMDMPolicyChanged() bool {
+	src := c.mdmSource.Load()
+	if src == nil {
+		return false
+	}
+	return src.detector.Changed()
+}
+
+// GetRestrictionsJSON returns the UI enforcement snapshot derived from the
+// active MDM policy, in the JSON shape shared with the desktop frontend.
+func (c *Client) GetRestrictionsJSON() (string, error) {
+	return mdm.BuildRestrictions(c.mdmLoader().Load()).JSON()
+}
+
+func (c *Client) applyMDMOverlay(cfg *profilemanager.Config) {
+	loader := c.mdmLoader()
+	if cfg == nil || loader == nil {
+		return
+	}
+	cfg.ApplyMDMPolicy(loader.Load())
+}
+
+func (c *Client) mdmLoader() *mdm.Loader {
+	if src := c.mdmSource.Load(); src != nil {
+		return src.loader
+	}
+	return nil
+}

client/android/connection_listener.go

@@ -0,0 +1,41 @@
+//go:build android
+
+package android
+
+import (
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+// Client state values delivered via ConnectionListener.OnStateChanged,
+// re-exported as basic constants so gomobile emits them into the generated
+// Java bindings. They mirror peer.ClientState*: append-only, never reorder.
+const (
+	ClientStateDisconnected  = int(peer.ClientStateDisconnected)
+	ClientStateConnected     = int(peer.ClientStateConnected)
+	ClientStateConnecting    = int(peer.ClientStateConnecting)
+	ClientStateDisconnecting = int(peer.ClientStateDisconnecting)
+	ClientStateNoNetwork     = int(peer.ClientStateNoNetwork)
+)
+
+// ConnectionListener export internal Listener for mobile. It mirrors
+// peer.Listener with OnStateChanged taking a plain int (one of the
+// ClientState* constants), because gomobile cannot bind named types.
+type ConnectionListener interface {
+	OnStateChanged(state int)
+	OnConnected()
+	OnDisconnected()
+	OnConnecting()
+	OnDisconnecting()
+	OnAddressChanged(string, string)
+	OnPeersListChanged(int)
+}
+
+// connectionListenerAdapter adapts the gomobile-facing ConnectionListener to
+// peer.Listener, converting the typed state to the int the binding carries.
+type connectionListenerAdapter struct {
+	ConnectionListener
+}
+
+func (a connectionListenerAdapter) OnStateChanged(state peer.ClientState) {
+	a.ConnectionListener.OnStateChanged(int(state))
+}

client/android/env_list.go

@@ -10,7 +10,7 @@ var (
 	EnvKeyNBForceRelay = peer.EnvKeyNBForceRelay
 
 	// EnvKeyNBLazyConn Exported for Android java client to configure lazy connection
-	EnvKeyNBLazyConn = lazyconn.EnvEnableLazyConn
+	EnvKeyNBLazyConn = lazyconn.EnvLazyConn
 
 	// EnvKeyNBInactivityThreshold Exported for Android java client to configure connection inactivity threshold
 	EnvKeyNBInactivityThreshold = lazyconn.EnvInactivityThreshold

client/android/login.go

@@ -4,8 +4,12 @@ import (
 	"context"
 	"fmt"
 
+	log "github.com/sirupsen/logrus"
+
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mdm"
+	"github.com/netbirdio/netbird/client/mobile"
 	"github.com/netbirdio/netbird/client/system"
 )
 
@@ -36,15 +40,31 @@ type Auth struct {
 }
 
 // NewAuth instantiate Auth struct and validate the management URL
-func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
-	inputCfg := profilemanager.ConfigInput{
-		ManagementURL: mgmURL,
+//
+// The configuration at cfgPath is reused when one is already there, and only created when it is
+// not. Building a fresh in-memory config unconditionally gives the client a new WireGuard key on
+// every call: the peer registers under that key, the key is written out, and any peer registered by
+// an earlier call is orphaned on the server. It also breaks a client that enrols and then runs from
+// the persisted config, because the identity it registered is not the one it runs with — the
+// management stream rejects it with "no peer auth method provided".
+//
+// Auth is constructed under the active MDM policy: the policy is overlaid on
+// the resolved config so the login runs against the enforced values, while
+// the persisted config keeps the caller-supplied ones; a caller-supplied
+// management URL is ignored while MDM manages that key. A nil fetcher
+// disables MDM enforcement.
+func NewAuth(cfgPath string, mgmURL string, fetcher PolicyFetcher) (*Auth, error) {
+	policy := loaderFor(fetcher).Load()
+	inputCfg := profilemanager.ConfigInput{ConfigPath: cfgPath}
+	if _, managed := policy.GetString(mdm.KeyManagementURL); !managed {
+		inputCfg.ManagementURL = mgmURL
 	}
 
-	cfg, err := profilemanager.CreateInMemoryConfig(inputCfg)
+	cfg, err := profilemanager.UpdateOrCreateConfig(inputCfg)
 	if err != nil {
 		return nil, err
 	}
+	cfg.ApplyMDMPolicy(policy)
 
 	return &Auth{
 		ctx:     context.Background(),
@@ -53,17 +73,18 @@ func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
 	}, nil
 }
 
-// NewAuthWithConfig instantiate Auth based on existing config
-func NewAuthWithConfig(ctx context.Context, config *profilemanager.Config) *Auth {
+// NewAuthWithConfig instantiate Auth based on existing config. cfgPath is the
+// file the config was loaded from; it identifies the profile whose account email
+// backs the login_hint.
+func NewAuthWithConfig(ctx context.Context, config *profilemanager.Config, cfgPath string) *Auth {
 	return &Auth{
-		ctx:    ctx,
-		config: config,
+		ctx:     ctx,
+		config:  config,
+		cfgPath: cfgPath,
 	}
 }
 
-// SaveConfigIfSSOSupported test the connectivity with the management server by retrieving the server device flow info.
-// If it returns a flow info than save the configuration and return true. If it gets a codes.NotFound, it means that SSO
-// is not supported and returns false without saving the configuration. For other errors return false.
+// SaveConfigIfSSOSupported reports whether the management server supports SSO; the config is already persisted by NewAuth.
 func (a *Auth) SaveConfigIfSSOSupported(listener SSOListener) {
 	go func() {
 		sso, err := a.saveConfigIfSSOSupported()
@@ -87,15 +108,10 @@ func (a *Auth) saveConfigIfSSOSupported() (bool, error) {
 		return false, fmt.Errorf("failed to check SSO support: %v", err)
 	}
 
-	if !supportsSSO {
-		return false, nil
-	}
-
-	err = profilemanager.WriteOutConfig(a.cfgPath, a.config)
-	return true, err
+	return supportsSSO, nil
 }
 
-// LoginWithSetupKeyAndSaveConfig test the connectivity with the management server with the setup key.
+// LoginWithSetupKeyAndSaveConfig registers the peer with the setup key; the config is already persisted by NewAuth.
 func (a *Auth) LoginWithSetupKeyAndSaveConfig(resultListener ErrListener, setupKey string, deviceName string) {
 	go func() {
 		err := a.loginWithSetupKeyAndSaveConfig(setupKey, deviceName)
@@ -120,8 +136,7 @@ func (a *Auth) loginWithSetupKeyAndSaveConfig(setupKey string, deviceName string
 	if err != nil {
 		return fmt.Errorf("login failed: %v", err)
 	}
-
-	return profilemanager.WriteOutConfig(a.cfgPath, a.config)
+	return nil
 }
 
 // Login try register the client on the server
@@ -150,12 +165,14 @@ func (a *Auth) login(urlOpener URLOpener, isAndroidTV bool) error {
 	}
 
 	jwtToken := ""
+	email := ""
 	if needsLogin {
 		tokenInfo, err := a.foregroundGetTokenInfo(authClient, urlOpener, isAndroidTV)
 		if err != nil {
 			return fmt.Errorf("interactive sso login failed: %v", err)
 		}
 		jwtToken = tokenInfo.GetTokenToUse()
+		email = tokenInfo.Email
 	}
 
 	err, _ = authClient.Login(a.ctx, "", jwtToken)
@@ -163,27 +180,62 @@ func (a *Auth) login(urlOpener URLOpener, isAndroidTV bool) error {
 		return fmt.Errorf("login failed: %v", err)
 	}
 
+	// Stored after Login, not before: a rejected token must not leave a hint
+	// pointing at an account that cannot be used.
+	if email != "" && a.cfgPath != "" {
+		if err := mobile.WriteProfileEmail(a.cfgPath, email); err != nil {
+			log.Warnf("failed to store profile account email: %v", err)
+		}
+	}
+
 	go urlOpener.OnLoginSuccess()
 
 	return nil
 }
 
 func (a *Auth) foregroundGetTokenInfo(authClient *auth.Auth, urlOpener URLOpener, isAndroidTV bool) (*auth.TokenInfo, error) {
-	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, isAndroidTV)
+	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, isAndroidTV, profileLoginHint(a.cfgPath))
 	if err != nil {
 		return nil, fmt.Errorf("failed to get OAuth flow: %v", err)
 	}
 
-	flowInfo, err := oAuthFlow.RequestAuthInfo(context.TODO())
+	return runOAuthFlow(a.ctx, oAuthFlow, urlOpener, nil)
+}
+
+// profileLoginHint returns the stored account email for the profile at cfgPath.
+// An empty hint is deliberate, not a fallback: a fresh profile leaves the
+// choice to the IdP. Switching accounts is done by switching or removing
+// profiles, not by logging out — logout keeps the email.
+func profileLoginHint(cfgPath string) string {
+	if cfgPath == "" {
+		return ""
+	}
+	return mobile.ReadProfileEmail(cfgPath)
+}
+
+// runOAuthFlow drives an already acquired OAuth flow to a token: requests the
+// flow info, presents the verification URL through the opener and waits for
+// the browser round-trip. Open is called synchronously — it is what marks the
+// surface as opened on the client side, and a fast token's OnLoginSuccess is
+// a no-op until it has, so the dismissal would be dropped rather than
+// delayed. Openers must therefore not block: they post their UI work and
+// return. onWaiting, when set, runs after the URL is shown, right before the
+// blocking wait.
+func runOAuthFlow(ctx context.Context, flow auth.OAuthFlow, urlOpener URLOpener, onWaiting func()) (*auth.TokenInfo, error) {
+	flowInfo, err := flow.RequestAuthInfo(ctx)
 	if err != nil {
-		return nil, fmt.Errorf("getting a request OAuth flow info failed: %v", err)
+		return nil, fmt.Errorf("request auth info: %w", err)
 	}
 
-	go urlOpener.Open(flowInfo.VerificationURIComplete, flowInfo.UserCode)
+	urlOpener.Open(flowInfo.VerificationURIComplete, flowInfo.UserCode)
 
-	tokenInfo, err := oAuthFlow.WaitToken(a.ctx, flowInfo)
+	if onWaiting != nil {
+		onWaiting()
+	}
+
+	tokenInfo, err := flow.WaitToken(ctx, flowInfo)
 	if err != nil {
-		return nil, fmt.Errorf("waiting for browser login failed: %v", err)
+		return nil, fmt.Errorf("wait for token: %w", err)
 	}
 
 	return &tokenInfo, nil

client/android/login_test.go

@@ -0,0 +1,51 @@
+package android
+
+import (
+	"path/filepath"
+	"testing"
+)
+
+// NewAuth must reuse the configuration already at cfgPath rather than building a fresh one.
+//
+// Creating a new in-memory config on every call gives the client a new WireGuard private key each
+// time. The peer registers under that key and the key is written out, so a peer registered by an
+// earlier call is orphaned on the server — a client that enrols twice leaves two entries and owns
+// neither. It also breaks enrol-then-run: RunWithoutLogin reloads the configuration from disk, so
+// the identity that registered is not the identity that runs, and the management stream rejects it
+// with "no peer auth method provided, please use a setup key or interactive SSO login".
+func TestNewAuth_ReusesPersistedIdentity(t *testing.T) {
+	cfgPath := filepath.Join(t.TempDir(), "config.json")
+
+	first, err := NewAuth(cfgPath, "https://api.example.com:443", nil)
+	if err != nil {
+		t.Fatalf("first NewAuth: %v", err)
+	}
+	if first.config.PrivateKey == "" {
+		t.Fatal("first NewAuth produced no private key")
+	}
+
+	second, err := NewAuth(cfgPath, "https://api.example.com:443", nil)
+	if err != nil {
+		t.Fatalf("second NewAuth: %v", err)
+	}
+
+	if second.config.PrivateKey != first.config.PrivateKey {
+		t.Errorf("private key changed between calls: a second enrolment would orphan the peer registered by the first")
+	}
+}
+
+// A missing configuration is still created, so a first enrolment works unchanged.
+func TestNewAuth_CreatesConfigWhenAbsent(t *testing.T) {
+	cfgPath := filepath.Join(t.TempDir(), "config.json")
+
+	auth, err := NewAuth(cfgPath, "https://api.example.com:443", nil)
+	if err != nil {
+		t.Fatalf("NewAuth: %v", err)
+	}
+	if auth.config == nil || auth.config.PrivateKey == "" {
+		t.Fatal("NewAuth did not create a usable configuration")
+	}
+	if auth.cfgPath != cfgPath {
+		t.Errorf("cfgPath = %q, want %q", auth.cfgPath, cfgPath)
+	}
+}

client/android/mdm.go

@@ -0,0 +1,19 @@
+package android
+
+import (
+	"github.com/netbirdio/netbird/client/mdm"
+)
+
+// PolicyFetcher is implemented by the native layer to return the current
+// managed configuration as a JSON-encoded object string; "" means no MDM
+// source is present.
+type PolicyFetcher interface {
+	FetchJSON() string
+}
+
+func loaderFor(p PolicyFetcher) *mdm.Loader {
+	if p == nil {
+		return mdm.NewJSONLoader(nil)
+	}
+	return mdm.NewJSONLoader(p.FetchJSON)
+}

client/android/peer_notifier.go

@@ -12,11 +12,30 @@ const (
 )
 
 // PeerInfo describe information about the peers. It designed for the UI usage
+//
+// The fields below ConnStatus back the peer detail screen. Durations and times
+// are pre-formatted into strings so the UI does not have to know Go's layouts;
+// Latency is additionally exposed as LatencyMs for colour coding.
 type PeerInfo struct {
 	IP         string
+	IPv6       string
 	FQDN       string
 	ConnStatus int
 	Routes     PeerRoutes
+
+	PubKey                     string
+	Latency                    string
+	LatencyMs                  int64
+	BytesRx                    int64
+	BytesTx                    int64
+	ConnStatusUpdate           string
+	Relayed                    bool
+	RosenpassEnabled           bool
+	LastWireguardHandshake     string
+	LocalIceCandidateType      string
+	RemoteIceCandidateType     string
+	LocalIceCandidateEndpoint  string
+	RemoteIceCandidateEndpoint string
 }
 
 func (p *PeerInfo) GetPeerRoutes() *PeerRoutes {

client/android/preferences.go

@@ -1,12 +1,16 @@
 package android
 
 import (
+	"sync/atomic"
+
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mdm"
 )
 
 // Preferences exports a subset of the internal config for gomobile
 type Preferences struct {
 	configInput profilemanager.ConfigInput
+	mdmLoader   atomic.Pointer[mdm.Loader]
 }
 
 // NewPreferences creates a new Preferences instance
@@ -14,11 +18,30 @@ func NewPreferences(configPath string) *Preferences {
 	ci := profilemanager.ConfigInput{
 		ConfigPath: configPath,
 	}
-	return &Preferences{ci}
+	return &Preferences{configInput: ci}
+}
+
+// SetMDMPolicyFetcher registers the native-provided MDM policy fetcher on
+// this Preferences instance; passing nil disables MDM enforcement.
+func (p *Preferences) SetMDMPolicyFetcher(f PolicyFetcher) {
+	p.mdmLoader.Store(loaderFor(f))
+}
+
+// GetRestrictionsJSON returns the UI enforcement snapshot derived from the
+// active MDM policy, in the JSON shape shared with the desktop frontend.
+func (p *Preferences) GetRestrictionsJSON() (string, error) {
+	return mdm.BuildRestrictions(p.policy()).JSON()
+}
+
+func (p *Preferences) policy() *mdm.Policy {
+	return p.mdmLoader.Load().Load()
 }
 
 // GetManagementURL reads URL from config file
 func (p *Preferences) GetManagementURL() (string, error) {
+	if v, ok := p.policy().GetString(mdm.KeyManagementURL); ok {
+		return mdm.CanonicalURL(v), nil
+	}
 	if p.configInput.ManagementURL != "" {
 		return p.configInput.ManagementURL, nil
 	}
@@ -27,7 +50,7 @@ func (p *Preferences) GetManagementURL() (string, error) {
 	if err != nil {
 		return "", err
 	}
-	return cfg.ManagementURL.String(), err
+	return cfg.ManagementURL.String(), nil
 }
 
 // SetManagementURL stores the given URL and waits for commit
@@ -53,17 +76,21 @@ func (p *Preferences) SetAdminURL(url string) {
 	p.configInput.AdminURL = url
 }
 
-// GetPreSharedKey reads pre-shared key from config file
-func (p *Preferences) GetPreSharedKey() (string, error) {
+// HasPreSharedKey reports whether a pre-shared key is staged, persisted, or
+// enforced by MDM; the key itself is never handed to the native layer.
+func (p *Preferences) HasPreSharedKey() (bool, error) {
+	if _, ok := p.policy().GetString(mdm.KeyPreSharedKey); ok {
+		return true, nil
+	}
 	if p.configInput.PreSharedKey != nil {
-		return *p.configInput.PreSharedKey, nil
+		return *p.configInput.PreSharedKey != "", nil
 	}
 
 	cfg, err := profilemanager.ReadConfig(p.configInput.ConfigPath)
 	if err != nil {
-		return "", err
+		return false, err
 	}
-	return cfg.PreSharedKey, err
+	return cfg.PreSharedKey != "", nil
 }
 
 // SetPreSharedKey stores the given key and waits for commit
@@ -78,6 +105,9 @@ func (p *Preferences) SetRosenpassEnabled(enabled bool) {
 
 // GetRosenpassEnabled reads Rosenpass enabled status from config file
 func (p *Preferences) GetRosenpassEnabled() (bool, error) {
+	if v, ok := p.policy().GetBool(mdm.KeyRosenpassEnabled); ok {
+		return v, nil
+	}
 	if p.configInput.RosenpassEnabled != nil {
 		return *p.configInput.RosenpassEnabled, nil
 	}
@@ -96,6 +126,9 @@ func (p *Preferences) SetRosenpassPermissive(permissive bool) {
 
 // GetRosenpassPermissive reads Rosenpass permissive setting from config file
 func (p *Preferences) GetRosenpassPermissive() (bool, error) {
+	if v, ok := p.policy().GetBool(mdm.KeyRosenpassPermissive); ok {
+		return v, nil
+	}
 	if p.configInput.RosenpassPermissive != nil {
 		return *p.configInput.RosenpassPermissive, nil
 	}
@@ -109,6 +142,9 @@ func (p *Preferences) GetRosenpassPermissive() (bool, error) {
 
 // GetDisableClientRoutes reads disable client routes setting from config file
 func (p *Preferences) GetDisableClientRoutes() (bool, error) {
+	if v, ok := p.policy().GetBool(mdm.KeyDisableClientRoutes); ok {
+		return v, nil
+	}
 	if p.configInput.DisableClientRoutes != nil {
 		return *p.configInput.DisableClientRoutes, nil
 	}
@@ -127,6 +163,9 @@ func (p *Preferences) SetDisableClientRoutes(disable bool) {
 
 // GetDisableServerRoutes reads disable server routes setting from config file
 func (p *Preferences) GetDisableServerRoutes() (bool, error) {
+	if v, ok := p.policy().GetBool(mdm.KeyDisableServerRoutes); ok {
+		return v, nil
+	}
 	if p.configInput.DisableServerRoutes != nil {
 		return *p.configInput.DisableServerRoutes, nil
 	}
@@ -181,6 +220,9 @@ func (p *Preferences) SetDisableFirewall(disable bool) {
 
 // GetServerSSHAllowed reads server SSH allowed setting from config file
 func (p *Preferences) GetServerSSHAllowed() (bool, error) {
+	if v, ok := p.policy().GetBool(mdm.KeyAllowServerSSH); ok {
+		return v, nil
+	}
 	if p.configInput.ServerSSHAllowed != nil {
 		return *p.configInput.ServerSSHAllowed, nil
 	}
@@ -291,6 +333,9 @@ func (p *Preferences) SetEnableSSHRemotePortForwarding(enabled bool) {
 
 // GetBlockInbound reads block inbound setting from config file
 func (p *Preferences) GetBlockInbound() (bool, error) {
+	if v, ok := p.policy().GetBool(mdm.KeyBlockInbound); ok {
+		return v, nil
+	}
 	if p.configInput.BlockInbound != nil {
 		return *p.configInput.BlockInbound, nil
 	}
@@ -307,8 +352,52 @@ func (p *Preferences) SetBlockInbound(block bool) {
 	p.configInput.BlockInbound = &block
 }
 
+// GetDisableIPv6 reads disable IPv6 setting from config file
+func (p *Preferences) GetDisableIPv6() (bool, error) {
+	if p.configInput.DisableIPv6 != nil {
+		return *p.configInput.DisableIPv6, nil
+	}
+
+	cfg, err := profilemanager.ReadConfig(p.configInput.ConfigPath)
+	if err != nil {
+		return false, err
+	}
+	return cfg.DisableIPv6, err
+}
+
+// SetDisableIPv6 stores the given value and waits for commit
+func (p *Preferences) SetDisableIPv6(disable bool) {
+	p.configInput.DisableIPv6 = &disable
+}
+
+// GetRemoteJobsAllowed reads the remote jobs opt-in from config file
+func (p *Preferences) GetRemoteJobsAllowed() (bool, error) {
+	policy := p.policy()
+	if !policy.HasKey(mdm.KeyRemoteJobsAllowed) && p.configInput.RemoteJobsAllowed != nil {
+		return *p.configInput.RemoteJobsAllowed, nil
+	}
+
+	cfg, err := profilemanager.ReadConfig(p.configInput.ConfigPath)
+	if err != nil {
+		return false, err
+	}
+	cfg.ApplyMDMPolicy(policy)
+	if cfg.RemoteJobsAllowed == nil {
+		return false, nil
+	}
+	return *cfg.RemoteJobsAllowed, nil
+}
+
+// SetRemoteJobsAllowed stores the given value and waits for commit
+func (p *Preferences) SetRemoteJobsAllowed(allowed bool) {
+	p.configInput.RemoteJobsAllowed = &allowed
+}
+
 // Commit writes out the changes to the config file
 func (p *Preferences) Commit() error {
+	if err := profilemanager.CheckMDMConflicts(p.configInput, p.policy()); err != nil {
+		return err
+	}
 	_, err := profilemanager.UpdateOrCreateConfig(p.configInput)
 	return err
 }

client/android/preferences_test.go

@@ -28,14 +28,13 @@ func TestPreferences_DefaultValues(t *testing.T) {
 		t.Errorf("invalid default management url: %s", defaultVar)
 	}
 
-	var preSharedKey string
-	preSharedKey, err = p.GetPreSharedKey()
+	hasPSK, err := p.HasPreSharedKey()
 	if err != nil {
-		t.Fatalf("failed to read default preshared key: %s", err)
+		t.Fatalf("failed to read default preshared key presence: %s", err)
 	}
 
-	if preSharedKey != "" {
-		t.Errorf("invalid preshared key: %s", preSharedKey)
+	if hasPSK {
+		t.Errorf("unexpected preshared key presence on fresh config")
 	}
 }
 
@@ -65,13 +64,13 @@ func TestPreferences_ReadUncommitedValues(t *testing.T) {
 	}
 
 	p.SetPreSharedKey(exampleString)
-	resp, err = p.GetPreSharedKey()
+	hasPSK, err := p.HasPreSharedKey()
 	if err != nil {
-		t.Fatalf("failed to read preshared key: %s", err)
+		t.Fatalf("failed to read preshared key presence: %s", err)
 	}
 
-	if resp != exampleString {
-		t.Errorf("unexpected preshared key: %s", resp)
+	if !hasPSK {
+		t.Errorf("expected preshared key presence after staging one")
 	}
 }
 
@@ -109,12 +108,12 @@ func TestPreferences_Commit(t *testing.T) {
 		t.Errorf("unexpected management url: %s", resp)
 	}
 
-	resp, err = p.GetPreSharedKey()
+	hasPSK, err := p.HasPreSharedKey()
 	if err != nil {
-		t.Fatalf("failed to read preshared key: %s", err)
+		t.Fatalf("failed to read preshared key presence: %s", err)
 	}
 
-	if resp != examplePresharedKey {
-		t.Errorf("unexpected preshared key: %s", resp)
+	if !hasPSK {
+		t.Errorf("expected preshared key presence after commit")
 	}
 }

client/android/profile_manager.go

@@ -3,42 +3,37 @@
 package android
 
 import (
-	"fmt"
-	"os"
-	"path/filepath"
-	"strings"
-
-	log "github.com/sirupsen/logrus"
-
-	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mobile"
 )
 
 const (
-	// Android-specific config filename (different from desktop default.json)
-	defaultConfigFilename = "netbird.cfg"
-	// Subdirectory for non-default profiles (must match Java Preferences.java)
-	profilesSubdir = "profiles"
-	// Android uses a single user context per app (non-empty username required by ServiceManager)
+	// Android uses a single user context per app.
 	androidUsername = "android"
 )
 
-// Profile represents a profile for gomobile
+// Profile represents a profile for gomobile.
 type Profile struct {
-	Name     string
+	ID   string
+	Name string
+	// Email is the account this profile last logged in with, "" if it never
+	// completed an SSO login. Kept across logouts; cleared when the profile is
+	// removed. See client/mobile/profile_state.go.
+	Email    string
 	IsActive bool
 }
 
-// ProfileArray wraps profiles for gomobile compatibility
+// ProfileArray wraps profiles for gomobile compatibility (gomobile cannot
+// bind Go slices directly).
 type ProfileArray struct {
 	items []*Profile
 }
 
-// Length returns the number of profiles
+// Length returns the number of profiles.
 func (p *ProfileArray) Length() int {
 	return len(p.items)
 }
 
-// Get returns the profile at index i
+// Get returns the profile at index i, or nil if out of range.
 func (p *ProfileArray) Get(i int) *Profile {
 	if i < 0 || i >= len(p.items) {
 		return nil
@@ -46,212 +41,104 @@ func (p *ProfileArray) Get(i int) *Profile {
 	return p.items[i]
 }
 
-/*
-
-/data/data/io.netbird.client/files/           ← configDir parameter
-├── netbird.cfg                                 ← Default profile config
-├── state.json                                  ← Default profile state
-├── active_profile.json                         ← Active profile tracker (JSON with Name + Username)
-└── profiles/                                   ← Subdirectory for non-default profiles
-    ├── work.json                              ← Work profile config
-    ├── work.state.json                        ← Work profile state
-    ├── personal.json                          ← Personal profile config
-    └── personal.state.json                    ← Personal profile state
-*/
-
-// ProfileManager manages profiles for Android
-// It wraps the internal profilemanager to provide Android-specific behavior
+// ProfileManager adapts the shared mobile profile manager (client/mobile) to
+// gomobile-friendly types. See that package for the on-disk layout and
+// semantics.
 type ProfileManager struct {
-	configDir  string
-	serviceMgr *profilemanager.ServiceManager
+	impl *mobile.ProfileManager
 }
 
-// NewProfileManager creates a new profile manager for Android
+// NewProfileManager creates a new profile manager for Android. configDir is
+// the app's files directory.
 func NewProfileManager(configDir string) *ProfileManager {
-	// Set the default config path for Android (stored in root configDir, not profiles/)
-	defaultConfigPath := filepath.Join(configDir, defaultConfigFilename)
-
-	// Set global paths for Android
-	profilemanager.DefaultConfigPathDir = configDir
-	profilemanager.DefaultConfigPath = defaultConfigPath
-	profilemanager.ActiveProfileStatePath = filepath.Join(configDir, "active_profile.json")
-
-	// Create ServiceManager with profiles/ subdirectory
-	// This avoids modifying the global ConfigDirOverride for profile listing
-	profilesDir := filepath.Join(configDir, profilesSubdir)
-	serviceMgr := profilemanager.NewServiceManagerWithProfilesDir(defaultConfigPath, profilesDir)
-
-	return &ProfileManager{
-		configDir:  configDir,
-		serviceMgr: serviceMgr,
-	}
+	return &ProfileManager{impl: mobile.NewProfileManager(configDir, androidUsername)}
 }
 
-// ListProfiles returns all available profiles
+// SetMDMPolicyFetcher registers the native-provided MDM policy fetcher on
+// this ProfileManager; passing nil disables MDM enforcement.
+func (pm *ProfileManager) SetMDMPolicyFetcher(f PolicyFetcher) {
+	pm.impl.SetMDMLoader(loaderFor(f))
+}
+
+// ListProfiles returns all available profiles, including the default profile,
+// with their active status set.
 func (pm *ProfileManager) ListProfiles() (*ProfileArray, error) {
-	// Use ServiceManager (looks in profiles/ directory, checks active_profile.json for IsActive)
-	internalProfiles, err := pm.serviceMgr.ListProfiles(androidUsername)
+	profiles, err := pm.impl.ListProfiles()
 	if err != nil {
-		return nil, fmt.Errorf("failed to list profiles: %w", err)
+		return nil, err
 	}
 
-	// Convert internal profiles to Android Profile type
-	var profiles []*Profile
-	for _, p := range internalProfiles {
-		profiles = append(profiles, &Profile{
-			Name:     p.Name,
-			IsActive: p.IsActive,
-		})
+	items := make([]*Profile, 0, len(profiles))
+	for i := range profiles {
+		items = append(items, fromMobileProfile(&profiles[i]))
 	}
-
-	return &ProfileArray{items: profiles}, nil
+	return &ProfileArray{items: items}, nil
 }
 
-// GetActiveProfile returns the currently active profile name
-func (pm *ProfileManager) GetActiveProfile() (string, error) {
-	// Use ServiceManager to stay consistent with ListProfiles
-	// ServiceManager uses active_profile.json
-	activeState, err := pm.serviceMgr.GetActiveProfileState()
+// GetActiveProfile returns the currently active profile.
+func (pm *ProfileManager) GetActiveProfile() (*Profile, error) {
+	p, err := pm.impl.GetActiveProfile()
 	if err != nil {
-		return "", fmt.Errorf("failed to get active profile: %w", err)
+		return nil, err
 	}
-	return activeState.Name, nil
+	return fromMobileProfile(p), nil
 }
 
-// SwitchProfile switches to a different profile
-func (pm *ProfileManager) SwitchProfile(profileName string) error {
-	// Use ServiceManager to stay consistent with ListProfiles
-	// ServiceManager uses active_profile.json
-	err := pm.serviceMgr.SetActiveProfileState(&profilemanager.ActiveProfileState{
-		Name:     profileName,
-		Username: androidUsername,
-	})
-	if err != nil {
-		return fmt.Errorf("failed to switch profile: %w", err)
-	}
-
-	log.Infof("switched to profile: %s", profileName)
-	return nil
+// SwitchProfile records the given profile ID as the active profile. The caller
+// must stop the VPN tunnel before switching.
+func (pm *ProfileManager) SwitchProfile(id string) error {
+	return pm.impl.SwitchProfile(id)
 }
 
-// AddProfile creates a new profile
+// AddProfile creates a new profile with the given display name and a
+// generated ID.
 func (pm *ProfileManager) AddProfile(profileName string) error {
-	// Use ServiceManager (creates profile in profiles/ directory)
-	if err := pm.serviceMgr.AddProfile(profileName, androidUsername); err != nil {
-		return fmt.Errorf("failed to add profile: %w", err)
-	}
-
-	log.Infof("created new profile: %s", profileName)
-	return nil
+	_, err := pm.impl.AddProfile(profileName)
+	return err
 }
 
-// LogoutProfile logs out from a profile (clears authentication)
-func (pm *ProfileManager) LogoutProfile(profileName string) error {
-	profileName = sanitizeProfileName(profileName)
-
-	configPath, err := pm.getProfileConfigPath(profileName)
-	if err != nil {
-		return err
-	}
-
-	// Check if profile exists
-	if _, err := os.Stat(configPath); os.IsNotExist(err) {
-		return fmt.Errorf("profile '%s' does not exist", profileName)
-	}
-
-	// Read current config using internal profilemanager
-	config, err := profilemanager.ReadConfig(configPath)
-	if err != nil {
-		return fmt.Errorf("failed to read profile config: %w", err)
-	}
-
-	// Clear authentication by removing private key and SSH key
-	config.PrivateKey = ""
-	config.SSHKey = ""
-
-	// Save config using internal profilemanager
-	if err := profilemanager.WriteOutConfig(configPath, config); err != nil {
-		return fmt.Errorf("failed to save config: %w", err)
-	}
-
-	log.Infof("logged out from profile: %s", profileName)
-	return nil
+// RenameProfile changes the display name of the profile identified by id. The
+// on-disk filename (the ID) is left unchanged.
+func (pm *ProfileManager) RenameProfile(id string, newName string) error {
+	return pm.impl.RenameProfile(id, newName)
 }
 
-// RemoveProfile deletes a profile
-func (pm *ProfileManager) RemoveProfile(profileName string) error {
-	// Use ServiceManager (removes profile from profiles/ directory)
-	if err := pm.serviceMgr.RemoveProfile(profileName, androidUsername); err != nil {
-		return fmt.Errorf("failed to remove profile: %w", err)
-	}
-
-	log.Infof("removed profile: %s", profileName)
-	return nil
+// LogoutProfile clears authentication data for a profile, forcing a re-login.
+// The management URL and other settings are preserved.
+func (pm *ProfileManager) LogoutProfile(id string) error {
+	return pm.impl.LogoutProfile(id)
 }
 
-// getProfileConfigPath returns the config file path for a profile
-// This is needed for Android-specific path handling (netbird.cfg for default profile)
-func (pm *ProfileManager) getProfileConfigPath(profileName string) (string, error) {
-	if profileName == "" || profileName == profilemanager.DefaultProfileName {
-		// Android uses netbird.cfg for default profile instead of default.json
-		// Default profile is stored in root configDir, not in profiles/
-		return filepath.Join(pm.configDir, defaultConfigFilename), nil
-	}
-
-	// Non-default profiles are stored in profiles subdirectory
-	// This matches the Java Preferences.java expectation
-	profileName = sanitizeProfileName(profileName)
-	profilesDir := filepath.Join(pm.configDir, profilesSubdir)
-	return filepath.Join(profilesDir, profileName+".json"), nil
+// RemoveProfile deletes a profile. The default profile and the active profile
+// cannot be removed.
+func (pm *ProfileManager) RemoveProfile(id string) error {
+	return pm.impl.RemoveProfile(id)
 }
 
-// GetConfigPath returns the config file path for a given profile
-// Java should call this instead of constructing paths with Preferences.configFile()
-func (pm *ProfileManager) GetConfigPath(profileName string) (string, error) {
-	return pm.getProfileConfigPath(profileName)
+// GetConfigPath returns the config file path for the given profile ID. Java
+// should call this instead of constructing paths with Preferences.configFile().
+func (pm *ProfileManager) GetConfigPath(id string) (string, error) {
+	return pm.impl.GetConfigPath(id)
 }
 
-// GetStateFilePath returns the state file path for a given profile
-// Java should call this instead of constructing paths with Preferences.stateFile()
-func (pm *ProfileManager) GetStateFilePath(profileName string) (string, error) {
-	if profileName == "" || profileName == profilemanager.DefaultProfileName {
-		return filepath.Join(pm.configDir, "state.json"), nil
-	}
-
-	profileName = sanitizeProfileName(profileName)
-	profilesDir := filepath.Join(pm.configDir, profilesSubdir)
-	return filepath.Join(profilesDir, profileName+".state.json"), nil
+// GetStateFilePath returns the state file path for the given profile ID. Java
+// should call this instead of constructing paths with Preferences.stateFile().
+func (pm *ProfileManager) GetStateFilePath(id string) (string, error) {
+	return pm.impl.GetStateFilePath(id)
 }
 
-// GetActiveConfigPath returns the config file path for the currently active profile
-// Java should call this instead of Preferences.getActiveProfileName() + Preferences.configFile()
+// GetActiveConfigPath returns the config file path for the currently active
+// profile.
 func (pm *ProfileManager) GetActiveConfigPath() (string, error) {
-	activeProfile, err := pm.GetActiveProfile()
-	if err != nil {
-		return "", fmt.Errorf("failed to get active profile: %w", err)
-	}
-	return pm.GetConfigPath(activeProfile)
+	return pm.impl.GetActiveConfigPath()
 }
 
-// GetActiveStateFilePath returns the state file path for the currently active profile
-// Java should call this instead of Preferences.getActiveProfileName() + Preferences.stateFile()
+// GetActiveStateFilePath returns the state file path for the currently active
+// profile.
 func (pm *ProfileManager) GetActiveStateFilePath() (string, error) {
-	activeProfile, err := pm.GetActiveProfile()
-	if err != nil {
-		return "", fmt.Errorf("failed to get active profile: %w", err)
-	}
-	return pm.GetStateFilePath(activeProfile)
+	return pm.impl.GetActiveStateFilePath()
 }
 
-// sanitizeProfileName removes invalid characters from profile name
-func sanitizeProfileName(name string) string {
-	// Keep only alphanumeric, underscore, and hyphen
-	var result strings.Builder
-	for _, r := range name {
-		if (r >= 'a' && r <= 'z') || (r >= 'A' && r <= 'Z') ||
-			(r >= '0' && r <= '9') || r == '_' || r == '-' {
-			result.WriteRune(r)
-		}
-	}
-	return result.String()
+func fromMobileProfile(p *mobile.Profile) *Profile {
+	return &Profile{ID: p.ID, Name: p.Name, Email: p.Email, IsActive: p.IsActive}
 }

client/android/profile_prefs.go

@@ -0,0 +1,37 @@
+//go:build android
+
+package android
+
+import (
+	"fmt"
+
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+)
+
+type prefsStore interface {
+	Get(namespace string, v any) (bool, error)
+	Put(namespace string, v any) error
+}
+
+type profilePrefs struct {
+	prefs *profilemanager.Prefs
+}
+
+func newProfilePrefs(configDir, profileID string) (*profilePrefs, error) {
+	if configDir == "" || profileID == "" {
+		return nil, fmt.Errorf("profile prefs require a config dir and profile ID")
+	}
+	prefs, err := NewProfileManager(configDir).impl.ProfilePrefs(profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &profilePrefs{prefs: prefs}, nil
+}
+
+func (p *profilePrefs) Get(namespace string, v any) (bool, error) {
+	return p.prefs.Get(namespace, v)
+}
+
+func (p *profilePrefs) Put(namespace string, v any) error {
+	return p.prefs.Put(namespace, v)
+}

client/android/route_command.go

@@ -1,67 +0,0 @@
-//go:build android
-
-package android
-
-import (
-	"fmt"
-
-	log "github.com/sirupsen/logrus"
-	"golang.org/x/exp/maps"
-
-	"github.com/netbirdio/netbird/client/internal/routemanager"
-	"github.com/netbirdio/netbird/route"
-)
-
-func executeRouteToggle(id string, manager routemanager.Manager,
-	operationName string,
-	routeOperation func(routes []route.NetID, allRoutes []route.NetID) error) error {
-	netID := route.NetID(id)
-	routes := []route.NetID{netID}
-
-	log.Debugf("%s with id: %s", operationName, id)
-
-	if err := routeOperation(routes, maps.Keys(manager.GetClientRoutesWithNetID())); err != nil {
-		log.Debugf("error when %s: %s", operationName, err)
-		return fmt.Errorf("error %s: %w", operationName, err)
-	}
-
-	manager.TriggerSelection(manager.GetClientRoutes())
-
-	return nil
-}
-
-type routeCommand interface {
-	toggleRoute() error
-}
-
-type selectRouteCommand struct {
-	route   string
-	manager routemanager.Manager
-}
-
-func (s selectRouteCommand) toggleRoute() error {
-	routeSelector := s.manager.GetRouteSelector()
-	if routeSelector == nil {
-		return fmt.Errorf("no route selector available")
-	}
-
-	routeOperation := func(routes []route.NetID, allRoutes []route.NetID) error {
-		return routeSelector.SelectRoutes(routes, true, allRoutes)
-	}
-
-	return executeRouteToggle(s.route, s.manager, "selecting route", routeOperation)
-}
-
-type deselectRouteCommand struct {
-	route   string
-	manager routemanager.Manager
-}
-
-func (d deselectRouteCommand) toggleRoute() error {
-	routeSelector := d.manager.GetRouteSelector()
-	if routeSelector == nil {
-		return fmt.Errorf("no route selector available")
-	}
-
-	return executeRouteToggle(d.route, d.manager, "deselecting route", routeSelector.DeselectRoutes)
-}

client/android/session.go

@@ -0,0 +1,312 @@
+//go:build android
+
+package android
+
+import (
+	"context"
+	"fmt"
+
+	log "github.com/sirupsen/logrus"
+
+	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/client/internal/auth"
+	"github.com/netbirdio/netbird/client/internal/auth/sessionwatch"
+	"github.com/netbirdio/netbird/client/internal/peer"
+	cProto "github.com/netbirdio/netbird/client/proto"
+)
+
+// StateChangeListener receives client state notifications.
+//
+// OnStateChanged is a payload-free wake-up whenever the state snapshot
+// changed: connection state, the run-loop status label (e.g. NeedsLogin) or
+// the session deadline. It mirrors the daemon's SubscribeStatus stream
+// trigger — on each signal the consumer pulls the fresh values via
+// Status() / SessionExpiresAtUnix().
+//
+// OnSessionExpiring forwards the engine's session-expiry warnings, fired at
+// sessionwatch.WarningLead before the deadline and again at FinalWarningLead
+// (finalWarning true). The second one is suppressed when the user dismissed
+// the first via DismissSessionWarning. The daemon turns the same events into
+// its tray notification.
+type StateChangeListener interface {
+	OnStateChanged()
+	OnSessionExpiring(expiresAtUnix int64, leadMinutes int64, finalWarning bool)
+}
+
+// Status returns the connect run-loop's status label — the same value the
+// desktop daemon serves in StatusResponse.Status. "NeedsLogin" means the
+// management server rejected the peer and an interactive login is required.
+//
+// The label is latched: the run loop keeps its status in a per-run context
+// state, which a restart replaces with a fresh Idle one, so an engine restart
+// (network change, always-on) would otherwise erase the fact that the peer
+// still needs to log in. Only a successful interactive login or extend clears
+// it — see clearLoginRequired.
+func (c *Client) Status() string {
+	latched, generation := c.loginRequiredState()
+	if latched {
+		return string(internal.StatusNeedsLogin)
+	}
+	cc := c.getConnectClient()
+	if cc == nil {
+		return string(internal.StatusIdle)
+	}
+	status := cc.Status()
+	if status == internal.StatusNeedsLogin {
+		c.latchLoginRequired(generation)
+	}
+	return string(status)
+}
+
+func (c *Client) loginRequiredState() (bool, uint64) {
+	c.loginRequiredMu.Lock()
+	defer c.loginRequiredMu.Unlock()
+	return c.loginRequired, c.loginCleared
+}
+
+// latchLoginRequired records a NeedsLogin observation, unless a clear landed
+// while the caller was reading the run loop's status: cc.Status() is read
+// outside the lock, so a login or extend completing in that window would
+// otherwise be undone by this stale observation, stranding the UI on
+// "login required" over a healthy session.
+func (c *Client) latchLoginRequired(observedGeneration uint64) {
+	c.loginRequiredMu.Lock()
+	defer c.loginRequiredMu.Unlock()
+	if c.loginCleared != observedGeneration {
+		return
+	}
+	c.loginRequired = true
+}
+
+// clearLoginRequired releases the latch after a successful interactive login
+// or session extend, and invalidates any observation already in flight.
+func (c *Client) clearLoginRequired() {
+	c.loginRequiredMu.Lock()
+	defer c.loginRequiredMu.Unlock()
+	c.loginRequired = false
+	c.loginCleared++
+}
+
+// SessionExpiresAtUnix returns the SSO session deadline as unix seconds, or 0
+// when no deadline is known (not SSO-registered, expiry disabled, or the
+// engine has not received one yet). A past value means the session expired.
+// Mirror of StatusResponse.sessionExpiresAt on the desktop daemon.
+func (c *Client) SessionExpiresAtUnix() int64 {
+	deadline := c.recorder.GetSessionExpiresAt()
+	if deadline.IsZero() {
+		return 0
+	}
+	return deadline.Unix()
+}
+
+// SetStateChangeListener registers the state notification listener.
+// Replaces any previously registered listener; remove it with
+// RemoveStateChangeListener.
+func (c *Client) SetStateChangeListener(listener StateChangeListener) {
+	c.stateChangeMu.Lock()
+	defer c.stateChangeMu.Unlock()
+	c.stopStateChangeWatchLocked()
+	if listener == nil {
+		return
+	}
+
+	// Both subscriptions are buffered (one pending tick, ten pending events),
+	// so unsubscribing is not enough to stop callbacks: the loops would drain
+	// what is already queued and deliver it to a listener the caller has
+	// already removed or replaced. Gate every callback on this registration's
+	// own signal, which is closed before unsubscribing.
+	done := make(chan struct{})
+	c.stateChangeDone = done
+
+	id, ch := c.recorder.SubscribeToStateChanges()
+	c.stateChangeSubID = id
+	// The channel is closed by UnsubscribeFromStateChanges, which ends the
+	// goroutine. Ticks are coalesced (buffer of one), so a burst of changes
+	// wakes the listener once.
+	go func() {
+		for range ch {
+			select {
+			case <-done:
+				return
+			default:
+			}
+			listener.OnStateChanged()
+		}
+	}()
+
+	c.eventSub = c.recorder.SubscribeToEvents()
+	go watchSessionWarnings(c.eventSub, listener, done)
+}
+
+// RemoveStateChangeListener unregisters the state notification listener.
+func (c *Client) RemoveStateChangeListener() {
+	c.stateChangeMu.Lock()
+	defer c.stateChangeMu.Unlock()
+	c.stopStateChangeWatchLocked()
+}
+
+// DismissSessionWarning records the user's "Dismiss" on the first expiry
+// warning and suppresses the final one for the current deadline. A refreshed
+// deadline re-arms both. No-op while the engine is not running.
+func (c *Client) DismissSessionWarning() {
+	cc := c.getConnectClient()
+	if cc == nil {
+		return
+	}
+	engine := cc.Engine()
+	if engine == nil {
+		return
+	}
+	engine.DismissSessionWarning()
+}
+
+// ExtendAuthSession runs the interactive SSO flow to obtain a fresh JWT and
+// asks the management server to extend the session deadline. The tunnel is
+// untouched: no resync, no reconnect. Async; the result arrives on the
+// listener. Mirror of the daemon's RequestExtendAuthSession /
+// WaitExtendAuthSession RPC pair, with URLOpener playing the "UI opens the
+// browser" role.
+//
+// Only one flow may be in flight: the PKCE step binds a fixed loopback port,
+// so a second concurrent flow would fail on that bind. Call
+// CancelExtendAuthSession when the user abandons the browser.
+func (c *Client) ExtendAuthSession(urlOpener URLOpener, isAndroidTV bool, resultListener ErrListener) {
+	ctx, err := c.beginExtend()
+	if err != nil {
+		resultListener.OnError(err)
+		return
+	}
+
+	go func() {
+		defer c.endExtend()
+		if err := c.extendAuthSession(ctx, urlOpener, isAndroidTV); err != nil {
+			resultListener.OnError(err)
+			return
+		}
+		resultListener.OnSuccess()
+	}()
+}
+
+// CancelExtendAuthSession aborts an in-flight ExtendAuthSession. The tunnel is
+// left alone — unlike the login flow, which cancels the whole client context
+// by stopping the engine. Without this the abandoned PKCE wait keeps its
+// loopback port for the full flow timeout and blocks every later attempt.
+// No-op when no flow is running.
+func (c *Client) CancelExtendAuthSession() {
+	c.extendMu.Lock()
+	defer c.extendMu.Unlock()
+	if c.extendCancel != nil {
+		c.extendCancel()
+	}
+}
+
+func (c *Client) stopStateChangeWatchLocked() {
+	// Signal first, unsubscribe second: closing the channels only stops new
+	// items, and the loops would still hand whatever is buffered to a listener
+	// that is no longer registered.
+	if c.stateChangeDone != nil {
+		close(c.stateChangeDone)
+		c.stateChangeDone = nil
+	}
+	if c.stateChangeSubID != "" {
+		c.recorder.UnsubscribeFromStateChanges(c.stateChangeSubID)
+		c.stateChangeSubID = ""
+	}
+	if c.eventSub != nil {
+		// Closes the channel, which ends watchSessionWarnings.
+		c.recorder.UnsubscribeFromEvents(c.eventSub)
+		c.eventSub = nil
+	}
+}
+
+// watchSessionWarnings forwards the engine's session-expiry warnings to the
+// listener. The event stream also carries unrelated traffic — network-map
+// updates on every sync, DNS and route errors — so everything but an
+// AUTHENTICATION event carrying the session-warning marker is dropped. Exits
+// when the subscription is closed by UnsubscribeFromEvents, or earlier when
+// done is closed — the stream buffers up to ten events, and a deregistered
+// listener must not receive the ones already queued.
+func watchSessionWarnings(sub *peer.EventSubscription, listener StateChangeListener, done <-chan struct{}) {
+	for ev := range sub.Events() {
+		select {
+		case <-done:
+			return
+		default:
+		}
+		if ev.GetCategory() != cProto.SystemEvent_AUTHENTICATION {
+			continue
+		}
+		meta := ev.GetMetadata()
+		if meta[sessionwatch.MetaSessionWarning] != "true" {
+			// Other AUTHENTICATION events exist (e.g. a deadline rejected as
+			// out of range); they carry no warning marker.
+			continue
+		}
+		deadline, err := sessionwatch.ParseExpiresAt(meta[sessionwatch.MetaSessionExpiresAt])
+		if err != nil {
+			log.Warnf("session warning event with unparsable deadline: %v", err)
+			continue
+		}
+		lead, err := sessionwatch.ParseLeadMinutes(meta[sessionwatch.MetaSessionLeadMinutes])
+		if err != nil {
+			// Informational only — the deadline above is what drives the UI.
+			lead = 0
+		}
+		listener.OnSessionExpiring(deadline.Unix(), int64(lead),
+			meta[sessionwatch.MetaSessionFinal] == "true")
+	}
+}
+
+func (c *Client) beginExtend() (context.Context, error) {
+	c.extendMu.Lock()
+	defer c.extendMu.Unlock()
+	if c.extendCancel != nil {
+		return nil, fmt.Errorf("session extend already in progress")
+	}
+	ctx, cancel := context.WithCancel(context.Background())
+	c.extendCancel = cancel
+	return ctx, nil
+}
+
+func (c *Client) endExtend() {
+	c.extendMu.Lock()
+	defer c.extendMu.Unlock()
+	if c.extendCancel != nil {
+		c.extendCancel()
+		c.extendCancel = nil
+	}
+}
+
+func (c *Client) extendAuthSession(ctx context.Context, urlOpener URLOpener, isAndroidTV bool) error {
+	cfg, cfgPath, cc := c.authSnapshot()
+	if cfg == nil || cc == nil {
+		return fmt.Errorf("engine is not running")
+	}
+	engine := cc.Engine()
+	if engine == nil {
+		return fmt.Errorf("engine is not initialized")
+	}
+
+	authClient, err := auth.NewAuth(ctx, cfg.PrivateKey, cfg.ManagementURL, cfg)
+	if err != nil {
+		return fmt.Errorf("failed to create auth client: %v", err)
+	}
+	defer authClient.Close()
+
+	// Passing the config path makes the flow pick up the login_hint: an extend
+	// renews the session of the account already signed in, so it must not stop to
+	// offer a choice.
+	a := NewAuthWithConfig(ctx, cfg, cfgPath)
+	tokenInfo, err := a.foregroundGetTokenInfo(authClient, urlOpener, isAndroidTV)
+	if err != nil {
+		return fmt.Errorf("interactive sso login failed: %v", err)
+	}
+
+	if _, err := engine.ExtendAuthSession(ctx, tokenInfo.GetTokenToUse()); err != nil {
+		return err
+	}
+	c.clearLoginRequired()
+
+	go urlOpener.OnLoginSuccess()
+	return nil
+}

client/android/split_tunnel.go

@@ -0,0 +1,125 @@
+package android
+
+// SplitTunnelMode is which of the two selections, if either, the tunnel applies.
+// Its values land in the profile's stored preferences, so the constants below
+// are append-only and must never be reordered.
+type SplitTunnelMode int
+
+const (
+	modeOff SplitTunnelMode = iota
+	modeExclude
+	modeInclude
+)
+
+// The same modes as basic ints. gomobile drops a constant whose type is not a
+// basic one, so these are what reaches the generated Java bindings, and they
+// keep the Android side tied to the values above instead of repeating 0, 1, 2.
+const (
+	SplitTunnelModeOff     = int(modeOff)
+	SplitTunnelModeExclude = int(modeExclude)
+	SplitTunnelModeInclude = int(modeInclude)
+)
+
+type splitTunnelSection struct {
+	Mode     SplitTunnelMode `json:"mode"`
+	Excluded []string        `json:"excluded"`
+	Included []string        `json:"included"`
+}
+
+// PackageList wraps []string for gomobile compatibility.
+type PackageList struct {
+	items []string
+}
+
+// NewPackageList creates an empty list to fill via Add.
+func NewPackageList() *PackageList {
+	return &PackageList{}
+}
+
+// Add appends a package name, ignoring empty ones.
+func (l *PackageList) Add(s string) {
+	if s == "" {
+		return
+	}
+	l.items = append(l.items, s)
+}
+
+// Size returns the number of entries.
+func (l *PackageList) Size() int {
+	return len(l.items)
+}
+
+// Get returns the entry at index i, or an empty string when out of range.
+func (l *PackageList) Get(i int) string {
+	if i < 0 || i >= len(l.items) {
+		return ""
+	}
+	return l.items[i]
+}
+
+// SplitTunnelSettings is one profile's choice of which applications the tunnel
+// carries. The two selections are kept apart because the platform applies one
+// or the other and never both, and so that switching mode does not throw away
+// the picks made in the other one.
+//
+// Mode is an int rather than a SplitTunnelMode because gomobile carries only
+// basic types across the binding. It holds one of the SplitTunnelMode*
+// constants.
+type SplitTunnelSettings struct {
+	Mode     int
+	Excluded *PackageList
+	Included *PackageList
+}
+
+// NewSplitTunnelSettings creates settings that carry every application.
+func NewSplitTunnelSettings() *SplitTunnelSettings {
+	return &SplitTunnelSettings{
+		Mode:     SplitTunnelModeOff,
+		Excluded: NewPackageList(),
+		Included: NewPackageList(),
+	}
+}
+
+func packagesOf(list *PackageList) []string {
+	if list == nil {
+		return nil
+	}
+	out := make([]string, 0, len(list.items))
+	out = append(out, list.items...)
+	return out
+}
+
+// normalizeSplitTunnelMode maps anything outside the known set to off, so a mode
+// written by a newer build degrades to carrying every application rather than to
+// some other mode's behaviour.
+func normalizeSplitTunnelMode(mode SplitTunnelMode) SplitTunnelMode {
+	switch mode {
+	case modeExclude, modeInclude:
+		return mode
+	default:
+		return modeOff
+	}
+}
+
+func settingsFromSection(section splitTunnelSection) *SplitTunnelSettings {
+	out := NewSplitTunnelSettings()
+	out.Mode = int(normalizeSplitTunnelMode(section.Mode))
+	for _, pkg := range section.Excluded {
+		out.Excluded.Add(pkg)
+	}
+	for _, pkg := range section.Included {
+		out.Included.Add(pkg)
+	}
+	return out
+}
+
+func sectionFromSettings(settings *SplitTunnelSettings) splitTunnelSection {
+	if settings == nil {
+		settings = NewSplitTunnelSettings()
+	}
+	return splitTunnelSection{
+		Mode:     normalizeSplitTunnelMode(SplitTunnelMode(settings.Mode)),
+		Excluded: packagesOf(settings.Excluded),
+		Included: packagesOf(settings.Included),
+	}
+}

client/android/split_tunnel_store.go

@@ -0,0 +1,34 @@
+//go:build android
+
+package android
+
+const splitTunnelNamespace = "split-tunnel"
+
+// SplitTunnelStore reads and writes a profile's split tunnelling settings.
+type SplitTunnelStore struct {
+	prefs prefsStore
+}
+
+// NewSplitTunnelStore opens the split tunnelling store of the given profile.
+func NewSplitTunnelStore(configDir, profileID string) (*SplitTunnelStore, error) {
+	prefs, err := newProfilePrefs(configDir, profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &SplitTunnelStore{prefs: prefs}, nil
+}
+
+// Load returns the stored settings, or settings that carry every application
+// when the profile has none saved.
+func (s *SplitTunnelStore) Load() (*SplitTunnelSettings, error) {
+	var section splitTunnelSection
+	if _, err := s.prefs.Get(splitTunnelNamespace, &section); err != nil {
+		return nil, err
+	}
+	return settingsFromSection(section), nil
+}
+
+// Save replaces the stored settings.
+func (s *SplitTunnelStore) Save(settings *SplitTunnelSettings) error {
+	return s.prefs.Put(splitTunnelNamespace, sectionFromSettings(settings))
+}

client/android/split_tunnel_test.go

@@ -0,0 +1,151 @@
+package android
+
+import (
+	"encoding/json"
+	"reflect"
+	"testing"
+)
+
+func TestNormalizeSplitTunnelMode(t *testing.T) {
+	tests := []struct {
+		name string
+		mode SplitTunnelMode
+		want SplitTunnelMode
+	}{
+		{name: "exclude is kept", mode: modeExclude, want: modeExclude},
+		{name: "include is kept", mode: modeInclude, want: modeInclude},
+		{name: "off is kept", mode: modeOff, want: modeOff},
+		{name: "a mode from a newer build falls back to off", mode: SplitTunnelMode(7), want: modeOff},
+		{name: "a negative mode falls back to off", mode: SplitTunnelMode(-1), want: modeOff},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := normalizeSplitTunnelMode(tt.mode); got != tt.want {
+				t.Errorf("normalizeSplitTunnelMode(%d) = %d, want %d", tt.mode, got, tt.want)
+			}
+		})
+	}
+}
+
+// The constants the Android side reads must stay the values the store writes:
+// gomobile carries the ints below, not the typed constants they mirror.
+func TestSplitTunnelModeConstantsMirrorTheTypedOnes(t *testing.T) {
+	if SplitTunnelModeOff != int(modeOff) {
+		t.Errorf("off = %d, want %d", SplitTunnelModeOff, modeOff)
+	}
+	if SplitTunnelModeExclude != int(modeExclude) {
+		t.Errorf("exclude = %d, want %d", SplitTunnelModeExclude, modeExclude)
+	}
+	if SplitTunnelModeInclude != int(modeInclude) {
+		t.Errorf("include = %d, want %d", SplitTunnelModeInclude, modeInclude)
+	}
+}
+
+func TestSettingsFromSection(t *testing.T) {
+	got := settingsFromSection(splitTunnelSection{
+		Mode:     modeExclude,
+		Excluded: []string{"com.example.a", "com.example.b"},
+		Included: []string{"com.example.c"},
+	})
+
+	if got.Mode != SplitTunnelModeExclude {
+		t.Errorf("mode = %d, want %d", got.Mode, SplitTunnelModeExclude)
+	}
+	if got.Excluded.Size() != 2 || got.Excluded.Get(0) != "com.example.a" {
+		t.Errorf("excluded = %v, want the two stored packages", packagesOf(got.Excluded))
+	}
+	if got.Included.Size() != 1 || got.Included.Get(0) != "com.example.c" {
+		t.Errorf("included = %v, want the stored package", packagesOf(got.Included))
+	}
+}
+
+// A profile that has never stored anything decodes into an empty section, and
+// must come back as settings that carry every application rather than as nil
+// lists the caller would have to guard against.
+func TestSettingsFromEmptySectionCarriesEverything(t *testing.T) {
+	got := settingsFromSection(splitTunnelSection{})
+
+	if got.Mode != SplitTunnelModeOff {
+		t.Errorf("mode = %d, want %d", got.Mode, SplitTunnelModeOff)
+	}
+	if got.Excluded == nil || got.Included == nil {
+		t.Fatal("both selections must be usable lists, not nil")
+	}
+	if got.Excluded.Size() != 0 || got.Included.Size() != 0 {
+		t.Errorf("selections = %v/%v, want both empty", packagesOf(got.Excluded), packagesOf(got.Included))
+	}
+}
+
+// The section is what the profile's preference file holds, so the mode has to
+// survive a JSON round trip as the number the constants name.
+func TestSectionEncodesTheModeAsItsNumber(t *testing.T) {
+	raw, err := json.Marshal(sectionFromSettings(&SplitTunnelSettings{Mode: SplitTunnelModeInclude}))
+	if err != nil {
+		t.Fatalf("marshal section: %v", err)
+	}
+
+	var back splitTunnelSection
+	if err := json.Unmarshal(raw, &back); err != nil {
+		t.Fatalf("unmarshal section: %v", err)
+	}
+	if back.Mode != modeInclude {
+		t.Errorf("mode = %d, want %d, from %s", back.Mode, modeInclude, raw)
+	}
+}
+
+func TestSectionFromSettingsRoundTrip(t *testing.T) {
+	settings := NewSplitTunnelSettings()
+	settings.Mode = SplitTunnelModeInclude
+	settings.Included.Add("com.example.a")
+	settings.Excluded.Add("com.example.b")
+
+	section := sectionFromSettings(settings)
+	back := settingsFromSection(section)
+
+	if back.Mode != SplitTunnelModeInclude {
+		t.Errorf("mode = %d, want %d", back.Mode, SplitTunnelModeInclude)
+	}
+	if !reflect.DeepEqual(packagesOf(back.Included), []string{"com.example.a"}) {
+		t.Errorf("included = %v, want [com.example.a]", packagesOf(back.Included))
+	}
+	// The inactive selection survives, so switching mode back does not make the
+	// user pick their applications again.
+	if !reflect.DeepEqual(packagesOf(back.Excluded), []string{"com.example.b"}) {
+		t.Errorf("excluded = %v, want [com.example.b]", packagesOf(back.Excluded))
+	}
+}
+
+// A mode the Java side never sets, such as one left by a newer build, must not
+// reach the stored section either.
+func TestSectionFromSettingsNormalizesAnUnknownMode(t *testing.T) {
+	section := sectionFromSettings(&SplitTunnelSettings{Mode: 7})
+
+	if section.Mode != modeOff {
+		t.Errorf("mode = %d, want %d", section.Mode, modeOff)
+	}
+}
+
+func TestSectionFromNilSettings(t *testing.T) {
+	section := sectionFromSettings(nil)
+
+	if section.Mode != modeOff {
+		t.Errorf("mode = %d, want %d", section.Mode, modeOff)
+	}
+	if len(section.Excluded) != 0 || len(section.Included) != 0 {
+		t.Errorf("selections = %v/%v, want both empty", section.Excluded, section.Included)
+	}
+}
+
+func TestPackageListIgnoresEmptyAndBounds(t *testing.T) {
+	list := NewPackageList()
+	list.Add("com.example.a")
+	list.Add("")
+
+	if list.Size() != 1 {
+		t.Errorf("size = %d, want 1", list.Size())
+	}
+	if list.Get(-1) != "" || list.Get(5) != "" {
+		t.Error("out of range access must return an empty string")
+	}
+}

client/android/ssh_client.go

@@ -0,0 +1,649 @@
+//go:build android
+
+package android
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	log "github.com/sirupsen/logrus"
+	gossh "golang.org/x/crypto/ssh"
+
+	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/client/internal/auth"
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	nbssh "github.com/netbirdio/netbird/client/ssh"
+	"github.com/netbirdio/netbird/client/ssh/detection"
+)
+
+const (
+	sshDialTimeout      = 30 * time.Second
+	sshDetectionTimeout = 5 * time.Second
+)
+
+// PasswordRequiredMarker tells Java to prompt for a password and retry. It is
+// a string because gomobile flattens errors to their message, so a sentinel
+// value would not survive the binding.
+const PasswordRequiredMarker = "netbird-ssh-password-required"
+
+// HostKeyUnknownMarker tells Java to show the fingerprint and, on confirmation,
+// retry with TrustHostKey set. The presented fingerprint is appended after the
+// marker so the prompt can display it and the retry can guard against a key
+// that changed between the two connects. Only regular (non-NetBird) servers
+// reach this: NetBird peers verify against the registry.
+const HostKeyUnknownMarker = "netbird-ssh-hostkey-unknown"
+
+var (
+	errPasswordRequired = errors.New(PasswordRequiredMarker)
+	errClientClosed     = errors.New("ssh client closed")
+)
+
+// errHostKeyUnknown carries the presented fingerprint so Connect can build the
+// marker message the Java side parses.
+type errHostKeyUnknown struct {
+	fingerprint string
+}
+
+func (e *errHostKeyUnknown) Error() string {
+	return HostKeyUnknownMarker + ":" + e.fingerprint
+}
+
+// SSHTerminalListener receives SSH session events. It is implemented in Java.
+//
+// All callbacks are invoked from goroutines and may run concurrently with each
+// other; the implementation must be safe to call from any thread.
+type SSHTerminalListener interface {
+	OnConnected()
+	OnData(data []byte)
+	OnClose(reason string)
+	OnError(message string)
+}
+
+// SSHClient is a NetBird-aware SSH client exposed to Java via gomobile.
+//
+// It dials through the running NetBird tunnel and runs a standard SSH session
+// on top with PTY enabled. Host-key verification uses the NetBird-provided
+// peer SSH host keys, identical to the desktop client.
+type SSHClient struct {
+	nb        *Client
+	mu        sync.Mutex
+	listener  SSHTerminalListener
+	urlOpener URLOpener
+
+	sshClient *gossh.Client
+	session   *gossh.Session
+	stdin     io.WriteCloser
+	closed    bool
+
+	// gen identifies the current connection attempt. Connect and Close bump it,
+	// so an in-flight dial or a reader left over from a previous connection
+	// finds itself stale and stays silent instead of publishing OnConnected or
+	// OnClose for a connection the caller already abandoned.
+	gen        uint64
+	dialCancel context.CancelFunc
+
+	// knownHostsConfigDir and knownHostsProfile locate the TOFU store for
+	// regular SSH servers in the profile's preferences. Java supplies them,
+	// since an overlay IP is a different host under a different profile. Empty
+	// until set: without them a regular server cannot be verified and Connect
+	// refuses one.
+	knownHostsConfigDir string
+	knownHostsProfile   string
+	// trustHostKey carries the fingerprint the user confirmed on a previous
+	// attempt, so the retry accepts exactly that key and persists it.
+	trustHostKey string
+}
+
+// NewSSHClient creates a new SSH client bound to the running NetBird Client.
+func NewSSHClient(c *Client) *SSHClient {
+	return &SSHClient{nb: c}
+}
+
+// SetListener registers the Java listener. Must be called before Connect to
+// receive any events.
+func (s *SSHClient) SetListener(l SSHTerminalListener) {
+	s.mu.Lock()
+	s.listener = l
+	s.mu.Unlock()
+}
+
+// SetURLOpener registers the Java URL opener used to display the device-code
+// authorization page in a Custom Tabs window when the target peer requires
+// JWT authentication. Must be set before Connect to be effective.
+func (s *SSHClient) SetURLOpener(opener URLOpener) {
+	s.mu.Lock()
+	s.urlOpener = opener
+	s.mu.Unlock()
+}
+
+// SetKnownHostsStore points the TOFU host-key store at a profile's preferences.
+// Must be set before connecting to a regular SSH server; without it such a
+// server cannot be verified and Connect refuses one.
+func (s *SSHClient) SetKnownHostsStore(configDir, profileID string) {
+	s.mu.Lock()
+	s.knownHostsConfigDir = configDir
+	s.knownHostsProfile = profileID
+	s.mu.Unlock()
+}
+
+// TrustHostKey records the fingerprint the user confirmed for a regular server,
+// so the next Connect accepts that exact key and adds it to the known-hosts
+// store. Passing a fingerprint that no longer matches makes the connect fail
+// rather than trust a key that changed since the prompt.
+func (s *SSHClient) TrustHostKey(fingerprint string) {
+	s.mu.Lock()
+	s.trustHostKey = fingerprint
+	s.mu.Unlock()
+}
+
+// Connect dials the SSH server through the NetBird tunnel and performs the
+// SSH handshake. It auto-detects the server type via SSH banner inspection
+// and selects the appropriate authentication path:
+//
+//   - NetBird-SSH server requiring JWT: launches the OAuth 2.0 device-code
+//     flow, opens the verification URL through the registered URLOpener, and
+//     uses the resulting token as the SSH password. Host-key verification
+//     uses the NetBird peer registry.
+//   - NetBird-SSH server without JWT: authenticates with the NetBird SSH
+//     private key. Host-key verification uses the NetBird peer registry.
+//   - Regular SSH server (e.g. OpenSSH): authenticates with the NetBird key
+//     first (so a user-installed NetBird public key works), then falls back
+//     to the supplied password if non-empty. Host-key verification is
+//     trust-on-first-use against the per-profile known-hosts store.
+//
+// The password parameter is only consulted for regular SSH servers.
+func (s *SSHClient) Connect(host string, port int, user, password string) error {
+	if port < 1 || port > 65535 {
+		return fmt.Errorf("invalid port: %d", port)
+	}
+
+	cfg, cfgPath, cc := s.nb.authSnapshot()
+	if cc == nil {
+		return errors.New("netbird client not running")
+	}
+	if cfg == nil {
+		return errors.New("netbird config not loaded")
+	}
+	engine := cc.Engine()
+	if engine == nil {
+		return errors.New("netbird engine not available")
+	}
+
+	s.mu.Lock()
+	s.gen++
+	gen := s.gen
+	s.mu.Unlock()
+
+	serverType := detectServerType(host, port)
+	log.Debugf("SSH server type: %s", serverType)
+
+	authMethods, hostKeyCallback, err := s.buildAuth(cfg, cfgPath, engine, serverType, password)
+	if err != nil {
+		return err
+	}
+
+	clientConfig := &gossh.ClientConfig{
+		User:            user,
+		Auth:            authMethods,
+		HostKeyCallback: hostKeyCallback,
+		Timeout:         sshDialTimeout,
+	}
+	err = s.dialAndHandshake(gen, host, port, clientConfig)
+
+	// An unknown host key is a prompt, not a failure: return the marker intact
+	// (rootCause would unwrap it) so Java can show the fingerprint and retry.
+	var unknownHost *errHostKeyUnknown
+	if errors.As(err, &unknownHost) {
+		return errors.New(unknownHost.Error())
+	}
+
+	// A regular server may still accept a password, so let the caller ask for
+	// one instead of failing. NetBird servers never use a password, so a
+	// failure there is genuine.
+	if err != nil && serverType != detection.ServerTypeNetBirdJWT &&
+		serverType != detection.ServerTypeNetBirdNoJWT && isAuthFailure(err) &&
+		passwordCouldHelp(err, password != "") {
+		return errPasswordRequired
+	}
+	if err != nil {
+		return rootCause(err)
+	}
+	return nil
+}
+
+// StartSession requests a PTY and starts an interactive shell. Output from
+// the session is forwarded to the listener via OnData.
+func (s *SSHClient) StartSession(cols, rows int) error {
+	err := s.startSession(cols, rows)
+	if err != nil {
+		log.Infof("SSH: start session failed: %v", err)
+		return rootCause(err)
+	}
+	return nil
+}
+
+// Write sends data to the SSH session stdin.
+func (s *SSHClient) Write(data []byte) error {
+	s.mu.Lock()
+	stdin := s.stdin
+	s.mu.Unlock()
+	if stdin == nil {
+		return errors.New("ssh session not started")
+	}
+	if _, err := stdin.Write(data); err != nil {
+		return fmt.Errorf("write stdin: %w", err)
+	}
+	return nil
+}
+
+// Resize updates the PTY window size.
+func (s *SSHClient) Resize(cols, rows int) error {
+	s.mu.Lock()
+	session := s.session
+	s.mu.Unlock()
+	if session == nil {
+		return errors.New("ssh session not started")
+	}
+	return session.WindowChange(rows, cols)
+}
+
+// Reset makes a closed client usable for another Connect: Close leaves the
+// one-shot guard set, and clearing it lets the same client back a reconnect.
+func (s *SSHClient) Reset() {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	s.closed = false
+}
+
+// Close terminates the SSH session and underlying connection. Safe to call
+// multiple times.
+func (s *SSHClient) Close() error {
+	s.mu.Lock()
+	s.gen++
+	if s.dialCancel != nil {
+		s.dialCancel()
+		s.dialCancel = nil
+	}
+	sshClient := s.sshClient
+	session := s.session
+	stdin := s.stdin
+	s.sshClient = nil
+	s.session = nil
+	s.stdin = nil
+	notify := !s.closed
+	s.closed = true
+	listener := s.listener
+	s.mu.Unlock()
+
+	if stdin != nil {
+		if err := stdin.Close(); err != nil {
+			log.Debugf("ssh: stdin close: %v", err)
+		}
+	}
+	if session != nil {
+		if err := session.Close(); err != nil && !errors.Is(err, io.EOF) {
+			log.Debugf("ssh: session close: %v", err)
+		}
+	}
+	var firstErr error
+	if sshClient != nil {
+		if err := sshClient.Close(); err != nil {
+			firstErr = err
+		}
+	}
+	if notify && listener != nil {
+		listener.OnClose("closed by client")
+	}
+	return firstErr
+}
+
+func (s *SSHClient) startSession(cols, rows int) error {
+	log.Debugf("SSH: starting session %dx%d", cols, rows)
+	s.mu.Lock()
+	sshClient := s.sshClient
+	gen := s.gen
+	s.mu.Unlock()
+
+	if sshClient == nil {
+		return errors.New("ssh client not connected")
+	}
+
+	pty, err := nbssh.StartPTYSession(sshClient, cols, rows)
+	if err != nil {
+		return err
+	}
+
+	s.mu.Lock()
+	if gen != s.gen {
+		s.mu.Unlock()
+		closeQuiet(pty.Session, "stale session")
+		return errClientClosed
+	}
+	s.session = pty.Session
+	s.stdin = pty.Stdin
+	s.mu.Unlock()
+
+	readerDone := make(chan string, 2)
+	go func() { readerDone <- s.readLoop(pty.Stdout, "stdout") }()
+	go func() { readerDone <- s.readLoop(pty.Stderr, "stderr") }()
+	go func() {
+		reason := <-readerDone
+		if second := <-readerDone; reason == "" {
+			reason = second
+		}
+		s.notifyClose(gen, reason)
+	}()
+	log.Debug("SSH: session started, shell running")
+	return nil
+}
+
+func (s *SSHClient) buildAuth(cfg *profilemanager.Config, cfgPath string, engine *internal.Engine,
+	serverType detection.ServerType, password string) ([]gossh.AuthMethod, gossh.HostKeyCallback, error) {
+
+	switch serverType {
+	case detection.ServerTypeNetBirdJWT:
+		token, err := s.requestJWTToken(cfg, cfgPath)
+		if err != nil {
+			return nil, nil, fmt.Errorf("jwt: %w", err)
+		}
+		auths := []gossh.AuthMethod{gossh.Password(token)}
+		return auths, nbssh.CreateHostKeyCallback(nbssh.PeerKeyLookup(engine.GetPeerSSHKey)), nil
+
+	case detection.ServerTypeNetBirdNoJWT:
+		if cfg.SSHKey == "" {
+			return nil, nil, errors.New("no NetBird SSH key available")
+		}
+		signer, err := gossh.ParsePrivateKey([]byte(cfg.SSHKey))
+		if err != nil {
+			return nil, nil, fmt.Errorf("parse netbird ssh key: %w", err)
+		}
+		auths := []gossh.AuthMethod{gossh.PublicKeys(signer)}
+		return auths, nbssh.CreateHostKeyCallback(nbssh.PeerKeyLookup(engine.GetPeerSSHKey)), nil
+
+	case detection.ServerTypeRegular:
+		var auths []gossh.AuthMethod
+		if cfg.SSHKey != "" {
+			if signer, err := gossh.ParsePrivateKey([]byte(cfg.SSHKey)); err == nil {
+				auths = append(auths, gossh.PublicKeys(signer))
+			} else {
+				log.Debugf("ssh: parse netbird key for regular auth: %v", err)
+			}
+		}
+		if password != "" {
+			pw := password
+			auths = append(auths, gossh.Password(pw))
+			auths = append(auths, gossh.KeyboardInteractive(func(_, _ string, questions []string, _ []bool) ([]string, error) {
+				answers := make([]string, len(questions))
+				for i := range questions {
+					answers[i] = pw
+				}
+				return answers, nil
+			}))
+		}
+		if len(auths) == 0 {
+			// Nothing to offer at all: ask for a password rather than failing,
+			// so the caller can retry once the user supplies one.
+			return nil, nil, errPasswordRequired
+		}
+		callback, err := s.tofuHostKeyCallback()
+		if err != nil {
+			return nil, nil, err
+		}
+		return auths, callback, nil
+
+	default:
+		return nil, nil, fmt.Errorf("unsupported SSH server type: %v", serverType)
+	}
+}
+
+// tofuHostKeyCallback verifies a regular server's host key against the
+// per-profile known-hosts store. An unknown host returns errHostKeyUnknown so
+// Java can show the fingerprint and, once confirmed, retry with the key
+// trusted; a changed key is rejected outright, as OpenSSH does. When the user
+// has confirmed a fingerprint, the callback accepts exactly that key and
+// appends it to the store.
+func (s *SSHClient) tofuHostKeyCallback() (gossh.HostKeyCallback, error) {
+	s.mu.Lock()
+	configDir := s.knownHostsConfigDir
+	profileID := s.knownHostsProfile
+	trusted := s.trustHostKey
+	s.mu.Unlock()
+
+	if configDir == "" || profileID == "" {
+		return nil, errors.New("no known-hosts store configured for regular SSH")
+	}
+
+	store, err := openKnownHostsStore(configDir, profileID)
+	if err != nil {
+		return nil, fmt.Errorf("load known-hosts store: %w", err)
+	}
+
+	return func(hostname string, remote net.Addr, key gossh.PublicKey) error {
+		verdict, err := store.verify(hostname, remote, key)
+		if err != nil {
+			return err
+		}
+		if verdict == hostKeyMatched {
+			return nil
+		}
+		if verdict == hostKeyChanged {
+			return fmt.Errorf("SSH host key changed for %s (possible attack)", hostname)
+		}
+
+		fingerprint := gossh.FingerprintSHA256(key)
+		if trusted == "" {
+			return &errHostKeyUnknown{fingerprint: fingerprint}
+		}
+		if trusted != fingerprint {
+			return fmt.Errorf("SSH host key changed since it was confirmed for %s", hostname)
+		}
+		if err := store.append(hostname, remote, key); err != nil {
+			return fmt.Errorf("persist trusted host key: %w", err)
+		}
+		// The confirmation is spent: now that the key is stored, a later
+		// reconnect must verify against the file, not re-accept this fingerprint.
+		s.mu.Lock()
+		s.trustHostKey = ""
+		s.mu.Unlock()
+		return nil
+	}, nil
+}
+
+func (s *SSHClient) requestJWTToken(cfg *profilemanager.Config, cfgPath string) (string, error) {
+	s.mu.Lock()
+	urlOpener := s.urlOpener
+	s.mu.Unlock()
+	if urlOpener == nil {
+		return "", errors.New("URL opener not configured for JWT auth")
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
+	defer cancel()
+
+	flow, err := auth.NewOAuthFlow(ctx, cfg, false, true, profileLoginHint(cfgPath))
+	if err != nil {
+		return "", fmt.Errorf("create oauth flow: %w", err)
+	}
+
+	// The status callback covers the browser round-trip, which would
+	// otherwise leave the terminal blank.
+	tokenInfo, err := runOAuthFlow(ctx, flow, urlOpener, func() {
+		s.notifyStatus("Waiting for browser authentication...")
+	})
+	if err != nil {
+		return "", err
+	}
+
+	token := tokenInfo.GetTokenToUse()
+	if token == "" {
+		return "", errors.New("empty token returned by IdP")
+	}
+
+	// Tells the client the browser round-trip is over so it can dismiss the
+	// surface it opened, the same way the login and session-extend flows do.
+	// Without it the Custom Tab stays in front of the terminal even though the
+	// token has already been collected.
+	urlOpener.OnLoginSuccess()
+
+	return token, nil
+}
+
+func (s *SSHClient) dialAndHandshake(gen uint64, host string, port int, clientConfig *gossh.ClientConfig) error {
+	addr := net.JoinHostPort(host, strconv.Itoa(port))
+	ctx, cancel := context.WithTimeout(context.Background(), sshDialTimeout)
+	defer cancel()
+
+	s.mu.Lock()
+	if gen != s.gen {
+		s.mu.Unlock()
+		return errClientClosed
+	}
+	s.dialCancel = cancel
+	s.mu.Unlock()
+
+	var dialer net.Dialer
+	conn, err := dialer.DialContext(ctx, "tcp", addr)
+	if err != nil {
+		return fmt.Errorf("dial %s: %w", addr, err)
+	}
+
+	client, err := nbssh.Handshake(ctx, conn, addr, clientConfig)
+	if err != nil {
+		return err
+	}
+
+	s.mu.Lock()
+	if gen != s.gen {
+		s.mu.Unlock()
+		closeQuiet(client, "stale ssh client")
+		return errClientClosed
+	}
+	s.sshClient = client
+	listener := s.listener
+	s.mu.Unlock()
+
+	if listener != nil {
+		listener.OnConnected()
+	}
+	return nil
+}
+
+func (s *SSHClient) readLoop(r io.Reader, name string) string {
+	buf := make([]byte, 4096)
+	for {
+		n, err := r.Read(buf)
+		if n > 0 {
+			s.mu.Lock()
+			listener := s.listener
+			s.mu.Unlock()
+			if listener != nil {
+				chunk := make([]byte, n)
+				copy(chunk, buf[:n])
+				listener.OnData(chunk)
+			}
+		}
+		if err != nil {
+			// EOF is a normal shell exit, so report it without a reason.
+			if errors.Is(err, io.EOF) {
+				return ""
+			}
+			log.Debugf("ssh %s read: %v", name, err)
+			return rootCause(err).Error()
+		}
+	}
+}
+
+// notifyStatus writes a progress line to the terminal through the normal
+// output path, so long steps are visible while nothing else is arriving.
+func (s *SSHClient) notifyStatus(text string) {
+	s.mu.Lock()
+	listener := s.listener
+	s.mu.Unlock()
+	if listener != nil {
+		listener.OnData([]byte("\r\n\x1b[33m" + text + "\x1b[0m\r\n"))
+	}
+}
+
+func (s *SSHClient) notifyClose(gen uint64, reason string) {
+	s.mu.Lock()
+	if gen != s.gen || s.closed {
+		s.mu.Unlock()
+		return
+	}
+	s.closed = true
+	listener := s.listener
+	s.mu.Unlock()
+	if listener != nil {
+		listener.OnClose(reason)
+	}
+}
+
+func closeQuiet(c io.Closer, label string) {
+	if c == nil {
+		return
+	}
+	if err := c.Close(); err != nil && !errors.Is(err, io.EOF) {
+		log.Debugf("ssh: close %s: %v", label, err)
+	}
+}
+
+func detectServerType(host string, port int) detection.ServerType {
+	ctx, cancel := context.WithTimeout(context.Background(), sshDetectionTimeout)
+	defer cancel()
+
+	dialer := &net.Dialer{}
+	serverType, err := detection.DetectSSHServerType(ctx, dialer, host, port)
+	if err != nil {
+		log.Debugf("ssh: server detection failed: %v (assuming regular SSH)", err)
+		return detection.ServerTypeRegular
+	}
+	return serverType
+}
+
+// rootCause returns the innermost error of a %w chain, so the terminal shows
+// "i/o timeout" rather than every layer that added context on the way up.
+func rootCause(err error) error {
+	for {
+		// A joined error has no single root, so keep it as-is.
+		if _, ok := err.(interface{ Unwrap() []error }); ok {
+			return err
+		}
+		next := errors.Unwrap(err)
+		if next == nil {
+			return err
+		}
+		err = next
+	}
+}
+
+// isAuthFailure distinguishes credential rejection from dial, timeout and
+// host-key errors, which retrying with a password would not fix.
+func isAuthFailure(err error) bool {
+	if errors.Is(err, errPasswordRequired) {
+		return true
+	}
+	var partial *gossh.PartialSuccessError
+	if errors.As(err, &partial) {
+		return true
+	}
+	return strings.Contains(err.Error(), "unable to authenticate")
+}
+
+// passwordCouldHelp reports whether prompting for a password again can change
+// the outcome. gossh lists a method under "attempted methods" only when the
+// server offered it, so a supplied password that was never attempted means the
+// server does not accept passwords and the real error should surface instead.
+func passwordCouldHelp(err error, passwordOffered bool) bool {
+	if !passwordOffered {
+		return true
+	}
+	msg := err.Error()
+	return strings.Contains(msg, "password") || strings.Contains(msg, "keyboard-interactive")
+}

client/android/ssh_known_hosts.go

@@ -0,0 +1,168 @@
+//go:build android
+
+package android
+
+import (
+	"bytes"
+	"net"
+	"strconv"
+	"strings"
+	"sync"
+
+	gossh "golang.org/x/crypto/ssh"
+	"golang.org/x/crypto/ssh/knownhosts"
+)
+
+const knownHostsNamespace = "ssh"
+
+const (
+	hostKeyUnknown hostKeyVerdict = iota
+	hostKeyMatched
+	hostKeyChanged
+)
+
+var knownHostsMu sync.Mutex
+
+type hostKeyVerdict uint8
+
+type knownHostsSection struct {
+	KnownHosts []string `json:"knownHosts"`
+}
+
+type knownHostsStore struct {
+	prefs prefsStore
+}
+
+// RemoveKnownHost deletes every known-hosts entry for host:port from the
+// profile's store, so a host trusted for a session that is being deleted does
+// not linger. Java calls this only once no session targets that host, so a
+// shared host stays trusted. A missing entry is not an error: the goal state
+// is "absent".
+func RemoveKnownHost(configDir, profileID, host string, port int) error {
+	store, err := openKnownHostsStore(configDir, profileID)
+	if err != nil {
+		return err
+	}
+	return store.removeHost(host, port)
+}
+
+func openKnownHostsStore(configDir, profileID string) (*knownHostsStore, error) {
+	prefs, err := newProfilePrefs(configDir, profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &knownHostsStore{prefs: prefs}, nil
+}
+
+func (st *knownHostsStore) verify(hostname string, remote net.Addr, key gossh.PublicKey) (hostKeyVerdict, error) {
+	lines, err := st.lines()
+	if err != nil {
+		return hostKeyUnknown, err
+	}
+	targets := knownHostsTargets(hostname, remote)
+
+	verdict := hostKeyUnknown
+	for _, line := range lines {
+		pubKey, ok := knownHostsLineKey(line, targets)
+		if !ok {
+			continue
+		}
+		if pubKey.Type() == key.Type() && bytes.Equal(pubKey.Marshal(), key.Marshal()) {
+			return hostKeyMatched, nil
+		}
+		verdict = hostKeyChanged
+	}
+	return verdict, nil
+}
+
+func (st *knownHostsStore) append(hostname string, remote net.Addr, key gossh.PublicKey) error {
+	line := knownhosts.Line(knownHostsTargets(hostname, remote), key)
+
+	knownHostsMu.Lock()
+	defer knownHostsMu.Unlock()
+
+	lines, err := st.lines()
+	if err != nil {
+		return err
+	}
+	return st.prefs.Put(knownHostsNamespace, knownHostsSection{KnownHosts: append(lines, line)})
+}
+
+func (st *knownHostsStore) removeHost(host string, port int) error {
+	target := knownhosts.Normalize(net.JoinHostPort(host, strconv.Itoa(port)))
+
+	knownHostsMu.Lock()
+	defer knownHostsMu.Unlock()
+
+	lines, err := st.lines()
+	if err != nil {
+		return err
+	}
+	kept := make([]string, 0, len(lines))
+	for _, line := range lines {
+		if knownHostsLineMatches(line, target) {
+			continue
+		}
+		kept = append(kept, line)
+	}
+	if len(kept) == len(lines) {
+		return nil
+	}
+	return st.prefs.Put(knownHostsNamespace, knownHostsSection{KnownHosts: kept})
+}
+
+func (st *knownHostsStore) lines() ([]string, error) {
+	var section knownHostsSection
+	if _, err := st.prefs.Get(knownHostsNamespace, &section); err != nil {
+		return nil, err
+	}
+	return section.KnownHosts, nil
+}
+
+func knownHostsTargets(hostname string, remote net.Addr) []string {
+	targets := []string{knownhosts.Normalize(hostname)}
+	if remote != nil {
+		if normalized := knownhosts.Normalize(remote.String()); normalized != targets[0] {
+			targets = append(targets, normalized)
+		}
+	}
+	return targets
+}
+
+func knownHostsLineKey(line string, targets []string) (gossh.PublicKey, bool) {
+	trimmed := strings.TrimSpace(line)
+	if trimmed == "" || strings.HasPrefix(trimmed, "#") {
+		return nil, false
+	}
+	_, hosts, pubKey, _, _, err := gossh.ParseKnownHosts([]byte(trimmed))
+	if err != nil {
+		return nil, false
+	}
+	for _, host := range hosts {
+		for _, target := range targets {
+			if host == target {
+				return pubKey, true
+			}
+		}
+	}
+	return nil, false
+}
+
+// knownHostsLineMatches reports whether a known-hosts line's address list
+// contains the normalized target. Comment and blank lines never match.
+func knownHostsLineMatches(line, target string) bool {
+	trimmed := strings.TrimSpace(line)
+	if trimmed == "" || strings.HasPrefix(trimmed, "#") {
+		return false
+	}
+	fields := strings.Fields(trimmed)
+	if len(fields) == 0 {
+		return false
+	}
+	for _, addr := range strings.Split(fields[0], ",") {
+		if addr == target {
+			return true
+		}
+	}
+	return false
+}

client/android/ssh_sessions.go

@@ -0,0 +1,104 @@
+//go:build android
+
+package android
+
+const (
+	sshSessionsNamespace = "ssh-sessions"
+	maxStoredSSHSessions = 50
+)
+
+type sshSessionRecord struct {
+	ID   string `json:"id"`
+	Host string `json:"host"`
+	Port int    `json:"port"`
+	User string `json:"user"`
+}
+
+type sshSessionsSection struct {
+	Sessions []sshSessionRecord `json:"sessions"`
+}
+
+// SSHSessionEntry is one stored SSH session, without any credential.
+type SSHSessionEntry struct {
+	ID   string
+	Host string
+	Port int
+	User string
+}
+
+// SSHSessionArray wraps stored SSH sessions for gomobile compatibility.
+type SSHSessionArray struct {
+	items []*SSHSessionEntry
+}
+
+// NewSSHSessionArray creates an empty session array to fill via Add.
+func NewSSHSessionArray() *SSHSessionArray {
+	return &SSHSessionArray{}
+}
+
+// Add appends a session entry, oldest first.
+func (a *SSHSessionArray) Add(id, host string, port int, user string) {
+	a.items = append(a.items, &SSHSessionEntry{ID: id, Host: host, Port: port, User: user})
+}
+
+// Length returns the number of entries.
+func (a *SSHSessionArray) Length() int {
+	return len(a.items)
+}
+
+// Get returns the entry at index i, or nil when out of range.
+func (a *SSHSessionArray) Get(i int) *SSHSessionEntry {
+	if i < 0 || i >= len(a.items) {
+		return nil
+	}
+	return a.items[i]
+}
+
+// SSHSessionStore reads and writes a profile's stored SSH sessions.
+type SSHSessionStore struct {
+	prefs prefsStore
+}
+
+// NewSSHSessionStore opens the session store of the given profile.
+func NewSSHSessionStore(configDir, profileID string) (*SSHSessionStore, error) {
+	prefs, err := newProfilePrefs(configDir, profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &SSHSessionStore{prefs: prefs}, nil
+}
+
+// Load returns the stored sessions, oldest first.
+func (s *SSHSessionStore) Load() (*SSHSessionArray, error) {
+	var section sshSessionsSection
+	if _, err := s.prefs.Get(sshSessionsNamespace, &section); err != nil {
+		return nil, err
+	}
+
+	out := NewSSHSessionArray()
+	for _, record := range section.Sessions {
+		if record.ID == "" || record.Host == "" {
+			continue
+		}
+		out.Add(record.ID, record.Host, record.Port, record.User)
+	}
+	return out, nil
+}
+
+// Save replaces the stored sessions, keeping only the newest entries when the
+// list exceeds the storage cap.
+func (s *SSHSessionStore) Save(sessions *SSHSessionArray) error {
+	var items []*SSHSessionEntry
+	if sessions != nil {
+		items = sessions.items
+	}
+	if len(items) > maxStoredSSHSessions {
+		items = items[len(items)-maxStoredSSHSessions:]
+	}
+
+	records := make([]sshSessionRecord, 0, len(items))
+	for _, item := range items {
+		records = append(records, sshSessionRecord{ID: item.ID, Host: item.Host, Port: item.Port, User: item.User})
+	}
+	return s.prefs.Put(sshSessionsNamespace, sshSessionsSection{Sessions: records})
+}

client/anonymize/anonymize.go

@@ -2,6 +2,7 @@ package anonymize
 
 import (
 	"crypto/rand"
+	"encoding/base64"
 	"fmt"
 	"math/big"
 	"net"
@@ -9,46 +10,148 @@ import (
 	"net/url"
 	"regexp"
 	"slices"
+	"strconv"
 	"strings"
 )
 
 const anonTLD = ".domain"
 
+// Level selects how much the anonymizer redacts. Levels are ordered: a higher
+// level redacts strictly more. On the wire (protos, flags) levels travel as
+// their string form.
+type Level int
+
+const (
+	// LevelDefault anonymizes public IP addresses, IPv6 ULA, domains, and MAC
+	// addresses. Internal IPv4 ranges (RFC 1918, CGNAT, link-local) are
+	// preserved so support can reason about the real topology.
+	LevelDefault Level = iota
+	// LevelStrict additionally anonymizes internal IP ranges, peer names, and
+	// WireGuard public keys.
+	LevelStrict
+)
+
+// LevelDefaultString and LevelStrictString are the wire forms of the levels,
+// for boundaries that pass levels as strings (flags, protos, mobile bindings).
+const (
+	LevelDefaultString = "default"
+	LevelStrictString  = "strict"
+)
+
+// ParseLevel maps s to a Level. Empty means LevelDefault; anything
+// unrecognized maps to LevelStrict so an unknown request never yields less
+// anonymization than intended.
+func ParseLevel(s string) Level {
+	switch strings.ToLower(s) {
+	case "", LevelDefaultString:
+		return LevelDefault
+	default:
+		return LevelStrict
+	}
+}
+
+// String returns the wire form of the level: "default" or "strict".
+func (l Level) String() string {
+	if l >= LevelStrict {
+		return LevelStrictString
+	}
+	return LevelDefaultString
+}
+
+// protectedDomains are NetBird-operated suffixes that stay recognizable in an
+// anonymized bundle. At LevelStrict the labels in front of them (the peer
+// name) are still replaced, except under netbird.io, which only hosts
+// NetBird infrastructure (api, signal, flow), never peer names.
+var protectedDomains = []string{"netbird.io", "netbird.selfhosted", "netbird.cloud", "netbird.stage"}
+
+const infraDomain = "netbird.io"
+
+var (
+	macColonRegex = regexp.MustCompile(`\b[0-9a-fA-F]{2}(?::[0-9a-fA-F]{2}){5}\b`)
+	macDashRegex  = regexp.MustCompile(`\b[0-9a-fA-F]{2}(?:-[0-9a-fA-F]{2}){5}\b`)
+	wgKeyRegex    = regexp.MustCompile(`\b[A-Za-z0-9+/]{43}=`)
+)
+
 type Anonymizer struct {
 	ipAnonymizer     map[netip.Addr]netip.Addr
 	domainAnonymizer map[string]string
-	currentAnonIPv4  netip.Addr
-	currentAnonIPv6  netip.Addr
-	startAnonIPv4    netip.Addr
-	startAnonIPv6    netip.Addr
+	// domainOrder caches the keys of domainAnonymizer sorted longest-first
+	// for AnonymizeString; it is rebuilt when the map gains entries.
+	domainOrder     []string
+	labelAnonymizer map[string]string
+	labelAnonymized map[string]struct{}
+	labelCounter    uint32
+	macAnonymizer   map[string]string
+	macCounter      uint32
+	wgKeyAnonymizer map[string]string
+	wgKeyAnonymized map[string]struct{}
+	currentAnonIPv4 netip.Addr
+	currentAnonIPv6 netip.Addr
+	startAnonIPv4   netip.Addr
+	startAnonIPv6   netip.Addr
+
+	// LevelStrict also anonymizes internal ranges (RFC 1918, CGNAT,
+	// link-local), replacing them from the dedicated internal pools below so
+	// a reader can still tell an internal address from a public one.
+	level                   Level
+	currentAnonInternalIPv4 netip.Addr
+	currentAnonInternalIPv6 netip.Addr
+	startAnonInternalIPv4   netip.Addr
+	startAnonInternalIPv6   netip.Addr
 
 	domainKeyRegex *regexp.Regexp
 }
 
 func DefaultAddresses() (netip.Addr, netip.Addr) {
-	// 198.51.100.0, 100::
-	return netip.AddrFrom4([4]byte{198, 51, 100, 0}), netip.AddrFrom16([16]byte{0x01})
+	// 198.51.100.0 (RFC 5737 TEST-NET-2), 2001:db8:ffff:: (RFC 3849 documentation, last /48)
+	// The old start 100:: (discard, RFC 6666) is now used for fake IPs on Android.
+	return netip.AddrFrom4([4]byte{198, 51, 100, 0}), netip.MustParseAddr("2001:db8:ffff::")
+}
+
+// InternalAddresses returns the pool starts used in strict mode for internal
+// ranges. Both are reserved ranges that cannot collide with real addressing:
+// 198.18.0.0 (RFC 2544 benchmarking), 2001:db8:1:: (RFC 3849 documentation).
+func InternalAddresses() (netip.Addr, netip.Addr) {
+	return netip.AddrFrom4([4]byte{198, 18, 0, 0}), netip.MustParseAddr("2001:db8:1::")
 }
 
 func NewAnonymizer(startIPv4, startIPv6 netip.Addr) *Anonymizer {
+	internalIPv4, internalIPv6 := InternalAddresses()
 	return &Anonymizer{
 		ipAnonymizer:     map[netip.Addr]netip.Addr{},
 		domainAnonymizer: map[string]string{},
+		labelAnonymizer:  map[string]string{},
+		labelAnonymized:  map[string]struct{}{},
+		macAnonymizer:    map[string]string{},
+		wgKeyAnonymizer:  map[string]string{},
+		wgKeyAnonymized:  map[string]struct{}{},
 		currentAnonIPv4:  startIPv4,
 		currentAnonIPv6:  startIPv6,
 		startAnonIPv4:    startIPv4,
 		startAnonIPv6:    startIPv6,
 
+		level:                   LevelDefault,
+		currentAnonInternalIPv4: internalIPv4,
+		currentAnonInternalIPv6: internalIPv6,
+		startAnonInternalIPv4:   internalIPv4,
+		startAnonInternalIPv6:   internalIPv6,
+
 		domainKeyRegex: regexp.MustCompile(`\bdomain=([^\s,:"]+)`),
 	}
 }
 
+// SetLevel selects the anonymization level. The zero value of a new
+// Anonymizer is LevelDefault.
+func (a *Anonymizer) SetLevel(level Level) {
+	a.level = level
+}
+
 func (a *Anonymizer) AnonymizeIP(ip netip.Addr) netip.Addr {
+	// Normalize 4-in-6 addresses so ::ffff:192.168.1.1 classifies and maps
+	// like 192.168.1.1.
+	ip = ip.Unmap()
+
 	if ip.IsLoopback() ||
-		ip.IsLinkLocalUnicast() ||
-		ip.IsLinkLocalMulticast() ||
-		ip.IsInterfaceLocalMulticast() ||
-		ip.IsPrivate() ||
 		ip.IsUnspecified() ||
 		ip.IsMulticast() ||
 		isWellKnown(ip) ||
@@ -57,18 +160,100 @@ func (a *Anonymizer) AnonymizeIP(ip netip.Addr) netip.Addr {
 		return ip
 	}
 
+	if isInternal(ip) && a.level < LevelStrict {
+		return ip
+	}
+
 	if _, ok := a.ipAnonymizer[ip]; !ok {
-		if ip.Is4() {
-			a.ipAnonymizer[ip] = a.currentAnonIPv4
-			a.currentAnonIPv4 = a.currentAnonIPv4.Next()
-		} else {
-			a.ipAnonymizer[ip] = a.currentAnonIPv6
-			a.currentAnonIPv6 = a.currentAnonIPv6.Next()
-		}
+		a.ipAnonymizer[ip] = a.nextAnonIP(ip)
 	}
 	return a.ipAnonymizer[ip]
 }
 
+func (a *Anonymizer) nextAnonIP(ip netip.Addr) netip.Addr {
+	// At the strict level, internal addresses (including IPv6 ULA, matched
+	// by IsPrivate) come from the internal pools so they remain recognizable
+	// as internal without disclosing the real values.
+	if a.level >= LevelStrict && (isInternal(ip) || ip.IsPrivate()) {
+		if ip.Is4() {
+			anon := a.currentAnonInternalIPv4
+			a.currentAnonInternalIPv4 = a.currentAnonInternalIPv4.Next()
+			return anon
+		}
+		anon := a.currentAnonInternalIPv6
+		a.currentAnonInternalIPv6 = a.currentAnonInternalIPv6.Next()
+		return anon
+	}
+
+	if ip.Is4() {
+		anon := a.currentAnonIPv4
+		a.currentAnonIPv4 = a.currentAnonIPv4.Next()
+		return anon
+	}
+	anon := a.currentAnonIPv6
+	a.currentAnonIPv6 = a.currentAnonIPv6.Next()
+	return anon
+}
+
+// AnonymizeMAC replaces a MAC address with a consistent placeholder from the
+// locally administered range starting at 02:00:00:00:00:01, at every
+// anonymization level. Broadcast, multicast, all-zero, and already assigned
+// placeholder addresses are preserved. The colon and dash spellings of the
+// same address share one placeholder; the output keeps the input's separator.
+func (a *Anonymizer) AnonymizeMAC(mac string) string {
+	hw, err := net.ParseMAC(mac)
+	if err != nil || len(hw) != 6 {
+		return mac
+	}
+
+	if isWellKnownMAC(hw) || a.isAnonymizedMAC(hw) {
+		return mac
+	}
+
+	key := hw.String()
+	anon, ok := a.macAnonymizer[key]
+	if !ok {
+		a.macCounter++
+		anon = fmt.Sprintf("02:00:00:%02x:%02x:%02x", byte(a.macCounter>>16), byte(a.macCounter>>8), byte(a.macCounter))
+		a.macAnonymizer[key] = anon
+	}
+
+	if strings.Contains(mac, "-") {
+		anon = strings.ReplaceAll(anon, ":", "-")
+	}
+	return anon
+}
+
+// isAnonymizedMAC reports whether hw is a placeholder this anonymizer already
+// handed out, so a second pass over anonymized output leaves it unchanged.
+func (a *Anonymizer) isAnonymizedMAC(hw net.HardwareAddr) bool {
+	if hw[0] != 0x02 || hw[1] != 0 || hw[2] != 0 {
+		return false
+	}
+	value := uint32(hw[3])<<16 | uint32(hw[4])<<8 | uint32(hw[5])
+	return value <= a.macCounter
+}
+
+// AnonymizeWGKey replaces a WireGuard public key with a consistent random
+// placeholder of the same shape. Keys are only anonymized at LevelStrict;
+// placeholders already handed out pass through unchanged.
+func (a *Anonymizer) AnonymizeWGKey(key string) string {
+	if a.level < LevelStrict || !looksLikeWGKey(key) {
+		return key
+	}
+	if _, ok := a.wgKeyAnonymized[key]; ok {
+		return key
+	}
+
+	anon, ok := a.wgKeyAnonymizer[key]
+	if !ok {
+		anon = generateAnonymousKey()
+		a.wgKeyAnonymizer[key] = anon
+		a.wgKeyAnonymized[anon] = struct{}{}
+	}
+	return anon
+}
+
 func (a *Anonymizer) AnonymizeUDPAddr(addr net.UDPAddr) net.UDPAddr {
 	// Convert IP to netip.Addr
 	ip, ok := netip.AddrFromSlice(addr.IP)
@@ -87,15 +272,20 @@ func (a *Anonymizer) AnonymizeUDPAddr(addr net.UDPAddr) net.UDPAddr {
 
 // isInAnonymizedRange checks if an IP is within the range of already assigned anonymized IPs
 func (a *Anonymizer) isInAnonymizedRange(ip netip.Addr) bool {
-	if ip.Is4() && ip.Compare(a.startAnonIPv4) >= 0 && ip.Compare(a.currentAnonIPv4) <= 0 {
-		return true
-	} else if !ip.Is4() && ip.Compare(a.startAnonIPv6) >= 0 && ip.Compare(a.currentAnonIPv6) <= 0 {
-		return true
+	if ip.Is4() {
+		return inPoolRange(ip, a.startAnonIPv4, a.currentAnonIPv4) ||
+			inPoolRange(ip, a.startAnonInternalIPv4, a.currentAnonInternalIPv4)
 	}
-	return false
+	return inPoolRange(ip, a.startAnonIPv6, a.currentAnonIPv6) ||
+		inPoolRange(ip, a.startAnonInternalIPv6, a.currentAnonInternalIPv6)
 }
 
 func (a *Anonymizer) AnonymizeIPString(ip string) string {
+	// Handle CIDR notation (e.g. "2001:db8::/32")
+	if prefix, err := netip.ParsePrefix(ip); err == nil {
+		return a.AnonymizeIP(prefix.Addr()).String() + "/" + strconv.Itoa(prefix.Bits())
+	}
+
 	addr, err := netip.ParseAddr(ip)
 	if err != nil {
 		return ip
@@ -111,14 +301,23 @@ func (a *Anonymizer) AnonymizeDomain(domain string) string {
 		baseDomain = domain[:len(domain)-1]
 	}
 
-	if strings.HasSuffix(baseDomain, "netbird.io") ||
-		strings.HasSuffix(baseDomain, "netbird.selfhosted") ||
-		strings.HasSuffix(baseDomain, "netbird.cloud") ||
-		strings.HasSuffix(baseDomain, "netbird.stage") ||
-		strings.HasSuffix(baseDomain, anonTLD) {
+	if strings.HasSuffix(baseDomain, anonTLD) {
 		return domain
 	}
 
+	// A reverse zone names an address prefix, so it follows the address rules,
+	// which also keeps its digit labels intact.
+	if zone, ok := a.anonymizeReverseZone(baseDomain); ok {
+		return withTrailingDot(zone, hasDot)
+	}
+
+	if suffix := protectedSuffix(baseDomain); suffix != "" {
+		if a.level < LevelStrict || baseDomain == suffix || suffix == infraDomain {
+			return domain
+		}
+		return withTrailingDot(a.anonymizePeerName(baseDomain, suffix), hasDot)
+	}
+
 	parts := strings.Split(baseDomain, ".")
 	if len(parts) < 2 {
 		return domain
@@ -134,12 +333,53 @@ func (a *Anonymizer) AnonymizeDomain(domain string) string {
 	}
 
 	result := strings.Replace(baseDomain, baseForLookup, anonymized, 1)
-	if hasDot {
-		result += "."
+	if a.level >= LevelStrict && len(parts) > 2 {
+		prefix := strings.TrimSuffix(baseDomain, "."+baseForLookup)
+		result = a.anonymizeLabels(prefix, "host") + "." + anonymized
+		// The full mapping feeds AnonymizeString so seeded FQDNs are caught
+		// in log lines as a whole, labels included.
+		a.domainAnonymizer[baseDomain] = result
+	}
+	return withTrailingDot(result, hasDot)
+}
+
+// anonymizePeerName replaces the labels in front of a protected suffix with
+// numbered peer placeholders, keeping the suffix, and records the full
+// mapping for string replacement in logs. The numbering keeps a peer
+// recognizable across the whole bundle without disclosing its name.
+func (a *Anonymizer) anonymizePeerName(baseDomain, suffix string) string {
+	prefix := strings.TrimSuffix(baseDomain, "."+suffix)
+	result := a.anonymizeLabels(prefix, "peer") + "." + suffix
+	if result != baseDomain {
+		a.domainAnonymizer[baseDomain] = result
 	}
 	return result
 }
 
+// anonymizeLabels replaces each dot-separated label with a consistent
+// numbered placeholder ("<placeholder>-<n>"). Wildcard labels and
+// placeholders already handed out pass through unchanged.
+func (a *Anonymizer) anonymizeLabels(prefix, placeholder string) string {
+	labels := strings.Split(prefix, ".")
+	for i, label := range labels {
+		if label == "*" {
+			continue
+		}
+		if _, ok := a.labelAnonymized[label]; ok {
+			continue
+		}
+		anon, ok := a.labelAnonymizer[label]
+		if !ok {
+			a.labelCounter++
+			anon = fmt.Sprintf("%s-%d", placeholder, a.labelCounter)
+			a.labelAnonymizer[label] = anon
+			a.labelAnonymized[anon] = struct{}{}
+		}
+		labels[i] = anon
+	}
+	return strings.Join(labels, ".")
+}
+
 func (a *Anonymizer) AnonymizeURI(uri string) string {
 	u, err := url.Parse(uri)
 	if err != nil {
@@ -150,7 +390,7 @@ func (a *Anonymizer) AnonymizeURI(uri string) string {
 	if u.Opaque != "" {
 		host, port, err := net.SplitHostPort(u.Opaque)
 		if err == nil {
-			anonymizedHost = fmt.Sprintf("%s:%s", a.AnonymizeDomain(host), port)
+			anonymizedHost = net.JoinHostPort(a.AnonymizeDomain(host), port)
 		} else {
 			anonymizedHost = a.AnonymizeDomain(u.Opaque)
 		}
@@ -158,7 +398,7 @@ func (a *Anonymizer) AnonymizeURI(uri string) string {
 	} else if u.Host != "" {
 		host, port, err := net.SplitHostPort(u.Host)
 		if err == nil {
-			anonymizedHost = fmt.Sprintf("%s:%s", a.AnonymizeDomain(host), port)
+			anonymizedHost = net.JoinHostPort(a.AnonymizeDomain(host), port)
 		} else {
 			anonymizedHost = a.AnonymizeDomain(u.Host)
 		}
@@ -171,17 +411,75 @@ func (a *Anonymizer) AnonymizeString(str string) string {
 	ipv4Regex := regexp.MustCompile(`\b(?:[0-9]{1,3}\.){3}[0-9]{1,3}\b`)
 	ipv6Regex := regexp.MustCompile(`\b([0-9a-fA-F:]+:+[0-9a-fA-F]{0,4})(?:%[0-9a-zA-Z]+)?(?:\/[0-9]{1,3})?(?::[0-9]{1,5})?\b`)
 
+	// Reverse zones go first and are then held out of the passes below: their
+	// labels are digits, which the address patterns would otherwise consume.
+	str, restoreZones := a.replaceReverseZones(str)
+
 	str = ipv4Regex.ReplaceAllStringFunc(str, a.AnonymizeIPString)
 	str = ipv6Regex.ReplaceAllStringFunc(str, a.AnonymizeIPString)
 
-	for domain, anonDomain := range a.domainAnonymizer {
-		str = strings.ReplaceAll(str, domain, anonDomain)
+	for _, domain := range a.sortedDomains() {
+		str = strings.ReplaceAll(str, domain, a.domainAnonymizer[domain])
 	}
 
 	str = a.AnonymizeSchemeURI(str)
 	str = a.AnonymizeDNSLogLine(str)
 
-	return str
+	// MAC handling runs after the IP passes so preserved IPv6 addresses are
+	// already out of the way; the separator guard skips matches embedded in a
+	// longer colon- or dash-separated sequence (such as an IPv6 tail).
+	str = a.anonymizeMACsInString(str, macColonRegex, ':')
+	str = a.anonymizeMACsInString(str, macDashRegex, '-')
+
+	if a.level >= LevelStrict {
+		str = wgKeyRegex.ReplaceAllStringFunc(str, a.AnonymizeWGKey)
+	}
+
+	return restoreZones(str)
+}
+
+// sortedDomains returns the domain mappings longest-first, so a full-FQDN
+// mapping (strict level) is applied before the base-domain mapping it
+// contains. The order is rebuilt only when domainAnonymizer has grown.
+func (a *Anonymizer) sortedDomains() []string {
+	if len(a.domainOrder) == len(a.domainAnonymizer) {
+		return a.domainOrder
+	}
+
+	a.domainOrder = a.domainOrder[:0]
+	for domain := range a.domainAnonymizer {
+		a.domainOrder = append(a.domainOrder, domain)
+	}
+	slices.SortFunc(a.domainOrder, func(x, y string) int {
+		if d := len(y) - len(x); d != 0 {
+			return d
+		}
+		return strings.Compare(x, y)
+	})
+	return a.domainOrder
+}
+
+// anonymizeMACsInString replaces MAC addresses matched by re, skipping
+// matches that directly adjoin another sep so a six-group run inside a longer
+// separated sequence is left alone.
+func (a *Anonymizer) anonymizeMACsInString(str string, re *regexp.Regexp, sep byte) string {
+	matches := re.FindAllStringIndex(str, -1)
+	if len(matches) == 0 {
+		return str
+	}
+
+	var b strings.Builder
+	last := 0
+	for _, m := range matches {
+		if (m[0] > 0 && str[m[0]-1] == sep) || (m[1] < len(str) && str[m[1]] == sep) {
+			continue
+		}
+		b.WriteString(str[last:m[0]])
+		b.WriteString(a.AnonymizeMAC(str[m[0]:m[1]]))
+		last = m[1]
+	}
+	b.WriteString(str[last:])
+	return b.String()
 }
 
 // AnonymizeSchemeURI finds and anonymizes URIs with ws, wss, rel, rels, stun, stuns, turn, and turns schemes.
@@ -232,10 +530,79 @@ func isWellKnown(addr netip.Addr) bool {
 		"128.0.0.0", "8000::", // 2nd split subnet for default routes
 	}
 
-	if slices.Contains(wellKnown, addr.String()) {
+	return slices.Contains(wellKnown, addr.String())
+}
+
+// isInternal reports whether ip identifies a host only within the local
+// network: IPv4 private (RFC 1918), CGNAT (RFC 6598), and link-local (v4 and
+// v6). These are preserved at the default level so support can reason about
+// the real topology, and replaced from the internal pools at the strict
+// level. IPv6 ULA is deliberately not internal: its random global ID uniquely
+// fingerprints the network, so it is anonymized at every level.
+func isInternal(ip netip.Addr) bool {
+	return (ip.Is4() && ip.IsPrivate()) ||
+		ip.IsLinkLocalUnicast() ||
+		isCGNAT(ip)
+}
+
+func inPoolRange(ip, start, current netip.Addr) bool {
+	return ip.Compare(start) >= 0 && ip.Compare(current) <= 0
+}
+
+// isWellKnownMAC reports whether hw carries no stable host identity: all-zero
+// or a group address (broadcast and multicast).
+func isWellKnownMAC(hw net.HardwareAddr) bool {
+	if hw[0]&1 == 1 {
 		return true
 	}
+	for _, b := range hw {
+		if b != 0 {
+			return false
+		}
+	}
+	return true
+}
 
+// looksLikeWGKey reports whether s has the shape of a WireGuard key:
+// 44 base64 characters decoding to 32 bytes.
+func looksLikeWGKey(s string) bool {
+	if len(s) != 44 || s[43] != '=' {
+		return false
+	}
+	decoded, err := base64.StdEncoding.DecodeString(s)
+	return err == nil && len(decoded) == 32
+}
+
+func generateAnonymousKey() string {
+	buf := make([]byte, 32)
+	if _, err := rand.Read(buf); err != nil {
+		return strings.Repeat("A", 43) + "="
+	}
+	return base64.StdEncoding.EncodeToString(buf)
+}
+
+// protectedSuffix returns the protected NetBird suffix baseDomain ends with,
+// or empty. The match is label-anchored so an unrelated domain that merely
+// ends in the same characters is not preserved.
+func protectedSuffix(baseDomain string) string {
+	for _, d := range protectedDomains {
+		if baseDomain == d || strings.HasSuffix(baseDomain, "."+d) {
+			return d
+		}
+	}
+	return ""
+}
+
+func withTrailingDot(domain string, hasDot bool) string {
+	if hasDot {
+		return domain + "."
+	}
+	return domain
+}
+
+// isCGNAT reports whether addr is in 100.64.0.0/10 (RFC 6598), the range
+// NetBird assigns overlay peer addresses from.
+func isCGNAT(addr netip.Addr) bool {
 	cgnatRangeStart := netip.AddrFrom4([4]byte{100, 64, 0, 0})
 	cgnatRange := netip.PrefixFrom(cgnatRangeStart, 10)
 

client/anonymize/anonymize_test.go

@@ -1,8 +1,11 @@
 package anonymize_test
 
 import (
+	"bytes"
+	"encoding/base64"
 	"net/netip"
 	"regexp"
+	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
@@ -13,7 +16,7 @@ import (
 
 func TestAnonymizeIP(t *testing.T) {
 	startIPv4 := netip.MustParseAddr("198.51.100.0")
-	startIPv6 := netip.MustParseAddr("100::")
+	startIPv6 := netip.MustParseAddr("2001:db8:ffff::")
 	anonymizer := anonymize.NewAnonymizer(startIPv4, startIPv6)
 
 	tests := []struct {
@@ -26,9 +29,9 @@ func TestAnonymizeIP(t *testing.T) {
 		{"Second Public IPv4", "4.3.2.1", "198.51.100.1"},
 		{"Repeated IPv4", "1.2.3.4", "198.51.100.0"},
 		{"Private IPv4", "192.168.1.1", "192.168.1.1"},
-		{"First Public IPv6", "2607:f8b0:4005:805::200e", "100::"},
-		{"Second Public IPv6", "a::b", "100::1"},
-		{"Repeated IPv6", "2607:f8b0:4005:805::200e", "100::"},
+		{"First Public IPv6", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
+		{"Second Public IPv6", "a::b", "2001:db8:ffff::1"},
+		{"Repeated IPv6", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
 		{"Private IPv6", "fe80::1", "fe80::1"},
 		{"In Range IPv4", "198.51.100.2", "198.51.100.2"},
 	}
@@ -44,6 +47,301 @@ func TestAnonymizeIP(t *testing.T) {
 	}
 }
 
+func TestParseLevel(t *testing.T) {
+	tests := []struct {
+		input  string
+		expect anonymize.Level
+	}{
+		{"", anonymize.LevelDefault},
+		{"default", anonymize.LevelDefault},
+		{"DEFAULT", anonymize.LevelDefault},
+		{"strict", anonymize.LevelStrict},
+		{"STRICT", anonymize.LevelStrict},
+		// Unknown values must never yield less anonymization than requested.
+		{"garbage", anonymize.LevelStrict},
+	}
+
+	for _, tc := range tests {
+		t.Run("input="+tc.input, func(t *testing.T) {
+			assert.Equal(t, tc.expect, anonymize.ParseLevel(tc.input), "parsed level should match")
+		})
+	}
+}
+
+func TestAnonymizeIP_DefaultLevelInternalRanges(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	tests := []struct {
+		name   string
+		ip     string
+		expect string
+	}{
+		{"RFC1918 10/8", "10.1.2.3", "10.1.2.3"},
+		{"RFC1918 172.16/12", "172.16.5.5", "172.16.5.5"},
+		{"RFC1918 192.168/16", "192.168.1.1", "192.168.1.1"},
+		{"CGNAT", "100.64.0.5", "100.64.0.5"},
+		{"IPv4 link-local", "169.254.1.1", "169.254.1.1"},
+		{"IPv6 link-local", "fe80::1", "fe80::1"},
+		// ULA is anonymized even at the default level: its random global ID
+		// uniquely fingerprints the network, unlike shared RFC 1918 space.
+		{"IPv6 ULA", "fd12:3456:789a::1", "2001:db8:ffff::"},
+		// 4-in-6 addresses classify like their unmapped IPv4 form.
+		{"4-in-6 RFC1918", "::ffff:192.168.1.1", "192.168.1.1"},
+		{"4-in-6 CGNAT", "::ffff:100.64.0.5", "100.64.0.5"},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := anonymizer.AnonymizeIP(netip.MustParseAddr(tc.ip))
+			assert.Equal(t, tc.expect, result.String(), "default level should preserve internal ranges except ULA")
+		})
+	}
+}
+
+func TestAnonymizeIP_StrictLevel(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	// Order matters: internal pool addresses are assigned sequentially.
+	tests := []struct {
+		name   string
+		ip     string
+		expect string
+	}{
+		{"RFC1918 192.168/16", "192.168.1.1", "198.18.0.0"},
+		{"Second RFC1918", "192.168.1.2", "198.18.0.1"},
+		{"Repeated RFC1918", "192.168.1.1", "198.18.0.0"},
+		{"RFC1918 10/8", "10.1.2.3", "198.18.0.2"},
+		{"RFC1918 172.16/12", "172.16.5.5", "198.18.0.3"},
+		{"CGNAT", "100.64.0.5", "198.18.0.4"},
+		{"IPv4 link-local", "169.254.1.1", "198.18.0.5"},
+		{"Public IPv4 uses public pool", "1.2.3.4", "198.51.100.0"},
+		{"IPv6 link-local", "fe80::1", "2001:db8:1::"},
+		{"IPv6 ULA", "fd12:3456:789a::1", "2001:db8:1::1"},
+		{"Public IPv6 uses public pool", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
+		{"Loopback IPv4", "127.0.0.1", "127.0.0.1"},
+		{"Loopback IPv6", "::1", "::1"},
+		{"Unspecified", "0.0.0.0", "0.0.0.0"},
+		{"Multicast", "224.0.0.251", "224.0.0.251"},
+		{"Well known resolver", "8.8.8.8", "8.8.8.8"},
+		{"Well known split marker", "128.0.0.0", "128.0.0.0"},
+		{"In internal pool range", "198.18.0.3", "198.18.0.3"},
+		{"In public pool range", "198.51.100.0", "198.51.100.0"},
+		{"4-in-6 repeated RFC1918", "::ffff:192.168.1.1", "198.18.0.0"},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := anonymizer.AnonymizeIP(netip.MustParseAddr(tc.ip))
+			assert.Equal(t, tc.expect, result.String(), "strict level should replace internal ranges from the internal pools")
+		})
+	}
+}
+
+func TestAnonymizeString_StrictInternalIPs(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	input := "route 10.20.30.0/24 via 192.168.1.1 dev eth0 src 100.64.0.7"
+	firstPass := anonymizer.AnonymizeString(input)
+	secondPass := anonymizer.AnonymizeString(firstPass)
+
+	assert.NotContains(t, firstPass, "10.20.30.0", "private network address should be anonymized")
+	assert.NotContains(t, firstPass, "192.168.1.1", "private gateway should be anonymized")
+	assert.NotContains(t, firstPass, "100.64.0.7", "CGNAT address should be anonymized")
+	assert.Contains(t, firstPass, "/24", "prefix length should be preserved")
+	assert.Equal(t, firstPass, secondPass, "second pass should not further anonymize the string")
+}
+
+func TestAnonymizeMAC(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	first := anonymizer.AnonymizeMAC("aa:bb:cc:dd:ee:0f")
+	assert.Equal(t, "02:00:00:00:00:01", first, "first MAC should get the first placeholder")
+	assert.Equal(t, first, anonymizer.AnonymizeMAC("aa:bb:cc:dd:ee:0f"), "repeated MAC should map to the same placeholder")
+	assert.Equal(t, first, anonymizer.AnonymizeMAC("AA:BB:CC:DD:EE:0F"), "case should not affect the mapping")
+	assert.Equal(t, "02-00-00-00-00-01", anonymizer.AnonymizeMAC("AA-BB-CC-DD-EE-0F"), "dash form should keep its separator but share the mapping")
+
+	second := anonymizer.AnonymizeMAC("10:22:33:44:55:66")
+	assert.Equal(t, "02:00:00:00:00:02", second, "second distinct MAC should get the next placeholder")
+
+	tests := []struct {
+		name string
+		mac  string
+	}{
+		{"Broadcast", "ff:ff:ff:ff:ff:ff"},
+		{"IPv4 multicast", "01:00:5e:00:00:fb"},
+		{"IPv6 multicast", "33:33:00:00:00:01"},
+		{"All zero", "00:00:00:00:00:00"},
+		{"Assigned placeholder", "02:00:00:00:00:01"},
+		{"Invalid", "not-a-mac"},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			assert.Equal(t, tc.mac, anonymizer.AnonymizeMAC(tc.mac), "should be preserved")
+		})
+	}
+}
+
+func TestAnonymizeString_MACAddresses(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	tests := []struct {
+		name   string
+		input  string
+		expect string
+	}{
+		{
+			name:   "nftables ether rule",
+			input:  "ether saddr aa:bb:cc:dd:ee:ff drop",
+			expect: "ether saddr 02:00:00:00:00:01 drop",
+		},
+		{
+			name:   "Windows dash form",
+			input:  "Physical Address : AA-BB-CC-DD-EE-FF",
+			expect: "Physical Address : 02-00-00-00-00-01",
+		},
+		{
+			name:   "IPv6 address tail is not treated as MAC",
+			input:  "addr fe80:0:11:22:33:44:55:66 scope link",
+			expect: "addr fe80:0:11:22:33:44:55:66 scope link",
+		},
+		{
+			name:   "broadcast MAC preserved",
+			input:  "dst ff:ff:ff:ff:ff:ff type ARP",
+			expect: "dst ff:ff:ff:ff:ff:ff type ARP",
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := anonymizer.AnonymizeString(tc.input)
+			assert.Equal(t, tc.expect, result, "MAC addresses should be anonymized at every level")
+			assert.Equal(t, result, anonymizer.AnonymizeString(result), "second pass should not change the result")
+		})
+	}
+}
+
+func TestAnonymizeWGKey(t *testing.T) {
+	key := base64.StdEncoding.EncodeToString(bytes.Repeat([]byte{0x42}, 32))
+
+	t.Run("default level preserves keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		assert.Equal(t, key, anonymizer.AnonymizeWGKey(key), "default level should not touch WireGuard keys")
+	})
+
+	t.Run("strict level replaces keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		anonymizer.SetLevel(anonymize.LevelStrict)
+
+		anon := anonymizer.AnonymizeWGKey(key)
+		assert.NotEqual(t, key, anon, "strict level should replace the key")
+		assert.Regexp(t, `^[A-Za-z0-9+/]{43}=$`, anon, "placeholder should keep the WireGuard key shape")
+		assert.Equal(t, anon, anonymizer.AnonymizeWGKey(key), "repeated key should map to the same placeholder")
+		assert.Equal(t, anon, anonymizer.AnonymizeWGKey(anon), "an assigned placeholder should pass through unchanged")
+
+		assert.Equal(t, "not-a-key", anonymizer.AnonymizeWGKey("not-a-key"), "non-key values should be preserved")
+	})
+}
+
+func TestAnonymizeString_WGKeys(t *testing.T) {
+	key := base64.StdEncoding.EncodeToString(bytes.Repeat([]byte{0x42}, 32))
+	input := "peer " + key + " handshake completed"
+
+	t.Run("default level preserves keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		assert.Equal(t, input, anonymizer.AnonymizeString(input), "default level should not touch WireGuard keys in strings")
+	})
+
+	t.Run("strict level replaces keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		anonymizer.SetLevel(anonymize.LevelStrict)
+
+		firstPass := anonymizer.AnonymizeString(input)
+		assert.NotContains(t, firstPass, key, "the key should not survive strict anonymization")
+		assert.Equal(t, anonymizer.AnonymizeWGKey(key), extractKey(t, firstPass), "string replacement should be consistent with AnonymizeWGKey")
+		assert.Equal(t, firstPass, anonymizer.AnonymizeString(firstPass), "second pass should not change the result")
+	})
+}
+
+func extractKey(t *testing.T, logLine string) string {
+	t.Helper()
+	fields := strings.Fields(logLine)
+	require.Len(t, fields, 4, "log line should keep its structure")
+	return fields[1]
+}
+
+func TestAnonymizeDomain_StrictLevel(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	t.Run("netbird peer name", func(t *testing.T) {
+		result := anonymizer.AnonymizeDomain("my-laptop.netbird.cloud")
+		assert.Regexp(t, `^peer-\d+\.netbird\.cloud$`, result, "peer name should be anonymized, suffix kept")
+		assert.NotContains(t, result, "my-laptop", "the peer name should not survive")
+		assert.Equal(t, result, anonymizer.AnonymizeDomain("my-laptop.netbird.cloud"), "repeated domain should map consistently")
+		assert.Equal(t, result, anonymizer.AnonymizeDomain(result), "an anonymized domain should pass through unchanged")
+	})
+
+	t.Run("bare netbird domain", func(t *testing.T) {
+		assert.Equal(t, "netbird.cloud", anonymizer.AnonymizeDomain("netbird.cloud"), "the bare protected suffix should be preserved")
+	})
+
+	t.Run("netbird infrastructure preserved", func(t *testing.T) {
+		assert.Equal(t, "api.netbird.io", anonymizer.AnonymizeDomain("api.netbird.io"),
+			"netbird.io hosts infrastructure, not peer names, and should stay readable")
+	})
+
+	t.Run("leading labels of other domains", func(t *testing.T) {
+		result := anonymizer.AnonymizeDomain("host1.corp.example.com")
+		assert.Regexp(t, `^host-\d+\.host-\d+\.anon-[a-zA-Z0-9]+\.domain$`, result, "every label should be anonymized")
+		for _, label := range []string{"host1", "corp", "example"} {
+			assert.NotContains(t, result, label, "no original label should survive")
+		}
+		assert.Equal(t, result, anonymizer.AnonymizeDomain("host1.corp.example.com"), "repeated domain should map consistently")
+	})
+
+	t.Run("same label maps consistently across domains", func(t *testing.T) {
+		first := anonymizer.AnonymizeDomain("shared.one.com")
+		second := anonymizer.AnonymizeDomain("shared.two.com")
+		assert.Equal(t, strings.Split(first, ".")[0], strings.Split(second, ".")[0], "the shared host label should get one placeholder")
+	})
+
+	t.Run("wildcard label preserved", func(t *testing.T) {
+		result := anonymizer.AnonymizeDomain("*.example.com")
+		assert.Regexp(t, `^\*\.anon-[a-zA-Z0-9]+\.domain$`, result, "the wildcard label should stay a wildcard")
+	})
+}
+
+func TestAnonymizeDomain_DefaultLevelKeepsPeerNames(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	assert.Equal(t, "my-laptop.netbird.cloud", anonymizer.AnonymizeDomain("my-laptop.netbird.cloud"),
+		"default level should preserve netbird FQDNs including the peer name")
+	assert.Regexp(t, `^sub\.anon-[a-zA-Z0-9]+\.domain$`, anonymizer.AnonymizeDomain("sub.example.com"),
+		"default level should keep subdomain labels")
+}
+
+func TestAnonymizeString_StrictPeerNames(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	// Seed like the bundle generator does from the status: base first, then
+	// the full FQDN, so replacement must prefer the longer mapping.
+	anonBase := anonymizer.AnonymizeDomain("example.com")
+	anonPeer := anonymizer.AnonymizeDomain("peer1.netbird.cloud")
+	anonHost := anonymizer.AnonymizeDomain("host1.example.com")
+
+	logLine := "connected to peer1.netbird.cloud via host1.example.com endpoint"
+	firstPass := anonymizer.AnonymizeString(logLine)
+	assert.NotContains(t, firstPass, "peer1", "the peer name should not survive in logs")
+	assert.NotContains(t, firstPass, "host1", "the host label should not survive in logs")
+	assert.Contains(t, firstPass, anonPeer, "the seeded peer mapping should be applied")
+	assert.Contains(t, firstPass, anonHost, "the seeded host mapping should be applied, not just the base mapping")
+	assert.NotContains(t, firstPass, "host1."+anonBase, "the base mapping must not preempt the longer FQDN mapping")
+	assert.Equal(t, firstPass, anonymizer.AnonymizeString(firstPass), "second pass should not change the result")
+}
+
 func TestAnonymizeDNSLogLine(t *testing.T) {
 	anonymizer := anonymize.NewAnonymizer(netip.Addr{}, netip.Addr{})
 	tests := []struct {
@@ -274,17 +572,27 @@ func TestAnonymizeString_IPAddresses(t *testing.T) {
 		{
 			name:   "IPv6 Address",
 			input:  "Access attempted from 2001:db8::ff00:42",
-			expect: "Access attempted from 100::",
+			expect: "Access attempted from 2001:db8:ffff::",
 		},
 		{
 			name:   "IPv6 Address with Port",
 			input:  "Access attempted from [2001:db8::ff00:42]:8080",
-			expect: "Access attempted from [100::]:8080",
+			expect: "Access attempted from [2001:db8:ffff::]:8080",
 		},
 		{
 			name:   "Both IPv4 and IPv6",
 			input:  "IPv4: 142.108.0.1 and IPv6: 2001:db8::ff00:43",
-			expect: "IPv4: 198.51.100.1 and IPv6: 100::1",
+			expect: "IPv4: 198.51.100.1 and IPv6: 2001:db8:ffff::1",
+		},
+		{
+			name:   "STUN URI with IPv6",
+			input:  "Connecting to stun:[2001:db8::ff00:42]:3478",
+			expect: "Connecting to stun:[2001:db8:ffff::]:3478",
+		},
+		{
+			name:   "HTTPS URI with IPv6",
+			input:  "Visit https://[2001:db8::ff00:42]:443/path",
+			expect: "Visit https://[2001:db8:ffff::]:443/path",
 		},
 	}
 

client/anonymize/reverse_zone.go

@@ -0,0 +1,174 @@
+package anonymize
+
+import (
+	"encoding/hex"
+	"net/netip"
+	"regexp"
+	"strconv"
+	"strings"
+)
+
+const (
+	reverseZoneSuffixV4 = ".in-addr.arpa"
+	reverseZoneSuffixV6 = ".ip6.arpa"
+
+	v6Nibbles = 32
+	v4Octets  = 4
+)
+
+// reverseZoneRegexes match a reverse zone or a full reverse name in free text.
+// They are applied before the address passes of AnonymizeString, whose IPv4
+// pattern would otherwise consume the digit labels of a zone and replace parts
+// of it with unrelated addresses.
+var reverseZoneRegexes = []*regexp.Regexp{
+	regexp.MustCompile(`(?:[0-9]{1,3}\.){1,4}in-addr\.arpa\b`),
+	regexp.MustCompile(`(?:[0-9a-fA-F]\.){1,32}ip6\.arpa\b`),
+}
+
+// anonymizeReverseZone maps a reverse zone to the zone of the anonymized form
+// of the prefix it encodes, so it follows the address rules rather than the
+// domain ones: the zone of an address that is preserved is preserved too, and
+// the zone of one that is replaced names the replacement. This keeps a reverse
+// zone recognizable as such, and consistent with the addresses it belongs to
+// elsewhere in the same output. It reports false for anything that is not a
+// reverse zone.
+func (a *Anonymizer) anonymizeReverseZone(domain string) (string, bool) {
+	prefix, labelCount, suffix, ok := parseReverseZone(domain)
+	if !ok {
+		return "", false
+	}
+
+	anonymized := a.AnonymizeIP(prefix)
+	if anonymized == prefix {
+		return domain, true
+	}
+
+	return reverseZoneName(anonymized, labelCount) + suffix, true
+}
+
+// replaceReverseZones anonymizes every reverse zone in str and swaps each one
+// for a placeholder, returning a function that puts the anonymized zones back.
+// The placeholders carry no dots, digits or colons, so no later pass matches
+// them.
+func (a *Anonymizer) replaceReverseZones(str string) (string, func(string) string) {
+	var zones []string
+
+	for _, re := range reverseZoneRegexes {
+		str = re.ReplaceAllStringFunc(str, func(match string) string {
+			zone, ok := a.anonymizeReverseZone(match)
+			if !ok {
+				return match
+			}
+
+			zones = append(zones, zone)
+			return reverseZonePlaceholder(len(zones) - 1)
+		})
+	}
+
+	if len(zones) == 0 {
+		return str, func(s string) string { return s }
+	}
+
+	return str, func(s string) string {
+		for i, zone := range zones {
+			s = strings.ReplaceAll(s, reverseZonePlaceholder(i), zone)
+		}
+		return s
+	}
+}
+
+func reverseZonePlaceholder(index int) string {
+	return "\x00reversezone" + strconv.Itoa(index) + "\x00"
+}
+
+// parseReverseZone turns a reverse zone into the address of the prefix its
+// labels spell backwards, padding the absent low-order part with zeroes, and
+// returns the label count and zone suffix so the name can be rebuilt.
+func parseReverseZone(domain string) (netip.Addr, int, string, bool) {
+	lower := strings.ToLower(domain)
+
+	switch {
+	case strings.HasSuffix(lower, reverseZoneSuffixV4):
+		labels := strings.Split(strings.TrimSuffix(lower, reverseZoneSuffixV4), ".")
+		addr, ok := reverseZoneAddrV4(labels)
+		return addr, len(labels), reverseZoneSuffixV4, ok
+	case strings.HasSuffix(lower, reverseZoneSuffixV6):
+		labels := strings.Split(strings.TrimSuffix(lower, reverseZoneSuffixV6), ".")
+		addr, ok := reverseZoneAddrV6(labels)
+		return addr, len(labels), reverseZoneSuffixV6, ok
+	default:
+		return netip.Addr{}, 0, "", false
+	}
+}
+
+func reverseZoneAddrV4(labels []string) (netip.Addr, bool) {
+	if len(labels) == 0 || len(labels) > v4Octets {
+		return netip.Addr{}, false
+	}
+
+	var octets [v4Octets]byte
+	for i, label := range labels {
+		octet, err := strconv.ParseUint(label, 10, 8)
+		if err != nil {
+			return netip.Addr{}, false
+		}
+		octets[len(labels)-1-i] = byte(octet)
+	}
+
+	return netip.AddrFrom4(octets), true
+}
+
+func reverseZoneAddrV6(labels []string) (netip.Addr, bool) {
+	if len(labels) == 0 || len(labels) > v6Nibbles {
+		return netip.Addr{}, false
+	}
+
+	nibbles := make([]byte, 0, v6Nibbles)
+	for i := len(labels) - 1; i >= 0; i-- {
+		if len(labels[i]) != 1 || !isHexDigit(labels[i][0]) {
+			return netip.Addr{}, false
+		}
+		nibbles = append(nibbles, labels[i][0])
+	}
+	for len(nibbles) < v6Nibbles {
+		nibbles = append(nibbles, '0')
+	}
+
+	var groups []string
+	for i := 0; i < len(nibbles); i += 4 {
+		groups = append(groups, string(nibbles[i:i+4]))
+	}
+
+	addr, err := netip.ParseAddr(strings.Join(groups, ":"))
+	if err != nil {
+		return netip.Addr{}, false
+	}
+
+	return addr, true
+}
+
+// reverseZoneName spells the first labelCount labels of addr backwards, the
+// inverse of parseReverseZone, without the zone suffix.
+func reverseZoneName(addr netip.Addr, labelCount int) string {
+	labels := make([]string, 0, labelCount)
+
+	if addr.Is4() {
+		octets := addr.As4()
+		for i := labelCount - 1; i >= 0; i-- {
+			labels = append(labels, strconv.Itoa(int(octets[i])))
+		}
+		return strings.Join(labels, ".")
+	}
+
+	address := addr.As16()
+	nibbles := hex.EncodeToString(address[:])
+	for i := labelCount - 1; i >= 0; i-- {
+		labels = append(labels, string(nibbles[i]))
+	}
+
+	return strings.Join(labels, ".")
+}
+
+func isHexDigit(c byte) bool {
+	return c >= '0' && c <= '9' || c >= 'a' && c <= 'f' || c >= 'A' && c <= 'F'
+}

client/anonymize/reverse_zone_test.go

@@ -0,0 +1,171 @@
+package anonymize
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func newLeveledAnonymizer(level Level) *Anonymizer {
+	a := NewAnonymizer(DefaultAddresses())
+	a.SetLevel(level)
+	return a
+}
+
+// TestAnonymizeDomainReverseZone covers reverse zones going through the address
+// rules instead of the domain ones, so a zone stays a zone and an address that
+// is preserved keeps the zone that names it.
+func TestAnonymizeDomainReverseZone(t *testing.T) {
+	// 100.64.0.0/10 is the overlay range, which is CGNAT: preserved at the
+	// default level and replaced from the internal pool at the strict one
+	const overlayZone = "64.100.in-addr.arpa"
+
+	t.Run("overlay zone preserved at the default level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+		assert.Equal(t, overlayZone, a.AnonymizeDomain(overlayZone), "should keep the zone of a preserved address")
+	})
+
+	t.Run("private zone preserved at the default level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+		assert.Equal(t, "168.192.in-addr.arpa", a.AnonymizeDomain("168.192.in-addr.arpa"), "should keep the zone of a private address")
+	})
+
+	t.Run("overlay zone replaced at the strict level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelStrict)
+
+		got := a.AnonymizeDomain(overlayZone)
+		require.True(t, strings.HasSuffix(got, reverseZoneSuffixV4), "should stay a reverse zone, got %q", got)
+		assert.NotEqual(t, overlayZone, got, "should replace the encoded prefix")
+		assert.Len(t, strings.Split(strings.TrimSuffix(got, reverseZoneSuffixV4), "."), 2,
+			"should keep the label count, got %q", got)
+	})
+
+	t.Run("public zone replaced at the default level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		got := a.AnonymizeDomain("113.0.203.in-addr.arpa")
+		require.True(t, strings.HasSuffix(got, reverseZoneSuffixV4), "should stay a reverse zone, got %q", got)
+		assert.NotEqual(t, "113.0.203.in-addr.arpa", got, "should replace a public prefix")
+	})
+
+	t.Run("zone of an address keeps that address mapping", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		anonymizedAddr := a.AnonymizeIPString("203.0.113.7")
+		got := a.AnonymizeDomain("7.113.0.203.in-addr.arpa")
+
+		octets := strings.Split(anonymizedAddr, ".")
+		want := octets[3] + "." + octets[2] + "." + octets[1] + "." + octets[0] + reverseZoneSuffixV4
+		assert.Equal(t, want, got, "should name the same replacement as the address itself")
+	})
+
+	t.Run("ipv6 nibble labels stay single digits", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		zone := "0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0" + reverseZoneSuffixV6
+		got := a.AnonymizeDomain(zone)
+
+		require.True(t, strings.HasSuffix(got, reverseZoneSuffixV6), "should stay a reverse zone, got %q", got)
+		labels := strings.Split(strings.TrimSuffix(got, reverseZoneSuffixV6), ".")
+		assert.Len(t, labels, 28, "should keep every nibble label, got %q", got)
+		for _, label := range labels {
+			assert.Len(t, label, 1, "nibble label %q should stay a single digit", label)
+		}
+	})
+
+	t.Run("trailing dot is kept", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+		assert.Equal(t, "64.100.in-addr.arpa.", a.AnonymizeDomain("64.100.in-addr.arpa."), "should keep the trailing dot")
+	})
+
+	t.Run("a domain that only looks like a zone is anonymized as a domain", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		got := a.AnonymizeDomain("not-a-zone.in-addr.arpa")
+		assert.NotContains(t, got, "in-addr.arpa", "should fall back to domain anonymization")
+	})
+}
+
+// TestAnonymizeStringReverseZone verifies that a zone inside free text, such as
+// a DNS log line, is not chewed up by the address passes. The IPv4 pattern
+// matches any run of dotted digits, which a reverse zone is made of.
+func TestAnonymizeStringReverseZone(t *testing.T) {
+	t.Run("ipv6 zone survives the address passes", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		zone := "0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0" + reverseZoneSuffixV6
+		got := a.AnonymizeString("question: domain=" + zone + " type=PTR")
+
+		assert.Contains(t, got, "type=PTR", "should keep the rest of the line")
+		assert.NotContains(t, got, "198.51.100", "should not rewrite nibble labels as an address")
+
+		labels := strings.Split(strings.TrimSuffix(strings.TrimPrefix(got, "question: domain="), reverseZoneSuffixV6+" type=PTR"), ".")
+		assert.Len(t, labels, 28, "should keep every nibble label, got %q", got)
+	})
+
+	t.Run("preserved ipv4 zone is untouched", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		line := "reverse zone 64.100.in-addr.arpa registered"
+		assert.Equal(t, line, a.AnonymizeString(line), "should keep the zone of a preserved address")
+	})
+
+	t.Run("public ipv4 zone is replaced consistently", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		got := a.AnonymizeString("zone 113.0.203.in-addr.arpa and address 203.0.113.7")
+		assert.NotContains(t, got, "113.0.203.in-addr.arpa", "should replace the zone")
+		assert.NotContains(t, got, "203.0.113.7", "should replace the address")
+		assert.Contains(t, got, reverseZoneSuffixV4, "should keep the zone suffix")
+	})
+}
+
+func TestParseReverseZone(t *testing.T) {
+	tests := []struct {
+		name   string
+		zone   string
+		addr   string
+		labels int
+	}{
+		{name: "v4 two labels", zone: "0.100" + reverseZoneSuffixV4, addr: "100.0.0.0", labels: 2},
+		{name: "v4 three labels", zone: "1.168.192" + reverseZoneSuffixV4, addr: "192.168.1.0", labels: 3},
+		{name: "v4 full address", zone: "7.113.0.203" + reverseZoneSuffixV4, addr: "203.0.113.7", labels: 4},
+		{
+			name:   "v6 prefix",
+			zone:   "0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0" + reverseZoneSuffixV6,
+			addr:   "2::",
+			labels: 28,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			addr, labels, suffix, ok := parseReverseZone(tc.zone)
+			require.True(t, ok, "should decode the reverse zone")
+			assert.Equal(t, tc.addr, addr.String(), "should decode to the encoded prefix")
+			assert.Equal(t, tc.labels, labels, "should count the labels")
+			assert.Equal(t, tc.zone, reverseZoneName(addr, labels)+suffix, "should re-encode to the original zone")
+		})
+	}
+}
+
+func TestParseReverseZoneRejectsNonZones(t *testing.T) {
+	tests := []string{
+		"example.com",
+		"in-addr.arpa",
+		"x.100" + reverseZoneSuffixV4,
+		"256" + reverseZoneSuffixV4,
+		"1.2.3.4.5" + reverseZoneSuffixV4,
+		"ab" + reverseZoneSuffixV6,
+		"g" + reverseZoneSuffixV6,
+	}
+
+	for _, zone := range tests {
+		t.Run(zone, func(t *testing.T) {
+			_, _, _, ok := parseReverseZone(zone)
+			assert.False(t, ok, "should reject %q", zone)
+		})
+	}
+}

client/cmd/capture.go

@@ -0,0 +1,196 @@
+package cmd
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"os"
+	"os/signal"
+	"path/filepath"
+	"strings"
+	"syscall"
+
+	"github.com/hashicorp/go-multierror"
+	"github.com/spf13/cobra"
+	"google.golang.org/grpc/status"
+	"google.golang.org/protobuf/types/known/durationpb"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	"github.com/netbirdio/netbird/client/proto"
+	"github.com/netbirdio/netbird/util/capture"
+)
+
+var captureCmd = &cobra.Command{
+	Use:   "capture",
+	Short: "Capture packets on the WireGuard interface",
+	Long: `Captures decrypted packets flowing through the WireGuard interface.
+
+Default output is human-readable text. Use --pcap or --output for pcap binary.
+Requires --enable-capture to be set at service install or reconfigure time.
+
+Examples:
+  netbird debug capture
+  netbird debug capture host 100.64.0.1 and port 443
+  netbird debug capture tcp
+  netbird debug capture icmp
+  netbird debug capture src host 10.0.0.1 and dst port 80
+  netbird debug capture -o capture.pcap
+  netbird debug capture --pcap | tshark -r -
+  netbird debug capture --pcap | tcpdump -r - -n`,
+	Args: cobra.ArbitraryArgs,
+	RunE: runCapture,
+}
+
+func init() {
+	debugCmd.AddCommand(captureCmd)
+
+	captureCmd.Flags().Bool("pcap", false, "Force pcap binary output (default when --output is set)")
+	captureCmd.Flags().BoolP("verbose", "v", false, "Show seq/ack, TTL, window, total length")
+	captureCmd.Flags().Bool("ascii", false, "Print payload as ASCII after each packet (useful for HTTP)")
+	captureCmd.Flags().Uint32("snap-len", 0, "Max bytes per packet (0 = full)")
+	captureCmd.Flags().DurationP("duration", "d", 0, "Capture duration (0 = until interrupted)")
+	captureCmd.Flags().StringP("output", "o", "", "Write pcap to file instead of stdout")
+}
+
+func runCapture(cmd *cobra.Command, args []string) error {
+	conn, err := getClient(cmd)
+	if err != nil {
+		return err
+	}
+	defer func() {
+		if err := conn.Close(); err != nil {
+			cmd.PrintErrf(errCloseConnection, err)
+		}
+	}()
+
+	client := proto.NewDaemonServiceClient(conn)
+
+	req, err := buildCaptureRequest(cmd, args)
+	if err != nil {
+		return err
+	}
+
+	ctx, cancel := signal.NotifyContext(cmd.Context(), syscall.SIGINT, syscall.SIGTERM)
+	defer cancel()
+
+	stream, err := client.StartCapture(ctx, req)
+	if err != nil {
+		return handleCaptureError(err)
+	}
+
+	// First Recv is the empty acceptance message from the server. If the
+	// device is unavailable (kernel WG, not connected, capture disabled),
+	// the server returns an error instead.
+	if _, err := stream.Recv(); err != nil {
+		return handleCaptureError(err)
+	}
+
+	out, cleanup, err := captureOutput(cmd)
+	if err != nil {
+		return err
+	}
+
+	if req.TextOutput {
+		cmd.PrintErrf("Capturing packets... Press Ctrl+C to stop.\n")
+	} else {
+		cmd.PrintErrf("Capturing packets (pcap)... Press Ctrl+C to stop.\n")
+	}
+
+	streamErr := streamCapture(ctx, cmd, stream, out)
+	cleanupErr := cleanup()
+	if streamErr != nil {
+		return streamErr
+	}
+	return cleanupErr
+}
+
+func buildCaptureRequest(cmd *cobra.Command, args []string) (*proto.StartCaptureRequest, error) {
+	req := &proto.StartCaptureRequest{}
+
+	if len(args) > 0 {
+		expr := strings.Join(args, " ")
+		if _, err := capture.ParseFilter(expr); err != nil {
+			return nil, fmt.Errorf("invalid filter: %w", err)
+		}
+		req.FilterExpr = expr
+	}
+
+	if snap, _ := cmd.Flags().GetUint32("snap-len"); snap > 0 {
+		req.SnapLen = snap
+	}
+	if d, _ := cmd.Flags().GetDuration("duration"); d != 0 {
+		if d < 0 {
+			return nil, fmt.Errorf("duration must not be negative")
+		}
+		req.Duration = durationpb.New(d)
+	}
+	req.Verbose, _ = cmd.Flags().GetBool("verbose")
+	req.Ascii, _ = cmd.Flags().GetBool("ascii")
+
+	outPath, _ := cmd.Flags().GetString("output")
+	forcePcap, _ := cmd.Flags().GetBool("pcap")
+	req.TextOutput = !forcePcap && outPath == ""
+
+	return req, nil
+}
+
+func streamCapture(ctx context.Context, cmd *cobra.Command, stream proto.DaemonService_StartCaptureClient, out io.Writer) error {
+	for {
+		pkt, err := stream.Recv()
+		if err != nil {
+			if ctx.Err() != nil {
+				cmd.PrintErrf("\nCapture stopped.\n")
+				return nil //nolint:nilerr // user interrupted
+			}
+			if err == io.EOF {
+				cmd.PrintErrf("\nCapture finished.\n")
+				return nil
+			}
+			return handleCaptureError(err)
+		}
+		if _, err := out.Write(pkt.GetData()); err != nil {
+			return fmt.Errorf("write output: %w", err)
+		}
+	}
+}
+
+// captureOutput returns the writer for capture data and a cleanup function
+// that finalizes the file. Errors from the cleanup must be propagated.
+func captureOutput(cmd *cobra.Command) (io.Writer, func() error, error) {
+	outPath, _ := cmd.Flags().GetString("output")
+	if outPath == "" {
+		return os.Stdout, func() error { return nil }, nil
+	}
+
+	f, err := os.CreateTemp(filepath.Dir(outPath), filepath.Base(outPath)+".*.tmp")
+	if err != nil {
+		return nil, nil, fmt.Errorf("create output file: %w", err)
+	}
+	tmpPath := f.Name()
+	return f, func() error {
+		var merr *multierror.Error
+		if err := f.Close(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("close output file: %w", err))
+		}
+		fi, statErr := os.Stat(tmpPath)
+		if statErr != nil || fi.Size() == 0 {
+			if rmErr := os.Remove(tmpPath); rmErr != nil && !os.IsNotExist(rmErr) {
+				merr = multierror.Append(merr, fmt.Errorf("remove empty output file: %w", rmErr))
+			}
+			return nberrors.FormatErrorOrNil(merr)
+		}
+		if err := os.Rename(tmpPath, outPath); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("rename output file: %w", err))
+			return nberrors.FormatErrorOrNil(merr)
+		}
+		cmd.PrintErrf("Wrote %s\n", outPath)
+		return nberrors.FormatErrorOrNil(merr)
+	}, nil
+}
+
+func handleCaptureError(err error) error {
+	if s, ok := status.FromError(err); ok {
+		return fmt.Errorf("%s", s.Message())
+	}
+	return err
+}

client/cmd/daemon_error.go

@@ -0,0 +1,66 @@
+package cmd
+
+import (
+	"errors"
+	"fmt"
+	"strings"
+
+	"google.golang.org/genproto/googleapis/rpc/errdetails"
+	gstatus "google.golang.org/grpc/status"
+
+	"github.com/netbirdio/netbird/client/internal/ipcauth"
+)
+
+// daemonCallError prepares a daemon error for display. A refusal the daemon
+// raised because the operation needs root/administrator is already guidance
+// written for the user, so it is surfaced on its own instead of buried under the
+// gRPC envelope and the name of the RPC that hit it. Anything else is wrapped
+// with context as usual.
+func daemonCallError(context string, err error) error {
+	if guidance, ok := privilegeGuidance(err); ok {
+		return errors.New(guidance)
+	}
+	return fmt.Errorf("%s: %w", context, err)
+}
+
+// privilegeGuidance renders the daemon's privilege refusal as a summary and the
+// command that performs the operation with the privileges it needs. It reports
+// false for any other error.
+func privilegeGuidance(err error) (string, bool) {
+	info, ok := privilegeErrorInfo(err)
+	if !ok {
+		return "", false
+	}
+
+	summary := info.GetMetadata()[ipcauth.ErrorMetaSummary]
+	command := info.GetMetadata()[ipcauth.ErrorMetaCommand]
+	if summary == "" {
+		// Detail without a summary: fall back to the status message, which
+		// carries the same text.
+		summary = strings.TrimSpace(gstatus.Convert(err).Message())
+	}
+	if command == "" {
+		return summary, true
+	}
+
+	return fmt.Sprintf("%s\n\n    %s\n", summary, command), true
+}
+
+// privilegeErrorInfo returns the daemon's privilege-refusal detail, if the error
+// carries one.
+func privilegeErrorInfo(err error) (*errdetails.ErrorInfo, bool) {
+	if err == nil {
+		return nil, false
+	}
+
+	for _, detail := range gstatus.Convert(err).Details() {
+		info, ok := detail.(*errdetails.ErrorInfo)
+		if !ok {
+			continue
+		}
+		if info.GetReason() == ipcauth.ErrorReasonPrivilegeRequired && info.GetDomain() == ipcauth.ErrorDomain {
+			return info, true
+		}
+	}
+	return nil, false
+}

client/cmd/debug.go

@@ -9,6 +9,8 @@ import (
 	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/status"
+	"google.golang.org/protobuf/encoding/protojson"
+	"google.golang.org/protobuf/types/known/durationpb"
 
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/debug"
@@ -18,15 +20,17 @@ import (
 	"github.com/netbirdio/netbird/client/server"
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
 	"github.com/netbirdio/netbird/upload-server/types"
+	"github.com/netbirdio/netbird/version"
 )
 
 const errCloseConnection = "Failed to close connection: %v"
 
 var (
-	logFileCount        uint32
-	systemInfoFlag      bool
-	uploadBundleFlag    bool
-	uploadBundleURLFlag string
+	logFileCount             uint32
+	systemInfoFlag           bool
+	uploadBundleFlag         bool
+	uploadBundleURLFlag      string
+	uploadBundleInsecureFlag bool
 )
 
 var debugCmd = &cobra.Command{
@@ -83,7 +87,79 @@ var persistenceCmd = &cobra.Command{
 	RunE:    setSyncResponsePersistence,
 }
 
+var debugConfigCmd = &cobra.Command{
+	Use:     "config",
+	Example: "  netbird debug config",
+	Short:   "Dump the effective configuration",
+	Long:    "Prints the daemon's resolved configuration (after applying defaults, file, env, CLI input, and MDM policy overrides) as JSON. Includes the list of MDM-managed fields.",
+	RunE:    debugConfigDump,
+}
+
+// debugConfigDump implements `netbird debug config`. It resolves the
+// active profile, queries the daemon for the effective configuration
+// via GetConfig, and prints the resulting GetConfigResponse as JSON
+// (via protojson with EmitUnpopulated=true so the output is stable
+// across runs and includes zero-valued fields).
+//
+// Useful for verifying MDM enforcement end-to-end: the response's
+// mDMManagedFields array is the single source of truth for "which
+// fields is the daemon currently enforcing from the MDM source", and
+// every config field side-by-side with that list confirms the merge
+// result. Secrets in the response (e.g. PreSharedKey) are already
+// redacted by the daemon-side handler.
+func debugConfigDump(cmd *cobra.Command, _ []string) error {
+	pm := profilemanager.NewProfileManager()
+	activeProf, err := pm.GetActiveProfile()
+	if err != nil {
+		return fmt.Errorf("get active profile: %v", err)
+	}
+	currUser, err := profilemanager.InvokingUser()
+	if err != nil {
+		return fmt.Errorf("get current user: %v", err)
+	}
+
+	conn, err := getClient(cmd)
+	if err != nil {
+		return err
+	}
+	defer func() {
+		if err := conn.Close(); err != nil {
+			log.Errorf(errCloseConnection, err)
+		}
+	}()
+
+	client := proto.NewDaemonServiceClient(conn)
+	resp, err := client.GetConfig(cmd.Context(), &proto.GetConfigRequest{
+		ProfileName: string(activeProf.ID),
+		Username:    currUser.Username,
+	})
+	if err != nil {
+		return fmt.Errorf("failed to get config: %v", status.Convert(err).Message())
+	}
+
+	// Use protojson so well-known fields render correctly; emit defaults so
+	// the operator sees every field even when zero/empty.
+	m := protojson.MarshalOptions{Multiline: true, Indent: "  ", EmitUnpopulated: true}
+	out, err := m.Marshal(resp)
+	if err != nil {
+		return fmt.Errorf("marshal config: %w", err)
+	}
+	cmd.Println(string(out))
+	return nil
+}
+
+// debugBundle requests the daemon to create a debug bundle and prints
+// the resulting local file path and, if uploaded, the uploaded file
+// key. It uses the package flags (anonymize, system info, log file
+// count, CLI version, optional upload URL) to configure the bundle
+// request. Returns an error if the RPC fails or if the daemon reports
+// an upload failure reason.
 func debugBundle(cmd *cobra.Command, _ []string) error {
+	anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
+	if err != nil {
+		return err
+	}
+
 	conn, err := getClient(cmd)
 	if err != nil {
 		return err
@@ -96,16 +172,19 @@ func debugBundle(cmd *cobra.Command, _ []string) error {
 
 	client := proto.NewDaemonServiceClient(conn)
 	request := &proto.DebugBundleRequest{
-		Anonymize:    anonymizeFlag,
-		SystemInfo:   systemInfoFlag,
-		LogFileCount: logFileCount,
+		Anonymize:      anonymizeEnabled,
+		AnonymizeLevel: anonymizeLevel.String(),
+		SystemInfo:     systemInfoFlag,
+		LogFileCount:   logFileCount,
+		CliVersion:     version.NetbirdVersion(),
 	}
 	if uploadBundleFlag {
 		request.UploadURL = uploadBundleURLFlag
+		request.UploadInsecure = uploadBundleInsecureFlag
 	}
 	resp, err := client.DebugBundle(cmd.Context(), request)
 	if err != nil {
-		return fmt.Errorf("failed to bundle debug: %v", status.Convert(err).Message())
+		return daemonCallError("bundle debug", err)
 	}
 	cmd.Printf("Local file:\n%s\n", resp.GetPath())
 
@@ -155,6 +234,11 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 		return fmt.Errorf("invalid duration format: %v", err)
 	}
 
+	anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
+	if err != nil {
+		return err
+	}
+
 	conn, err := getClient(cmd)
 	if err != nil {
 		return err
@@ -239,11 +323,50 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 		}()
 	}
 
+	captureStarted := false
+	if wantCapture, _ := cmd.Flags().GetBool("capture"); wantCapture {
+		captureTimeout := duration + 30*time.Second
+		const maxBundleCapture = 10 * time.Minute
+		if captureTimeout > maxBundleCapture {
+			captureTimeout = maxBundleCapture
+		}
+		_, err := client.StartBundleCapture(cmd.Context(), &proto.StartBundleCaptureRequest{
+			Timeout: durationpb.New(captureTimeout),
+		})
+		if err != nil {
+			cmd.PrintErrf("Failed to start packet capture: %v\n", status.Convert(err).Message())
+		} else {
+			captureStarted = true
+			cmd.Println("Packet capture started.")
+			// Safety: always stop on exit, even if the normal stop below runs too.
+			defer func() {
+				if captureStarted {
+					stopCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+					defer cancel()
+					if _, err := client.StopBundleCapture(stopCtx, &proto.StopBundleCaptureRequest{}); err != nil {
+						cmd.PrintErrf("Failed to stop packet capture: %v\n", err)
+					}
+				}
+			}()
+		}
+	}
+
 	if waitErr := waitForDurationOrCancel(cmd.Context(), duration, cmd); waitErr != nil {
 		return waitErr
 	}
 	cmd.Println("\nDuration completed")
 
+	if captureStarted {
+		stopCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+		defer cancel()
+		if _, err := client.StopBundleCapture(stopCtx, &proto.StopBundleCaptureRequest{}); err != nil {
+			cmd.PrintErrf("Failed to stop packet capture: %v\n", err)
+		} else {
+			captureStarted = false
+			cmd.Println("Packet capture stopped.")
+		}
+	}
+
 	if cpuProfilingStarted {
 		if _, err := client.StopCPUProfile(cmd.Context(), &proto.StopCPUProfileRequest{}); err != nil {
 			cmd.PrintErrf("Failed to stop CPU profiling: %v\n", err)
@@ -255,16 +378,19 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 	cmd.Println("Creating debug bundle...")
 
 	request := &proto.DebugBundleRequest{
-		Anonymize:    anonymizeFlag,
-		SystemInfo:   systemInfoFlag,
-		LogFileCount: logFileCount,
+		Anonymize:      anonymizeEnabled,
+		AnonymizeLevel: anonymizeLevel.String(),
+		SystemInfo:     systemInfoFlag,
+		LogFileCount:   logFileCount,
+		CliVersion:     version.NetbirdVersion(),
 	}
 	if uploadBundleFlag {
 		request.UploadURL = uploadBundleURLFlag
+		request.UploadInsecure = uploadBundleInsecureFlag
 	}
 	resp, err := client.DebugBundle(cmd.Context(), request)
 	if err != nil {
-		return fmt.Errorf("failed to bundle debug: %v", status.Convert(err).Message())
+		return daemonCallError("bundle debug", err)
 	}
 
 	if needsRestoreUp {
@@ -392,6 +518,7 @@ func generateDebugBundle(config *profilemanager.Config, recorder *peer.Status, c
 			SyncResponse:   syncResponse,
 			LogPath:        logFilePath,
 			CPUProfile:     nil,
+			DaemonVersion:  version.NetbirdVersion(), // acting as daemon
 		},
 		debug.BundleConfig{
 			IncludeSystemInfo: true,
@@ -411,9 +538,12 @@ func init() {
 	debugBundleCmd.Flags().BoolVarP(&systemInfoFlag, "system-info", "S", true, "Adds system information to the debug bundle")
 	debugBundleCmd.Flags().BoolVarP(&uploadBundleFlag, "upload-bundle", "U", false, "Uploads the debug bundle to a server")
 	debugBundleCmd.Flags().StringVar(&uploadBundleURLFlag, "upload-bundle-url", types.DefaultBundleURL, "Service URL to get an URL to upload the debug bundle")
+	debugBundleCmd.Flags().BoolVar(&uploadBundleInsecureFlag, "upload-bundle-insecure", false, "Allow uploading to an http or untrusted-TLS upload server (self-hosted); requires root")
 
 	forCmd.Flags().Uint32VarP(&logFileCount, "log-file-count", "C", 1, "Number of rotated log files to include in debug bundle")
 	forCmd.Flags().BoolVarP(&systemInfoFlag, "system-info", "S", true, "Adds system information to the debug bundle")
 	forCmd.Flags().BoolVarP(&uploadBundleFlag, "upload-bundle", "U", false, "Uploads the debug bundle to a server")
 	forCmd.Flags().StringVar(&uploadBundleURLFlag, "upload-bundle-url", types.DefaultBundleURL, "Service URL to get an URL to upload the debug bundle")
+	forCmd.Flags().BoolVar(&uploadBundleInsecureFlag, "upload-bundle-insecure", false, "Allow uploading to an http or untrusted-TLS upload server (self-hosted); requires root")
+	forCmd.Flags().Bool("capture", false, "Capture packets during the debug duration and include in bundle")
 }

client/cmd/jobs.go

@@ -0,0 +1,13 @@
+package cmd
+
+// remoteJobsAllowedFlag opts this peer into running remote jobs (e.g. debug
+// bundles) requested by the management server. It defaults to false: remote
+// jobs are an explicit opt-in, and enabling it is a privileged change (see the
+// daemon gate in client/server), mirroring the SSH server opt-in.
+const remoteJobsAllowedFlag = "allow-remote-jobs"
+
+var remoteJobsAllowed bool
+
+func init() {
+	upCmd.PersistentFlags().BoolVar(&remoteJobsAllowed, remoteJobsAllowedFlag, false, "Allow the management server to run remote jobs (e.g. debug bundles) on this peer")
+}

client/cmd/kubernetes.go

@@ -0,0 +1,301 @@
+package cmd
+
+import (
+	"context"
+	"crypto/tls"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"net/http"
+	"net/url"
+	"os"
+	"path/filepath"
+	"slices"
+	"strings"
+
+	"github.com/goccy/go-yaml"
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+
+	"github.com/netbirdio/netbird/client/proto"
+)
+
+const (
+	KubernetesDNSSuffix = "netbird-kubeapi-proxy"
+)
+
+var kubernetesCmd = &cobra.Command{
+	Use:   "kubernetes",
+	Short: "Kubernetes cluster commands.",
+	Long:  "Kubernetes cluster commands.",
+}
+
+var kubernetesListCmd = &cobra.Command{
+	Use:   "list",
+	RunE:  kubernetesList,
+	Short: "List Kubernetes clusters.",
+	Long:  "List Kubernetes clusters by discovering NetBird peers running netbird-kubeapi-proxy.",
+}
+
+var kubernetesWriteKubeconfigCmd = &cobra.Command{
+	Use:   "write-kubeconfig",
+	RunE:  kubernetesWriteKubeconfig,
+	Args:  cobra.ExactArgs(1),
+	Short: "Write kubeconfig for a Kubernetes cluster.",
+	Long:  "Updates kubeconfig in place to allow token-less access to the Kubernetes cluster through NetBird.",
+}
+
+func init() {
+	kubernetesWriteKubeconfigCmd.Flags().String("kubeconfig", "", "path to kubeconfig file")
+}
+
+func kubernetesList(cmd *cobra.Command, _ []string) error {
+	conn, err := getClient(cmd)
+	if err != nil {
+		return err
+	}
+	defer conn.Close()
+	client := proto.NewDaemonServiceClient(conn)
+	statusResp, err := client.Status(cmd.Context(), &proto.StatusRequest{GetFullPeerStatus: true})
+	if err != nil {
+		return err
+	}
+
+	kcs, err := getKubernetesClusters(cmd.Context(), statusResp.FullStatus.Peers, "")
+	if err != nil {
+		return err
+	}
+	if len(kcs) == 0 {
+		cmd.Println("No Kubernetes clusters available.")
+		return nil
+	}
+	cmd.Println("Available Kubernetes clusters:")
+	for _, k := range kcs {
+		cmd.Printf("\n  - Name: %s\n    FQDN: %s\n    Version: %s\n", k.name, k.url.Host, k.version)
+	}
+	return nil
+}
+
+func kubernetesWriteKubeconfig(cmd *cobra.Command, args []string) error {
+	kubeconfigPath, err := resolveKubeconfigPath(cmd)
+	if err != nil {
+		return err
+	}
+
+	conn, err := getClient(cmd)
+	if err != nil {
+		return err
+	}
+	defer conn.Close()
+	client := proto.NewDaemonServiceClient(conn)
+	statusResp, err := client.Status(cmd.Context(), &proto.StatusRequest{GetFullPeerStatus: true})
+	if err != nil {
+		return err
+	}
+
+	clusterName := args[0]
+	kcs, err := getKubernetesClusters(cmd.Context(), statusResp.FullStatus.Peers, clusterName)
+	if err != nil {
+		return err
+	}
+	if len(kcs) == 0 {
+		return fmt.Errorf("kubernetes cluster named %s not found", clusterName)
+	}
+	if len(kcs) > 1 {
+		return fmt.Errorf("too many Kubernetes clusters returned")
+	}
+	err = writeKubeconfig(kubeconfigPath, kcs[0])
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+type kubernetesCluster struct {
+	name    string
+	url     *url.URL
+	version string
+}
+
+func getKubernetesClusters(ctx context.Context, peers []*proto.PeerState, nameFilter string) ([]kubernetesCluster, error) {
+	transport := http.DefaultTransport.(*http.Transport).Clone()
+	transport.TLSClientConfig = &tls.Config{
+		InsecureSkipVerify: true,
+	}
+	httpClient := &http.Client{
+		Transport: transport,
+	}
+	resolver := net.Resolver{
+		// Required so both DNS records are returned.
+		// https://github.com/golang/go/issues/17093
+		PreferGo: true,
+	}
+
+	kcs := []kubernetesCluster{}
+	attempted := map[string]struct{}{}
+	for _, peer := range peers {
+		fqdns, err := resolver.LookupAddr(ctx, peer.IP)
+		if err != nil {
+			return nil, err
+		}
+		for _, fqdn := range fqdns {
+			if _, ok := attempted[fqdn]; ok {
+				continue
+			}
+			attempted[fqdn] = struct{}{}
+			comps := strings.Split(fqdn, ".")
+			if len(comps) < 2 {
+				continue
+			}
+			if comps[1] != KubernetesDNSSuffix {
+				continue
+			}
+			if nameFilter != "" && nameFilter != comps[0] {
+				continue
+			}
+			clusterURL, clusterVersion, err := fingerprintClusters(ctx, httpClient, fqdn)
+			if err != nil {
+				log.Debugf("could not fingerprint Kubernetes cluster %s %q", fqdn, err)
+				continue
+			}
+			kc := kubernetesCluster{
+				name:    comps[0],
+				url:     clusterURL,
+				version: clusterVersion,
+			}
+			if nameFilter != "" {
+				return []kubernetesCluster{kc}, nil
+			}
+			kcs = append(kcs, kc)
+		}
+	}
+	return kcs, nil
+}
+
+func fingerprintClusters(ctx context.Context, httpClient *http.Client, fqdn string) (*url.URL, string, error) {
+	clusterURL, err := url.Parse("https://" + fqdn)
+	if err != nil {
+		return nil, "", err
+	}
+	versionURL, err := clusterURL.Parse("/version")
+	if err != nil {
+		return nil, "", err
+	}
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, versionURL.String(), nil)
+	if err != nil {
+		return nil, "", err
+	}
+	resp, err := httpClient.Do(req)
+	if err != nil {
+		return nil, "", err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode != http.StatusOK {
+		return nil, "", fmt.Errorf("expected %d response but got %s", http.StatusOK, resp.Status)
+	}
+	b, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return nil, "", err
+	}
+	versionData := map[string]string{}
+	err = json.Unmarshal(b, &versionData)
+	if err != nil {
+		return nil, "", err
+	}
+	version, ok := versionData["gitVersion"]
+	if !ok {
+		return nil, "", errors.New("no version found in response")
+	}
+	return clusterURL, version, nil
+}
+
+func resolveKubeconfigPath(cmd *cobra.Command) (string, error) {
+	if cmd.Flags().Changed("kubeconfig") {
+		path, err := cmd.Flags().GetString("kubeconfig")
+		if err != nil {
+			return "", err
+		}
+		return path, nil
+	}
+	if env := os.Getenv("KUBECONFIG"); env != "" {
+		return env, nil
+	}
+	home, err := os.UserHomeDir()
+	if err != nil {
+		return "", fmt.Errorf("could not determine home directory: %w", err)
+	}
+	return filepath.Join(home, ".kube", "config"), nil
+}
+
+func writeKubeconfig(kubeconfigPath string, kc kubernetesCluster) error {
+	b, err := os.ReadFile(kubeconfigPath)
+	if err != nil && !errors.Is(err, os.ErrNotExist) {
+		return err
+	}
+	var cfg map[string]any
+	if err := yaml.Unmarshal(b, &cfg); err != nil {
+		return err
+	}
+	if cfg == nil {
+		cfg = map[string]any{
+			"apiVersion": "v1",
+			"kind":       "Config",
+		}
+	}
+
+	cfg["clusters"] = appendWithName(cfg["clusters"], map[string]any{
+		"name": kc.name,
+		"cluster": map[string]any{
+			"server":                   kc.url.String(),
+			"insecure-skip-tls-verify": true,
+		},
+	})
+	cfg["users"] = appendWithName(cfg["users"], map[string]any{
+		"name": "netbird",
+		"user": map[string]any{
+			"token": "none",
+		},
+	})
+	cfg["contexts"] = appendWithName(cfg["contexts"], map[string]any{
+		"name": kc.name,
+		"context": map[string]any{
+			"cluster":   kc.name,
+			"user":      "netbird",
+			"namespace": "default",
+		},
+	})
+	cfg["current-context"] = kc.name
+
+	out, err := yaml.Marshal(cfg)
+	if err != nil {
+		return err
+	}
+	if err := os.WriteFile(kubeconfigPath, out, 0o600); err != nil {
+		return err
+	}
+	return nil
+}
+
+func appendWithName(data any, add map[string]any) any {
+	if data == nil {
+		return []any{add}
+	}
+	v, ok := data.([]any)
+	if !ok {
+		return []any{add}
+	}
+	i := slices.IndexFunc(v, func(item any) bool {
+		m, ok := item.(map[string]any)
+		if !ok {
+			return false
+		}
+		return m["name"] == add["name"]
+	})
+	if i == -1 {
+		return append(v, add)
+	}
+	v[i] = add
+	return v
+}

client/cmd/kubernetes_test.go

@@ -0,0 +1,120 @@
+package cmd
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"net/url"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/spf13/cobra"
+	"github.com/stretchr/testify/require"
+)
+
+func TestFingerprintClusters(t *testing.T) {
+	t.Parallel()
+
+	srv := httptest.NewTLSServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		//nolint: errcheck
+		w.Write([]byte(`{"gitVersion": "foobar"}`))
+	}))
+	defer srv.Close()
+
+	clusterURL, clusterVersion, err := fingerprintClusters(t.Context(), srv.Client(), srv.Listener.Addr().String())
+	require.NoError(t, err)
+	require.Equal(t, srv.URL, clusterURL.String())
+	require.Equal(t, "foobar", clusterVersion)
+}
+
+func TestResolveKubeconfigPath(t *testing.T) {
+	home, err := os.UserHomeDir()
+	if err != nil {
+		t.Fatalf("could not determine home directory: %v", err)
+	}
+	defaultPath := filepath.Join(home, ".kube", "config")
+	path, err := resolveKubeconfigPath(&cobra.Command{})
+	require.NoError(t, err)
+	require.Equal(t, defaultPath, path)
+
+	flagPath := "flag-path"
+	cmd := &cobra.Command{}
+	cmd.Flags().String("kubeconfig", "", "")
+	err = cmd.Flags().Set("kubeconfig", flagPath)
+	require.NoError(t, err)
+	path, err = resolveKubeconfigPath(cmd)
+	require.NoError(t, err)
+	require.Equal(t, flagPath, path)
+
+	envPath := "env-path"
+	t.Setenv("KUBECONFIG", envPath)
+	path, err = resolveKubeconfigPath(&cobra.Command{})
+	require.NoError(t, err)
+	require.Equal(t, envPath, path)
+}
+
+func TestWriteKubeconfig(t *testing.T) {
+	t.Parallel()
+
+	tests := []struct {
+		name     string
+		existing string
+	}{
+		{
+			name: "empty file",
+		},
+		{
+			name: "existing content",
+			existing: `apiVersion: v1
+clusters:
+- cluster:
+    insecure-skip-tls-verify: true
+    server: https://foobar.com
+  name: foo
+current-context: test
+kind: Config
+users: []
+`,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Parallel()
+
+			kubeconfigPath := filepath.Join(t.TempDir(), "config")
+			err := os.WriteFile(kubeconfigPath, []byte(tt.existing), 0o644)
+			require.NoError(t, err)
+
+			kc := kubernetesCluster{
+				name: "foo",
+				url:  &url.URL{Scheme: "https", Host: "example.com"},
+			}
+			err = writeKubeconfig(kubeconfigPath, kc)
+			require.NoError(t, err)
+
+			b, err := os.ReadFile(kubeconfigPath)
+			require.NoError(t, err)
+			expected := `apiVersion: v1
+clusters:
+- cluster:
+    insecure-skip-tls-verify: true
+    server: https://example.com
+  name: foo
+contexts:
+- context:
+    cluster: foo
+    namespace: default
+    user: netbird
+  name: foo
+current-context: foo
+kind: Config
+users:
+- name: netbird
+  user:
+    token: none
+`
+			require.Equal(t, expected, string(b))
+		})
+	}
+
+}

client/cmd/login.go

@@ -4,27 +4,38 @@ import (
 	"context"
 	"fmt"
 	"os"
-	"os/user"
-	"runtime"
 	"strings"
 
 	log "github.com/sirupsen/logrus"
 	"github.com/spf13/cobra"
+	"golang.org/x/term"
 	"google.golang.org/grpc/codes"
 	gstatus "google.golang.org/grpc/status"
 
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mdm"
+	nbnet "github.com/netbirdio/netbird/client/net"
 	"github.com/netbirdio/netbird/client/proto"
+	"github.com/netbirdio/netbird/client/server"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/util"
 )
 
+// extendSessionFlag drives the `netbird login --extend` flow: refresh the
+// SSO session expiry on the management server without tearing down the
+// tunnel. Mutually exclusive with setup-key login (a setup-key cannot
+// refresh an SSO-tracked peer — see auth.errSetupKeyOnSSOExpiredPeer).
+var extendSessionFlag bool
+
 func init() {
 	loginCmd.PersistentFlags().BoolVar(&noBrowser, noBrowserFlag, false, noBrowserDesc)
+	loginCmd.PersistentFlags().BoolVar(&showQR, showQRFlag, false, showQRDesc)
 	loginCmd.PersistentFlags().StringVar(&profileName, profileNameFlag, "", profileNameDesc)
 	loginCmd.PersistentFlags().StringVarP(&configPath, "config", "c", "", "(DEPRECATED) Netbird config file location")
+	loginCmd.PersistentFlags().BoolVar(&extendSessionFlag, "extend", false,
+		"refresh the SSO session expiry without tearing down the tunnel (requires an active connection)")
 }
 
 var loginCmd = &cobra.Command{
@@ -42,7 +53,7 @@ var loginCmd = &cobra.Command{
 			// nolint
 			ctx = context.WithValue(ctx, system.DeviceNameCtxKey, hostName)
 		}
-		username, err := user.Current()
+		username, err := profilemanager.InvokingUser()
 		if err != nil {
 			return fmt.Errorf("get current user: %v", err)
 		}
@@ -59,6 +70,16 @@ var loginCmd = &cobra.Command{
 			return err
 		}
 
+		if extendSessionFlag {
+			if providedSetupKey != "" {
+				return fmt.Errorf("--extend cannot be combined with a setup key; setup keys can only enrol new peers")
+			}
+			if err := doExtendSession(ctx, cmd, activeProf); err != nil {
+				return fmt.Errorf("extend session failed: %v", err)
+			}
+			return nil
+		}
+
 		// workaround to run without service
 		if util.FindFirstLogPath(logFiles) == "" {
 			if err := doForegroundLogin(ctx, cmd, providedSetupKey, activeProf); err != nil {
@@ -71,7 +92,7 @@ var loginCmd = &cobra.Command{
 			return fmt.Errorf("daemon login failed: %v", err)
 		}
 
-		cmd.Println("Logging successfully")
+		cmd.Println("Login successful")
 
 		return nil
 	},
@@ -94,17 +115,19 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
 		dnsLabelsReq = dnsLabelsValidated.ToSafeStringList()
 	}
 
+	handle := activeProf.ID.String()
+
 	loginRequest := proto.LoginRequest{
 		SetupKey:            providedSetupKey,
 		ManagementUrl:       managementURL,
-		IsUnixDesktopClient: isUnixRunningDesktop(),
+		IsUnixDesktopClient: util.HasGraphicalSession(),
 		Hostname:            hostName,
 		DnsLabels:           dnsLabelsReq,
-		ProfileName:         &activeProf.Name,
+		ProfileName:         &handle,
 		Username:            &username,
 	}
 
-	profileState, err := pm.GetProfileState(activeProf.Name)
+	profileState, err := pm.GetProfileState(activeProf.ID)
 	if err != nil {
 		log.Debugf("failed to get profile state for login hint: %v", err)
 	} else if profileState.Email != "" {
@@ -148,13 +171,73 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
 	return nil
 }
 
+// doExtendSession drives the daemon's RequestExtendAuthSession /
+// WaitExtendAuthSession pair. The user is sent through a regular SSO flow
+// (browser + verification URL) and the resulting JWT is forwarded to the
+// management server's ExtendAuthSession RPC. The tunnel stays up
+// throughout — no Down/Up, no network-map resync.
+func doExtendSession(ctx context.Context, cmd *cobra.Command, activeProf *profilemanager.Profile) error {
+	conn, err := DialClientGRPCServer(ctx, daemonAddr)
+	if err != nil {
+		//nolint
+		return fmt.Errorf("failed to connect to daemon error: %v\n"+
+			"If the daemon is not running please run: "+
+			"\nnetbird service install \nnetbird service start\n", err)
+	}
+	defer conn.Close()
+
+	client := proto.NewDaemonServiceClient(conn)
+
+	// the CLI runs in the user's session, the daemon does not: tell it what we can see
+	req := &proto.RequestExtendAuthSessionRequest{HasGraphicalSession: util.HasGraphicalSession()}
+	// Pre-fill the IdP login hint from the resolved profile so the user
+	// doesn't have to retype their email. Best-effort: we still proceed
+	// without a hint if the lookup fails.
+	pm := profilemanager.NewProfileManager()
+	if profState, perr := pm.GetProfileState(activeProf.ID); perr == nil && profState.Email != "" {
+		req.Hint = &profState.Email
+	}
+
+	startResp, err := client.RequestExtendAuthSession(ctx, req)
+	if err != nil {
+		return fmt.Errorf("start extend session: %v", err)
+	}
+
+	uri := startResp.GetVerificationURIComplete()
+	if uri == "" {
+		uri = startResp.GetVerificationURI()
+	}
+	openURL(cmd, uri, startResp.GetUserCode(), noBrowser, showQR)
+
+	waitResp, err := client.WaitExtendAuthSession(ctx, &proto.WaitExtendAuthSessionRequest{
+		DeviceCode: startResp.GetDeviceCode(),
+		UserCode:   startResp.GetUserCode(),
+	})
+	if err != nil {
+		return fmt.Errorf("wait for extend session: %v", err)
+	}
+
+	if ts := waitResp.GetSessionExpiresAt(); ts.IsValid() && !ts.AsTime().IsZero() {
+		deadline := ts.AsTime().Local()
+		cmd.Printf("Session extended. New expiry: %s\n", deadline.Format("2006-01-02 15:04:05 MST"))
+	} else {
+		// Management reported the peer is not eligible (e.g. login
+		// expiration disabled on the account). Surface that fact
+		// instead of pretending the call succeeded.
+		cmd.Println("Session extension call completed, but the management server did not return a new deadline (peer may not be SSO-tracked or login expiration is disabled).")
+	}
+	return nil
+}
+
 func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, profileName string, username string) (*profilemanager.Profile, error) {
 	// switch profile if provided
 
 	if profileName != "" {
-		if err := switchProfileOnDaemon(ctx, pm, profileName, username); err != nil {
+		prof, err := switchProfileOnDaemon(ctx, pm, profileName, username)
+		if err != nil {
 			return nil, fmt.Errorf("switch profile: %v", err)
 		}
+		return prof, nil
 	}
 
 	activeProf, err := pm.GetActiveProfile()
@@ -168,21 +251,19 @@ func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, pr
 	return activeProf, nil
 }
 
-func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManager, profileName string, username string) error {
-	err := switchProfile(context.Background(), profileName, username)
+func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManager, handle string, username string) (*profilemanager.Profile, error) {
+	resolvedID, err := switchProfile(ctx, handle, username)
 	if err != nil {
-		return fmt.Errorf("switch profile on daemon: %v", err)
+		return nil, fmt.Errorf("switch profile on daemon: %v", err)
 	}
 
-	err = pm.SwitchProfile(profileName)
-	if err != nil {
-		return fmt.Errorf("switch profile: %v", err)
+	if err := pm.SwitchProfile(resolvedID); err != nil {
+		return nil, fmt.Errorf("switch profile: %v", err)
 	}
 
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
-		log.Errorf("failed to connect to service CLI interface %v", err)
-		return err
+		return nil, fmt.Errorf("connect to service CLI interface: %w", err)
 	}
 	defer conn.Close()
 
@@ -190,24 +271,28 @@ func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManage
 
 	status, err := client.Status(ctx, &proto.StatusRequest{})
 	if err != nil {
-		return fmt.Errorf("unable to get daemon status: %v", err)
+		return nil, fmt.Errorf("unable to get daemon status: %v", err)
 	}
 
 	if status.Status == string(internal.StatusConnected) {
 		if _, err := client.Down(ctx, &proto.DownRequest{}); err != nil {
 			log.Errorf("call service down method: %v", err)
-			return err
+			return nil, err
 		}
 	}
 
-	return nil
+	return &profilemanager.Profile{ID: resolvedID}, nil
 }
 
-func switchProfile(ctx context.Context, profileName string, username string) error {
+// switchProfile asks the daemon to switch to the profile identified by
+// handle (a name, ID, or unique ID prefix). Returns the resolved profile
+// ID so the caller can update the local active-profile state without
+// re-resolving the handle.
+func switchProfile(ctx context.Context, handle string, username string) (profilemanager.ID, error) {
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
 		//nolint
-		return fmt.Errorf("failed to connect to daemon error: %v\n"+
+		return "", fmt.Errorf("failed to connect to daemon error: %v\n"+
 			"If the daemon is not running please run: "+
 			"\nnetbird service install \nnetbird service start\n", err)
 	}
@@ -215,15 +300,15 @@ func switchProfile(ctx context.Context, profileName string, username string) err
 
 	client := proto.NewDaemonServiceClient(conn)
 
-	_, err = client.SwitchProfile(ctx, &proto.SwitchProfileRequest{
-		ProfileName: &profileName,
+	resp, err := client.SwitchProfile(ctx, &proto.SwitchProfileRequest{
+		ProfileName: &handle,
 		Username:    &username,
 	})
 	if err != nil {
-		return fmt.Errorf("switch profile failed: %v", err)
+		return "", fmt.Errorf("switch profile failed: %w", err)
 	}
 
-	return nil
+	return profilemanager.ID(resp.Id), nil
 }
 
 func doForegroundLogin(ctx context.Context, cmd *cobra.Command, setupKey string, activeProf *profilemanager.Profile) error {
@@ -246,17 +331,30 @@ func doForegroundLogin(ctx context.Context, cmd *cobra.Command, setupKey string,
 	if err != nil {
 		return fmt.Errorf("read config file %s: %v", configFilePath, err)
 	}
+	// CLI standalone login: profilemanager no longer auto-applies MDM,
+	// so layer in the OS-native policy here. Desktop builds construct
+	// a Loader with no fetcher — the build-tagged loadPlatform reads
+	// the registry/plist directly.
+	config.ApplyMDMPolicy(mdm.NewLoader(nil).Load())
 
-	err = foregroundLogin(ctx, cmd, config, setupKey, activeProf.Name)
+	// Mirror runInForegroundMode: recover residual state (DNS, firewall,
+	// ssh config, legacy routing) from a previous unclean shutdown and
+	// enable advanced routing before dialing management.
+	if err := server.RestoreResidualState(ctx, profilemanager.NewServiceManager(configFilePath).GetStatePath()); err != nil {
+		log.Warnf("failed to restore residual state: %v", err)
+	}
+	nbnet.Init()
+
+	err = foregroundLogin(ctx, cmd, config, setupKey, activeProf.ID)
 	if err != nil {
 		return fmt.Errorf("foreground login failed: %v", err)
 	}
-	cmd.Println("Logging successfully")
+	cmd.Println("Login successful")
 	return nil
 }
 
 func handleSSOLogin(ctx context.Context, cmd *cobra.Command, loginResp *proto.LoginResponse, client proto.DaemonServiceClient, pm *profilemanager.ProfileManager) error {
-	openURL(cmd, loginResp.VerificationURIComplete, loginResp.UserCode, noBrowser)
+	openURL(cmd, loginResp.VerificationURIComplete, loginResp.UserCode, noBrowser, showQR)
 
 	resp, err := client.WaitSSOLogin(ctx, &proto.WaitSSOLoginRequest{UserCode: loginResp.UserCode, Hostname: hostName})
 	if err != nil {
@@ -275,7 +373,7 @@ func handleSSOLogin(ctx context.Context, cmd *cobra.Command, loginResp *proto.Lo
 	return nil
 }
 
-func foregroundLogin(ctx context.Context, cmd *cobra.Command, config *profilemanager.Config, setupKey, profileName string) error {
+func foregroundLogin(ctx context.Context, cmd *cobra.Command, config *profilemanager.Config, setupKey string, profileID profilemanager.ID) error {
 	authClient, err := auth.NewAuth(ctx, config.PrivateKey, config.ManagementURL, config)
 	if err != nil {
 		return fmt.Errorf("failed to create auth client: %v", err)
@@ -289,7 +387,7 @@ func foregroundLogin(ctx context.Context, cmd *cobra.Command, config *profileman
 
 	jwtToken := ""
 	if setupKey == "" && needsLogin {
-		tokenInfo, err := foregroundGetTokenInfo(ctx, cmd, config, profileName)
+		tokenInfo, err := foregroundGetTokenInfo(ctx, cmd, config, profileID)
 		if err != nil {
 			return fmt.Errorf("interactive sso login failed: %v", err)
 		}
@@ -304,17 +402,17 @@ func foregroundLogin(ctx context.Context, cmd *cobra.Command, config *profileman
 	return nil
 }
 
-func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *profilemanager.Config, profileName string) (*auth.TokenInfo, error) {
+func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *profilemanager.Config, profileID profilemanager.ID) (*auth.TokenInfo, error) {
 	hint := ""
 	pm := profilemanager.NewProfileManager()
-	profileState, err := pm.GetProfileState(profileName)
+	profileState, err := pm.GetProfileState(profileID)
 	if err != nil {
 		log.Debugf("failed to get profile state for login hint: %v", err)
 	} else if profileState.Email != "" {
 		hint = profileState.Email
 	}
 
-	oAuthFlow, err := auth.NewOAuthFlow(ctx, config, isUnixRunningDesktop(), false, hint)
+	oAuthFlow, err := auth.NewOAuthFlow(ctx, config, util.HasGraphicalSession(), false, hint)
 	if err != nil {
 		return nil, err
 	}
@@ -324,7 +422,7 @@ func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *pro
 		return nil, fmt.Errorf("getting a request OAuth flow info failed: %v", err)
 	}
 
-	openURL(cmd, flowInfo.VerificationURIComplete, flowInfo.UserCode, noBrowser)
+	openURL(cmd, flowInfo.VerificationURIComplete, flowInfo.UserCode, noBrowser, showQR)
 
 	tokenInfo, err := oAuthFlow.WaitToken(context.TODO(), flowInfo)
 	if err != nil {
@@ -334,7 +432,7 @@ func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *pro
 	return &tokenInfo, nil
 }
 
-func openURL(cmd *cobra.Command, verificationURIComplete, userCode string, noBrowser bool) {
+func openURL(cmd *cobra.Command, verificationURIComplete, userCode string, noBrowser, showQR bool) {
 	var codeMsg string
 	if userCode != "" && !strings.Contains(verificationURIComplete, userCode) {
 		codeMsg = fmt.Sprintf("and enter the code %s to authenticate.", userCode)
@@ -348,6 +446,12 @@ func openURL(cmd *cobra.Command, verificationURIComplete, userCode string, noBro
 			verificationURIComplete + " " + codeMsg)
 	}
 
+	if showQR {
+		if f, ok := cmd.OutOrStdout().(*os.File); ok && term.IsTerminal(int(f.Fd())) {
+			printQRCode(f, verificationURIComplete)
+		}
+	}
+
 	cmd.Println("")
 
 	if !noBrowser {
@@ -358,14 +462,6 @@ func openURL(cmd *cobra.Command, verificationURIComplete, userCode string, noBro
 	}
 }
 
-// isUnixRunningDesktop checks if a Linux OS is running desktop environment
-func isUnixRunningDesktop() bool {
-	if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
-		return false
-	}
-	return os.Getenv("DESKTOP_SESSION") != "" || os.Getenv("XDG_CURRENT_DESKTOP") != ""
-}
-
 func setEnvAndFlags(cmd *cobra.Command) error {
 	SetFlagsFromEnvVars(rootCmd)
 

client/cmd/login_test.go

@@ -27,7 +27,7 @@ func TestLogin(t *testing.T) {
 	profilemanager.ActiveProfileStatePath = tempDir + "/active_profile.json"
 	sm := profilemanager.ServiceManager{}
 	err = sm.SetActiveProfileState(&profilemanager.ActiveProfileState{
-		Name:     "default",
+		ID:       "default",
 		Username: currUser.Username,
 	})
 	if err != nil {

client/cmd/logout.go

@@ -3,11 +3,11 @@ package cmd
 import (
 	"context"
 	"fmt"
-	"os/user"
 	"time"
 
 	"github.com/spf13/cobra"
 
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/proto"
 )
 
@@ -37,7 +37,7 @@ var logoutCmd = &cobra.Command{
 		if profileName != "" {
 			req.ProfileName = &profileName
 
-			currUser, err := user.Current()
+			currUser, err := profilemanager.InvokingUser()
 			if err != nil {
 				return fmt.Errorf("get current user: %v", err)
 			}
@@ -46,7 +46,7 @@ var logoutCmd = &cobra.Command{
 		}
 
 		if _, err := daemonClient.Logout(ctx, req); err != nil {
-			return fmt.Errorf("deregister: %v", err)
+			return daemonCallError("deregister", err)
 		}
 
 		cmd.Println("Deregistered successfully")

client/cmd/profile.go

@@ -2,11 +2,15 @@ package cmd
 
 import (
 	"context"
+	"errors"
 	"fmt"
-	"os/user"
+	"strings"
+	"text/tabwriter"
 	"time"
 
 	"github.com/spf13/cobra"
+	"google.golang.org/grpc/codes"
+	gstatus "google.golang.org/grpc/status"
 
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
@@ -14,6 +18,8 @@ import (
 	"github.com/netbirdio/netbird/util"
 )
 
+var profileListShowID bool
+
 var profileCmd = &cobra.Command{
 	Use:   "profile",
 	Short: "Manage NetBird client profiles",
@@ -31,27 +37,40 @@ var profileListCmd = &cobra.Command{
 var profileAddCmd = &cobra.Command{
 	Use:   "add <profile_name>",
 	Short: "Add a new profile",
-	Long:  `Add a new profile to the NetBird client. The profile name must be unique.`,
+	Long:  `Add a new profile. Profile name is free-form, a unique ID is generated for the on-disk config file.`,
 	Args:  cobra.ExactArgs(1),
 	RunE:  addProfileFunc,
 }
 
+var profileRenameCmd = &cobra.Command{
+	Use:   "rename <profile> <new_profile_name>",
+	Short: "Renames an existing profile",
+	Long:  `Renames an existing profile (by a name, ID, or unique ID prefix). Profile name is free-form.`,
+	Args:  cobra.ExactArgs(2),
+	RunE:  renameProfileFunc,
+}
+
 var profileRemoveCmd = &cobra.Command{
-	Use:   "remove <profile_name>",
-	Short: "Remove a profile",
-	Long:  `Remove a profile from the NetBird client. The profile must not be inactive.`,
-	Args:  cobra.ExactArgs(1),
-	RunE:  removeProfileFunc,
+	Use:     "remove <profile>",
+	Short:   "Remove a profile",
+	Long:    `Remove a profile by name, ID, or unique ID prefix.`,
+	Aliases: []string{"rm"},
+	Args:    cobra.ExactArgs(1),
+	RunE:    removeProfileFunc,
 }
 
 var profileSelectCmd = &cobra.Command{
-	Use:   "select <profile_name>",
+	Use:   "select <profile>",
 	Short: "Select a profile",
-	Long:  `Make the specified profile active. This will switch the client to use the selected profile's configuration.`,
+	Long:  `Make the specified profile active. Accepts a name, ID, or unique ID prefix.`,
 	Args:  cobra.ExactArgs(1),
 	RunE:  selectProfileFunc,
 }
 
+func init() {
+	profileListCmd.Flags().BoolVar(&profileListShowID, "show-id", false, "show the profile ID column")
+}
+
 func setupCmd(cmd *cobra.Command) error {
 	SetFlagsFromEnvVars(rootCmd)
 	SetFlagsFromEnvVars(cmd)
@@ -65,6 +84,7 @@ func setupCmd(cmd *cobra.Command) error {
 
 	return nil
 }
+
 func listProfilesFunc(cmd *cobra.Command, _ []string) error {
 	if err := setupCmd(cmd); err != nil {
 		return err
@@ -76,32 +96,40 @@ func listProfilesFunc(cmd *cobra.Command, _ []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
 
 	daemonClient := proto.NewDaemonServiceClient(conn)
 
-	profiles, err := daemonClient.ListProfiles(cmd.Context(), &proto.ListProfilesRequest{
+	resp, err := daemonClient.ListProfiles(cmd.Context(), &proto.ListProfilesRequest{
 		Username: currUser.Username,
 	})
 	if err != nil {
 		return err
 	}
 
-	// list profiles, add a tick if the profile is active
-	cmd.Println("Found", len(profiles.Profiles), "profiles:")
-	for _, profile := range profiles.Profiles {
-		// use a cross to indicate the passive profiles
-		activeMarker := "✗"
-		if profile.IsActive {
-			activeMarker = "✓"
-		}
-		cmd.Println(activeMarker, profile.Name)
+	tw := tabwriter.NewWriter(cmd.OutOrStdout(), 0, 0, 2, ' ', 0)
+	if profileListShowID {
+		fmt.Fprintln(tw, "ID\tNAME\tACTIVE")
+	} else {
+		fmt.Fprintln(tw, "NAME\tACTIVE")
 	}
-
-	return nil
+	for _, profile := range resp.Profiles {
+		marker := ""
+		if profile.IsActive {
+			marker = "✓"
+		}
+		name := profilemanager.StripCtrlChars(profile.Name)
+		id := profilemanager.ID(profile.Id)
+		if profileListShowID {
+			fmt.Fprintf(tw, "%s\t%s\t%s\n", id.ShortID(), name, marker)
+		} else {
+			fmt.Fprintf(tw, "%s\t%s\n", name, marker)
+		}
+	}
+	return tw.Flush()
 }
 
 func addProfileFunc(cmd *cobra.Command, args []string) error {
@@ -109,33 +137,90 @@ func addProfileFunc(cmd *cobra.Command, args []string) error {
 		return err
 	}
 
+	currUser, err := profilemanager.InvokingUser()
+	if err != nil {
+		return fmt.Errorf("get current user: %w", err)
+	}
+
+	conn, err := DialClientGRPCServer(cmd.Context(), daemonAddr)
+	if err != nil {
+		return fmt.Errorf("connect to service CLI interface: %w", err)
+	}
+	defer conn.Close()
+
+	daemonClient := proto.NewDaemonServiceClient(conn)
+	profileName := args[0]
+
+	id, err := addProfileOnDaemon(cmd.Context(), daemonClient, profileName, currUser.Username)
+	if err != nil {
+		return err
+	}
+
+	dupCount, _ := countProfilesWithName(cmd.Context(), daemonClient, currUser.Username, profileName)
+	if dupCount > 1 {
+		cmd.Printf("Warning: %d other profile(s) already use the name %q.\n", dupCount-1, profileName)
+		cmd.Println("Use `netbird profile list --show-id` to disambiguate later.")
+	}
+
+	cmd.Printf("Profile added: %s  %s\n", id.ShortID(), profilemanager.StripCtrlChars(profileName))
+	return nil
+
+}
+
+func renameProfileFunc(cmd *cobra.Command, args []string) error {
+	if err := setupCmd(cmd); err != nil {
+		return err
+	}
+
 	conn, err := DialClientGRPCServer(cmd.Context(), daemonAddr)
 	if err != nil {
 		return fmt.Errorf("connect to service CLI interface: %w", err)
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
 
 	daemonClient := proto.NewDaemonServiceClient(conn)
+	handle := args[0]
+	newProfilename := args[1]
 
-	profileName := args[0]
-
-	_, err = daemonClient.AddProfile(cmd.Context(), &proto.AddProfileRequest{
-		ProfileName: profileName,
-		Username:    currUser.Username,
+	resp, err := daemonClient.RenameProfile(cmd.Context(), &proto.RenameProfileRequest{
+		Handle:         handle,
+		Username:       currUser.Username,
+		NewProfileName: newProfilename,
 	})
 	if err != nil {
-		return err
+		return wrapAmbiguityError(err, handle)
 	}
 
-	cmd.Println("Profile added successfully:", profileName)
+	dupCount, _ := countProfilesWithName(cmd.Context(), daemonClient, currUser.Username, newProfilename)
+	if dupCount > 1 {
+		cmd.Printf("Warning: %d other profile(s) already use the name %q.\n", dupCount-1, newProfilename)
+		cmd.Println("Use `netbird profile list --show-id` to disambiguate later.")
+	}
+
+	cmd.Printf("Profile renamed from %s to %s\n", profilemanager.StripCtrlChars(resp.OldProfileName), profilemanager.StripCtrlChars(newProfilename))
+
 	return nil
 }
 
+func countProfilesWithName(ctx context.Context, c proto.DaemonServiceClient, username, name string) (int, error) {
+	resp, err := c.ListProfiles(ctx, &proto.ListProfilesRequest{Username: username})
+	if err != nil {
+		return 0, err
+	}
+	n := 0
+	for _, p := range resp.Profiles {
+		if p.Name == name {
+			n++
+		}
+	}
+	return n, nil
+}
+
 func removeProfileFunc(cmd *cobra.Command, args []string) error {
 	if err := setupCmd(cmd); err != nil {
 		return err
@@ -147,24 +232,23 @@ func removeProfileFunc(cmd *cobra.Command, args []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
 
 	daemonClient := proto.NewDaemonServiceClient(conn)
+	handle := args[0]
 
-	profileName := args[0]
-
-	_, err = daemonClient.RemoveProfile(cmd.Context(), &proto.RemoveProfileRequest{
-		ProfileName: profileName,
+	resp, err := daemonClient.RemoveProfile(cmd.Context(), &proto.RemoveProfileRequest{
+		ProfileName: handle,
 		Username:    currUser.Username,
 	})
 	if err != nil {
-		return err
+		return wrapAmbiguityError(err, handle)
 	}
 
-	cmd.Println("Profile removed successfully:", profileName)
+	cmd.Printf("Profile removed: %s\n", resp.Id)
 	return nil
 }
 
@@ -174,9 +258,9 @@ func selectProfileFunc(cmd *cobra.Command, args []string) error {
 	}
 
 	profileManager := profilemanager.NewProfileManager()
-	profileName := args[0]
+	handle := args[0]
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -191,32 +275,15 @@ func selectProfileFunc(cmd *cobra.Command, args []string) error {
 
 	daemonClient := proto.NewDaemonServiceClient(conn)
 
-	profiles, err := daemonClient.ListProfiles(ctx, &proto.ListProfilesRequest{
-		Username: currUser.Username,
+	switchResp, err := daemonClient.SwitchProfile(ctx, &proto.SwitchProfileRequest{
+		ProfileName: &handle,
+		Username:    &currUser.Username,
 	})
 	if err != nil {
-		return fmt.Errorf("list profiles: %w", err)
+		return wrapAmbiguityError(err, handle)
 	}
 
-	var profileExists bool
-
-	for _, profile := range profiles.Profiles {
-		if profile.Name == profileName {
-			profileExists = true
-			break
-		}
-	}
-
-	if !profileExists {
-		return fmt.Errorf("profile %s does not exist", profileName)
-	}
-
-	if err := switchProfile(cmd.Context(), profileName, currUser.Username); err != nil {
-		return err
-	}
-
-	err = profileManager.SwitchProfile(profileName)
-	if err != nil {
+	if err := profileManager.SwitchProfile(profilemanager.ID(switchResp.Id)); err != nil {
 		return err
 	}
 
@@ -231,6 +298,46 @@ func selectProfileFunc(cmd *cobra.Command, args []string) error {
 		}
 	}
 
-	cmd.Println("Profile switched successfully to:", profileName)
+	id := profilemanager.ID(switchResp.Id)
+	cmd.Printf("Profile switched to: %s\n", id.ShortID())
 	return nil
 }
+
+// wrapAmbiguityError turns the daemon's gRPC InvalidArgument errors
+// (which carry the resolver's message verbatim) into CLI-friendly text
+// that points the user at --show-id.
+func wrapAmbiguityError(err error, handle string) error {
+	if err == nil {
+		return nil
+	}
+	st, ok := gstatus.FromError(err)
+	if !ok {
+		return err
+	}
+	switch st.Code() {
+	case codes.InvalidArgument:
+		msg := st.Message()
+		if strings.Contains(msg, "ambiguous") {
+			return errors.New(msg + "\nRun `netbird profile list --show-id` to see IDs, then select by ID prefix:\n  netbird profile select|remove <id-prefix>")
+		}
+	case codes.NotFound:
+		return fmt.Errorf("profile %q not found", handle)
+	}
+	return err
+}
+
+// addProfileOnDaemon issues the AddProfile RPC on an existing daemon client
+// and returns the new profile's ID. It is the single entry point for profile
+// creation, shared by `netbird profile add` and the `netbird up --profile
+// <name>` auto-create path.
+func addProfileOnDaemon(ctx context.Context, client proto.DaemonServiceClient, profileName, username string) (profilemanager.ID, error) {
+	resp, err := client.AddProfile(ctx, &proto.AddProfileRequest{
+		ProfileName: profileName,
+		Username:    username,
+	})
+	if err != nil {
+		return "", fmt.Errorf("add profile failed: %w", err)
+	}
+
+	return profilemanager.ID(resp.Id), nil
+}

client/cmd/qr.go

@@ -0,0 +1,25 @@
+package cmd
+
+import (
+	"io"
+
+	"github.com/mdp/qrterminal/v3"
+)
+
+// printQRCode prints a QR code for the given URL to the writer.
+// Called only when the user explicitly requests QR output via --qr.
+func printQRCode(w io.Writer, url string) {
+	if url == "" {
+		return
+	}
+	qrterminal.GenerateWithConfig(url, qrterminal.Config{
+		Level:          qrterminal.M,
+		Writer:         w,
+		HalfBlocks:     true,
+		BlackChar:      qrterminal.BLACK_BLACK,
+		WhiteChar:      qrterminal.WHITE_WHITE,
+		BlackWhiteChar: qrterminal.BLACK_WHITE,
+		WhiteBlackChar: qrterminal.WHITE_BLACK,
+		QuietZone:      qrterminal.QUIET_ZONE,
+	})
+}

client/cmd/qr_test.go

@@ -0,0 +1,26 @@
+package cmd
+
+import (
+	"bytes"
+	"testing"
+)
+
+func TestPrintQRCode_EmptyURL(t *testing.T) {
+	var buf bytes.Buffer
+
+	printQRCode(&buf, "")
+
+	if buf.Len() != 0 {
+		t.Error("expected no output for empty URL")
+	}
+}
+
+func TestPrintQRCode_WritesOutput(t *testing.T) {
+	var buf bytes.Buffer
+
+	printQRCode(&buf, "https://example.com/auth")
+
+	if buf.Len() == 0 {
+		t.Error("expected QR code output for non-empty URL")
+	}
+}

client/cmd/root.go

@@ -20,9 +20,10 @@ import (
 	"github.com/spf13/cobra"
 	"github.com/spf13/pflag"
 	"google.golang.org/grpc"
-	"google.golang.org/grpc/credentials/insecure"
 
+	"github.com/netbirdio/netbird/client/anonymize"
 	daddr "github.com/netbirdio/netbird/client/internal/daemonaddr"
+	"github.com/netbirdio/netbird/client/internal/localmetrics"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 )
 
@@ -31,6 +32,8 @@ const (
 	dnsResolverAddress       = "dns-resolver-address"
 	enableRosenpassFlag      = "enable-rosenpass"
 	rosenpassPermissiveFlag  = "rosenpass-permissive"
+	enableLocalMetricsFlag   = "enable-local-metrics"
+	localMetricsAddressFlag  = "local-metrics-address"
 	preSharedKeyFlag         = "preshared-key"
 	interfaceNameFlag        = "interface-name"
 	wireguardPortFlag        = "wireguard-port"
@@ -70,12 +73,18 @@ var (
 	autoConnectDisabled     bool
 	extraIFaceBlackList     []string
 	anonymizeFlag           bool
+	anonymizeLevelFlag      string
 	dnsRouteInterval        time.Duration
-	lazyConnEnabled         bool
-	mtu                     uint16
-	profilesDisabled        bool
-	updateSettingsDisabled  bool
-	networksDisabled        bool
+	// lazyConnEnabled is the parse target for the deprecated --enable-lazy-connection
+	// flag. The flag is inert; the value is no longer read (use NB_LAZY_CONN instead).
+	lazyConnEnabled        bool
+	mtu                    uint16
+	profilesDisabled       bool
+	updateSettingsDisabled bool
+	captureEnabled         bool
+	networksDisabled       bool
+	localMetricsEnabled    bool
+	localMetricsAddr       string
 
 	rootCmd = &cobra.Command{
 		Use:          "netbird",
@@ -88,13 +97,16 @@ var (
 			// Don't resolve for service commands — they create the socket, not connect to it.
 			if !isServiceCmd(cmd) {
 				daemonAddr = daddr.ResolveUnixDaemonAddr(daemonAddr)
+				daemonAddr = daddr.ResolveDaemonAddr(daemonAddr)
 			}
 			return nil
 		},
 	}
 )
 
-// Execute executes the root command.
+// Execute runs the appropriate Cobra command for the CLI.
+// If the process is the update binary it delegates to updateCmd; otherwise it runs the root command.
+// It returns any error produced during command execution.
 func Execute() error {
 	if isUpdateBinary() {
 		return updateCmd.Execute()
@@ -102,6 +114,16 @@ func Execute() error {
 	return rootCmd.Execute()
 }
 
+// init initialises package-level defaults and configures the root
+// Cobra command tree. Sets platform-specific config / log directory
+// paths (including legacy Wiretrustee fallbacks) and a default daemon
+// address; registers persistent CLI flags (daemon address,
+// management / admin URLs, logging, setup key (file and inline,
+// mutually exclusive), preshared key, hostname, anonymise, config
+// path); attaches top-level and nested subcommands to the root
+// command; and registers `up`-specific persistent flags (external IP
+// maps, custom DNS resolver address, Rosenpass options, auto-connect
+// disabling, lazy connection).
 func init() {
 	defaultConfigPathDir = "/etc/netbird/"
 	defaultLogFileDir = "/var/log/netbird/"
@@ -128,10 +150,10 @@ func init() {
 
 	defaultDaemonAddr := "unix:///var/run/netbird.sock"
 	if runtime.GOOS == "windows" {
-		defaultDaemonAddr = "tcp://127.0.0.1:41731"
+		defaultDaemonAddr = daddr.WindowsPipeAddr
 	}
 
-	rootCmd.PersistentFlags().StringVar(&daemonAddr, "daemon-addr", defaultDaemonAddr, "Daemon service address to serve CLI requests [unix|tcp]://[path|host:port]")
+	rootCmd.PersistentFlags().StringVar(&daemonAddr, "daemon-addr", defaultDaemonAddr, "Daemon service address to serve CLI requests [unix|tcp|npipe]://[path|host:port|name]")
 	rootCmd.PersistentFlags().StringVarP(&managementURL, "management-url", "m", "", fmt.Sprintf("Management Service URL [http|https]://[host]:[port] (default \"%s\")", profilemanager.DefaultManagementURL))
 	rootCmd.PersistentFlags().StringVar(&adminURL, "admin-url", "", fmt.Sprintf("Admin Panel URL [http|https]://[host]:[port] (default \"%s\")", profilemanager.DefaultAdminURL))
 	rootCmd.PersistentFlags().StringVarP(&logLevel, "log-level", "l", "info", "sets NetBird log level")
@@ -141,8 +163,9 @@ func init() {
 	rootCmd.MarkFlagsMutuallyExclusive("setup-key", "setup-key-file")
 	rootCmd.PersistentFlags().StringVar(&preSharedKey, preSharedKeyFlag, "", "Sets WireGuard PreSharedKey property. If set, then only peers that have the same key can communicate.")
 	rootCmd.PersistentFlags().StringVarP(&hostName, "hostname", "n", "", "Sets a custom hostname for the device")
-	rootCmd.PersistentFlags().BoolVarP(&anonymizeFlag, "anonymize", "A", false, "anonymize IP addresses and non-netbird.io domains in logs and status output")
-	rootCmd.PersistentFlags().StringVarP(&configPath, "config", "c", defaultConfigPath, "Overrides the default profile file location")
+	rootCmd.PersistentFlags().BoolVarP(&anonymizeFlag, "anonymize", "A", false, "anonymize public IP addresses, MAC addresses, and non-netbird.io domains in logs and status output; private, CGNAT, and link-local IP ranges are kept (see --anonymize-level strict)")
+	rootCmd.PersistentFlags().StringVar(&anonymizeLevelFlag, "anonymize-level", "", "anonymization level: \"default\" or \"strict\"; strict also anonymizes private, CGNAT, and link-local IP ranges, peer names, and WireGuard public keys. Setting this flag implies --anonymize")
+	rootCmd.PersistentFlags().StringVarP(&configPath, "config", "c", profilemanager.DefaultConfigPath, "Overrides the default profile file location")
 
 	rootCmd.AddCommand(upCmd)
 	rootCmd.AddCommand(downCmd)
@@ -167,10 +190,17 @@ func init() {
 	logCmd.AddCommand(logLevelCmd)
 	debugCmd.AddCommand(forCmd)
 	debugCmd.AddCommand(persistenceCmd)
+	debugCmd.AddCommand(debugConfigCmd)
+
+	// kubernetes commands
+	rootCmd.AddCommand(kubernetesCmd)
+	kubernetesCmd.AddCommand(kubernetesListCmd)
+	kubernetesCmd.AddCommand(kubernetesWriteKubeconfigCmd)
 
 	// profile commands
 	profileCmd.AddCommand(profileListCmd)
 	profileCmd.AddCommand(profileAddCmd)
+	profileCmd.AddCommand(profileRenameCmd)
 	profileCmd.AddCommand(profileRemoveCmd)
 	profileCmd.AddCommand(profileSelectCmd)
 
@@ -190,7 +220,10 @@ func init() {
 	upCmd.PersistentFlags().BoolVar(&rosenpassEnabled, enableRosenpassFlag, false, "[Experimental] Enable Rosenpass feature. If enabled, the connection will be post-quantum secured via Rosenpass.")
 	upCmd.PersistentFlags().BoolVar(&rosenpassPermissive, rosenpassPermissiveFlag, false, "[Experimental] Enable Rosenpass in permissive mode to allow this peer to accept WireGuard connections without requiring Rosenpass functionality from peers that do not have Rosenpass enabled.")
 	upCmd.PersistentFlags().BoolVar(&autoConnectDisabled, disableAutoConnectFlag, false, "Disables auto-connect feature. If enabled, then the client won't connect automatically when the service starts.")
-	upCmd.PersistentFlags().BoolVar(&lazyConnEnabled, enableLazyConnectionFlag, false, "[Experimental] Enable the lazy connection feature. If enabled, the client will establish connections on-demand. Note: this setting may be overridden by management configuration.")
+	upCmd.PersistentFlags().BoolVar(&localMetricsEnabled, enableLocalMetricsFlag, false, "Enables a local Prometheus /metrics endpoint exposing connection state (peers, latency, P2P vs relay).")
+	upCmd.PersistentFlags().StringVar(&localMetricsAddr, localMetricsAddressFlag, localmetrics.DefaultListenAddress, "Listen address of the local Prometheus /metrics endpoint.")
+	upCmd.PersistentFlags().BoolVar(&lazyConnEnabled, enableLazyConnectionFlag, false, "Deprecated: no longer used. Lazy connections are controlled by the server and the NB_LAZY_CONN environment variable.")
+	_ = upCmd.PersistentFlags().MarkDeprecated(enableLazyConnectionFlag, "no longer used; lazy connections are controlled by the server and the NB_LAZY_CONN environment variable")
 
 }
 
@@ -246,12 +279,10 @@ func DialClientGRPCServer(ctx context.Context, addr string) (*grpc.ClientConn, e
 	ctx, cancel := context.WithTimeout(ctx, time.Second*10)
 	defer cancel()
 
-	return grpc.DialContext(
-		ctx,
-		strings.TrimPrefix(addr, "tcp://"),
-		grpc.WithTransportCredentials(insecure.NewCredentials()),
-		grpc.WithBlock(),
-	)
+	target, opts := daddr.DialTarget(addr)
+	opts = append(opts, grpc.WithBlock())
+
+	return grpc.DialContext(ctx, target, opts...)
 }
 
 // WithBackOff execute function in backoff cycle.
@@ -272,6 +303,19 @@ var CLIBackOffSettings = &backoff.ExponentialBackOff{
 	Clock:               backoff.SystemClock,
 }
 
+// effectiveAnonymize resolves the --anonymize and --anonymize-level flags:
+// setting a level implies anonymization, and an invalid level is rejected.
+func effectiveAnonymize() (bool, anonymize.Level, error) {
+	if anonymizeLevelFlag == "" {
+		return anonymizeFlag, anonymize.LevelDefault, nil
+	}
+	level := anonymize.ParseLevel(anonymizeLevelFlag)
+	if !strings.EqualFold(anonymizeLevelFlag, level.String()) {
+		return false, anonymize.LevelDefault, fmt.Errorf("invalid anonymize level %q: use %q or %q", anonymizeLevelFlag, anonymize.LevelDefault.String(), anonymize.LevelStrict.String())
+	}
+	return true, level, nil
+}
+
 func getSetupKey() (string, error) {
 	if setupKeyPath != "" && setupKey == "" {
 		return getSetupKeyFromFile(setupKeyPath)

client/cmd/service.go

@@ -5,6 +5,7 @@ package cmd
 import (
 	"context"
 	"fmt"
+	"net/http"
 	"runtime"
 	"strings"
 	"sync"
@@ -22,15 +23,26 @@ var serviceCmd = &cobra.Command{
 	Short: "Manage the NetBird daemon service",
 }
 
+const defaultJSONSocket = "unix:///var/run/netbird-http.sock"
+
 var (
-	serviceName    string
-	serviceEnvVars []string
+	serviceName      string
+	serviceEnvVars   []string
+	jsonSocket       string
+	enableJSONSocket bool
 )
 
 type program struct {
-	ctx              context.Context
-	cancel           context.CancelFunc
-	serv             *grpc.Server
+	ctx      context.Context
+	cancel   context.CancelFunc
+	serv     *grpc.Server
+	jsonServ *http.Server
+	// jsonClient is the gateway's own connection to the daemon. It is held so
+	// shutting the gateway down also closes it: nothing else references it once
+	// the handlers are registered, so its transport goroutines would otherwise
+	// outlive the server.
+	jsonClient       *grpc.ClientConn
+	jsonServMu       sync.Mutex
 	serverInstance   *server.Server
 	serverInstanceMu sync.Mutex
 }
@@ -44,7 +56,10 @@ func init() {
 	serviceCmd.AddCommand(runCmd, startCmd, stopCmd, restartCmd, svcStatusCmd, installCmd, uninstallCmd, reconfigureCmd, resetParamsCmd)
 	serviceCmd.PersistentFlags().BoolVar(&profilesDisabled, "disable-profiles", false, "Disables profiles feature. If enabled, the client will not be able to change or edit any profile. To persist this setting, use: netbird service install --disable-profiles")
 	serviceCmd.PersistentFlags().BoolVar(&updateSettingsDisabled, "disable-update-settings", false, "Disables update settings feature. If enabled, the client will not be able to change or edit any settings. To persist this setting, use: netbird service install --disable-update-settings")
+	serviceCmd.PersistentFlags().BoolVar(&captureEnabled, "enable-capture", false, "Enables packet capture via 'netbird debug capture'. To persist, use: netbird service install --enable-capture")
 	serviceCmd.PersistentFlags().BoolVar(&networksDisabled, "disable-networks", false, "Disables network selection. If enabled, the client will not allow listing, selecting, or deselecting networks. To persist, use: netbird service install --disable-networks")
+	serviceCmd.PersistentFlags().BoolVar(&enableJSONSocket, "enable-json-socket", false, "Enables the HTTP/JSON API socket served by grpc-gateway. To persist, use: netbird service install --enable-json-socket")
+	serviceCmd.PersistentFlags().StringVar(&jsonSocket, "json-socket", defaultJSONSocket, "HTTP/JSON API socket address [unix|tcp]://[path|host:port]. Requires --enable-json-socket to serve. To persist, use: netbird service install --enable-json-socket --json-socket")
 
 	rootCmd.PersistentFlags().StringVarP(&serviceName, "service", "s", defaultServiceName, "Netbird system service name")
 	serviceEnvDesc := `Sets extra environment variables for the service. ` +

client/cmd/service_controller.go

@@ -5,9 +5,7 @@ package cmd
 import (
 	"context"
 	"fmt"
-	"net"
-	"os"
-	"strings"
+	"runtime"
 	"time"
 
 	"github.com/kardianos/service"
@@ -16,69 +14,159 @@ import (
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc"
 
+	"github.com/netbirdio/netbird/client/internal/daemonaddr"
+	"github.com/netbirdio/netbird/client/internal/ipcauth"
 	"github.com/netbirdio/netbird/client/proto"
 	"github.com/netbirdio/netbird/client/server"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/util"
 )
 
+func validateJSONSocketFlags() error {
+	if serviceCmd.PersistentFlags().Changed("json-socket") && !enableJSONSocket {
+		return fmt.Errorf("--json-socket requires --enable-json-socket to configure the daemon JSON gateway")
+	}
+	return nil
+}
+
+// daemonServerOptions installs the transport credentials that expose each
+// caller's kernel-authenticated identity to the handlers, which is what lets
+// the daemon require root/administrator for privileged operations.
+//
+// The handshake exchanges no bytes, so older CLI and UI binaries still
+// interoperate. Callers on a TCP socket carry no identity at all: the daemon
+// keeps serving them, and the privileged operations deny them, so a warning is
+// logged to make the loss of functionality visible.
+func daemonServerOptions(network string) []grpc.ServerOption {
+	if network == "tcp" {
+		log.Warnf("daemon is listening on TCP (%s): callers carry no verifiable identity over TCP, "+
+			"so privileged operations (SSH root login, SSH auth, enabling the SSH server, management URL changes, "+
+			"deregistration) will be denied, and the SSH JWT cache is neither filled nor served. "+
+			"Use a unix socket, or npipe:// on Windows", daemonAddr)
+		return nil
+	}
+
+	creds := ipcauth.NewTransportCredentials() //nolint:staticcheck
+	if creds == nil {                          //nolint:staticcheck // nil only on platforms without a peer-identity primitive
+		log.Warnf("daemon IPC has no peer-identity primitive on %s: privileged operations will be denied "+
+			"and the SSH JWT cache is neither filled nor served", runtime.GOOS)
+		return nil
+	}
+
+	return []grpc.ServerOption{grpc.Creds(creds)}
+}
+
 func (p *program) Start(svc service.Service) error {
 	// Start should not block. Do the actual work async.
 	log.Info("starting NetBird service") //nolint
 
+	if err := validateJSONSocketFlags(); err != nil {
+		return err
+	}
+
 	// Collect static system and platform information
 	system.UpdateStaticInfoAsync()
 
-	// in any case, even if configuration does not exists we run daemon to serve CLI gRPC API.
-	p.serv = grpc.NewServer()
-
-	split := strings.Split(daemonAddr, "://")
-	switch split[0] {
-	case "unix":
-		// cleanup failed close
-		stat, err := os.Stat(split[1])
-		if err == nil && !stat.IsDir() {
-			if err := os.Remove(split[1]); err != nil {
-				log.Debugf("remove socket file: %v", err)
-			}
-		}
-	case "tcp":
-	default:
-		return fmt.Errorf("unsupported daemon address protocol: %v", split[0])
+	// A daemon installed before named-pipe support has the loopback TCP address
+	// persisted. Move it to the named pipe so an upgraded daemon can identify
+	// its callers instead of silently serving an unauthenticated socket.
+	if migrated, ok := daemonaddr.MigrateLegacy(daemonAddr); ok {
+		log.Infof("daemon address %q predates named-pipe support, listening on %q so callers can be identified", daemonAddr, migrated)
+		daemonAddr = migrated
 	}
 
-	listen, err := net.Listen(split[0], split[1])
+	network, _, err := parseListenAddress(daemonAddr)
 	if err != nil {
-		return fmt.Errorf("listen daemon interface: %w", err)
+		return fmt.Errorf("parse daemon address: %w", err)
 	}
+
+	// in any case, even if configuration does not exists we run daemon to serve CLI gRPC API.
+	p.serv = grpc.NewServer(daemonServerOptions(network)...)
+
+	daemonListener, jsonListener, err := listenDaemonSockets()
+	if err != nil {
+		return err
+	}
+
 	go func() {
-		defer listen.Close()
-
-		if split[0] == "unix" {
-			if err := os.Chmod(split[1], 0666); err != nil {
-				log.Errorf("failed setting daemon permissions: %v", split[1])
-				return
-			}
-		}
-
-		serverInstance := server.New(p.ctx, util.FindFirstLogPath(logFiles), configPath, profilesDisabled, updateSettingsDisabled, networksDisabled)
-		if err := serverInstance.Start(); err != nil {
-			log.Fatalf("failed to start daemon: %v", err)
-		}
-		proto.RegisterDaemonServiceServer(p.serv, serverInstance)
-
-		p.serverInstanceMu.Lock()
-		p.serverInstance = serverInstance
-		p.serverInstanceMu.Unlock()
-
-		log.Printf("started daemon server: %v", split[1])
-		if err := p.serv.Serve(listen); err != nil {
-			log.Errorf("failed to serve daemon requests: %v", err)
+		// Fatal here rather than inside serve, so serve's deferred listener
+		// closes run before the process exits.
+		if err := p.serve(daemonListener, jsonListener); err != nil {
+			log.Fatalf("failed to %v", err)
 		}
 	}()
 	return nil
 }
 
+// listenDaemonSockets opens the daemon control socket and, when it is enabled, the
+// JSON gateway socket. The control socket is closed again if the second one fails,
+// so a failed start leaves nothing listening. The returned JSON listener is nil
+// when the socket is disabled.
+func listenDaemonSockets() (*socketListener, *socketListener, error) {
+	daemonListener, err := listenOnAddress(daemonAddr)
+	if err != nil {
+		return nil, nil, fmt.Errorf("listen daemon interface: %w", err)
+	}
+
+	if !enableJSONSocket {
+		removeStaleUnixSocketForAddress(jsonSocket)
+		return daemonListener, nil, nil
+	}
+
+	jsonListener, err := listenOnAddress(jsonSocket)
+	if err != nil {
+		if cerr := daemonListener.Close(); cerr != nil {
+			log.Debugf("close daemon listener: %v", cerr)
+		}
+		return nil, nil, fmt.Errorf("listen daemon JSON interface: %w", err)
+	}
+
+	return daemonListener, jsonListener, nil
+}
+
+// serve brings up the daemon server on an already-open control socket and blocks
+// until it stops. jsonListener is nil when the JSON socket is disabled. A returned
+// error means the daemon cannot run at all and the caller is expected to exit; the
+// failures it recovers from on its own are logged here.
+func (p *program) serve(daemonListener, jsonListener *socketListener) error {
+	defer daemonListener.Close()
+	if jsonListener != nil {
+		defer jsonListener.Close()
+	}
+
+	// chmodUnixSocket is a no-op for a nil listener and for a non-unix one.
+	if err := daemonListener.chmodUnixSocket("daemon"); err != nil {
+		log.Error(err)
+		return nil
+	}
+	if err := jsonListener.chmodUnixSocket("daemon JSON"); err != nil {
+		log.Error(err)
+		return nil
+	}
+
+	serverInstance := server.New(p.ctx, util.FindFirstLogPath(logFiles), configPath, profilesDisabled, updateSettingsDisabled, captureEnabled, networksDisabled)
+	if err := serverInstance.Start(); err != nil {
+		return fmt.Errorf("start daemon: %w", err)
+	}
+	proto.RegisterDaemonServiceServer(p.serv, serverInstance)
+
+	p.serverInstanceMu.Lock()
+	p.serverInstance = serverInstance
+	p.serverInstanceMu.Unlock()
+
+	if jsonListener == nil {
+		log.Debug("daemon JSON socket disabled")
+	} else if err := p.startJSONGateway(jsonListener, daemonAddr); err != nil {
+		return fmt.Errorf("start daemon JSON server: %w", err)
+	}
+
+	log.Printf("started daemon server: %v", daemonListener.address)
+	if err := p.serv.Serve(daemonListener.Listener); err != nil {
+		log.Errorf("failed to serve daemon requests: %v", err)
+	}
+	return nil
+}
+
 func (p *program) Stop(srv service.Service) error {
 	p.serverInstanceMu.Lock()
 	if p.serverInstance != nil {
@@ -92,6 +180,25 @@ func (p *program) Stop(srv service.Service) error {
 
 	p.cancel()
 
+	p.jsonServMu.Lock()
+	jsonServ, jsonClient := p.jsonServ, p.jsonClient
+	p.jsonServMu.Unlock()
+	if jsonClient != nil {
+		if err := jsonClient.Close(); err != nil {
+			log.Debugf("close daemon JSON gateway client: %v", err)
+		}
+	}
+	if jsonServ != nil {
+		shutdownCtx, shutdownCancel := context.WithTimeout(context.Background(), 2*time.Second)
+		if err := jsonServ.Shutdown(shutdownCtx); err != nil {
+			log.Errorf("failed to stop daemon JSON server gracefully: %v", err)
+			if err := jsonServ.Close(); err != nil {
+				log.Errorf("failed to close daemon JSON server: %v", err)
+			}
+		}
+		shutdownCancel()
+	}
+
 	if p.serv != nil {
 		p.serv.Stop()
 	}
@@ -102,7 +209,7 @@ func (p *program) Stop(srv service.Service) error {
 }
 
 // Common setup for service control commands
-func setupServiceControlCommand(cmd *cobra.Command, ctx context.Context, cancel context.CancelFunc) (service.Service, error) {
+func setupServiceControlCommand(cmd *cobra.Command, ctx context.Context, cancel context.CancelFunc, consoleLog bool) (service.Service, error) {
 	// rootCmd env vars are already applied by PersistentPreRunE.
 	SetFlagsFromEnvVars(serviceCmd)
 
@@ -112,8 +219,14 @@ func setupServiceControlCommand(cmd *cobra.Command, ctx context.Context, cancel
 		return nil, err
 	}
 
-	if err := util.InitLog(logLevel, logFiles...); err != nil {
-		return nil, fmt.Errorf("init log: %w", err)
+	if consoleLog {
+		if err := util.InitLog(logLevel, util.LogConsole); err != nil {
+			return nil, fmt.Errorf("init log: %w", err)
+		}
+	} else {
+		if err := util.InitLog(logLevel, logFiles...); err != nil {
+			return nil, fmt.Errorf("init log: %w", err)
+		}
 	}
 
 	cfg, err := newSVCConfig()
@@ -138,10 +251,13 @@ var runCmd = &cobra.Command{
 		SetupCloseHandler(ctx, cancel)
 		SetupDebugHandler(ctx, nil, nil, nil, util.FindFirstLogPath(logFiles))
 
-		s, err := setupServiceControlCommand(cmd, ctx, cancel)
+		s, err := setupServiceControlCommand(cmd, ctx, cancel, false)
 		if err != nil {
 			return err
 		}
+		if err := validateJSONSocketFlags(); err != nil {
+			return err
+		}
 
 		return s.Run()
 	},
@@ -152,10 +268,13 @@ var startCmd = &cobra.Command{
 	Short: "starts NetBird service",
 	RunE: func(cmd *cobra.Command, args []string) error {
 		ctx, cancel := context.WithCancel(cmd.Context())
-		s, err := setupServiceControlCommand(cmd, ctx, cancel)
+		s, err := setupServiceControlCommand(cmd, ctx, cancel, false)
 		if err != nil {
 			return err
 		}
+		if err := validateJSONSocketFlags(); err != nil {
+			return err
+		}
 
 		if err := s.Start(); err != nil {
 			return fmt.Errorf("start service: %w", err)
@@ -170,7 +289,7 @@ var stopCmd = &cobra.Command{
 	Short: "stops NetBird service",
 	RunE: func(cmd *cobra.Command, args []string) error {
 		ctx, cancel := context.WithCancel(cmd.Context())
-		s, err := setupServiceControlCommand(cmd, ctx, cancel)
+		s, err := setupServiceControlCommand(cmd, ctx, cancel, false)
 		if err != nil {
 			return err
 		}
@@ -188,10 +307,13 @@ var restartCmd = &cobra.Command{
 	Short: "restarts NetBird service",
 	RunE: func(cmd *cobra.Command, args []string) error {
 		ctx, cancel := context.WithCancel(cmd.Context())
-		s, err := setupServiceControlCommand(cmd, ctx, cancel)
+		s, err := setupServiceControlCommand(cmd, ctx, cancel, false)
 		if err != nil {
 			return err
 		}
+		if err := validateJSONSocketFlags(); err != nil {
+			return err
+		}
 
 		if err := s.Restart(); err != nil {
 			return fmt.Errorf("restart service: %w", err)
@@ -206,7 +328,7 @@ var svcStatusCmd = &cobra.Command{
 	Short: "shows NetBird service status",
 	RunE: func(cmd *cobra.Command, args []string) error {
 		ctx, cancel := context.WithCancel(cmd.Context())
-		s, err := setupServiceControlCommand(cmd, ctx, cancel)
+		s, err := setupServiceControlCommand(cmd, ctx, cancel, true)
 		if err != nil {
 			return err
 		}

client/cmd/service_installer.go

@@ -59,10 +59,18 @@ func buildServiceArguments() []string {
 		args = append(args, "--disable-update-settings")
 	}
 
+	if captureEnabled {
+		args = append(args, "--enable-capture")
+	}
+
 	if networksDisabled {
 		args = append(args, "--disable-networks")
 	}
 
+	if enableJSONSocket {
+		args = append(args, "--enable-json-socket", "--json-socket", jsonSocket)
+	}
+
 	return args
 }
 
@@ -102,6 +110,10 @@ func configurePlatformSpecificSettings(svcConfig *service.Config) error {
 
 // Create fully configured service config for install/reconfigure
 func createServiceConfigForInstall() (*service.Config, error) {
+	if err := validateJSONSocketFlags(); err != nil {
+		return nil, err
+	}
+
 	svcConfig, err := newSVCConfig()
 	if err != nil {
 		return nil, fmt.Errorf("create service config: %w", err)

client/cmd/service_json_gateway.go

@@ -0,0 +1,150 @@
+//go:build !ios && !android
+
+package cmd
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"net"
+	"net/http"
+	"sync"
+	"time"
+
+	"github.com/grpc-ecosystem/grpc-gateway/v2/runtime"
+	log "github.com/sirupsen/logrus"
+	"google.golang.org/grpc"
+
+	"github.com/netbirdio/netbird/client/internal/daemonaddr"
+	"github.com/netbirdio/netbird/client/internal/ipcauth"
+	"github.com/netbirdio/netbird/client/proto"
+)
+
+// jsonPeerIdentity is the context key under which the connecting HTTP client's
+// identity is stashed for the lifetime of its connection.
+type jsonPeerIdentity struct{}
+
+// jsonPeerIdentityValue pairs the identity with whether it could be read at
+// all, so an unreadable identity is forwarded as "unknown" rather than omitted.
+type jsonPeerIdentityValue struct {
+	id    ipcauth.Identity
+	known bool
+}
+
+// jsonConnContext reads the identity of the client connecting to the JSON
+// socket and stashes it on the connection's context. The gateway re-dials the
+// daemon in-process, so the daemon would otherwise see every JSON request as
+// coming from the daemon itself.
+func jsonConnContext(ctx context.Context, c net.Conn) context.Context {
+	value := jsonPeerIdentityValue{}
+	id, err := ipcauth.ConnIdentity(c)
+	if err != nil {
+		log.Warnf("json gateway: cannot read HTTP client identity, privileged operations will be denied for this connection: %v", err)
+	} else {
+		value.id = id
+		value.known = true
+	}
+	return context.WithValue(ctx, jsonPeerIdentity{}, value)
+}
+
+// forwardIdentity stamps the HTTP client's identity onto every call the gateway
+// makes to the daemon.
+//
+// It is an interceptor on the gateway's client connection rather than a
+// runtime.WithMetadata annotator because grpc-gateway skips annotators when no
+// request header maps to metadata, which an HTTP/1.0 request with no Host header
+// over a unix socket achieves. The daemon would then receive no marker, see its own
+// identity as the transport peer, and authorize the request as the daemon itself.
+// An interceptor runs for every RPC whatever the request looked like.
+func forwardIdentity(ctx context.Context) context.Context {
+	value, ok := ctx.Value(jsonPeerIdentity{}).(jsonPeerIdentityValue)
+	if !ok {
+		// No ConnContext ran for this request, so forward an unknown identity:
+		// the daemon must not mistake its own identity for the client's.
+		return ipcauth.WithForwardedIdentity(ctx, ipcauth.Identity{}, false)
+	}
+	return ipcauth.WithForwardedIdentity(ctx, value.id, value.known)
+}
+
+func forwardIdentityUnary(ctx context.Context, method string, req, reply any, cc *grpc.ClientConn, invoker grpc.UnaryInvoker, opts ...grpc.CallOption) error {
+	return invoker(forwardIdentity(ctx), method, req, reply, cc, opts...)
+}
+
+func forwardIdentityStream(ctx context.Context, desc *grpc.StreamDesc, cc *grpc.ClientConn, method string, streamer grpc.Streamer, opts ...grpc.CallOption) (grpc.ClientStream, error) {
+	return streamer(forwardIdentity(ctx), desc, cc, method, opts...)
+}
+
+// reservedHeaderWarning limits the dropped-header warning to the first occurrence.
+var reservedHeaderWarning sync.Once
+
+// jsonIncomingHeaderMatcher keeps an HTTP client from supplying the metadata the
+// gateway uses to forward its identity. grpc-gateway turns "Grpc-Metadata-<key>"
+// headers into gRPC metadata and joins them ahead of what its annotators add, so
+// without this filter a JSON client could send its own x-netbird-fwd-uid and the
+// daemon would authorize that instead of the client's real identity.
+func jsonIncomingHeaderMatcher(key string) (string, bool) {
+	mapped, ok := runtime.DefaultHeaderMatcher(key)
+	if !ok {
+		return "", false
+	}
+	if ipcauth.IsReservedForwardKey(mapped) {
+		// Warn once: any client can send these on every request, so warning each
+		// time hands it a way to fill the log. The rest are debug-level.
+		reservedHeaderWarning.Do(func() {
+			log.Warnf("json gateway: dropping reserved header %q from a request: only the gateway may set the caller's identity", key)
+		})
+		log.Debugf("json gateway: dropping reserved header %q", key)
+		return "", false
+	}
+	return mapped, true
+}
+
+func (p *program) startJSONGateway(jsonListener *socketListener, daemonEndpoint string) error {
+	if jsonListener.network == "tcp" {
+		log.Warnf("daemon JSON socket is listening on TCP (%s): callers carry no verifiable identity over TCP, "+
+			"so privileged operations will be denied for JSON clients", jsonListener.address)
+	}
+
+	mux := runtime.NewServeMux(runtime.WithIncomingHeaderMatcher(jsonIncomingHeaderMatcher))
+
+	// grpc.NewClient does not connect until the first request, so registering
+	// the handler here cannot block daemon startup.
+	target, opts := daemonaddr.DialTarget(daemonEndpoint)
+	opts = append(opts,
+		grpc.WithChainUnaryInterceptor(forwardIdentityUnary),
+		grpc.WithChainStreamInterceptor(forwardIdentityStream),
+	)
+	conn, err := grpc.NewClient(target, opts...)
+	if err != nil {
+		return fmt.Errorf("create daemon client for JSON gateway: %w", err)
+	}
+	if err := proto.RegisterDaemonServiceHandler(p.ctx, mux, conn); err != nil {
+		if cerr := conn.Close(); cerr != nil {
+			log.Debugf("close daemon client after failed JSON gateway registration: %v", cerr)
+		}
+		return err
+	}
+
+	jsonServer := &http.Server{
+		Handler:           mux,
+		ReadHeaderTimeout: 5 * time.Second,
+		BaseContext: func(net.Listener) context.Context {
+			return p.ctx
+		},
+		ConnContext: jsonConnContext,
+	}
+
+	p.jsonServMu.Lock()
+	p.jsonServ = jsonServer
+	p.jsonClient = conn
+	p.jsonServMu.Unlock()
+
+	go func() {
+		log.Printf("started daemon JSON server: %v", jsonListener.address)
+		if err := jsonServer.Serve(jsonListener.Listener); err != nil && !errors.Is(err, http.ErrServerClosed) {
+			log.Errorf("failed to serve daemon JSON requests: %v", err)
+		}
+	}()
+
+	return nil
+}

client/cmd/service_json_gateway_test.go

@@ -0,0 +1,261 @@
+//go:build !windows && !ios && !android
+
+package cmd
+
+import (
+	"context"
+	"net"
+	"net/http"
+	"path/filepath"
+	"testing"
+	"time"
+
+	"github.com/grpc-ecosystem/grpc-gateway/v2/runtime"
+	"google.golang.org/grpc/credentials"
+	"google.golang.org/grpc/metadata"
+	"google.golang.org/grpc/peer"
+
+	"github.com/netbirdio/netbird/client/internal/ipcauth"
+)
+
+// The JSON gateway runs inside the daemon and re-dials it locally, so every JSON
+// request reaches a handler with the daemon's own identity as the transport peer.
+// The gateway therefore forwards its HTTP client's identity as metadata, and the
+// daemon authorizes that instead of itself. These tests drive the real wiring
+// (jsonConnContext, forwardIdentity, jsonIncomingHeaderMatcher) and check the
+// identity a handler would end up authorizing.
+
+// daemonSideCtx is what a handler sees for a gateway-relayed call. The transport
+// peer must be this process's own identity: the gateway is the daemon, so the two
+// cannot differ, and hardcoding root here instead would describe a state that
+// never occurs.
+func daemonSideCtx(t *testing.T, md metadata.MD) context.Context {
+	t.Helper()
+	self, err := ipcauth.CurrentProcessIdentity()
+	if err != nil {
+		t.Skipf("cannot read this process's identity: %v", err)
+	}
+	ctx := peer.NewContext(context.Background(), &peer.Peer{
+		AuthInfo: ipcauth.AuthInfo{
+			CommonAuthInfo: credentials.CommonAuthInfo{SecurityLevel: credentials.NoSecurity},
+			Identity:       self,
+		},
+	})
+	return metadata.NewIncomingContext(ctx, md)
+}
+
+// gatewayMetadata reproduces what the daemon receives for a JSON request: the
+// mux annotates the context from the request's headers, then the interceptor on the
+// gateway's client connection stamps the caller's identity. The order matters,
+// since the interceptor must win over anything a header put there.
+func gatewayMetadata(t *testing.T, req *http.Request, ctx context.Context) metadata.MD {
+	t.Helper()
+	mux := runtime.NewServeMux(runtime.WithIncomingHeaderMatcher(jsonIncomingHeaderMatcher))
+	annotated, err := runtime.AnnotateContext(ctx, mux, req,
+		"/daemon.DaemonService/SetConfig",
+		runtime.WithHTTPPathPattern("/daemon.DaemonService/SetConfig"))
+	if err != nil {
+		t.Fatalf("annotate: %v", err)
+	}
+
+	md, ok := metadata.FromOutgoingContext(forwardIdentity(annotated))
+	if !ok {
+		t.Fatal("the interceptor produced no metadata")
+	}
+	return md
+}
+
+// clientCtx is the connection context jsonConnContext would have produced for an
+// HTTP client whose identity the gateway could read.
+func clientCtx(id ipcauth.Identity, known bool) context.Context {
+	return context.WithValue(context.Background(), jsonPeerIdentity{},
+		jsonPeerIdentityValue{id: id, known: known})
+}
+
+// An HTTP client must not be able to name its own identity. grpc-gateway turns
+// Grpc-Metadata-<key> headers into gRPC metadata, so without the header filter and
+// the interceptor overwriting the reserved keys, this request would authorize as
+// uid 0.
+func TestJSONGateway_ForgedIdentityHeaderIsDropped(t *testing.T) {
+	req, err := http.NewRequest(http.MethodPost, "http://localhost/daemon.DaemonService/SetConfig", nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	req.Header.Set("Grpc-Metadata-X-Netbird-Fwd-Uid", "0")
+	req.Header.Set("Grpc-Metadata-X-Netbird-Fwd-Gid", "0")
+	req.Header.Set("Grpc-Metadata-X-Netbird-Fwd", "1")
+	req.Header.Set("Grpc-Metadata-X-Netbird-Fwd-Sid", "S-1-5-18")
+
+	caller := ipcauth.Identity{UID: 31000, GID: 31000}
+	md := gatewayMetadata(t, req, clientCtx(caller, true))
+
+	id, ok := ipcauth.CallerIdentity(daemonSideCtx(t, md))
+	if !ok {
+		t.Fatal("the forwarded identity should be usable")
+	}
+	if id.IsPrivileged() {
+		t.Errorf("forged header was believed: authorized as %v", id)
+	}
+	if id.UID != caller.UID {
+		t.Errorf("authorized as uid %d, want the real client %d", id.UID, caller.UID)
+	}
+}
+
+// A request with no headers at all (HTTP/1.0 needs no Host, and a unix socket
+// yields no host:port) makes grpc-gateway produce no metadata whatsoever and skip
+// its annotators: "if len(pairs) == 0 { return ctx, nil, nil }" in
+// runtime/context.go. That is why the identity is stamped by an interceptor
+// instead. This is the case that previously reached the gate as the daemon itself.
+func TestJSONGateway_HeaderlessRequestIsStillMarkedForwarded(t *testing.T) {
+	req, err := http.NewRequest(http.MethodPost, "http://localhost/daemon.DaemonService/SetConfig", nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	req.Header = http.Header{}
+	req.Host = ""
+
+	caller := ipcauth.Identity{UID: 31000, GID: 31000}
+	ctx := clientCtx(caller, true)
+
+	// Pin the skip path itself: if grpc-gateway ever produced a pair here, this
+	// test would still pass below while no longer covering what it was written for.
+	mux := runtime.NewServeMux(runtime.WithIncomingHeaderMatcher(jsonIncomingHeaderMatcher))
+	annotated, err := runtime.AnnotateContext(ctx, mux, req,
+		"/daemon.DaemonService/SetConfig",
+		runtime.WithHTTPPathPattern("/daemon.DaemonService/SetConfig"))
+	if err != nil {
+		t.Fatalf("annotate: %v", err)
+	}
+	if md, ok := metadata.FromOutgoingContext(annotated); ok {
+		t.Fatalf("grpc-gateway produced metadata %v for a headerless request; "+
+			"this test no longer covers the annotator-skip path", md)
+	}
+
+	md := gatewayMetadata(t, req, ctx)
+
+	id, ok := ipcauth.CallerIdentity(daemonSideCtx(t, md))
+	if !ok {
+		t.Fatal("the forwarded identity should be usable")
+	}
+	if id.UID != caller.UID || id.IsPrivileged() {
+		t.Errorf("authorized as %v, want the real client uid %d", id, caller.UID)
+	}
+}
+
+// When the gateway cannot read its client's identity (a TCP JSON socket, say) it
+// forwards the marker alone. The daemon must then report "unidentified" so the
+// privileged operations refuse, rather than falling back to the gateway's own
+// identity.
+func TestJSONGateway_UnreadableClientIdentityIsUnidentified(t *testing.T) {
+	req, err := http.NewRequest(http.MethodPost, "http://localhost/daemon.DaemonService/SetConfig", nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	md := gatewayMetadata(t, req, clientCtx(ipcauth.Identity{}, false))
+
+	if id, ok := ipcauth.CallerIdentity(daemonSideCtx(t, md)); ok {
+		t.Errorf("a request with no client identity was authorized as %v", id)
+	}
+}
+
+// A request that never passed through jsonConnContext (no stashed identity) must
+// also come out unidentified rather than as the daemon.
+func TestJSONGateway_MissingConnContextIsUnidentified(t *testing.T) {
+	req, err := http.NewRequest(http.MethodPost, "http://localhost/daemon.DaemonService/SetConfig", nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	md := gatewayMetadata(t, req, context.Background())
+
+	if id, ok := ipcauth.CallerIdentity(daemonSideCtx(t, md)); ok {
+		t.Errorf("a request with no connection context was authorized as %v", id)
+	}
+}
+
+// End to end over a real unix socket: the gateway reads the connecting client's
+// identity from the socket itself, so a client cannot present anything else.
+func TestJSONGateway_IdentityComesFromTheSocket(t *testing.T) {
+	mux := runtime.NewServeMux(runtime.WithIncomingHeaderMatcher(jsonIncomingHeaderMatcher))
+
+	type observed struct {
+		md metadata.MD
+	}
+	seen := make(chan observed, 1)
+
+	srv := &http.Server{
+		Handler: http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+			ctx, err := runtime.AnnotateContext(r.Context(), mux, r,
+				"/daemon.DaemonService/SetConfig",
+				runtime.WithHTTPPathPattern("/daemon.DaemonService/SetConfig"))
+			if err != nil {
+				t.Errorf("annotate: %v", err)
+				return
+			}
+			md, _ := metadata.FromOutgoingContext(forwardIdentity(ctx))
+			seen <- observed{md: md}
+			w.WriteHeader(http.StatusOK)
+		}),
+		ReadHeaderTimeout: 5 * time.Second,
+		ConnContext:       jsonConnContext,
+	}
+
+	sock := filepath.Join(t.TempDir(), "http.sock")
+	ln, err := net.Listen("unix", sock)
+	if err != nil {
+		t.Fatal(err)
+	}
+	t.Cleanup(func() {
+		if err := srv.Close(); err != nil {
+			t.Logf("close server: %v", err)
+		}
+	})
+	go func() {
+		if err := srv.Serve(ln); err != nil && err != http.ErrServerClosed {
+			t.Logf("serve: %v", err)
+		}
+	}()
+
+	conn, err := net.Dial("unix", sock)
+	if err != nil {
+		t.Fatal(err)
+	}
+	t.Cleanup(func() {
+		if err := conn.Close(); err != nil {
+			t.Logf("close conn: %v", err)
+		}
+	})
+
+	// Forge the identity headers on the wire as well.
+	request := "POST /daemon.DaemonService/SetConfig HTTP/1.1\r\n" +
+		"Host: localhost\r\n" +
+		"Grpc-Metadata-X-Netbird-Fwd: 1\r\n" +
+		"Grpc-Metadata-X-Netbird-Fwd-Uid: 0\r\n" +
+		"Content-Length: 0\r\n\r\n"
+	if _, err := conn.Write([]byte(request)); err != nil {
+		t.Fatal(err)
+	}
+
+	select {
+	case got := <-seen:
+		self, err := ipcauth.CurrentProcessIdentity()
+		if err != nil {
+			t.Skipf("cannot read this process's identity: %v", err)
+		}
+		// The socket peer is this test process, so that is the identity the
+		// gateway must forward, not the uid 0 the request asked for.
+		if uids := got.md.Get("x-netbird-fwd-uid"); len(uids) != 1 {
+			t.Fatalf("x-netbird-fwd-uid = %v, want exactly the gateway's own value", uids)
+		}
+		id, ok := ipcauth.CallerIdentity(daemonSideCtx(t, got.md))
+		if !ok {
+			t.Fatal("the forwarded identity should be usable")
+		}
+		if id.UID != self.UID {
+			t.Errorf("authorized as uid %d, want the socket peer %d", id.UID, self.UID)
+		}
+	case <-time.After(5 * time.Second):
+		t.Fatal("the gateway never handled the request")
+	}
+}

client/cmd/service_json_socket_test.go

@@ -0,0 +1,176 @@
+//go:build !ios && !android
+
+package cmd
+
+import (
+	"net"
+	"os"
+	"path/filepath"
+	"runtime"
+	"testing"
+
+	"github.com/spf13/cobra"
+	"github.com/spf13/pflag"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func preserveJSONSocketTestState(t *testing.T) {
+	t.Helper()
+
+	origJSONSocket := jsonSocket
+	origEnableJSONSocket := enableJSONSocket
+	origChanged := map[string]bool{}
+	serviceCmd.PersistentFlags().VisitAll(func(flag *pflag.Flag) {
+		origChanged[flag.Name] = flag.Changed
+	})
+
+	t.Cleanup(func() {
+		jsonSocket = origJSONSocket
+		enableJSONSocket = origEnableJSONSocket
+		serviceCmd.PersistentFlags().VisitAll(func(flag *pflag.Flag) {
+			flag.Changed = origChanged[flag.Name]
+		})
+	})
+}
+
+func TestJSONSocketFlagsArePositiveEnableOnly(t *testing.T) {
+	assert.NotNil(t, serviceCmd.PersistentFlags().Lookup("enable-json-socket"))
+	assert.NotNil(t, serviceCmd.PersistentFlags().Lookup("json-socket"))
+	assert.Nil(t, serviceCmd.PersistentFlags().Lookup("disable-json-socket"))
+	assert.Equal(t, "false", serviceCmd.PersistentFlags().Lookup("enable-json-socket").DefValue)
+}
+
+func TestBuildServiceArgumentsDefaultDisablesJSONSocket(t *testing.T) {
+	preserveJSONSocketTestState(t)
+
+	enableJSONSocket = false
+	jsonSocket = "tcp://127.0.0.1:8080"
+
+	args := buildServiceArguments()
+
+	assert.NotContains(t, args, "--enable-json-socket")
+	assert.NotContains(t, args, "--json-socket")
+}
+
+func TestBuildServiceArgumentsIncludesJSONSocketWhenEnabled(t *testing.T) {
+	preserveJSONSocketTestState(t)
+
+	enableJSONSocket = true
+	jsonSocket = "tcp://127.0.0.1:8080"
+
+	args := buildServiceArguments()
+
+	enableIndex := indexOfArg(args, "--enable-json-socket")
+	jsonIndex := indexOfArg(args, "--json-socket")
+	require.NotEqual(t, -1, enableIndex)
+	require.NotEqual(t, -1, jsonIndex)
+	require.Less(t, enableIndex, jsonIndex)
+	require.Less(t, jsonIndex+1, len(args))
+	assert.Equal(t, "tcp://127.0.0.1:8080", args[jsonIndex+1])
+}
+
+func TestJSONSocketWithoutEnableValidation(t *testing.T) {
+	preserveJSONSocketTestState(t)
+
+	enableJSONSocket = false
+	require.NoError(t, serviceCmd.PersistentFlags().Set("json-socket", "tcp://127.0.0.1:8080"))
+
+	err := validateJSONSocketFlags()
+
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "--enable-json-socket")
+}
+
+func TestJSONSocketWithEnableValidation(t *testing.T) {
+	preserveJSONSocketTestState(t)
+
+	require.NoError(t, serviceCmd.PersistentFlags().Set("enable-json-socket", "true"))
+	require.NoError(t, serviceCmd.PersistentFlags().Set("json-socket", "tcp://127.0.0.1:8080"))
+
+	assert.NoError(t, validateJSONSocketFlags())
+}
+
+func TestJSONSocketServiceParamsPersistEnableAndAddress(t *testing.T) {
+	preserveJSONSocketTestState(t)
+	serviceCmd.PersistentFlags().VisitAll(func(flag *pflag.Flag) {
+		flag.Changed = false
+	})
+
+	enableJSONSocket = true
+	jsonSocket = "tcp://127.0.0.1:8080"
+
+	params := currentServiceParams()
+	require.True(t, params.EnableJSONSocket)
+	require.Equal(t, "tcp://127.0.0.1:8080", params.JSONSocket)
+
+	enableJSONSocket = false
+	jsonSocket = defaultJSONSocket
+	applyServiceParams(testServiceEnvCommand(), params)
+
+	assert.True(t, enableJSONSocket)
+	assert.Equal(t, "tcp://127.0.0.1:8080", jsonSocket)
+}
+
+func TestRemoveStaleUnixSocketDoesNotRemoveRegularFile(t *testing.T) {
+	path := filepath.Join(t.TempDir(), "netbird-http.sock")
+	require.NoError(t, os.WriteFile(path, []byte("not a socket"), 0600))
+
+	removeStaleUnixSocket(path)
+
+	data, err := os.ReadFile(path)
+	require.NoError(t, err)
+	assert.Equal(t, []byte("not a socket"), data)
+}
+
+func TestRemoveStaleUnixSocketRemovesSocket(t *testing.T) {
+	if runtime.GOOS == "windows" {
+		t.Skip("unix sockets are not available on Windows")
+	}
+
+	path := filepath.Join(t.TempDir(), "netbird-http.sock")
+	addr := &net.UnixAddr{Name: path, Net: "unix"}
+	listener, err := net.ListenUnix("unix", addr)
+	require.NoError(t, err)
+	listener.SetUnlinkOnClose(false)
+	require.NoError(t, listener.Close())
+
+	_, err = os.Lstat(path)
+	require.NoError(t, err, "test setup must leave a stale Unix socket path")
+
+	removeStaleUnixSocket(path)
+
+	_, err = os.Lstat(path)
+	assert.True(t, os.IsNotExist(err), "expected stale Unix socket to be removed, got %v", err)
+}
+
+func TestRemoveStaleUnixSocketDoesNotRemoveLiveSocket(t *testing.T) {
+	if runtime.GOOS == "windows" {
+		t.Skip("unix sockets are not available on Windows")
+	}
+
+	path := filepath.Join(t.TempDir(), "netbird-http.sock")
+	listener, err := net.Listen("unix", path)
+	require.NoError(t, err)
+	defer listener.Close()
+
+	removeStaleUnixSocket(path)
+
+	_, err = os.Lstat(path)
+	assert.NoError(t, err, "expected live Unix socket to be preserved")
+}
+
+func testServiceEnvCommand() *cobra.Command {
+	cmd := &cobra.Command{}
+	cmd.Flags().StringSlice("service-env", nil, "")
+	return cmd
+}
+
+func indexOfArg(args []string, arg string) int {
+	for i, candidate := range args {
+		if candidate == arg {
+			return i
+		}
+	}
+	return -1
+}

client/cmd/service_params.go

@@ -13,6 +13,7 @@ import (
 	"github.com/spf13/cobra"
 
 	"github.com/netbirdio/netbird/client/configs"
+	"github.com/netbirdio/netbird/client/internal/daemonaddr"
 	"github.com/netbirdio/netbird/util"
 )
 
@@ -23,12 +24,15 @@ const serviceParamsFile = "service.json"
 type serviceParams struct {
 	LogLevel              string            `json:"log_level"`
 	DaemonAddr            string            `json:"daemon_addr"`
+	JSONSocket            string            `json:"json_socket"`
 	ManagementURL         string            `json:"management_url,omitempty"`
 	ConfigPath            string            `json:"config_path,omitempty"`
 	LogFiles              []string          `json:"log_files,omitempty"`
 	DisableProfiles       bool              `json:"disable_profiles,omitempty"`
 	DisableUpdateSettings bool              `json:"disable_update_settings,omitempty"`
+	EnableCapture         bool              `json:"enable_capture,omitempty"`
 	DisableNetworks       bool              `json:"disable_networks,omitempty"`
+	EnableJSONSocket      bool              `json:"enable_json_socket,omitempty"`
 	ServiceEnvVars        map[string]string `json:"service_env_vars,omitempty"`
 }
 
@@ -74,12 +78,15 @@ func currentServiceParams() *serviceParams {
 	params := &serviceParams{
 		LogLevel:              logLevel,
 		DaemonAddr:            daemonAddr,
+		JSONSocket:            jsonSocket,
 		ManagementURL:         managementURL,
 		ConfigPath:            configPath,
 		LogFiles:              logFiles,
 		DisableProfiles:       profilesDisabled,
 		DisableUpdateSettings: updateSettingsDisabled,
+		EnableCapture:         captureEnabled,
 		DisableNetworks:       networksDisabled,
+		EnableJSONSocket:      enableJSONSocket,
 	}
 
 	if len(serviceEnvVars) > 0 {
@@ -111,15 +118,29 @@ func applyServiceParams(cmd *cobra.Command, params *serviceParams) {
 		return
 	}
 
-	// For fields with non-empty defaults (log-level, daemon-addr), keep the
-	// != "" guard so that an older service.json missing the field doesn't
-	// clobber the default with an empty string.
+	// For fields with non-empty defaults, keep the != "" guard so that an older
+	// service.json missing the field doesn't clobber the default with an empty string.
 	if !rootCmd.PersistentFlags().Changed("log-level") && params.LogLevel != "" {
 		logLevel = params.LogLevel
 	}
 
 	if !rootCmd.PersistentFlags().Changed("daemon-addr") && params.DaemonAddr != "" {
 		daemonAddr = params.DaemonAddr
+		// An install that predates named-pipe support has the loopback TCP
+		// address saved. Callers carry no identity over TCP, so move it to the
+		// pipe instead of restoring a socket the daemon cannot authorize on.
+		if migrated, ok := daemonaddr.MigrateLegacy(daemonAddr); ok {
+			cmd.Printf("Moving the saved daemon address from %s to %s so the daemon can identify its callers\n", daemonAddr, migrated)
+			daemonAddr = migrated
+		}
+	}
+
+	if !serviceCmd.PersistentFlags().Changed("json-socket") && params.JSONSocket != "" {
+		jsonSocket = params.JSONSocket
+	}
+
+	if !serviceCmd.PersistentFlags().Changed("enable-json-socket") {
+		enableJSONSocket = params.EnableJSONSocket
 	}
 
 	// For optional fields where empty means "use default", always apply so
@@ -144,6 +165,10 @@ func applyServiceParams(cmd *cobra.Command, params *serviceParams) {
 		updateSettingsDisabled = params.DisableUpdateSettings
 	}
 
+	if !serviceCmd.PersistentFlags().Changed("enable-capture") {
+		captureEnabled = params.EnableCapture
+	}
+
 	if !serviceCmd.PersistentFlags().Changed("disable-networks") {
 		networksDisabled = params.DisableNetworks
 	}

client/cmd/service_params_test.go

@@ -41,6 +41,8 @@ func TestSaveAndLoadServiceParams(t *testing.T) {
 	params := &serviceParams{
 		LogLevel:              "debug",
 		DaemonAddr:            "unix:///var/run/netbird.sock",
+		JSONSocket:            "tcp://127.0.0.1:8080",
+		EnableJSONSocket:      true,
 		ManagementURL:         "https://my.server.com",
 		ConfigPath:            "/etc/netbird/config.json",
 		LogFiles:              []string{"/var/log/netbird/client.log", "console"},
@@ -63,6 +65,8 @@ func TestSaveAndLoadServiceParams(t *testing.T) {
 
 	assert.Equal(t, params.LogLevel, loaded.LogLevel)
 	assert.Equal(t, params.DaemonAddr, loaded.DaemonAddr)
+	assert.Equal(t, params.JSONSocket, loaded.JSONSocket)
+	assert.Equal(t, params.EnableJSONSocket, loaded.EnableJSONSocket)
 	assert.Equal(t, params.ManagementURL, loaded.ManagementURL)
 	assert.Equal(t, params.ConfigPath, loaded.ConfigPath)
 	assert.Equal(t, params.LogFiles, loaded.LogFiles)
@@ -101,6 +105,8 @@ func TestLoadServiceParams_InvalidJSON(t *testing.T) {
 func TestCurrentServiceParams(t *testing.T) {
 	origLogLevel := logLevel
 	origDaemonAddr := daemonAddr
+	origJSONSocket := jsonSocket
+	origEnableJSONSocket := enableJSONSocket
 	origManagementURL := managementURL
 	origConfigPath := configPath
 	origLogFiles := logFiles
@@ -110,6 +116,8 @@ func TestCurrentServiceParams(t *testing.T) {
 	t.Cleanup(func() {
 		logLevel = origLogLevel
 		daemonAddr = origDaemonAddr
+		jsonSocket = origJSONSocket
+		enableJSONSocket = origEnableJSONSocket
 		managementURL = origManagementURL
 		configPath = origConfigPath
 		logFiles = origLogFiles
@@ -120,6 +128,8 @@ func TestCurrentServiceParams(t *testing.T) {
 
 	logLevel = "trace"
 	daemonAddr = "tcp://127.0.0.1:9999"
+	jsonSocket = "tcp://127.0.0.1:8080"
+	enableJSONSocket = true
 	managementURL = "https://mgmt.example.com"
 	configPath = "/tmp/test-config.json"
 	logFiles = []string{"/tmp/test.log"}
@@ -131,6 +141,8 @@ func TestCurrentServiceParams(t *testing.T) {
 
 	assert.Equal(t, "trace", params.LogLevel)
 	assert.Equal(t, "tcp://127.0.0.1:9999", params.DaemonAddr)
+	assert.Equal(t, "tcp://127.0.0.1:8080", params.JSONSocket)
+	assert.True(t, params.EnableJSONSocket)
 	assert.Equal(t, "https://mgmt.example.com", params.ManagementURL)
 	assert.Equal(t, "/tmp/test-config.json", params.ConfigPath)
 	assert.Equal(t, []string{"/tmp/test.log"}, params.LogFiles)
@@ -142,6 +154,8 @@ func TestCurrentServiceParams(t *testing.T) {
 func TestApplyServiceParams_OnlyUnchangedFlags(t *testing.T) {
 	origLogLevel := logLevel
 	origDaemonAddr := daemonAddr
+	origJSONSocket := jsonSocket
+	origEnableJSONSocket := enableJSONSocket
 	origManagementURL := managementURL
 	origConfigPath := configPath
 	origLogFiles := logFiles
@@ -151,6 +165,8 @@ func TestApplyServiceParams_OnlyUnchangedFlags(t *testing.T) {
 	t.Cleanup(func() {
 		logLevel = origLogLevel
 		daemonAddr = origDaemonAddr
+		jsonSocket = origJSONSocket
+		enableJSONSocket = origEnableJSONSocket
 		managementURL = origManagementURL
 		configPath = origConfigPath
 		logFiles = origLogFiles
@@ -162,6 +178,8 @@ func TestApplyServiceParams_OnlyUnchangedFlags(t *testing.T) {
 	// Reset all flags to defaults.
 	logLevel = "info"
 	daemonAddr = "unix:///var/run/netbird.sock"
+	jsonSocket = defaultJSONSocket
+	enableJSONSocket = false
 	managementURL = ""
 	configPath = "/etc/netbird/config.json"
 	logFiles = []string{"/var/log/netbird/client.log"}
@@ -184,6 +202,8 @@ func TestApplyServiceParams_OnlyUnchangedFlags(t *testing.T) {
 	saved := &serviceParams{
 		LogLevel:              "debug",
 		DaemonAddr:            "tcp://127.0.0.1:5555",
+		JSONSocket:            "tcp://127.0.0.1:8080",
+		EnableJSONSocket:      true,
 		ManagementURL:         "https://saved.example.com",
 		ConfigPath:            "/saved/config.json",
 		LogFiles:              []string{"/saved/client.log"},
@@ -201,6 +221,8 @@ func TestApplyServiceParams_OnlyUnchangedFlags(t *testing.T) {
 
 	// All other fields were not Changed, so they should use saved values.
 	assert.Equal(t, "tcp://127.0.0.1:5555", daemonAddr)
+	assert.Equal(t, "tcp://127.0.0.1:8080", jsonSocket)
+	assert.True(t, enableJSONSocket)
 	assert.Equal(t, "https://saved.example.com", managementURL)
 	assert.Equal(t, "/saved/config.json", configPath)
 	assert.Equal(t, []string{"/saved/client.log"}, logFiles)
@@ -212,14 +234,17 @@ func TestApplyServiceParams_OnlyUnchangedFlags(t *testing.T) {
 func TestApplyServiceParams_BooleanRevertToFalse(t *testing.T) {
 	origProfilesDisabled := profilesDisabled
 	origUpdateSettingsDisabled := updateSettingsDisabled
+	origEnableJSONSocket := enableJSONSocket
 	t.Cleanup(func() {
 		profilesDisabled = origProfilesDisabled
 		updateSettingsDisabled = origUpdateSettingsDisabled
+		enableJSONSocket = origEnableJSONSocket
 	})
 
 	// Simulate current state where booleans are true (e.g. set by previous install).
 	profilesDisabled = true
 	updateSettingsDisabled = true
+	enableJSONSocket = true
 
 	// Reset Changed state so flags appear unset.
 	serviceCmd.PersistentFlags().VisitAll(func(f *pflag.Flag) {
@@ -238,6 +263,7 @@ func TestApplyServiceParams_BooleanRevertToFalse(t *testing.T) {
 
 	assert.False(t, profilesDisabled, "saved false should override current true")
 	assert.False(t, updateSettingsDisabled, "saved false should override current true")
+	assert.False(t, enableJSONSocket, "saved false should override current true")
 }
 
 func TestApplyServiceParams_ClearManagementURL(t *testing.T) {
@@ -530,12 +556,15 @@ func fieldToGlobalVar(field string) string {
 	m := map[string]string{
 		"LogLevel":              "logLevel",
 		"DaemonAddr":            "daemonAddr",
+		"JSONSocket":            "jsonSocket",
 		"ManagementURL":         "managementURL",
 		"ConfigPath":            "configPath",
 		"LogFiles":              "logFiles",
 		"DisableProfiles":       "profilesDisabled",
 		"DisableUpdateSettings": "updateSettingsDisabled",
+		"EnableCapture":         "captureEnabled",
 		"DisableNetworks":       "networksDisabled",
+		"EnableJSONSocket":      "enableJSONSocket",
 		"ServiceEnvVars":        "serviceEnvVars",
 	}
 	if v, ok := m[field]; ok {

client/cmd/service_pipe_other.go

@@ -0,0 +1,14 @@
+//go:build !windows
+
+package cmd
+
+import (
+	"fmt"
+	"net"
+)
+
+// listenNamedPipe is Windows-only: no other platform serves the daemon on a
+// named pipe.
+func listenNamedPipe(string) (net.Listener, string, error) {
+	return nil, "", fmt.Errorf("named pipes are only supported on Windows")
+}

client/cmd/service_pipe_windows.go

@@ -0,0 +1,41 @@
+//go:build windows
+
+package cmd
+
+import (
+	"errors"
+	"fmt"
+	"net"
+
+	"github.com/Microsoft/go-winio"
+	log "github.com/sirupsen/logrus"
+
+	"github.com/netbirdio/netbird/client/internal/daemonaddr"
+	"github.com/netbirdio/netbird/client/internal/ipcauth"
+)
+
+// listenNamedPipe creates the daemon control pipe and reports the path it ended
+// up on. The security descriptor lets any local caller connect, as a Unix socket
+// at 0666 does, and the privileged operations are authorized separately from the
+// caller's token.
+//
+// The protected name comes first so that an unprivileged process cannot take the
+// name before the service does. Creating it requires being an administrator or
+// LocalSystem, so a daemon an ordinary user runs themselves, as in netstack mode,
+// falls back to the plain name; clients try both and check who serves them.
+func listenNamedPipe(name string) (net.Listener, string, error) {
+	var errs []error
+	for _, path := range daemonaddr.PipePaths(name) {
+		listener, err := winio.ListenPipe(path, &winio.PipeConfig{
+			SecurityDescriptor: ipcauth.DefaultPipeSDDL(),
+		})
+		if err != nil {
+			log.Debugf("not serving the daemon on %s: %v", path, err)
+			errs = append(errs, fmt.Errorf("%s: %w", path, err))
+			continue
+		}
+		return listener, path, nil
+	}
+
+	return nil, "", errors.Join(errs...)
+}

client/cmd/service_privileged_test.go

@@ -0,0 +1,196 @@
+//go:build privileged
+
+package cmd
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"runtime"
+	"testing"
+	"time"
+
+	"github.com/kardianos/service"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+const (
+	serviceStartTimeout = 10 * time.Second
+	serviceStopTimeout  = 5 * time.Second
+	statusPollInterval  = 500 * time.Millisecond
+)
+
+// waitForServiceStatus waits for service to reach expected status with timeout
+func waitForServiceStatus(expectedStatus service.Status, timeout time.Duration) (bool, error) {
+	cfg, err := newSVCConfig()
+	if err != nil {
+		return false, err
+	}
+
+	ctxSvc, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+	if err != nil {
+		return false, err
+	}
+
+	ctx, timeoutCancel := context.WithTimeout(context.Background(), timeout)
+	defer timeoutCancel()
+
+	ticker := time.NewTicker(statusPollInterval)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return false, fmt.Errorf("timeout waiting for service status %v", expectedStatus)
+		case <-ticker.C:
+			status, err := s.Status()
+			if err != nil {
+				// Continue polling on transient errors
+				continue
+			}
+			if status == expectedStatus {
+				return true, nil
+			}
+		}
+	}
+}
+
+// TestServiceLifecycle tests the complete service lifecycle
+func TestServiceLifecycle(t *testing.T) {
+	// TODO: Add support for Windows and macOS
+	if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
+		t.Skipf("Skipping service lifecycle test on unsupported OS: %s", runtime.GOOS)
+	}
+
+	if os.Getenv("CONTAINER") == "true" {
+		t.Skip("Skipping service lifecycle test in container environment")
+	}
+
+	originalServiceName := serviceName
+	serviceName = "netbirdtest" + fmt.Sprintf("%d", time.Now().Unix())
+	defer func() {
+		serviceName = originalServiceName
+	}()
+
+	tempDir := t.TempDir()
+	configPath = fmt.Sprintf("%s/netbird-test-config.json", tempDir)
+	logLevel = "info"
+	daemonAddr = fmt.Sprintf("unix://%s/netbird-test.sock", tempDir)
+
+	// Ensure cleanup even if a subtest fails and Stop/Uninstall subtests don't run.
+	t.Cleanup(func() {
+		cfg, err := newSVCConfig()
+		if err != nil {
+			t.Errorf("cleanup: create service config: %v", err)
+			return
+		}
+		ctxSvc, cancel := context.WithCancel(context.Background())
+		defer cancel()
+		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+		if err != nil {
+			t.Errorf("cleanup: create service: %v", err)
+			return
+		}
+
+		// If the subtests already cleaned up, there's nothing to do.
+		if _, err := s.Status(); err != nil {
+			return
+		}
+
+		if err := s.Stop(); err != nil {
+			t.Errorf("cleanup: stop service: %v", err)
+		}
+		if err := s.Uninstall(); err != nil {
+			t.Errorf("cleanup: uninstall service: %v", err)
+		}
+	})
+
+	ctx := context.Background()
+
+	t.Run("Install", func(t *testing.T) {
+		installCmd.SetContext(ctx)
+		err := installCmd.RunE(installCmd, []string{})
+		require.NoError(t, err)
+
+		cfg, err := newSVCConfig()
+		require.NoError(t, err)
+
+		ctxSvc, cancel := context.WithCancel(context.Background())
+		defer cancel()
+
+		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+		require.NoError(t, err)
+
+		status, err := s.Status()
+		assert.NoError(t, err)
+		assert.NotEqual(t, service.StatusUnknown, status)
+	})
+
+	t.Run("Start", func(t *testing.T) {
+		startCmd.SetContext(ctx)
+		err := startCmd.RunE(startCmd, []string{})
+		require.NoError(t, err)
+
+		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
+		require.NoError(t, err)
+		assert.True(t, running)
+	})
+
+	t.Run("Restart", func(t *testing.T) {
+		restartCmd.SetContext(ctx)
+		err := restartCmd.RunE(restartCmd, []string{})
+		require.NoError(t, err)
+
+		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
+		require.NoError(t, err)
+		assert.True(t, running)
+	})
+
+	t.Run("Reconfigure", func(t *testing.T) {
+		originalLogLevel := logLevel
+		logLevel = "debug"
+		defer func() {
+			logLevel = originalLogLevel
+		}()
+
+		reconfigureCmd.SetContext(ctx)
+		err := reconfigureCmd.RunE(reconfigureCmd, []string{})
+		require.NoError(t, err)
+
+		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
+		require.NoError(t, err)
+		assert.True(t, running)
+	})
+
+	t.Run("Stop", func(t *testing.T) {
+		stopCmd.SetContext(ctx)
+		err := stopCmd.RunE(stopCmd, []string{})
+		require.NoError(t, err)
+
+		stopped, err := waitForServiceStatus(service.StatusStopped, serviceStopTimeout)
+		require.NoError(t, err)
+		assert.True(t, stopped)
+	})
+
+	t.Run("Uninstall", func(t *testing.T) {
+		uninstallCmd.SetContext(ctx)
+		err := uninstallCmd.RunE(uninstallCmd, []string{})
+		require.NoError(t, err)
+
+		cfg, err := newSVCConfig()
+		require.NoError(t, err)
+
+		ctxSvc, cancel := context.WithCancel(context.Background())
+		defer cancel()
+
+		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
+		require.NoError(t, err)
+
+		_, err = s.Status()
+		assert.Error(t, err)
+	})
+}

client/cmd/service_socket.go

@@ -0,0 +1,119 @@
+//go:build !ios && !android
+
+package cmd
+
+import (
+	"errors"
+	"fmt"
+	"net"
+	"os"
+	"strings"
+	"syscall"
+	"time"
+
+	log "github.com/sirupsen/logrus"
+)
+
+type socketListener struct {
+	net.Listener
+	network string
+	address string
+}
+
+func listenOnAddress(addr string) (*socketListener, error) {
+	network, address, err := parseListenAddress(addr)
+	if err != nil {
+		return nil, err
+	}
+
+	if network == "npipe" {
+		listener, path, err := listenNamedPipe(address) //nolint:staticcheck
+		if err != nil {                                 //nolint:staticcheck // always errors on non-Windows builds
+			return nil, err
+		}
+		return &socketListener{Listener: listener, network: network, address: path}, nil
+	}
+
+	if network == "unix" {
+		removeStaleUnixSocket(address)
+	}
+
+	listener, err := net.Listen(network, address)
+	if err != nil {
+		return nil, err
+	}
+
+	return &socketListener{Listener: listener, network: network, address: address}, nil
+}
+
+func parseListenAddress(addr string) (string, string, error) {
+	network, address, ok := strings.Cut(addr, "://")
+	if !ok || network == "" || address == "" {
+		return "", "", fmt.Errorf("address must be in [unix|tcp|npipe]://[path|host:port|name] format: %q", addr)
+	}
+
+	switch network {
+	case "unix", "tcp", "npipe":
+		return network, address, nil
+	default:
+		return "", "", fmt.Errorf("unsupported daemon address protocol: %v", network)
+	}
+}
+
+func removeStaleUnixSocket(path string) {
+	stat, err := os.Lstat(path)
+	if err != nil {
+		if !os.IsNotExist(err) {
+			log.Debugf("stat socket file: %v", err)
+		}
+		return
+	}
+
+	if stat.Mode()&os.ModeSocket == 0 {
+		return
+	}
+
+	if !isStaleUnixSocket(path) {
+		return
+	}
+
+	if err := os.Remove(path); err != nil {
+		log.Debugf("remove socket file: %v", err)
+	}
+}
+
+func isStaleUnixSocket(path string) bool {
+	conn, err := net.DialTimeout("unix", path, 100*time.Millisecond)
+	if err == nil {
+		if closeErr := conn.Close(); closeErr != nil {
+			log.Debugf("close unix socket probe: %v", closeErr)
+		}
+		return false
+	}
+
+	if os.IsNotExist(err) || os.IsPermission(err) || os.IsTimeout(err) {
+		log.Debugf("not removing unix socket %s after probe error: %v", path, err)
+		return false
+	}
+
+	return errors.Is(err, syscall.ECONNREFUSED)
+}
+
+func removeStaleUnixSocketForAddress(addr string) {
+	network, address, err := parseListenAddress(addr)
+	if err != nil || network != "unix" {
+		return
+	}
+	removeStaleUnixSocket(address)
+}
+
+func (l *socketListener) chmodUnixSocket(description string) error {
+	if l == nil || l.network != "unix" {
+		return nil
+	}
+
+	if err := os.Chmod(l.address, 0666); err != nil {
+		return fmt.Errorf("failed setting %s permissions for %s: %w", description, l.address, err)
+	}
+	return nil
+}

client/cmd/service_test.go

@@ -1,16 +1,12 @@
 package cmd
 
 import (
-	"context"
-	"fmt"
 	"os"
 	"os/signal"
 	"runtime"
 	"syscall"
 	"testing"
-	"time"
 
-	"github.com/kardianos/service"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -31,186 +27,6 @@ func TestMain(m *testing.M) {
 	os.Exit(m.Run())
 }
 
-const (
-	serviceStartTimeout = 10 * time.Second
-	serviceStopTimeout  = 5 * time.Second
-	statusPollInterval  = 500 * time.Millisecond
-)
-
-// waitForServiceStatus waits for service to reach expected status with timeout
-func waitForServiceStatus(expectedStatus service.Status, timeout time.Duration) (bool, error) {
-	cfg, err := newSVCConfig()
-	if err != nil {
-		return false, err
-	}
-
-	ctxSvc, cancel := context.WithCancel(context.Background())
-	defer cancel()
-
-	s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-	if err != nil {
-		return false, err
-	}
-
-	ctx, timeoutCancel := context.WithTimeout(context.Background(), timeout)
-	defer timeoutCancel()
-
-	ticker := time.NewTicker(statusPollInterval)
-	defer ticker.Stop()
-
-	for {
-		select {
-		case <-ctx.Done():
-			return false, fmt.Errorf("timeout waiting for service status %v", expectedStatus)
-		case <-ticker.C:
-			status, err := s.Status()
-			if err != nil {
-				// Continue polling on transient errors
-				continue
-			}
-			if status == expectedStatus {
-				return true, nil
-			}
-		}
-	}
-}
-
-// TestServiceLifecycle tests the complete service lifecycle
-func TestServiceLifecycle(t *testing.T) {
-	// TODO: Add support for Windows and macOS
-	if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
-		t.Skipf("Skipping service lifecycle test on unsupported OS: %s", runtime.GOOS)
-	}
-
-	if os.Getenv("CONTAINER") == "true" {
-		t.Skip("Skipping service lifecycle test in container environment")
-	}
-
-	originalServiceName := serviceName
-	serviceName = "netbirdtest" + fmt.Sprintf("%d", time.Now().Unix())
-	defer func() {
-		serviceName = originalServiceName
-	}()
-
-	tempDir := t.TempDir()
-	configPath = fmt.Sprintf("%s/netbird-test-config.json", tempDir)
-	logLevel = "info"
-	daemonAddr = fmt.Sprintf("unix://%s/netbird-test.sock", tempDir)
-
-	// Ensure cleanup even if a subtest fails and Stop/Uninstall subtests don't run.
-	t.Cleanup(func() {
-		cfg, err := newSVCConfig()
-		if err != nil {
-			t.Errorf("cleanup: create service config: %v", err)
-			return
-		}
-		ctxSvc, cancel := context.WithCancel(context.Background())
-		defer cancel()
-		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-		if err != nil {
-			t.Errorf("cleanup: create service: %v", err)
-			return
-		}
-
-		// If the subtests already cleaned up, there's nothing to do.
-		if _, err := s.Status(); err != nil {
-			return
-		}
-
-		if err := s.Stop(); err != nil {
-			t.Errorf("cleanup: stop service: %v", err)
-		}
-		if err := s.Uninstall(); err != nil {
-			t.Errorf("cleanup: uninstall service: %v", err)
-		}
-	})
-
-	ctx := context.Background()
-
-	t.Run("Install", func(t *testing.T) {
-		installCmd.SetContext(ctx)
-		err := installCmd.RunE(installCmd, []string{})
-		require.NoError(t, err)
-
-		cfg, err := newSVCConfig()
-		require.NoError(t, err)
-
-		ctxSvc, cancel := context.WithCancel(context.Background())
-		defer cancel()
-
-		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-		require.NoError(t, err)
-
-		status, err := s.Status()
-		assert.NoError(t, err)
-		assert.NotEqual(t, service.StatusUnknown, status)
-	})
-
-	t.Run("Start", func(t *testing.T) {
-		startCmd.SetContext(ctx)
-		err := startCmd.RunE(startCmd, []string{})
-		require.NoError(t, err)
-
-		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
-		require.NoError(t, err)
-		assert.True(t, running)
-	})
-
-	t.Run("Restart", func(t *testing.T) {
-		restartCmd.SetContext(ctx)
-		err := restartCmd.RunE(restartCmd, []string{})
-		require.NoError(t, err)
-
-		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
-		require.NoError(t, err)
-		assert.True(t, running)
-	})
-
-	t.Run("Reconfigure", func(t *testing.T) {
-		originalLogLevel := logLevel
-		logLevel = "debug"
-		defer func() {
-			logLevel = originalLogLevel
-		}()
-
-		reconfigureCmd.SetContext(ctx)
-		err := reconfigureCmd.RunE(reconfigureCmd, []string{})
-		require.NoError(t, err)
-
-		running, err := waitForServiceStatus(service.StatusRunning, serviceStartTimeout)
-		require.NoError(t, err)
-		assert.True(t, running)
-	})
-
-	t.Run("Stop", func(t *testing.T) {
-		stopCmd.SetContext(ctx)
-		err := stopCmd.RunE(stopCmd, []string{})
-		require.NoError(t, err)
-
-		stopped, err := waitForServiceStatus(service.StatusStopped, serviceStopTimeout)
-		require.NoError(t, err)
-		assert.True(t, stopped)
-	})
-
-	t.Run("Uninstall", func(t *testing.T) {
-		uninstallCmd.SetContext(ctx)
-		err := uninstallCmd.RunE(uninstallCmd, []string{})
-		require.NoError(t, err)
-
-		cfg, err := newSVCConfig()
-		require.NoError(t, err)
-
-		ctxSvc, cancel := context.WithCancel(context.Background())
-		defer cancel()
-
-		s, err := newSVC(newProgram(ctxSvc, cancel), cfg)
-		require.NoError(t, err)
-
-		_, err = s.Status()
-		assert.Error(t, err)
-	})
-}
-
 // TestServiceEnvVars tests environment variable parsing
 func TestServiceEnvVars(t *testing.T) {
 	tests := []struct {

client/cmd/ssh.go

@@ -523,7 +523,7 @@ func parseHostnameAndCommand(args []string) error {
 }
 
 func runSSH(ctx context.Context, addr string, cmd *cobra.Command) error {
-	target := fmt.Sprintf("%s:%d", addr, port)
+	target := net.JoinHostPort(strings.Trim(addr, "[]"), strconv.Itoa(port))
 	c, err := sshclient.Dial(ctx, target, username, sshclient.DialOptions{
 		KnownHostsFile:     knownHostsFile,
 		IdentityFile:       identityFile,
@@ -787,10 +787,10 @@ func isUnixSocket(path string) bool {
 	return strings.HasPrefix(path, "/") || strings.HasPrefix(path, "./")
 }
 
-// normalizeLocalHost converts "*" to "0.0.0.0" for binding to all interfaces.
+// normalizeLocalHost converts "*" to "" for binding to all interfaces (dual-stack).
 func normalizeLocalHost(host string) string {
 	if host == "*" {
-		return "0.0.0.0"
+		return ""
 	}
 	return host
 }

client/cmd/ssh_test.go

@@ -527,10 +527,10 @@ func TestParsePortForward(t *testing.T) {
 		{
 			name:           "wildcard bind all interfaces",
 			spec:           "*:8080:localhost:80",
-			expectedLocal:  "0.0.0.0:8080",
+			expectedLocal:  ":8080",
 			expectedRemote: "localhost:80",
 			expectError:    false,
-			description:    "Wildcard * should bind to all interfaces (0.0.0.0)",
+			description:    "Wildcard * should bind to all interfaces (dual-stack)",
 		},
 		{
 			name:           "wildcard for port only",

client/cmd/status.go

@@ -6,12 +6,12 @@ import (
 	"net"
 	"net/netip"
 	"strings"
+	"time"
 
 	"github.com/spf13/cobra"
 	"google.golang.org/grpc/status"
 
 	"github.com/netbirdio/netbird/client/internal"
-	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/proto"
 	nbstatus "github.com/netbirdio/netbird/client/status"
 	"github.com/netbirdio/netbird/util"
@@ -20,6 +20,7 @@ import (
 var (
 	detailFlag           bool
 	ipv4Flag             bool
+	ipv6Flag             bool
 	jsonFlag             bool
 	yamlFlag             bool
 	ipsFilter            []string
@@ -42,15 +43,16 @@ func init() {
 	ipsFilterMap = make(map[string]struct{})
 	prefixNamesFilterMap = make(map[string]struct{})
 	statusCmd.PersistentFlags().BoolVarP(&detailFlag, "detail", "d", false, "display detailed status information in human-readable format")
-	statusCmd.PersistentFlags().BoolVar(&jsonFlag, "json", false, "display detailed status information in json format")
-	statusCmd.PersistentFlags().BoolVar(&yamlFlag, "yaml", false, "display detailed status information in yaml format")
-	statusCmd.PersistentFlags().BoolVar(&ipv4Flag, "ipv4", false, "display only NetBird IPv4 of this peer, e.g., --ipv4 will output 100.64.0.33")
-	statusCmd.MarkFlagsMutuallyExclusive("detail", "json", "yaml", "ipv4")
-	statusCmd.PersistentFlags().StringSliceVar(&ipsFilter, "filter-by-ips", []string{}, "filters the detailed output by a list of one or more IPs, e.g., --filter-by-ips 100.64.0.100,100.64.0.200")
-	statusCmd.PersistentFlags().StringSliceVar(&prefixNamesFilter, "filter-by-names", []string{}, "filters the detailed output by a list of one or more peer FQDN or hostnames, e.g., --filter-by-names peer-a,peer-b.netbird.cloud")
-	statusCmd.PersistentFlags().StringVar(&statusFilter, "filter-by-status", "", "filters the detailed output by connection status(idle|connecting|connected), e.g., --filter-by-status connected")
-	statusCmd.PersistentFlags().StringVar(&connectionTypeFilter, "filter-by-connection-type", "", "filters the detailed output by connection type (P2P|Relayed), e.g., --filter-by-connection-type P2P")
-	statusCmd.PersistentFlags().StringVar(&checkFlag, "check", "", "run a health check and exit with code 0 on success, 1 on failure (live|ready|startup)")
+	statusCmd.PersistentFlags().BoolVarP(&jsonFlag, "json", "j", false, "display detailed status information in json format")
+	statusCmd.PersistentFlags().BoolVarP(&yamlFlag, "yaml", "y", false, "display detailed status information in yaml format")
+	statusCmd.PersistentFlags().BoolVarP(&ipv4Flag, "ipv4", "4", false, "display only NetBird IPv4 of this peer, e.g., --ipv4 will output 100.64.0.33")
+	statusCmd.PersistentFlags().BoolVarP(&ipv6Flag, "ipv6", "6", false, "display only NetBird IPv6 of this peer")
+	statusCmd.MarkFlagsMutuallyExclusive("detail", "json", "yaml", "ipv4", "ipv6")
+	statusCmd.PersistentFlags().StringSliceVarP(&ipsFilter, "filter-by-ips", "I", []string{}, "filters the detailed output by a list of one or more IPs (v4 or v6), e.g., --filter-by-ips 100.64.0.100,fd00::1")
+	statusCmd.PersistentFlags().StringSliceVarP(&prefixNamesFilter, "filter-by-names", "N", []string{}, "filters the detailed output by a list of one or more peer FQDN or hostnames, e.g., --filter-by-names peer-a,peer-b.netbird.cloud")
+	statusCmd.PersistentFlags().StringVarP(&statusFilter, "filter-by-status", "S", "", "filters the detailed output by connection status(idle|connecting|connected), e.g., --filter-by-status connected")
+	statusCmd.PersistentFlags().StringVarP(&connectionTypeFilter, "filter-by-connection-type", "T", "", "filters the detailed output by connection type (P2P|Relayed), e.g., --filter-by-connection-type P2P")
+	statusCmd.PersistentFlags().StringVarP(&checkFlag, "check", "C", "", "run a health check and exit with code 0 on success, 1 on failure (live|ready|startup)")
 }
 
 func statusFunc(cmd *cobra.Command, args []string) error {
@@ -101,14 +103,32 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 		return nil
 	}
 
-	pm := profilemanager.NewProfileManager()
-	var profName string
-	if activeProf, err := pm.GetActiveProfile(); err == nil {
-		profName = activeProf.Name
+	if ipv6Flag {
+		ipv6 := resp.GetFullStatus().GetLocalPeerState().GetIpv6()
+		if ipv6 != "" {
+			cmd.Print(parseInterfaceIP(ipv6))
+		}
+		return nil
+	}
+
+	// Resolve the active profile's display name via the daemon, which runs
+	// as root and can read the per-user profile files. The local profile
+	// manager only knows the active profile ID, not its display name.
+	profName := getActiveProfileName(ctx)
+
+	var sessionExpiresAt time.Time
+	if ts := resp.GetSessionExpiresAt(); ts.IsValid() {
+		sessionExpiresAt = ts.AsTime().UTC()
+	}
+
+	anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
+	if err != nil {
+		return err
 	}
 
 	var outputInformationHolder = nbstatus.ConvertToStatusOutputOverview(resp.GetFullStatus(), nbstatus.ConvertOptions{
-		Anonymize:            anonymizeFlag,
+		Anonymize:            anonymizeEnabled,
+		AnonymizeLevel:       anonymizeLevel,
 		DaemonVersion:        resp.GetDaemonVersion(),
 		DaemonStatus:         nbstatus.ParseDaemonStatus(status),
 		StatusFilter:         statusFilter,
@@ -117,6 +137,7 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 		IPsFilter:            ipsFilterMap,
 		ConnectionTypeFilter: connectionTypeFilter,
 		ProfileName:          profName,
+		SessionExpiresAt:     sessionExpiresAt,
 	})
 	var statusOutputString string
 	switch {
@@ -157,6 +178,25 @@ func getStatus(ctx context.Context, fullPeerStatus bool, shouldRunProbes bool) (
 	return resp, nil
 }
 
+// getActiveProfileName asks the daemon for the active profile's display
+// name. The daemon runs as root and can read the per-user profile files to
+// resolve the ID to its human-readable name. Returns an empty string on any
+// error so status output degrades gracefully.
+func getActiveProfileName(ctx context.Context) string {
+	conn, err := DialClientGRPCServer(ctx, daemonAddr)
+	if err != nil {
+		return ""
+	}
+	defer conn.Close()
+
+	resp, err := proto.NewDaemonServiceClient(conn).GetActiveProfile(ctx, &proto.GetActiveProfileRequest{})
+	if err != nil {
+		return ""
+	}
+
+	return resp.GetProfileName()
+}
+
 func parseFilters() error {
 	switch strings.ToLower(statusFilter) {
 	case "", "idle", "connecting", "connected":

client/cmd/system.go

@@ -8,6 +8,7 @@ const (
 	disableFirewallFlag     = "disable-firewall"
 	blockLANAccessFlag      = "block-lan-access"
 	blockInboundFlag        = "block-inbound"
+	disableIPv6Flag         = "disable-ipv6"
 )
 
 var (
@@ -17,6 +18,7 @@ var (
 	disableFirewall     bool
 	blockLANAccess      bool
 	blockInbound        bool
+	disableIPv6         bool
 )
 
 func init() {
@@ -39,4 +41,7 @@ func init() {
 	upCmd.PersistentFlags().BoolVar(&blockInbound, blockInboundFlag, false,
 		"Block inbound connections. If enabled, the client will not allow any inbound connections to the local machine nor routed networks.\n"+
 			"This overrides any policies received from the management service.")
+
+	upCmd.PersistentFlags().BoolVar(&disableIPv6, disableIPv6Flag, false,
+		"Disable IPv6 overlay. If enabled, the client won't request or use an IPv6 overlay address.")
 }

client/cmd/testutil_test.go

@@ -6,12 +6,12 @@ import (
 	"testing"
 	"time"
 
-	"github.com/golang/mock/gomock"
+	"go.uber.org/mock/gomock"
 	"github.com/stretchr/testify/require"
 	"go.opentelemetry.io/otel"
 	"google.golang.org/grpc"
 
-	"github.com/netbirdio/management-integrations/integrations"
+	"github.com/netbirdio/netbird/management/server/integrations/integrated_validator/validator"
 
 	"github.com/netbirdio/netbird/management/internals/modules/permissions"
 
@@ -110,7 +110,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 		t.Fatal(err)
 	}
 
-	iv, _ := integrations.NewIntegratedValidator(ctx, peersmanager, settingsManagerMock, eventStore, cacheStore)
+	iv, _ := validator.NewIntegratedValidator(ctx, peersmanager, settingsManagerMock, eventStore, cacheStore)
 
 	metrics, err := telemetry.NewDefaultAppMetrics(ctx)
 	require.NoError(t, err)
@@ -125,7 +125,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 
 	updateManager := update_channel.NewPeersUpdateManager(metrics)
 	requestBuffer := mgmt.NewAccountRequestBuffer(ctx, store)
-	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config)
+	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config, nil)
 
 	accountManager, err := mgmt.BuildManager(ctx, config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false, cacheStore)
 	if err != nil {
@@ -136,7 +136,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 	if err != nil {
 		t.Fatal(err)
 	}
-	mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, &mgmt.MockIntegratedValidator{}, networkMapController, nil)
+	mgmtServer, err := nbgrpc.NewServer(config, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, &mgmt.MockIntegratedValidator{}, networkMapController, nil, nil)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -161,7 +161,7 @@ func startClientDaemon(
 	s := grpc.NewServer()
 
 	server := client.New(ctx,
-		"", "", false, false, false)
+		"", "", false, false, false, false)
 	if err := server.Start(); err != nil {
 		t.Fatal(err)
 	}

client/cmd/up.go

@@ -2,10 +2,10 @@ package cmd
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"net"
 	"net/netip"
-	"os/user"
 	"runtime"
 	"strings"
 	"time"
@@ -21,7 +21,10 @@ import (
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mdm"
+	nbnet "github.com/netbirdio/netbird/client/net"
 	"github.com/netbirdio/netbird/client/proto"
+	"github.com/netbirdio/netbird/client/server"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/shared/management/domain"
 	"github.com/netbirdio/netbird/util"
@@ -39,15 +42,21 @@ const (
 	noBrowserFlag = "no-browser"
 	noBrowserDesc = "do not open the browser for SSO login"
 
+	showQRFlag = "qr"
+	showQRDesc = "show QR code for the SSO login URL (useful for headless machines without browser access)"
+
 	profileNameFlag = "profile"
 	profileNameDesc = "profile name to use for the login. If not specified, the last used profile will be used."
 )
 
+var errDaemonActiveProfileUnsupported = errors.New("daemon does not support active profile lookup")
+
 var (
 	foregroundMode     bool
 	dnsLabels          []string
 	dnsLabelsValidated domain.List
 	noBrowser          bool
+	showQR             bool
 	profileName        string
 	configPath         string
 
@@ -80,6 +89,7 @@ func init() {
 	)
 
 	upCmd.PersistentFlags().BoolVar(&noBrowser, noBrowserFlag, false, noBrowserDesc)
+	upCmd.PersistentFlags().BoolVar(&showQR, showQRFlag, false, showQRDesc)
 	upCmd.PersistentFlags().StringVar(&profileName, profileNameFlag, "", profileNameDesc)
 	upCmd.PersistentFlags().StringVarP(&configPath, "config", "c", "", "(DEPRECATED) NetBird config file location. ")
 
@@ -115,30 +125,25 @@ func upFunc(cmd *cobra.Command, args []string) error {
 
 	pm := profilemanager.NewProfileManager()
 
-	username, err := user.Current()
+	username, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
 
+	var activeProf *profilemanager.Profile
 	var profileSwitched bool
 	// switch profile if provided
 	if profileName != "" {
-		err = switchProfile(cmd.Context(), profileName, username.Username)
+		activeProf, err = switchOrCreateProfile(cmd.Context(), pm, profileName, username.Username)
 		if err != nil {
 			return fmt.Errorf("switch profile: %v", err)
 		}
-
-		err = pm.SwitchProfile(profileName)
-		if err != nil {
-			return fmt.Errorf("switch profile: %v", err)
-		}
-
 		profileSwitched = true
-	}
-
-	activeProf, err := pm.GetActiveProfile()
-	if err != nil {
-		return fmt.Errorf("get active profile: %v", err)
+	} else {
+		activeProf, err = pm.GetActiveProfile()
+		if err != nil {
+			return fmt.Errorf("get active profile: %v", err)
+		}
 	}
 
 	if foregroundMode {
@@ -147,6 +152,54 @@ func upFunc(cmd *cobra.Command, args []string) error {
 	return runInDaemonMode(ctx, cmd, pm, activeProf, profileSwitched)
 }
 
+// switchOrCreateProfile switches the active profile to the one identified by
+// handle, creating it first when it does not exist yet. This restores the
+// pre-0.73 behaviour where `netbird up --profile <name>` auto-creates a
+// missing profile instead of failing. Returns the daemon-resolved profile so
+// callers act on it directly instead of re-reading the local state, which is
+// not updated under sudo.
+func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManager, handle, username string) (*profilemanager.Profile, error) {
+	resolvedID, err := switchProfile(ctx, handle, username)
+	if err != nil {
+		st, ok := gstatus.FromError(err)
+		if !ok || st.Code() != codes.NotFound {
+			return nil, err
+		}
+		// Don't fail immediately on a create error: a concurrent run may
+		// have created the profile between the NotFound above and this
+		// call, in which case the retried switch still succeeds. Only
+		// surface the create error if the switch also fails.
+		_, createErr := createProfile(ctx, handle, username)
+		if resolvedID, err = switchProfile(ctx, handle, username); err != nil {
+			if createErr != nil {
+				return nil, fmt.Errorf("create profile: %w", createErr)
+			}
+			return nil, err
+		}
+	}
+
+	if err := pm.SwitchProfile(resolvedID); err != nil {
+		return nil, err
+	}
+	return &profilemanager.Profile{ID: resolvedID}, nil
+}
+
+// createProfile dials the daemon and creates a new profile with the given
+// display name, returning its generated ID. Use addProfileOnDaemon directly
+// when a daemon client is already available to reuse the connection.
+func createProfile(ctx context.Context, profileName, username string) (profilemanager.ID, error) {
+	conn, err := DialClientGRPCServer(ctx, daemonAddr)
+	if err != nil {
+		//nolint
+		return "", fmt.Errorf("failed to connect to daemon error: %v\n"+
+			"If the daemon is not running please run: "+
+			"\nnetbird service install \nnetbird service start\n", err)
+	}
+	defer conn.Close()
+
+	return addProfileOnDaemon(ctx, proto.NewDaemonServiceClient(conn), profileName, username)
+}
+
 func runInForegroundMode(ctx context.Context, cmd *cobra.Command, activeProf *profilemanager.Profile) error {
 	// override the default profile filepath if provided
 	if configPath != "" {
@@ -182,10 +235,32 @@ func runInForegroundMode(ctx context.Context, cmd *cobra.Command, activeProf *pr
 	if err != nil {
 		return fmt.Errorf("get config file: %v", err)
 	}
+	// CLI foreground path runs without the daemon Server: layer in the
+	// active MDM policy explicitly so a forced ManagementURL / PSK /
+	// other managed key actually takes effect on this run.
+	config.ApplyMDMPolicy(mdm.NewLoader(nil).Load())
 
 	_, _ = profilemanager.UpdateOldManagementURL(ctx, config, configFilePath)
 
-	err = foregroundLogin(ctx, cmd, config, providedSetupKey, activeProf.Name)
+	// Restore residual state left by a previous run that did not shut down
+	// cleanly, mirroring what the daemon does before connecting: it recovers
+	// DNS config (a stale resolv.conf takeover can make the management
+	// hostname unresolvable), firewall rules, ssh config and legacy routing.
+	// Route cleanup itself happens at engine start; nbnet.Init() below lets
+	// the management dial bypass a leftover fwmark rule until then.
+	// Foreground mode is particularly exposed in containers: a crashed
+	// container restarts inside the same (pod) network namespace, so stale
+	// state survives while the process does not.
+	if err := server.RestoreResidualState(ctx, profilemanager.NewServiceManager(configPath).GetStatePath()); err != nil {
+		log.Warnf("failed to restore residual state: %v", err)
+	}
+
+	// Enable advanced routing (as the daemon does on startup) so the
+	// management dial bypasses a leftover fwmark rule instead of being
+	// shunted into a stale routing table.
+	nbnet.Init()
+
+	err = foregroundLogin(ctx, cmd, config, providedSetupKey, activeProf.ID)
 	if err != nil {
 		return fmt.Errorf("foreground login failed: %v", err)
 	}
@@ -238,6 +313,30 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 		return fmt.Errorf("unable to get daemon status: %v", err)
 	}
 
+	// Under sudo the invoking user's local active-profile mirror is never
+	// written (the SwitchProfile write is a no-op), and plain root has no
+	// invoking user at all — so the mirror read into activeProf above is stale
+	// or defaulted and must not drive the daemon. With no --profile to make the
+	// choice explicit, take the profile the daemon already holds for this user
+	// instead: it stays on the user's current profile rather than silently
+	// switching to the mirror's default, and refuses when the daemon is on
+	// another user's profile.
+	if profileName == "" && !profilemanager.MirrorIsAuthoritative() {
+		u, err := profilemanager.InvokingUser()
+		if err != nil {
+			return fmt.Errorf("get current user: %v", err)
+		}
+		resolved, err := daemonActiveProfileForUser(ctx, client, u.Username)
+		switch {
+		case errors.Is(err, errDaemonActiveProfileUnsupported):
+			log.Warnf("keeping the locally resolved profile: %v", err)
+		case err != nil:
+			return err
+		default:
+			activeProf = resolved
+		}
+	}
+
 	if status.Status == string(internal.StatusConnected) {
 		if !profileSwitched {
 			cmd.Println("Already connected")
@@ -250,18 +349,18 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 		}
 	}
 
-	username, err := user.Current()
+	username, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
 
 	// set the new config
-	req := setupSetConfigReq(customDNSAddressConverted, cmd, activeProf.Name, username.Username)
+	req := setupSetConfigReq(customDNSAddressConverted, cmd, activeProf.ID.String(), username.Username)
 	if _, err := client.SetConfig(ctx, req); err != nil {
 		if st, ok := gstatus.FromError(err); ok && st.Code() == codes.Unavailable {
-			log.Warnf("setConfig method is not available in the daemon")
+			log.Warnf("setConfig method is not available in the daemon: %s", st.Message())
 		} else {
-			return fmt.Errorf("call service setConfig method: %v", err)
+			return daemonCallError("call service setConfig method", err)
 		}
 	}
 
@@ -284,10 +383,11 @@ func doDaemonUp(ctx context.Context, cmd *cobra.Command, client proto.DaemonServ
 		return fmt.Errorf("setup login request: %v", err)
 	}
 
-	loginRequest.ProfileName = &activeProf.Name
+	profileID := activeProf.ID.String()
+	loginRequest.ProfileName = &profileID
 	loginRequest.Username = &username
 
-	profileState, err := pm.GetProfileState(activeProf.Name)
+	profileState, err := pm.GetProfileState(activeProf.ID)
 	if err != nil {
 		log.Debugf("failed to get profile state for login hint: %v", err)
 	} else if profileState.Email != "" {
@@ -314,7 +414,7 @@ func doDaemonUp(ctx context.Context, cmd *cobra.Command, client proto.DaemonServ
 	}
 
 	if loginErr != nil {
-		return fmt.Errorf("login failed: %v", loginErr)
+		return daemonCallError("login failed", loginErr)
 	}
 
 	if loginResp.NeedsSSOLogin {
@@ -324,35 +424,30 @@ func doDaemonUp(ctx context.Context, cmd *cobra.Command, client proto.DaemonServ
 	}
 
 	if _, err := client.Up(ctx, &proto.UpRequest{
-		ProfileName: &activeProf.Name,
+		ProfileName: &profileID,
 		Username:    &username,
 	}); err != nil {
-		return fmt.Errorf("call service up method: %v", err)
+		return daemonCallError("call service up method", err)
 	}
 
 	return nil
 }
 
-func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, profileName, username string) *proto.SetConfigRequest {
-	var req proto.SetConfigRequest
-	req.ProfileName = profileName
-	req.Username = username
-
-	req.ManagementUrl = managementURL
-	req.AdminURL = adminURL
-	req.NatExternalIPs = natExternalIPs
-	req.CustomDNSAddress = customDNSAddressConverted
-	req.ExtraIFaceBlacklist = extraIFaceBlackList
-	req.DnsLabels = dnsLabelsValidated.ToPunycodeList()
-	req.CleanDNSLabels = dnsLabels != nil && len(dnsLabels) == 0
-	req.CleanNATExternalIPs = natExternalIPs != nil && len(natExternalIPs) == 0
-
-	if cmd.Flag(enableRosenpassFlag).Changed {
-		req.RosenpassEnabled = &rosenpassEnabled
-	}
-	if cmd.Flag(rosenpassPermissiveFlag).Changed {
-		req.RosenpassPermissive = &rosenpassPermissive
+// setBoolPtrIfChanged points dst at a copy of val when the named bool flag was
+// explicitly set on cmd. It collapses the repeated
+// "if cmd.Flag(x).Changed { field = &val }" pattern in the request builders into
+// a single call, keeping their cognitive complexity within bounds.
+func setBoolPtrIfChanged(cmd *cobra.Command, name string, dst **bool, val bool) {
+	if cmd.Flag(name).Changed {
+		dst2 := val
+		*dst = &dst2
 	}
+}
+
+// setSSHSetConfigFields copies the SSH server flags the user actually
+// passed into req, leaving the rest unset so the daemon keeps the
+// persisted values.
+func setSSHSetConfigFields(req *proto.SetConfigRequest, cmd *cobra.Command) {
 	if cmd.Flag(serverSSHAllowedFlag).Changed {
 		req.ServerSSHAllowed = &serverSSHAllowed
 	}
@@ -375,6 +470,31 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
 		req.SshJWTCacheTTL = &sshJWTCacheTTL32
 	}
+}
+
+func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, profileName, username string) *proto.SetConfigRequest {
+	var req proto.SetConfigRequest
+	req.ProfileName = profileName
+	req.Username = username
+
+	req.ManagementUrl = managementURL
+	req.AdminURL = adminURL
+	req.NatExternalIPs = natExternalIPs
+	req.CustomDNSAddress = customDNSAddressConverted
+	req.ExtraIFaceBlacklist = extraIFaceBlackList
+	req.DnsLabels = dnsLabelsValidated.ToPunycodeList()
+	req.CleanDNSLabels = dnsLabels != nil && len(dnsLabels) == 0
+	req.CleanNATExternalIPs = natExternalIPs != nil && len(natExternalIPs) == 0
+
+	if cmd.Flag(enableRosenpassFlag).Changed {
+		req.RosenpassEnabled = &rosenpassEnabled
+	}
+	if cmd.Flag(rosenpassPermissiveFlag).Changed {
+		req.RosenpassPermissive = &rosenpassPermissive
+	}
+	setSSHSetConfigFields(&req, cmd)
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &req.RemoteJobsAllowed, remoteJobsAllowed)
+
 	if cmd.Flag(interfaceNameFlag).Changed {
 		if err := parseInterfaceName(interfaceName); err != nil {
 			log.Errorf("parse interface name: %v", err)
@@ -430,8 +550,15 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		req.BlockInbound = &blockInbound
 	}
 
-	if cmd.Flag(enableLazyConnectionFlag).Changed {
-		req.LazyConnectionEnabled = &lazyConnEnabled
+	if cmd.Flag(disableIPv6Flag).Changed {
+		req.DisableIpv6 = &disableIPv6
+	}
+
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		req.EnableLocalMetrics = &localMetricsEnabled
+	}
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		req.LocalMetricsAddress = &localMetricsAddr
 	}
 
 	return &req
@@ -458,6 +585,7 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 	if cmd.Flag(serverSSHAllowedFlag).Changed {
 		ic.ServerSSHAllowed = &serverSSHAllowed
 	}
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &ic.RemoteJobsAllowed, remoteJobsAllowed)
 
 	if cmd.Flag(enableSSHRootFlag).Changed {
 		ic.EnableSSHRoot = &enableSSHRoot
@@ -547,12 +675,49 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 		ic.BlockInbound = &blockInbound
 	}
 
-	if cmd.Flag(enableLazyConnectionFlag).Changed {
-		ic.LazyConnectionEnabled = &lazyConnEnabled
+	if cmd.Flag(disableIPv6Flag).Changed {
+		ic.DisableIPv6 = &disableIPv6
 	}
+
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		ic.LocalMetricsEnabled = &localMetricsEnabled
+	}
+
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		ic.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	return &ic, nil
 }
 
+// setSSHLoginFields copies the SSH server flags the user actually passed
+// into req, leaving the rest unset so the daemon keeps the persisted
+// values.
+func setSSHLoginFields(req *proto.LoginRequest, cmd *cobra.Command) {
+	if cmd.Flag(serverSSHAllowedFlag).Changed {
+		req.ServerSSHAllowed = &serverSSHAllowed
+	}
+	if cmd.Flag(enableSSHRootFlag).Changed {
+		req.EnableSSHRoot = &enableSSHRoot
+	}
+	if cmd.Flag(enableSSHSFTPFlag).Changed {
+		req.EnableSSHSFTP = &enableSSHSFTP
+	}
+	if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
+		req.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
+	}
+	if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
+		req.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
+	}
+	if cmd.Flag(disableSSHAuthFlag).Changed {
+		req.DisableSSHAuth = &disableSSHAuth
+	}
+	if cmd.Flag(sshJWTCacheTTLFlag).Changed {
+		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
+		req.SshJWTCacheTTL = &sshJWTCacheTTL32
+	}
+}
+
 func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte, cmd *cobra.Command) (*proto.LoginRequest, error) {
 	loginRequest := proto.LoginRequest{
 		SetupKey:            providedSetupKey,
@@ -560,7 +725,7 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		NatExternalIPs:      natExternalIPs,
 		CleanNATExternalIPs: natExternalIPs != nil && len(natExternalIPs) == 0,
 		CustomDNSAddress:    customDNSAddressConverted,
-		IsUnixDesktopClient: isUnixRunningDesktop(),
+		IsUnixDesktopClient: util.HasGraphicalSession(),
 		Hostname:            hostName,
 		ExtraIFaceBlacklist: extraIFaceBlackList,
 		DnsLabels:           dnsLabels,
@@ -579,39 +744,21 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		loginRequest.RosenpassPermissive = &rosenpassPermissive
 	}
 
-	if cmd.Flag(serverSSHAllowedFlag).Changed {
-		loginRequest.ServerSSHAllowed = &serverSSHAllowed
-	}
-
-	if cmd.Flag(enableSSHRootFlag).Changed {
-		loginRequest.EnableSSHRoot = &enableSSHRoot
-	}
-
-	if cmd.Flag(enableSSHSFTPFlag).Changed {
-		loginRequest.EnableSSHSFTP = &enableSSHSFTP
-	}
-
-	if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
-		loginRequest.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
-	}
-
-	if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
-		loginRequest.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
-	}
-
-	if cmd.Flag(disableSSHAuthFlag).Changed {
-		loginRequest.DisableSSHAuth = &disableSSHAuth
-	}
-
-	if cmd.Flag(sshJWTCacheTTLFlag).Changed {
-		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
-		loginRequest.SshJWTCacheTTL = &sshJWTCacheTTL32
-	}
+	setSSHLoginFields(&loginRequest, cmd)
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &loginRequest.RemoteJobsAllowed, remoteJobsAllowed)
 
 	if cmd.Flag(disableAutoConnectFlag).Changed {
 		loginRequest.DisableAutoConnect = &autoConnectDisabled
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		loginRequest.EnableLocalMetrics = &localMetricsEnabled
+	}
+
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		loginRequest.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	if cmd.Flag(interfaceNameFlag).Changed {
 		if err := parseInterfaceName(interfaceName); err != nil {
 			return nil, err
@@ -661,9 +808,10 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		loginRequest.BlockInbound = &blockInbound
 	}
 
-	if cmd.Flag(enableLazyConnectionFlag).Changed {
-		loginRequest.LazyConnectionEnabled = &lazyConnEnabled
+	if cmd.Flag(disableIPv6Flag).Changed {
+		loginRequest.DisableIpv6 = &disableIPv6
 	}
+
 	return &loginRequest, nil
 }
 
@@ -782,3 +930,31 @@ func isValidAddrPort(input string) bool {
 	_, err := netip.ParseAddrPort(input)
 	return err == nil
 }
+
+// daemonActiveProfileForUser returns the profile the daemon currently holds for
+// username, for the no --profile case where the local mirror is not
+// authoritative (sudo or plain root). It returns that profile when the daemon
+// owns it for this user or when the profile is unowned (empty username, as on a
+// fresh install), so the caller acts on the daemon's real state instead of the
+// stale mirror. It denies with a --profile hint when the daemon is on another
+// user's profile, when the lookup fails, or when the daemon reports no active
+// profile. Returns errDaemonActiveProfileUnsupported when the daemon predates
+// the RPC; the caller keeps the mirror-derived profile in that case.
+func daemonActiveProfileForUser(ctx context.Context, client proto.DaemonServiceClient, username string) (*profilemanager.Profile, error) {
+	active, err := client.GetActiveProfile(ctx, &proto.GetActiveProfileRequest{})
+	if err != nil {
+		if st, ok := gstatus.FromError(err); ok && st.Code() == codes.Unimplemented {
+			return nil, fmt.Errorf("%w: %v", errDaemonActiveProfileUnsupported, err)
+		}
+		return nil, fmt.Errorf("pass --profile to choose the profile explicitly: the daemon's active profile could not be verified: %v", err)
+	}
+	if active.GetId() == "" {
+		return nil, fmt.Errorf("pass --profile to choose the profile explicitly: the daemon reported no active profile")
+	}
+	if active.GetUsername() != "" && active.GetUsername() != username {
+		return nil, fmt.Errorf(
+			"pass --profile to choose the profile explicitly: the daemon's active profile is %q (user %q) but this invocation runs for %q",
+			active.GetProfileName(), active.GetUsername(), username)
+	}
+	return &profilemanager.Profile{ID: profilemanager.ID(active.GetId())}, nil
+}

client/cmd/up_daemon_test.go

@@ -29,14 +29,14 @@ func TestUpDaemon(t *testing.T) {
 	}
 
 	sm := profilemanager.ServiceManager{}
-	err = sm.AddProfile("test1", currUser.Username)
+	created, err := sm.AddProfile("test1", currUser.Username)
 	if err != nil {
 		t.Fatalf("failed to add profile: %v", err)
 		return
 	}
 
 	err = sm.SetActiveProfileState(&profilemanager.ActiveProfileState{
-		Name:     "test1",
+		ID:       created.ID,
 		Username: currUser.Username,
 	})
 	if err != nil {

client/cmd/up_test.go

@@ -0,0 +1,88 @@
+package cmd
+
+import (
+	"context"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+	gstatus "google.golang.org/grpc/status"
+
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/proto"
+)
+
+type fakeActiveProfileClient struct {
+	proto.DaemonServiceClient
+	resp *proto.GetActiveProfileResponse
+	err  error
+}
+
+func (f *fakeActiveProfileClient) GetActiveProfile(_ context.Context, _ *proto.GetActiveProfileRequest, _ ...grpc.CallOption) (*proto.GetActiveProfileResponse, error) {
+	return f.resp, f.err
+}
+
+func TestDaemonActiveProfileForUserReturnsOwnProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "default", ProfileName: "default", Username: "root"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.NoError(t, err)
+	require.NotNil(t, prof)
+	assert.Equal(t, profilemanager.ID("default"), prof.ID)
+}
+
+func TestDaemonActiveProfileForUserReturnsUnownedProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "default", ProfileName: "default", Username: ""}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.NoError(t, err)
+	require.NotNil(t, prof)
+	assert.Equal(t, profilemanager.ID("default"), prof.ID)
+}
+
+func TestDaemonActiveProfileForUserKeepsDaemonProfileOverStaleMirror(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "ab12", ProfileName: "work", Username: "misha"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "misha")
+	require.NoError(t, err)
+	require.NotNil(t, prof)
+	assert.Equal(t, profilemanager.ID("ab12"), prof.ID)
+}
+
+func TestDaemonActiveProfileForUserRejectsOtherUsersProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "ab12", ProfileName: "work", Username: "misha"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserRejectsOtherUsersDefaultProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "default", ProfileName: "default", Username: "misha"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserRejectsLookupError(t *testing.T) {
+	client := &fakeActiveProfileClient{err: gstatus.Error(codes.Internal, "boom")}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserRejectsEmptyResponse(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserKeepsMirrorWhenDaemonWithoutRPC(t *testing.T) {
+	client := &fakeActiveProfileClient{err: gstatus.Error(codes.Unimplemented, "unknown method")}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.ErrorIs(t, err, errDaemonActiveProfileUnsupported)
+	assert.Nil(t, prof)
+}

client/cmd/version.go

@@ -12,7 +12,13 @@ var (
 		Short: "Print the NetBird's client application version",
 		Run: func(cmd *cobra.Command, args []string) {
 			cmd.SetOut(cmd.OutOrStdout())
-			cmd.Println(version.NetbirdVersion())
+			out := version.NetbirdVersion()
+			if version.IsDevelopmentVersion(out) {
+				if commit := version.NetbirdCommit(); commit != "" {
+					out += "-" + commit
+				}
+			}
+			cmd.Println(out)
 		},
 	}
 )

client/configs/configs.go

@@ -6,6 +6,11 @@ import (
 	"runtime"
 )
 
+// UILogFile is the file name the desktop UI writes its log to. It is defined
+// here so the UI (writer), the daemon's RegisterUILog validation, and the debug
+// bundle collector all share one definition.
+const UILogFile = "gui-client.log"
+
 var StateDir string
 
 func init() {

client/embed/capture.go

@@ -0,0 +1,65 @@
+package embed
+
+import (
+	"io"
+
+	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/util/capture"
+)
+
+// CaptureOptions configures a packet capture session.
+type CaptureOptions struct {
+	// Output receives pcap-formatted data. Nil disables pcap output.
+	Output io.Writer
+	// TextOutput receives human-readable packet summaries. Nil disables text output.
+	TextOutput io.Writer
+	// Filter is a BPF-like filter expression (e.g. "host 10.0.0.1 and tcp port 443").
+	// Empty captures all packets.
+	Filter string
+	// Verbose adds seq/ack, TTL, window, and total length to text output.
+	Verbose bool
+	// ASCII dumps transport payload as printable ASCII after each packet line.
+	ASCII bool
+}
+
+// CaptureStats reports capture session counters.
+type CaptureStats struct {
+	Packets int64
+	Bytes   int64
+	Dropped int64
+}
+
+// CaptureSession represents an active packet capture. Call Stop to end the
+// capture and flush buffered packets.
+type CaptureSession struct {
+	sess   *capture.Session
+	engine *internal.Engine
+}
+
+// Stop ends the capture, flushes remaining packets, and detaches from the device.
+// Safe to call multiple times.
+func (cs *CaptureSession) Stop() {
+	if cs.engine != nil {
+		_ = cs.engine.SetCapture(nil)
+		cs.engine = nil
+	}
+	if cs.sess != nil {
+		cs.sess.Stop()
+	}
+}
+
+// Stats returns current capture counters.
+func (cs *CaptureSession) Stats() CaptureStats {
+	s := cs.sess.Stats()
+	return CaptureStats{
+		Packets: s.Packets,
+		Bytes:   s.Bytes,
+		Dropped: s.Dropped,
+	}
+}
+
+// Done returns a channel that is closed when the capture's writer goroutine
+// has fully exited and all buffered packets have been flushed.
+func (cs *CaptureSession) Done() <-chan struct{} {
+	return cs.sess.Done()
+}

client/embed/embed.go

@@ -12,6 +12,7 @@ import (
 	"sync"
 
 	"github.com/sirupsen/logrus"
+	wgdevice "golang.zx2c4.com/wireguard/device"
 	wgnetstack "golang.zx2c4.com/wireguard/tun/netstack"
 
 	"github.com/netbirdio/netbird/client/iface"
@@ -20,10 +21,12 @@ import (
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
-	sshcommon "github.com/netbirdio/netbird/client/ssh"
+	"github.com/netbirdio/netbird/client/mdm"
+	nbssh "github.com/netbirdio/netbird/client/ssh"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/shared/management/domain"
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
+	"github.com/netbirdio/netbird/util/capture"
 )
 
 var (
@@ -65,7 +68,7 @@ type Options struct {
 	PrivateKey string
 	// ManagementURL overrides the default management server URL
 	ManagementURL string
-	// PreSharedKey is the pre-shared key for the WireGuard interface
+	// PreSharedKey is the pre-shared key for the tunnel interface
 	PreSharedKey string
 	// LogOutput is the output destination for logs (defaults to os.Stderr if nil)
 	LogOutput io.Writer
@@ -79,11 +82,31 @@ type Options struct {
 	StatePath string
 	// DisableClientRoutes disables the client routes
 	DisableClientRoutes bool
+	// DisableIPv6 disables IPv6 overlay addressing
+	DisableIPv6 bool
 	// BlockInbound blocks all inbound connections from peers
 	BlockInbound bool
-	// WireguardPort is the port for the WireGuard interface. Use 0 for a random port.
+	// EnableRosenpass enables the Rosenpass post-quantum key exchange.
+	EnableRosenpass bool
+	// RosenpassPermissive lets a Rosenpass-enabled peer still connect to peers
+	// that do not run Rosenpass (falling back to the plain WireGuard PSK).
+	RosenpassPermissive bool
+	// BlockLANAccess blocks the embedded peer from reaching the host's
+	// LAN (RFC 1918, link-local, loopback) when it's used as a routing
+	// peer. Mirrors profilemanager.ConfigInput.BlockLANAccess. Useful
+	// when the embedded client must never act as a stepping stone into
+	// the host's local network (e.g. the proxy's overlay peer).
+	BlockLANAccess bool
+	// LazyConnectionEnabled is a tri-state local override for lazy connections,
+	// mirroring the NB_LAZY_CONN env var. Nil defers to the management feature
+	// flag; a set value overrides it in both directions. A short-lived client
+	// that reaches only a few known peers can set this to false, so its peers
+	// connect eagerly and the first request does not wait for the connection to
+	// be established.
+	LazyConnectionEnabled *bool
+	// WireguardPort is the port for the tunnel interface. Use 0 for a random port.
 	WireguardPort *int
-	// MTU is the MTU for the WireGuard interface.
+	// MTU is the MTU for the tunnel interface.
 	// Valid values are in the range 576..8192 bytes.
 	// If non-nil, this value overrides any value stored in the config file.
 	// If nil, the existing config MTU (if non-zero) is preserved; otherwise it defaults to 1280.
@@ -91,6 +114,26 @@ type Options struct {
 	MTU *uint16
 	// DNSLabels defines additional DNS labels configured in the peer.
 	DNSLabels []string
+	// Performance configures the tunnel's buffer pool cap and batch size.
+	Performance Performance
+}
+
+// Performance configures the embedded client's tunnel memory/throughput knobs.
+//
+// These settings are process-global: any non-nil field also becomes the
+// default for Clients constructed by later embed.New calls in the same
+// process. Nil fields are ignored.
+type Performance struct {
+	// PreallocatedBuffersPerPool caps the per-tunnel buffer pool. Zero
+	// leaves the pool unbounded. Lower values trade throughput for a
+	// tighter memory ceiling. May also be changed on a running Client via
+	// Client.SetPerformance, provided this field was nonzero at construction.
+	PreallocatedBuffersPerPool *uint32
+	// MaxBatchSize overrides the number of packets the tunnel reads or
+	// writes per syscall, which also bounds eager buffer allocation per
+	// worker. Zero uses the platform default. Applied at construction
+	// only; ignored by Client.SetPerformance.
+	MaxBatchSize *uint32
 }
 
 // validateCredentials checks that exactly one credential type is provided
@@ -170,7 +213,11 @@ func New(opts Options) (*Client, error) {
 		PreSharedKey:        &opts.PreSharedKey,
 		DisableServerRoutes: &t,
 		DisableClientRoutes: &opts.DisableClientRoutes,
+		DisableIPv6:         &opts.DisableIPv6,
 		BlockInbound:        &opts.BlockInbound,
+		BlockLANAccess:      &opts.BlockLANAccess,
+		RosenpassEnabled:    &opts.EnableRosenpass,
+		RosenpassPermissive: &opts.RosenpassPermissive,
 		WireguardPort:       opts.WireguardPort,
 		MTU:                 opts.MTU,
 		DNSLabels:           parsedLabels,
@@ -183,11 +230,31 @@ func New(opts Options) (*Client, error) {
 	if err != nil {
 		return nil, fmt.Errorf("create config: %w", err)
 	}
+	// Embedded path runs without the daemon Server: apply the active
+	// MDM policy explicitly so a forced ManagementURL / PSK / other
+	// managed key takes effect on this embedded engine instance.
+	config.ApplyMDMPolicy(mdm.NewLoader(nil).Load())
 
 	if opts.PrivateKey != "" {
 		config.PrivateKey = opts.PrivateKey
 	}
 
+	if opts.LazyConnectionEnabled != nil {
+		// Runtime-only override, read back through lazyconn.ParseState; a set value
+		// wins over the management feature flag in both directions.
+		config.LazyConnection = "off"
+		if *opts.LazyConnectionEnabled {
+			config.LazyConnection = "on"
+		}
+	}
+
+	if opts.Performance.PreallocatedBuffersPerPool != nil {
+		wgdevice.SetPreallocatedBuffersPerPool(*opts.Performance.PreallocatedBuffersPerPool)
+	}
+	if opts.Performance.MaxBatchSize != nil {
+		wgdevice.SetMaxBatchSizeOverride(*opts.Performance.MaxBatchSize)
+	}
+
 	return &Client{
 		deviceName: opts.DeviceName,
 		setupKey:   opts.SetupKey,
@@ -240,6 +307,12 @@ func (c *Client) Start(startCtx context.Context) error {
 
 	select {
 	case <-startCtx.Done():
+		// ConnectClient.Stop now cancels its own run context and waits for the
+		// run loop to tear the engine down, so this cancel() is no longer
+		// required to break the deadlock and could be removed. It is kept as a
+		// defensive belt-and-suspenders: cancelling the parent context first
+		// guarantees the run loop is unblocked even if Stop's contract regresses.
+		cancel()
 		if stopErr := client.Stop(); stopErr != nil {
 			return fmt.Errorf("stop error after context done. Stop error: %w. Context done: %w", stopErr, startCtx.Err())
 		}
@@ -332,7 +405,7 @@ func (c *Client) ListenTCP(address string) (net.Listener, error) {
 	if err != nil {
 		return nil, fmt.Errorf("split host port: %w", err)
 	}
-	listenAddr := fmt.Sprintf("%s:%s", addr, port)
+	listenAddr := net.JoinHostPort(addr.String(), port)
 
 	tcpAddr, err := net.ResolveTCPAddr("tcp", listenAddr)
 	if err != nil {
@@ -353,7 +426,7 @@ func (c *Client) ListenUDP(address string) (net.PacketConn, error) {
 	if err != nil {
 		return nil, fmt.Errorf("split host port: %w", err)
 	}
-	listenAddr := fmt.Sprintf("%s:%s", addr, port)
+	listenAddr := net.JoinHostPort(addr.String(), port)
 
 	udpAddr, err := net.ResolveUDPAddr("udp", listenAddr)
 	if err != nil {
@@ -401,6 +474,21 @@ func (c *Client) Expose(ctx context.Context, req ExposeRequest) (*ExposeSession,
 	}, nil
 }
 
+// IdentityForIP looks up a remote peer by its tunnel IP using the
+// embedded client's status recorder. Returns the peer's WireGuard public
+// key and FQDN. ok=false means the IP doesn't belong to an active peer
+// — offline roster peers are treated as unknown, same as foreign IPs.
+func (c *Client) IdentityForIP(ip netip.Addr) (pubKey, fqdn string, ok bool) {
+	if !ip.IsValid() || c.recorder == nil {
+		return "", "", false
+	}
+	state, found := c.recorder.PeerStateByIP(ip.String())
+	if !found {
+		return "", "", false
+	}
+	return state.PubKey, state.FQDN, true
+}
+
 // Status returns the current status of the client.
 func (c *Client) Status() (peer.FullStatus, error) {
 	c.mu.Lock()
@@ -410,7 +498,7 @@ func (c *Client) Status() (peer.FullStatus, error) {
 	if connect != nil {
 		engine := connect.Engine()
 		if engine != nil {
-			_ = engine.RunHealthProbes(false)
+			_ = engine.RunHealthProbes(context.Background(), false)
 		}
 	}
 
@@ -461,12 +549,72 @@ func (c *Client) VerifySSHHostKey(peerAddress string, key []byte) error {
 		return err
 	}
 
-	storedKey, found := engine.GetPeerSSHKey(peerAddress)
-	if !found {
-		return sshcommon.ErrPeerNotFound
+	return nbssh.PeerKeyLookup(engine.GetPeerSSHKey).VerifySSHHostKey(peerAddress, key)
+}
+
+// SetPerformance retunes a running Client. Only PreallocatedBuffersPerPool
+// takes effect, and only when it was nonzero at construction;
+// MaxBatchSize is construction-only and returns an error if set here.
+//
+// Returns ErrClientNotStarted / ErrEngineNotStarted if the Client is not
+// running yet.
+func (c *Client) SetPerformance(t Performance) error {
+	if t.MaxBatchSize != nil {
+		return errors.New("MaxBatchSize is construction-only and cannot be changed at runtime")
+	}
+	engine, err := c.getEngine()
+	if err != nil {
+		return err
+	}
+	return engine.SetPerformance(internal.Performance{
+		PreallocatedBuffersPerPool: t.PreallocatedBuffersPerPool,
+	})
+}
+
+// StartCapture begins capturing packets on this client's tunnel device.
+// Only one capture can be active at a time; starting a new one stops the previous.
+// Call StopCapture (or CaptureSession.Stop) to end it.
+func (c *Client) StartCapture(opts CaptureOptions) (*CaptureSession, error) {
+	engine, err := c.getEngine()
+	if err != nil {
+		return nil, err
 	}
 
-	return sshcommon.VerifyHostKey(storedKey, key, peerAddress)
+	var matcher capture.Matcher
+	if opts.Filter != "" {
+		m, err := capture.ParseFilter(opts.Filter)
+		if err != nil {
+			return nil, fmt.Errorf("parse filter: %w", err)
+		}
+		matcher = m
+	}
+
+	sess, err := capture.NewSession(capture.Options{
+		Output:     opts.Output,
+		TextOutput: opts.TextOutput,
+		Matcher:    matcher,
+		Verbose:    opts.Verbose,
+		ASCII:      opts.ASCII,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("create capture session: %w", err)
+	}
+
+	if err := engine.SetCapture(sess); err != nil {
+		sess.Stop()
+		return nil, fmt.Errorf("set capture: %w", err)
+	}
+
+	return &CaptureSession{sess: sess, engine: engine}, nil
+}
+
+// StopCapture stops the active capture session if one is running.
+func (c *Client) StopCapture() error {
+	engine, err := c.getEngine()
+	if err != nil {
+		return err
+	}
+	return engine.SetCapture(nil)
 }
 
 // getEngine safely retrieves the engine from the client with proper locking.

client/embed/embed_test.go

@@ -0,0 +1,168 @@
+package embed
+
+import (
+	"context"
+	"net"
+	"testing"
+	"time"
+
+	"go.uber.org/mock/gomock"
+	"github.com/stretchr/testify/require"
+	"google.golang.org/grpc"
+
+	"github.com/netbirdio/netbird/management/internals/controllers/network_map/controller"
+	"github.com/netbirdio/netbird/management/internals/controllers/network_map/update_channel"
+	"github.com/netbirdio/netbird/management/internals/modules/peers"
+	"github.com/netbirdio/netbird/management/internals/modules/peers/ephemeral/manager"
+	"github.com/netbirdio/netbird/management/internals/server/config"
+	nbgrpc "github.com/netbirdio/netbird/management/internals/shared/grpc"
+	mgmt "github.com/netbirdio/netbird/management/server"
+	"github.com/netbirdio/netbird/management/server/activity"
+	nbcache "github.com/netbirdio/netbird/management/server/cache"
+	"github.com/netbirdio/netbird/management/server/groups"
+	"github.com/netbirdio/netbird/management/server/integrations/integrated_validator/validator"
+	"github.com/netbirdio/netbird/management/server/integrations/port_forwarding"
+	"github.com/netbirdio/netbird/management/server/job"
+	"github.com/netbirdio/netbird/management/internals/modules/permissions"
+	"github.com/netbirdio/netbird/management/server/settings"
+	"github.com/netbirdio/netbird/management/server/store"
+	"github.com/netbirdio/netbird/management/server/telemetry"
+	"github.com/netbirdio/netbird/management/server/types"
+	mgmtProto "github.com/netbirdio/netbird/shared/management/proto"
+	"github.com/netbirdio/netbird/util"
+)
+
+const testSetupKey = "A2C8E62B-38F5-4553-B31E-DD66C696CEBB"
+
+// TestClientStartTimeoutRollback reproduces a deadlock between Engine.Start and
+// Engine.Stop. The signal endpoint accepts gRPC connections but never serves the
+// SignalExchange service, so Engine.Start parks in WaitStreamConnected while
+// holding the engine mutex. When the Start context expires, the rollback path
+// calls ConnectClient.Stop, which must not block forever acquiring that mutex.
+func TestClientStartTimeoutRollback(t *testing.T) {
+	signalAddr := startBlackholeSignal(t)
+	mgmAddr := startManagement(t, signalAddr)
+
+	wgPort := 0
+	client, err := New(Options{
+		DeviceName:    "embed-rollback-test",
+		SetupKey:      testSetupKey,
+		ManagementURL: "http://" + mgmAddr,
+		WireguardPort: &wgPort,
+	})
+	require.NoError(t, err, "embed client creation must succeed")
+
+	startCtx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	startErr := make(chan error, 1)
+	go func() {
+		startErr <- client.Start(startCtx)
+	}()
+
+	select {
+	case err := <-startErr:
+		require.ErrorIs(t, err, context.DeadlineExceeded)
+	case <-time.After(60 * time.Second):
+		t.Fatal("client.Start did not return after its context expired: Engine.Stop deadlocked against Engine.Start waiting for the signal stream")
+	}
+}
+
+// startBlackholeSignal starts a gRPC server without the SignalExchange service
+// registered. Connections succeed, but the signal stream can never be
+// established, which keeps Engine.Start parked in WaitStreamConnected.
+func startBlackholeSignal(t *testing.T) string {
+	t.Helper()
+
+	lis, err := net.Listen("tcp", "localhost:0")
+	require.NoError(t, err)
+
+	s := grpc.NewServer()
+	go func() {
+		if err := s.Serve(lis); err != nil {
+			t.Error(err)
+		}
+	}()
+	t.Cleanup(s.Stop)
+
+	return lis.Addr().String()
+}
+
+func startManagement(t *testing.T, signalAddr string) string {
+	t.Helper()
+
+	cfg := &config.Config{
+		Stuns:      []*config.Host{},
+		TURNConfig: &config.TURNConfig{},
+		Relay: &config.Relay{
+			Addresses:      []string{"127.0.0.1:1234"},
+			CredentialsTTL: util.Duration{Duration: time.Hour},
+			Secret:         "222222222222222222",
+		},
+		Signal: &config.Host{
+			Proto: "http",
+			URI:   signalAddr,
+		},
+		Datadir:    t.TempDir(),
+		HttpConfig: nil,
+	}
+
+	lis, err := net.Listen("tcp", "localhost:0")
+	require.NoError(t, err)
+
+	s := grpc.NewServer()
+
+	testStore, cleanUp, err := store.NewTestStoreFromSQL(context.Background(), "../testdata/store.sql", cfg.Datadir)
+	require.NoError(t, err)
+	t.Cleanup(cleanUp)
+
+	eventStore := &activity.InMemoryEventStore{}
+
+	permissionsManager := permissions.NewManager(testStore)
+	peersManager := peers.NewManager(testStore)
+	jobManager := job.NewJobManager(nil, testStore, peersManager)
+
+	cacheStore, err := nbcache.NewStore(context.Background(), 100*time.Millisecond, 300*time.Millisecond, 100)
+	require.NoError(t, err)
+
+	iv, err := validator.NewIntegratedValidator(context.Background(), peersManager, nil, eventStore, cacheStore)
+	require.NoError(t, err)
+	metrics, err := telemetry.NewDefaultAppMetrics(context.Background())
+	require.NoError(t, err)
+
+	ctrl := gomock.NewController(t)
+	t.Cleanup(ctrl.Finish)
+	settingsMockManager := settings.NewMockManager(ctrl)
+	settingsMockManager.EXPECT().
+		GetSettings(gomock.Any(), gomock.Any(), gomock.Any()).
+		Return(&types.Settings{}, nil).
+		AnyTimes()
+	settingsMockManager.EXPECT().
+		GetExtraSettings(gomock.Any(), gomock.Any()).
+		Return(&types.ExtraSettings{}, nil).
+		AnyTimes()
+
+	groupsManager := groups.NewManagerMock()
+
+	updateManager := update_channel.NewPeersUpdateManager(metrics)
+	requestBuffer := mgmt.NewAccountRequestBuffer(context.Background(), testStore)
+	networkMapController := controller.NewController(context.Background(), testStore, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(testStore, peersManager), cfg, nil)
+	accountManager, err := mgmt.BuildManager(context.Background(), cfg, testStore, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManager, false, cacheStore)
+	require.NoError(t, err)
+
+	secretsManager, err := nbgrpc.NewTimeBasedAuthSecretsManager(updateManager, cfg.TURNConfig, cfg.Relay, settingsMockManager, groupsManager)
+	require.NoError(t, err)
+
+	mgmtServer, err := nbgrpc.NewServer(cfg, accountManager, settingsMockManager, jobManager, secretsManager, nil, nil, &mgmt.MockIntegratedValidator{}, networkMapController, nil, nil)
+	require.NoError(t, err)
+	mgmtProto.RegisterManagementServiceServer(s, mgmtServer)
+
+	go func() {
+		if err := s.Serve(lis); err != nil {
+			t.Error(err)
+		}
+	}()
+	t.Cleanup(s.Stop)
+
+	return lis.Addr().String()
+}

client/firewall/allower_other.go

@@ -0,0 +1,11 @@
+//go:build android || (!linux && !windows)
+
+package firewall
+
+import "github.com/netbirdio/netbird/client/firewall/uspfilter"
+
+// interfaceAllower returns no allower: these platforms have no host firewall to
+// open for the interface.
+func interfaceAllower(IFaceMapper, uint16) uspfilter.InterfaceAllower {
+	return nil
+}

client/firewall/allower_windows.go

@@ -0,0 +1,10 @@
+//go:build windows
+
+package firewall
+
+import "github.com/netbirdio/netbird/client/firewall/uspfilter"
+
+// interfaceAllower returns the Windows netsh-based interface allower.
+func interfaceAllower(iface IFaceMapper, _ uint16) uspfilter.InterfaceAllower {
+	return uspfilter.NewWindowsInterfaceAllower(iface)
+}

client/firewall/create.go

@@ -6,8 +6,6 @@ import (
 	"fmt"
 	"runtime"
 
-	log "github.com/sirupsen/logrus"
-
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 	"github.com/netbirdio/netbird/client/firewall/uspfilter"
 	nftypes "github.com/netbirdio/netbird/client/internal/netflow/types"
@@ -21,13 +19,11 @@ func NewFirewall(iface IFaceMapper, _ *statemanager.Manager, flowLogger nftypes.
 	}
 
 	// use userspace packet filtering firewall
-	fm, err := uspfilter.Create(iface, disableServerRoutes, flowLogger, mtu)
-	if err != nil {
-		return nil, err
-	}
-	err = fm.AllowNetbird()
-	if err != nil {
-		log.Warnf("failed to allow netbird interface traffic: %v", err)
-	}
-	return fm, nil
+	return uspfilter.Create(uspfilter.Config{
+		IFace:               iface,
+		DisableServerRoutes: disableServerRoutes,
+		FlowLogger:          flowLogger,
+		MTU:                 mtu,
+		InterfaceAllower:    interfaceAllower(iface, mtu),
+	})
 }

client/firewall/create_linux.go

@@ -16,6 +16,7 @@ import (
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 	nbnftables "github.com/netbirdio/netbird/client/firewall/nftables"
 	"github.com/netbirdio/netbird/client/firewall/uspfilter"
+	"github.com/netbirdio/netbird/client/iface/netstack"
 	nftypes "github.com/netbirdio/netbird/client/internal/netflow/types"
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 )
@@ -29,47 +30,107 @@ const (
 	NFTABLES
 )
 
-// SKIP_NFTABLES_ENV is the environment variable to skip nftables check
-const SKIP_NFTABLES_ENV = "NB_SKIP_NFTABLES_CHECK"
+// SkipNftablesEnv is the environment variable to skip nftables check
+const SkipNftablesEnv = "NB_SKIP_NFTABLES_CHECK"
+
+// errNoFirewallManager indicates no kernel firewall backend is present,
+// as opposed to a backend that exists but failed to create or initialize.
+var errNoFirewallManager = errors.New("no firewall manager found")
 
 // FWType is the type for the firewall type
 type FWType int
 
 func NewFirewall(iface IFaceMapper, stateManager *statemanager.Manager, flowLogger nftypes.FlowLogger, disableServerRoutes bool, mtu uint16) (firewall.Manager, error) {
-	// We run in userspace mode and force userspace firewall was requested. We don't attempt native firewall.
-	if iface.IsUserspaceBind() && forceUserspaceFirewall() {
-		log.Info("forcing userspace firewall")
-		return createUserspaceFirewall(iface, nil, disableServerRoutes, flowLogger, mtu)
+	// Userspace firewall without a native counterpart: routing is handled
+	// entirely in userspace. The interface is opened in the kernel's foreign
+	// filter chains via a table-less allower, except in netstack mode where no
+	// kernel interface exists.
+	if netstack.IsEnabled() || (iface.IsUserspaceBind() && forceUserspaceFirewall()) {
+		if netstack.IsEnabled() {
+			log.Info("netstack mode, using userspace firewall")
+		} else {
+			log.Info("forcing userspace firewall")
+		}
+		cfg := uspfilter.Config{
+			IFace:               iface,
+			DisableServerRoutes: disableServerRoutes,
+			FlowLogger:          flowLogger,
+			MTU:                 mtu,
+			InterfaceAllower:    interfaceAllower(iface, mtu),
+		}
+
+		return uspfilter.Create(cfg)
 	}
 
 	// Use native firewall for either kernel or userspace, the interface appears identical to netfilter
-	fm, err := createNativeFirewall(iface, stateManager, disableServerRoutes, mtu)
-
-	// Kernel cannot fall back to anything else, need to return error
-	if !iface.IsUserspaceBind() {
-		return fm, err
-	}
-
-	// Fall back to the userspace packet filter if native is unavailable
-	if err != nil {
-		log.Warnf("failed to create native firewall: %v. Proceeding with userspace", err)
-		return createUserspaceFirewall(iface, nil, disableServerRoutes, flowLogger, mtu)
-	}
-
-	// Native firewall handles packet filtering, but the userspace WireGuard bind
-	// needs a device filter for DNS interception hooks. Install a minimal
-	// hooks-only filter that passes all traffic through to the kernel firewall.
-	if err := iface.SetFilter(&uspfilter.HooksFilter{}); err != nil {
-		log.Warnf("failed to set hooks filter, DNS via memory hooks will not work: %v", err)
+	fm, err := createNativeFirewall(iface, stateManager, mtu)
+	switch {
+	case err == nil && !iface.IsUserspaceBind():
+		// Nothing to do, fall through
+	case err == nil && iface.IsUserspaceBind():
+		// Native firewall handles packet filtering, but the userspace WireGuard bind
+		// needs a device filter for DNS interception hooks. Install a minimal
+		// hooks-only filter that passes all traffic through to the kernel firewall.
+		if err := iface.SetFilter(&uspfilter.HooksFilter{}); err != nil {
+			log.Warnf("failed to set hooks filter, DNS via memory hooks will not work: %v", err)
+		}
+	case err != nil && !iface.IsUserspaceBind():
+		// Kernel cannot fall back to anything else, need to return error
+		return nil, err
+	case err != nil && iface.IsUserspaceBind():
+		// Fall back to the userspace packet filter if native is unavailable
+		logNativeFirewallUnavailable(err)
+		return uspfilter.Create(uspfilter.Config{
+			IFace:               iface,
+			DisableServerRoutes: disableServerRoutes,
+			FlowLogger:          flowLogger,
+			MTU:                 mtu,
+			InterfaceAllower:    interfaceAllower(iface, mtu),
+		})
 	}
 
 	return fm, nil
 }
 
-func createNativeFirewall(iface IFaceMapper, stateManager *statemanager.Manager, routes bool, mtu uint16) (firewall.Manager, error) {
+// interfaceAllower selects how the userspace firewall opens the interface in
+// foreign kernel chains: nftables when available (which also opens foreign nft
+// tables), else iptables (the legacy fallback, filter INPUT only), else nil.
+// firewalld trust is applied separately by the manager. Netstack has no kernel
+// interface to open.
+func interfaceAllower(iface IFaceMapper, mtu uint16) uspfilter.InterfaceAllower {
+	if netstack.IsEnabled() {
+		return nil
+	}
+
+	nftAllower, err := nbnftables.NewInterfaceAllower(iface, mtu)
+	if err == nil {
+		return nftAllower
+	}
+	log.Infof("no nftables interface allower: %v", err)
+
+	iptAllower, err := nbiptables.NewInterfaceAllower(iface)
+	if err == nil {
+		return iptAllower
+	}
+	log.Infof("no iptables interface allower: %v", err)
+
+	return nil
+}
+
+// logNativeFirewallUnavailable logs the fallback to userspace at info level
+// when no kernel firewall backend exists, and at warn level otherwise.
+func logNativeFirewallUnavailable(err error) {
+	if errors.Is(err, errNoFirewallManager) {
+		log.Infof("no native firewall backend available: %v. Proceeding with userspace", err)
+	} else {
+		log.Warnf("failed to create native firewall: %v. Proceeding with userspace", err)
+	}
+}
+
+func createNativeFirewall(iface IFaceMapper, stateManager *statemanager.Manager, mtu uint16) (firewall.Manager, error) {
 	fm, err := createFW(iface, mtu)
 	if err != nil {
-		return nil, fmt.Errorf("create firewall: %s", err)
+		return nil, fmt.Errorf("create firewall: %w", err)
 	}
 
 	if err = fm.Init(stateManager); err != nil {
@@ -88,29 +149,10 @@ func createFW(iface IFaceMapper, mtu uint16) (firewall.Manager, error) {
 		log.Info("creating an nftables firewall manager")
 		return nbnftables.Create(iface, mtu)
 	default:
-		log.Info("no firewall manager found, trying to use userspace packet filtering firewall")
-		return nil, errors.New("no firewall manager found")
+		return nil, errNoFirewallManager
 	}
 }
 
-func createUserspaceFirewall(iface IFaceMapper, fm firewall.Manager, disableServerRoutes bool, flowLogger nftypes.FlowLogger, mtu uint16) (firewall.Manager, error) {
-	var errUsp error
-	if fm != nil {
-		fm, errUsp = uspfilter.CreateWithNativeFirewall(iface, fm, disableServerRoutes, flowLogger, mtu)
-	} else {
-		fm, errUsp = uspfilter.Create(iface, disableServerRoutes, flowLogger, mtu)
-	}
-
-	if errUsp != nil {
-		return nil, fmt.Errorf("create userspace firewall: %s", errUsp)
-	}
-
-	if err := fm.AllowNetbird(); err != nil {
-		log.Errorf("failed to allow netbird interface traffic: %v", err)
-	}
-	return fm, nil
-}
-
 // check returns the firewall type based on common lib checks. It returns UNKNOWN if no firewall is found.
 func check() FWType {
 	useIPTABLES := false
@@ -132,35 +174,38 @@ func check() FWType {
 		}
 	}
 
-	nf := nftables.Conn{}
-	if chains, err := nf.ListChains(); err == nil && os.Getenv(SKIP_NFTABLES_ENV) != "true" {
-		if !useIPTABLES {
-			return NFTABLES
-		}
-
-		// search for chains where table is filter
-		// if we find one, we assume that nftables manager can be used with iptables
-		for _, chain := range chains {
-			if chain.Table.Name == "filter" {
+	// Honor the skip env before probing nftables at all.
+	if os.Getenv(SkipNftablesEnv) != "true" {
+		nf := nftables.Conn{}
+		if chains, err := nf.ListChains(); err == nil {
+			if !useIPTABLES {
 				return NFTABLES
 			}
-		}
 
-		// check tables for the following constraints:
-		// 1. there is no chain in nftables for the filter table and there is at least one chain in iptables, we assume that nftables manager can not be used
-		// 2. there is no tables or more than one table, we assume that nftables manager can be used
-		// 3. there is only one table and its name is filter, we assume that nftables manager can not be used, since there was no chain in it
-		// 4. if we find an error we log and continue with iptables check
-		nbTablesList, err := nf.ListTables()
-		switch {
-		case err == nil && len(iptablesChains) > 0:
-			return IPTABLES
-		case err == nil && len(nbTablesList) != 1:
-			return NFTABLES
-		case err == nil && len(nbTablesList) == 1 && nbTablesList[0].Name == "filter":
-			return IPTABLES
-		case err != nil:
-			log.Errorf("failed to list nftables tables on fw manager discovery: %s", err)
+			// search for chains where table is filter
+			// if we find one, we assume that nftables manager can be used with iptables
+			for _, chain := range chains {
+				if chain.Table.Name == "filter" {
+					return NFTABLES
+				}
+			}
+
+			// check tables for the following constraints:
+			// 1. there is no chain in nftables for the filter table and there is at least one chain in iptables, we assume that nftables manager can not be used
+			// 2. there is no tables or more than one table, we assume that nftables manager can be used
+			// 3. there is only one table and its name is filter, we assume that nftables manager can not be used, since there was no chain in it
+			// 4. if we find an error we log and continue with iptables check
+			nbTablesList, err := nf.ListTables()
+			switch {
+			case err == nil && len(iptablesChains) > 0:
+				return IPTABLES
+			case err == nil && len(nbTablesList) != 1:
+				return NFTABLES
+			case err == nil && len(nbTablesList) == 1 && nbTablesList[0].Name == "filter":
+				return IPTABLES
+			case err != nil:
+				log.Errorf("failed to list nftables tables on fw manager discovery: %s", err)
+			}
 		}
 	}
 
@@ -176,15 +221,21 @@ func isIptablesClientAvailable(client *iptables.IPTables) bool {
 	return err == nil
 }
 
+// forceUserspaceFirewall reports whether the userspace firewall is forced.
+// NB_FORCE_USERSPACE_ROUTER is an alias: forcing userspace routing implies the
+// userspace firewall, since the two are no longer separable.
 func forceUserspaceFirewall() bool {
-	val := os.Getenv(EnvForceUserspaceFirewall)
+	return envForceBool(EnvForceUserspaceFirewall) || envForceBool(uspfilter.EnvForceUserspaceRouter)
+}
+
+func envForceBool(name string) bool {
+	val := os.Getenv(name)
 	if val == "" {
 		return false
 	}
-
 	force, err := strconv.ParseBool(val)
 	if err != nil {
-		log.Warnf("failed to parse %s: %v", EnvForceUserspaceFirewall, err)
+		log.Warnf("failed to parse %s: %v", name, err)
 		return false
 	}
 	return force

client/firewall/firewalld/firewalld.go

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

client/firewall/iptables/acl_linux.go

@@ -1,529 +0,0 @@
-package iptables
-
-import (
-	"errors"
-	"fmt"
-	"net"
-	"slices"
-
-	"github.com/coreos/go-iptables/iptables"
-	"github.com/google/uuid"
-	ipset "github.com/lrh3321/ipset-go"
-	log "github.com/sirupsen/logrus"
-
-	firewall "github.com/netbirdio/netbird/client/firewall/manager"
-	"github.com/netbirdio/netbird/client/internal/statemanager"
-	nbnet "github.com/netbirdio/netbird/client/net"
-)
-
-const (
-	tableName = "filter"
-
-	// rules chains contains the effective ACL rules
-	chainNameInputRules = "NETBIRD-ACL-INPUT"
-
-	// mangleFwdKey is the entries map key for mangle FORWARD guard rules that prevent
-	// external DNAT from bypassing ACL rules.
-	mangleFwdKey = "MANGLE-FORWARD"
-)
-
-type aclEntries map[string][][]string
-
-type entry struct {
-	spec     []string
-	position int
-}
-
-type aclManager struct {
-	iptablesClient  *iptables.IPTables
-	wgIface         iFaceMapper
-	entries         aclEntries
-	optionalEntries map[string][]entry
-	ipsetStore      *ipsetStore
-
-	stateManager *statemanager.Manager
-}
-
-func newAclManager(iptablesClient *iptables.IPTables, wgIface iFaceMapper) (*aclManager, error) {
-	return &aclManager{
-		iptablesClient:  iptablesClient,
-		wgIface:         wgIface,
-		entries:         make(map[string][][]string),
-		optionalEntries: make(map[string][]entry),
-		ipsetStore:      newIpsetStore(),
-	}, nil
-}
-
-func (m *aclManager) init(stateManager *statemanager.Manager) error {
-	m.stateManager = stateManager
-
-	m.seedInitialEntries()
-	m.seedInitialOptionalEntries()
-
-	if err := m.cleanChains(); err != nil {
-		return fmt.Errorf("clean chains: %w", err)
-	}
-
-	if err := m.createDefaultChains(); err != nil {
-		return fmt.Errorf("create default chains: %w", err)
-	}
-
-	m.updateState()
-
-	return nil
-}
-
-func (m *aclManager) AddPeerFiltering(
-	id []byte,
-	ip net.IP,
-	protocol firewall.Protocol,
-	sPort *firewall.Port,
-	dPort *firewall.Port,
-	action firewall.Action,
-	ipsetName string,
-) ([]firewall.Rule, error) {
-	chain := chainNameInputRules
-
-	ipsetName = transformIPsetName(ipsetName, sPort, dPort, action)
-	specs := filterRuleSpecs(ip, string(protocol), sPort, dPort, action, ipsetName)
-
-	mangleSpecs := slices.Clone(specs)
-	mangleSpecs = append(mangleSpecs,
-		"-i", m.wgIface.Name(),
-		"-m", "addrtype", "--dst-type", "LOCAL",
-		"-j", "MARK", "--set-xmark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkRedirected),
-	)
-
-	specs = append(specs, "-j", actionToStr(action))
-	if ipsetName != "" {
-		if ipList, ipsetExists := m.ipsetStore.ipset(ipsetName); ipsetExists {
-			if err := m.addToIPSet(ipsetName, ip); err != nil {
-				return nil, fmt.Errorf("add IP to ipset: %w", err)
-			}
-			// if ruleset already exists it means we already have the firewall rule
-			// so we need to update IPs in the ruleset and return new fw.Rule object for ACL manager.
-			ipList.addIP(ip.String())
-			return []firewall.Rule{&Rule{
-				ruleID:    uuid.New().String(),
-				ipsetName: ipsetName,
-				ip:        ip.String(),
-				chain:     chain,
-				specs:     specs,
-			}}, nil
-		}
-
-		if err := m.flushIPSet(ipsetName); err != nil {
-			if errors.Is(err, ipset.ErrSetNotExist) {
-				log.Debugf("flush ipset %s before use: %v", ipsetName, err)
-			} else {
-				log.Errorf("flush ipset %s before use: %v", ipsetName, err)
-			}
-		}
-		if err := m.createIPSet(ipsetName); err != nil {
-			return nil, fmt.Errorf("create ipset: %w", err)
-		}
-		if err := m.addToIPSet(ipsetName, ip); err != nil {
-			return nil, fmt.Errorf("add IP to ipset: %w", err)
-		}
-
-		ipList := newIpList(ip.String())
-		m.ipsetStore.addIpList(ipsetName, ipList)
-	}
-
-	ok, err := m.iptablesClient.Exists(tableFilter, chain, specs...)
-	if err != nil {
-		return nil, fmt.Errorf("failed to check rule: %w", err)
-	}
-	if ok {
-		return nil, fmt.Errorf("rule already exists")
-	}
-
-	// Insert DROP rules at the beginning, append ACCEPT rules at the end
-	if action == firewall.ActionDrop {
-		// Insert at the beginning of the chain (position 1)
-		err = m.iptablesClient.Insert(tableFilter, chain, 1, specs...)
-	} else {
-		err = m.iptablesClient.Append(tableFilter, chain, specs...)
-	}
-	if err != nil {
-		return nil, err
-	}
-
-	if err := m.iptablesClient.Append(tableMangle, chainRTPRE, mangleSpecs...); err != nil {
-		log.Errorf("failed to add mangle rule: %v", err)
-		mangleSpecs = nil
-	}
-
-	rule := &Rule{
-		ruleID:      uuid.New().String(),
-		specs:       specs,
-		mangleSpecs: mangleSpecs,
-		ipsetName:   ipsetName,
-		ip:          ip.String(),
-		chain:       chain,
-	}
-
-	m.updateState()
-
-	return []firewall.Rule{rule}, nil
-}
-
-// DeletePeerRule from the firewall by rule definition
-func (m *aclManager) DeletePeerRule(rule firewall.Rule) error {
-	r, ok := rule.(*Rule)
-	if !ok {
-		return fmt.Errorf("invalid rule type")
-	}
-
-	shouldDestroyIpset := false
-	if ipsetList, ok := m.ipsetStore.ipset(r.ipsetName); ok {
-		// delete IP from ruleset IPs list and ipset
-		if _, ok := ipsetList.ips[r.ip]; ok {
-			ip := net.ParseIP(r.ip)
-			if ip == nil {
-				return fmt.Errorf("parse IP %s", r.ip)
-			}
-			if err := m.delFromIPSet(r.ipsetName, ip); err != nil {
-				return fmt.Errorf("delete ip from ipset: %w", err)
-			}
-			delete(ipsetList.ips, r.ip)
-		}
-
-		// if after delete, set still contains other IPs,
-		// no need to delete firewall rule and we should exit here
-		if len(ipsetList.ips) != 0 {
-			return nil
-		}
-
-		// we delete last IP from the set, that means we need to delete
-		// set itself and associated firewall rule too
-		m.ipsetStore.deleteIpset(r.ipsetName)
-		shouldDestroyIpset = true
-	}
-
-	if err := m.iptablesClient.Delete(tableName, r.chain, r.specs...); err != nil {
-		return fmt.Errorf("failed to delete rule: %s, %v: %w", r.chain, r.specs, err)
-	}
-
-	if r.mangleSpecs != nil {
-		if err := m.iptablesClient.Delete(tableMangle, chainRTPRE, r.mangleSpecs...); err != nil {
-			log.Errorf("failed to delete mangle rule: %v", err)
-		}
-	}
-
-	if shouldDestroyIpset {
-		if err := m.destroyIPSet(r.ipsetName); err != nil {
-			if errors.Is(err, ipset.ErrBusy) || errors.Is(err, ipset.ErrSetNotExist) {
-				log.Debugf("destroy empty ipset: %v", err)
-			} else {
-				log.Errorf("destroy empty ipset: %v", err)
-			}
-		}
-	}
-
-	m.updateState()
-
-	return nil
-}
-
-func (m *aclManager) Reset() error {
-	if err := m.cleanChains(); err != nil {
-		return fmt.Errorf("clean chains: %w", err)
-	}
-
-	m.updateState()
-
-	return nil
-}
-
-// todo write less destructive cleanup mechanism
-func (m *aclManager) cleanChains() error {
-	ok, err := m.iptablesClient.ChainExists(tableName, chainNameInputRules)
-	if err != nil {
-		log.Debugf("failed to list chains: %s", err)
-		return err
-	}
-	if ok {
-		for _, rule := range m.entries["INPUT"] {
-			err := m.iptablesClient.DeleteIfExists(tableName, "INPUT", rule...)
-			if err != nil {
-				log.Errorf("failed to delete rule: %v, %s", rule, err)
-			}
-		}
-
-		for _, rule := range m.entries["FORWARD"] {
-			err := m.iptablesClient.DeleteIfExists(tableName, "FORWARD", rule...)
-			if err != nil {
-				log.Errorf("failed to delete rule: %v, %s", rule, err)
-			}
-		}
-
-		err = m.iptablesClient.ClearAndDeleteChain(tableName, chainNameInputRules)
-		if err != nil {
-			log.Debugf("failed to clear and delete %s chain: %s", chainNameInputRules, err)
-			return err
-		}
-	}
-
-	ok, err = m.iptablesClient.ChainExists("mangle", "PREROUTING")
-	if err != nil {
-		return fmt.Errorf("list chains: %w", err)
-	}
-	if ok {
-		for _, rule := range m.entries["PREROUTING"] {
-			err := m.iptablesClient.DeleteIfExists("mangle", "PREROUTING", rule...)
-			if err != nil {
-				log.Errorf("failed to delete rule: %v, %s", rule, err)
-			}
-		}
-	}
-
-	for _, rule := range m.entries[mangleFwdKey] {
-		if err := m.iptablesClient.DeleteIfExists(tableMangle, chainFORWARD, rule...); err != nil {
-			log.Errorf("failed to delete mangle FORWARD guard rule: %v, %s", rule, err)
-		}
-	}
-
-	for _, ipsetName := range m.ipsetStore.ipsetNames() {
-		if err := m.flushIPSet(ipsetName); err != nil {
-			if errors.Is(err, ipset.ErrSetNotExist) {
-				log.Debugf("flush ipset %q during reset: %v", ipsetName, err)
-			} else {
-				log.Errorf("flush ipset %q during reset: %v", ipsetName, err)
-			}
-		}
-		if err := m.destroyIPSet(ipsetName); err != nil {
-			if errors.Is(err, ipset.ErrBusy) || errors.Is(err, ipset.ErrSetNotExist) {
-				log.Debugf("destroy ipset %q during reset: %v", ipsetName, err)
-			} else {
-				log.Errorf("destroy ipset %q during reset: %v", ipsetName, err)
-			}
-		}
-		m.ipsetStore.deleteIpset(ipsetName)
-	}
-
-	return nil
-}
-
-func (m *aclManager) createDefaultChains() error {
-	// chain netbird-acl-input-rules
-	if err := m.iptablesClient.NewChain(tableName, chainNameInputRules); err != nil {
-		log.Debugf("failed to create '%s' chain: %s", chainNameInputRules, err)
-		return err
-	}
-
-	for chainName, rules := range m.entries {
-		// mangle FORWARD guard rules are handled separately below
-		if chainName == mangleFwdKey {
-			continue
-		}
-		for _, rule := range rules {
-			if err := m.iptablesClient.InsertUnique(tableName, chainName, 1, rule...); err != nil {
-				log.Debugf("failed to create input chain jump rule: %s", err)
-				return err
-			}
-		}
-	}
-
-	for chainName, entries := range m.optionalEntries {
-		for _, entry := range entries {
-			if err := m.iptablesClient.InsertUnique(tableName, chainName, entry.position, entry.spec...); err != nil {
-				log.Errorf("failed to insert optional entry %v: %v", entry.spec, err)
-				continue
-			}
-			m.entries[chainName] = append(m.entries[chainName], entry.spec)
-		}
-	}
-	clear(m.optionalEntries)
-
-	// Insert mangle FORWARD guard rules to prevent external DNAT bypass.
-	for _, rule := range m.entries[mangleFwdKey] {
-		if err := m.iptablesClient.AppendUnique(tableMangle, chainFORWARD, rule...); err != nil {
-			log.Errorf("failed to add mangle FORWARD guard rule: %v", err)
-		}
-	}
-
-	return nil
-}
-
-// seedInitialEntries adds default rules to the entries map, rules are inserted on pos 1, hence the order is reversed.
-// We want to make sure our traffic is not dropped by existing rules.
-
-// The existing FORWARD rules/policies decide outbound traffic towards our interface.
-// In case the FORWARD policy is set to "drop", we add an established/related rule to allow return traffic for the inbound rule.
-func (m *aclManager) seedInitialEntries() {
-	established := getConntrackEstablished()
-
-	m.appendToEntries("INPUT", []string{"-i", m.wgIface.Name(), "-j", "DROP"})
-	m.appendToEntries("INPUT", []string{"-i", m.wgIface.Name(), "-j", chainNameInputRules})
-	m.appendToEntries("INPUT", append([]string{"-i", m.wgIface.Name()}, established...))
-
-	// Inbound is handled by our ACLs, the rest is dropped.
-	// For outbound we respect the FORWARD policy. However, we need to allow established/related traffic for inbound rules.
-	m.appendToEntries("FORWARD", []string{"-i", m.wgIface.Name(), "-j", "DROP"})
-
-	m.appendToEntries("FORWARD", []string{"-o", m.wgIface.Name(), "-j", chainRTFWDOUT})
-	m.appendToEntries("FORWARD", []string{"-i", m.wgIface.Name(), "-j", chainRTFWDIN})
-
-	// Mangle FORWARD guard: when external DNAT redirects traffic from the wg interface, it
-	// traverses FORWARD instead of INPUT, bypassing ACL rules. ACCEPT rules in filter FORWARD
-	// can be inserted above ours. Mangle runs before filter, so these guard rules enforce the
-	// ACL mark check where it cannot be overridden.
-	m.appendToEntries(mangleFwdKey, []string{
-		"-i", m.wgIface.Name(),
-		"-m", "conntrack", "--ctstate", "RELATED,ESTABLISHED",
-		"-j", "ACCEPT",
-	})
-	m.appendToEntries(mangleFwdKey, []string{
-		"-i", m.wgIface.Name(),
-		"-m", "conntrack", "--ctstate", "DNAT",
-		"-m", "mark", "!", "--mark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkRedirected),
-		"-j", "DROP",
-	})
-}
-
-func (m *aclManager) seedInitialOptionalEntries() {
-	m.optionalEntries["FORWARD"] = []entry{
-		{
-			spec:     []string{"-m", "mark", "--mark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkRedirected), "-j", "ACCEPT"},
-			position: 2,
-		},
-	}
-}
-
-func (m *aclManager) appendToEntries(chainName string, spec []string) {
-	m.entries[chainName] = append(m.entries[chainName], spec)
-}
-
-func (m *aclManager) updateState() {
-	if m.stateManager == nil {
-		return
-	}
-
-	var currentState *ShutdownState
-	if existing := m.stateManager.GetState(currentState); existing != nil {
-		if existingState, ok := existing.(*ShutdownState); ok {
-			currentState = existingState
-		}
-	}
-	if currentState == nil {
-		currentState = &ShutdownState{}
-	}
-
-	currentState.Lock()
-	defer currentState.Unlock()
-
-	currentState.ACLEntries = m.entries
-	currentState.ACLIPsetStore = m.ipsetStore
-
-	if err := m.stateManager.UpdateState(currentState); err != nil {
-		log.Errorf("failed to update state: %v", err)
-	}
-}
-
-// filterRuleSpecs returns the specs of a filtering rule
-func filterRuleSpecs(ip net.IP, protocol string, sPort, dPort *firewall.Port, action firewall.Action, ipsetName string) (specs []string) {
-	// don't use IP matching if IP is 0.0.0.0
-	matchByIP := !ip.IsUnspecified()
-
-	if matchByIP {
-		if ipsetName != "" {
-			specs = append(specs, "-m", "set", "--set", ipsetName, "src")
-		} else {
-			specs = append(specs, "-s", ip.String())
-		}
-	}
-	if protocol != "all" {
-		specs = append(specs, "-p", protocol)
-	}
-	specs = append(specs, applyPort("--sport", sPort)...)
-	specs = append(specs, applyPort("--dport", dPort)...)
-	return specs
-}
-
-func actionToStr(action firewall.Action) string {
-	if action == firewall.ActionAccept {
-		return "ACCEPT"
-	}
-	return "DROP"
-}
-
-func transformIPsetName(ipsetName string, sPort, dPort *firewall.Port, action firewall.Action) string {
-	if ipsetName == "" {
-		return ""
-	}
-
-	actionSuffix := ""
-	if action == firewall.ActionDrop {
-		actionSuffix = "-drop"
-	}
-
-	switch {
-	case sPort != nil && dPort != nil:
-		return ipsetName + "-sport-dport" + actionSuffix
-	case sPort != nil:
-		return ipsetName + "-sport" + actionSuffix
-	case dPort != nil:
-		return ipsetName + "-dport" + actionSuffix
-	default:
-		return ipsetName + actionSuffix
-	}
-}
-
-func (m *aclManager) createIPSet(name string) error {
-	opts := ipset.CreateOptions{
-		Replace: true,
-	}
-
-	if err := ipset.Create(name, ipset.TypeHashNet, opts); err != nil {
-		return fmt.Errorf("create ipset %s: %w", name, err)
-	}
-
-	log.Debugf("created ipset %s with type hash:net", name)
-	return nil
-}
-
-func (m *aclManager) addToIPSet(name string, ip net.IP) error {
-	cidr := uint8(32)
-	if ip.To4() == nil {
-		cidr = 128
-	}
-
-	entry := &ipset.Entry{
-		IP:      ip,
-		CIDR:    cidr,
-		Replace: true,
-	}
-
-	if err := ipset.Add(name, entry); err != nil {
-		return fmt.Errorf("add IP to ipset %s: %w", name, err)
-	}
-
-	return nil
-}
-
-func (m *aclManager) delFromIPSet(name string, ip net.IP) error {
-	cidr := uint8(32)
-	if ip.To4() == nil {
-		cidr = 128
-	}
-
-	entry := &ipset.Entry{
-		IP:   ip,
-		CIDR: cidr,
-	}
-
-	if err := ipset.Del(name, entry); err != nil {
-		return fmt.Errorf("delete IP from ipset %s: %w", name, err)
-	}
-
-	return nil
-}
-
-func (m *aclManager) flushIPSet(name string) error {
-	return ipset.Flush(name)
-}
-
-func (m *aclManager) destroyIPSet(name string) error {
-	return ipset.Destroy(name)
-}

client/firewall/iptables/chains_linux.go

@@ -0,0 +1,346 @@
+//go:build !android
+
+package iptables
+
+import (
+	"fmt"
+
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+	nbnet "github.com/netbirdio/netbird/client/net"
+)
+
+func (r *family) createContainers() error {
+	for _, chainInfo := range []struct {
+		chain string
+		table string
+	}{
+		{chainRTFwdIn, tableFilter},
+		{chainRTFwdOut, tableFilter},
+		{chainRTPre, tableMangle},
+		{chainRTNAT, tableNat},
+		{chainRTRdr, tableNat},
+		{chainRTMSSClamp, tableMangle},
+	} {
+		// Fallback: clear chains that survived an unclean shutdown.
+		if ok, _ := r.iptablesClient.ChainExists(chainInfo.table, chainInfo.chain); ok {
+			if err := r.iptablesClient.ClearAndDeleteChain(chainInfo.table, chainInfo.chain); err != nil {
+				log.Warnf("clear stale chain %s in %s: %v", chainInfo.chain, chainInfo.table, err)
+			}
+		}
+		if err := r.iptablesClient.NewChain(chainInfo.table, chainInfo.chain); err != nil {
+			return fmt.Errorf("create chain %s in table %s: %w", chainInfo.chain, chainInfo.table, err)
+		}
+	}
+
+	if err := r.insertEstablishedRule(chainRTFwdIn); err != nil {
+		return fmt.Errorf("insert established rule: %w", err)
+	}
+
+	if err := r.insertEstablishedRule(chainRTFwdOut); err != nil {
+		return fmt.Errorf("insert established rule: %w", err)
+	}
+
+	if err := r.addPostroutingRules(); err != nil {
+		return fmt.Errorf("add static nat rules: %w", err)
+	}
+
+	if err := r.addJumpRules(); err != nil {
+		return fmt.Errorf("add jump rules: %w", err)
+	}
+
+	if err := r.addMSSClampingRules(); err != nil {
+		log.Errorf("failed to add MSS clamping rules: %s", err)
+	}
+
+	return nil
+}
+
+func (r *family) addJumpRules() error {
+	// Jump to nat chain
+	natRule := jumpRuleSpec(chainRTNAT)
+	if err := r.iptablesClient.Insert(tableNat, chainPostrouting, 1, natRule...); err != nil {
+		return fmt.Errorf("add nat postrouting jump rule: %w", err)
+	}
+	r.rules[jumpNATPost] = natRule
+
+	// Jump to mangle prerouting chain
+	preRule := jumpRuleSpec(chainRTPre)
+	if err := r.iptablesClient.Insert(tableMangle, chainPrerouting, 1, preRule...); err != nil {
+		return fmt.Errorf("add mangle prerouting jump rule: %w", err)
+	}
+	r.rules[jumpManglePre] = preRule
+
+	// Jump to nat prerouting chain
+	rdrRule := jumpRuleSpec(chainRTRdr)
+	if err := r.iptablesClient.Insert(tableNat, chainPrerouting, 1, rdrRule...); err != nil {
+		return fmt.Errorf("add nat prerouting jump rule: %w", err)
+	}
+	r.rules[jumpNATPre] = rdrRule
+
+	return nil
+}
+
+func (r *family) setupDataPlaneMark() error {
+	var merr *multierror.Error
+	preRule := []string{
+		"-i", r.wgIface.Name(),
+		"-m", "conntrack", "--ctstate", "NEW",
+		"-j", "CONNMARK", "--set-mark", fmt.Sprintf("%#x", nbnet.DataPlaneMarkIn),
+	}
+
+	if err := r.iptablesClient.AppendUnique(tableMangle, chainPrerouting, preRule...); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("add mangle prerouting rule: %w", err))
+	} else {
+		r.rules[markManglePre] = preRule
+	}
+
+	postRule := []string{
+		"-o", r.wgIface.Name(),
+		"-m", "conntrack", "--ctstate", "NEW",
+		"-j", "CONNMARK", "--set-mark", fmt.Sprintf("%#x", nbnet.DataPlaneMarkOut),
+	}
+
+	if err := r.iptablesClient.AppendUnique(tableMangle, chainPostrouting, postRule...); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("add mangle postrouting rule: %w", err))
+	} else {
+		r.rules[markManglePost] = postRule
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// seedInitialEntries adds default rules to the entries map. Rules are
+// inserted at position 1, so the order here is reversed.
+//
+// Existing FORWARD policy decides outbound traffic towards our
+// interface. If FORWARD policy is "drop", we add an
+// established/related rule to allow return traffic for inbound rules.
+func (r *family) seedInitialEntries() {
+	established := getConntrackEstablished()
+
+	r.appendToEntries(chainInput, []string{"-i", r.wgIface.Name(), "-j", "DROP"})
+	r.appendToEntries(chainInput, []string{"-i", r.wgIface.Name(), "-j", chainACLInput})
+	r.appendToEntries(chainInput, append([]string{"-i", r.wgIface.Name()}, established...))
+
+	r.appendToEntries(chainForward, []string{"-i", r.wgIface.Name(), "-j", "DROP"})
+	r.appendToEntries(chainForward, []string{"-o", r.wgIface.Name(), "-j", chainRTFwdOut})
+	r.appendToEntries(chainForward, []string{"-i", r.wgIface.Name(), "-j", chainRTFwdIn})
+
+	// Mangle FORWARD guard: when external DNAT redirects traffic from
+	// the wg interface, it traverses FORWARD instead of INPUT,
+	// bypassing ACL rules. ACCEPT rules in filter FORWARD can be
+	// inserted above ours. Mangle runs before filter, so these guard
+	// rules enforce the ACL mark check where it cannot be overridden.
+	r.appendToEntries(mangleForwardKey, []string{
+		"-i", r.wgIface.Name(),
+		"-m", "conntrack", "--ctstate", "RELATED,ESTABLISHED",
+		"-j", "ACCEPT",
+	})
+	r.appendToEntries(mangleForwardKey, []string{
+		"-i", r.wgIface.Name(),
+		"-m", "conntrack", "--ctstate", "DNAT",
+		"-m", "mark", "!", "--mark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkRedirected),
+		"-j", "DROP",
+	})
+}
+
+func (r *family) seedInitialOptionalEntries() {
+	r.optionalEntries[chainForward] = []entry{
+		{
+			spec:     []string{"-m", "mark", "--mark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkRedirected), "-j", "ACCEPT"},
+			position: 2,
+		},
+	}
+}
+
+func (r *family) appendToEntries(chain chainKey, spec ruleSpec) {
+	r.entries[chain] = append(r.entries[chain], spec)
+}
+
+func (r *family) createDefaultChains() error {
+	if err := r.iptablesClient.NewChain(tableFilter, chainACLInput); err != nil {
+		return fmt.Errorf("create %s chain: %w", chainACLInput, err)
+	}
+
+	for chain, rules := range r.entries {
+		// mangle FORWARD guard rules are handled separately below
+		if chain == mangleForwardKey {
+			continue
+		}
+		for _, rule := range rules {
+			if err := r.iptablesClient.InsertUnique(tableFilter, string(chain), 1, rule...); err != nil {
+				return fmt.Errorf("insert jump rule into %s: %w", chain, err)
+			}
+		}
+	}
+
+	for chain, entries := range r.optionalEntries {
+		for _, entry := range entries {
+			if err := r.iptablesClient.InsertUnique(tableFilter, string(chain), entry.position, entry.spec...); err != nil {
+				log.Errorf("failed to insert optional entry %v: %v", entry.spec, err)
+				continue
+			}
+			r.entries[chain] = append(r.entries[chain], entry.spec)
+		}
+	}
+	clear(r.optionalEntries)
+
+	// Insert mangle FORWARD guard rules to prevent external DNAT bypass.
+	for _, rule := range r.entries[mangleForwardKey] {
+		if err := r.iptablesClient.AppendUnique(tableMangle, chainForward, rule...); err != nil {
+			log.Errorf("failed to add mangle FORWARD guard rule: %v", err)
+		}
+	}
+
+	return nil
+}
+
+func (r *family) cleanUpDefaultForwardRules() error {
+	var merr *multierror.Error
+
+	// cleanJumpRules removes the OUTPUT jump to NETBIRD-NAT-OUTPUT among
+	// the others, so the chain below deletes cleanly instead of failing
+	// with "device or resource busy".
+	if err := r.cleanJumpRules(); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("clean jump rules: %w", err))
+	}
+
+	for _, chainInfo := range []struct {
+		chain string
+		table string
+	}{
+		{chainRTFwdIn, tableFilter},
+		{chainRTFwdOut, tableFilter},
+		{chainRTPre, tableMangle},
+		{chainRTNAT, tableNat},
+		{chainRTRdr, tableNat},
+		{chainNATOutput, tableNat},
+		{chainRTMSSClamp, tableMangle},
+	} {
+		ok, err := r.iptablesClient.ChainExists(chainInfo.table, chainInfo.chain)
+		if err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("check chain %s in table %s: %w", chainInfo.chain, chainInfo.table, err))
+			continue
+		}
+		if ok {
+			if err := r.iptablesClient.ClearAndDeleteChain(chainInfo.table, chainInfo.chain); err != nil {
+				merr = multierror.Append(merr, fmt.Errorf("clear and delete chain %s in table %s: %w", chainInfo.chain, chainInfo.table, err))
+			}
+		}
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) cleanJumpRules() error {
+	// locations maps each jump rule to the built-in table and chain it
+	// was inserted into, plus the netbird chain it targets.
+	locations := map[firewall.RuleID]struct{ table, chain, target string }{
+		jumpNATPost:   {tableNat, chainPostrouting, chainRTNAT},
+		jumpManglePre: {tableMangle, chainPrerouting, chainRTPre},
+		jumpNATPre:    {tableNat, chainPrerouting, chainRTRdr},
+		jumpMSSClamp:  {tableMangle, chainForward, chainRTMSSClamp},
+		jumpNATOutput: {tableNat, chainOutput, chainNATOutput},
+	}
+
+	var merr *multierror.Error
+	for ruleID, loc := range locations {
+		rule, exists := r.rules[ruleID]
+		if !exists {
+			// Untracked (e.g. fresh start after an unclean shutdown with no
+			// restored state): if the target chain survived, remove the stale
+			// jump to it so the chain can be deleted.
+			ok, err := r.iptablesClient.ChainExists(loc.table, loc.target)
+			if err != nil {
+				merr = multierror.Append(merr, fmt.Errorf("check chain %s in table %s: %w", loc.target, loc.table, err))
+				continue
+			}
+			if !ok {
+				continue
+			}
+			rule = jumpRuleSpec(loc.target)
+		}
+		if err := r.iptablesClient.DeleteIfExists(loc.table, loc.chain, rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete rule from chain %s in table %s: %w", loc.chain, loc.table, err))
+			continue
+		}
+		delete(r.rules, ruleID)
+	}
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// jumpRuleSpec builds the iptables rule spec that jumps to target. Create
+// and cleanup sites share it so the installed and deleted specs cannot drift.
+func jumpRuleSpec(target string) []string {
+	return []string{"-j", target}
+}
+
+func (r *family) cleanAclChains() error {
+	var merr *multierror.Error
+
+	if err := r.cleanInputAclChain(); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	for _, rule := range r.entries[mangleForwardKey] {
+		if err := r.iptablesClient.DeleteIfExists(tableMangle, chainForward, rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete mangle %s guard rule %v: %w", chainForward, rule, err))
+		}
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) cleanInputAclChain() error {
+	ok, err := r.iptablesClient.ChainExists(tableFilter, chainACLInput)
+	if err != nil {
+		return fmt.Errorf("check chain %s: %w", chainACLInput, err)
+	}
+	if !ok {
+		return nil
+	}
+
+	var merr *multierror.Error
+	for _, rule := range r.entries[chainInput] {
+		if err := r.iptablesClient.DeleteIfExists(tableFilter, chainInput, rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete %s rule %v: %w", chainInput, rule, err))
+		}
+	}
+
+	for _, rule := range r.entries[chainForward] {
+		if err := r.iptablesClient.DeleteIfExists(tableFilter, chainForward, rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete %s rule %v: %w", chainForward, rule, err))
+		}
+	}
+
+	if err := r.iptablesClient.ClearAndDeleteChain(tableFilter, chainACLInput); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("clear and delete %s chain: %w", chainACLInput, err))
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) cleanupDataPlaneMark() error {
+	var merr *multierror.Error
+	if preRule, exists := r.rules[markManglePre]; exists {
+		if err := r.iptablesClient.DeleteIfExists(tableMangle, chainPrerouting, preRule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove mangle prerouting rule: %w", err))
+		} else {
+			delete(r.rules, markManglePre)
+		}
+	}
+
+	if postRule, exists := r.rules[markManglePost]; exists {
+		if err := r.iptablesClient.DeleteIfExists(tableMangle, chainPostrouting, postRule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove mangle postrouting rule: %w", err))
+		} else {
+			delete(r.rules, markManglePost)
+		}
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}

client/firewall/iptables/dnat_linux.go

@@ -0,0 +1,302 @@
+//go:build !android
+
+package iptables
+
+import (
+	"fmt"
+	"net/netip"
+	"strconv"
+	"strings"
+
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+)
+
+func (r *family) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
+	ruleID := rule.ID()
+	if _, exists := r.rules[ruleID+dnatSuffix]; exists {
+		return rule, nil
+	}
+
+	toDestination := rule.TranslatedAddress.String()
+	switch {
+	case len(rule.TranslatedPort.Values) == 0:
+		// no translated port, use original port
+	case len(rule.TranslatedPort.Values) == 1:
+		toDestination += fmt.Sprintf(":%d", rule.TranslatedPort.Values[0])
+	case rule.TranslatedPort.IsRange && len(rule.TranslatedPort.Values) == 2:
+		// need the "/originalport" suffix to avoid dnat port randomization
+		toDestination += fmt.Sprintf(":%d-%d/%d", rule.TranslatedPort.Values[0], rule.TranslatedPort.Values[1], rule.DestinationPort.Values[0])
+	default:
+		return nil, fmt.Errorf("invalid translated port: %v", rule.TranslatedPort)
+	}
+
+	proto := strings.ToLower(string(rule.Protocol))
+
+	rules := make(map[firewall.RuleID]ruleInfo, 3)
+
+	// DNAT rule
+	dnatRule := []string{
+		"!", "-i", r.wgIface.Name(),
+		"-p", proto,
+		"-j", "DNAT",
+		"--to-destination", toDestination,
+	}
+	dnatRule = append(dnatRule, applyPort("--dport", &rule.DestinationPort)...)
+	rules[ruleID+dnatSuffix] = ruleInfo{
+		table: tableNat,
+		chain: chainRTRdr,
+		rule:  dnatRule,
+	}
+
+	// SNAT rule
+	snatRule := []string{
+		"-o", r.wgIface.Name(),
+		"-p", proto,
+		"-d", rule.TranslatedAddress.String(),
+		"-j", "MASQUERADE",
+	}
+	snatRule = append(snatRule, applyPort("--dport", &rule.TranslatedPort)...)
+	rules[ruleID+snatSuffix] = ruleInfo{
+		table: tableNat,
+		chain: chainRTNAT,
+		rule:  snatRule,
+	}
+
+	// Forward filtering rule, if fwd policy is DROP
+	forwardRule := []string{
+		"-o", r.wgIface.Name(),
+		"-p", proto,
+		"-d", rule.TranslatedAddress.String(),
+		"-j", "ACCEPT",
+	}
+	forwardRule = append(forwardRule, applyPort("--dport", &rule.TranslatedPort)...)
+	rules[ruleID+fwdSuffix] = ruleInfo{
+		table: tableFilter,
+		chain: chainRTFwdOut,
+		rule:  forwardRule,
+	}
+
+	for key, ruleInfo := range rules {
+		if err := r.iptablesClient.Append(ruleInfo.table, ruleInfo.chain, ruleInfo.rule...); err != nil {
+			r.cleanupFailedDNATAdd(rules)
+			return nil, fmt.Errorf("add rule %s: %w", key, err)
+		}
+		r.rules[key] = ruleInfo.rule
+	}
+
+	if err := r.ipFwdState.RequestForwarding(r.v6); err != nil {
+		r.cleanupFailedDNATAdd(rules)
+		return nil, fmt.Errorf("enable forwarding: %w", err)
+	}
+
+	r.updateState()
+	return rule, nil
+}
+
+// cleanupFailedDNATAdd removes the bookkeeping written by a partially applied
+// AddDNATRule before rolling back the kernel rules, so no entries remain that
+// never got a forwarding refcount. rollbackRules re-adds entries it failed to
+// remove from the kernel.
+func (r *family) cleanupFailedDNATAdd(rules map[firewall.RuleID]ruleInfo) {
+	for key := range rules {
+		delete(r.rules, key)
+	}
+	if err := r.rollbackRules(rules); err != nil {
+		log.Errorf("rollback failed: %v", err)
+	}
+}
+
+func (r *family) rollbackRules(rules map[firewall.RuleID]ruleInfo) error {
+	var merr *multierror.Error
+	for key, ruleInfo := range rules {
+		if err := r.iptablesClient.DeleteIfExists(ruleInfo.table, ruleInfo.chain, ruleInfo.rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("rollback rule %s: %w", key, err))
+			// On rollback error, add to rules map for next cleanup
+			r.rules[key] = ruleInfo.rule
+		}
+	}
+	if merr != nil {
+		r.updateState()
+	}
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) DeleteDNATRule(rule firewall.Rule) error {
+	ruleID := rule.ID()
+
+	_, hadDNAT := r.rules[ruleID+dnatSuffix]
+	_, hadSNAT := r.rules[ruleID+snatSuffix]
+	_, hadFWD := r.rules[ruleID+fwdSuffix]
+	if !hadDNAT && !hadSNAT && !hadFWD {
+		return nil
+	}
+
+	var merr *multierror.Error
+	if dnatRule, exists := r.rules[ruleID+dnatSuffix]; exists {
+		if err := r.iptablesClient.Delete(tableNat, chainRTRdr, dnatRule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete DNAT rule: %w", err))
+		} else {
+			delete(r.rules, ruleID+dnatSuffix)
+		}
+	}
+
+	if snatRule, exists := r.rules[ruleID+snatSuffix]; exists {
+		if err := r.iptablesClient.Delete(tableNat, chainRTNAT, snatRule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete SNAT rule: %w", err))
+		} else {
+			delete(r.rules, ruleID+snatSuffix)
+		}
+	}
+
+	if fwdRule, exists := r.rules[ruleID+fwdSuffix]; exists {
+		if err := r.iptablesClient.Delete(tableFilter, chainRTFwdOut, fwdRule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete forward rule: %w", err))
+		} else {
+			delete(r.rules, ruleID+fwdSuffix)
+		}
+	}
+
+	// Release the refcount only once all rules are gone from the kernel. On
+	// partial failure the failed entries stay in r.rules so a retry can remove
+	// them and release then.
+	if merr == nil {
+		r.releaseForwarding()
+	}
+
+	r.updateState()
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// releaseForwarding drops one IP forwarding reference, logging any error.
+func (r *family) releaseForwarding() {
+	if err := r.ipFwdState.ReleaseForwarding(r.v6); err != nil {
+		log.Errorf("release IP forwarding: %v", err)
+	}
+}
+
+func (r *family) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	ruleID := firewall.RuleID(fmt.Sprintf("inbound-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	if _, exists := r.rules[ruleID]; exists {
+		return nil
+	}
+
+	dnatRule := []string{
+		"-i", r.wgIface.Name(),
+		"-p", strings.ToLower(protoForFamily(protocol, r.v6)),
+		"--dport", strconv.Itoa(int(originalPort)),
+		"-d", localAddr.String(),
+		"-m", "addrtype", "--dst-type", "LOCAL",
+		"-j", "DNAT",
+		"--to-destination", ":" + strconv.Itoa(int(translatedPort)),
+	}
+
+	info := ruleInfo{
+		table: tableNat,
+		chain: chainRTRdr,
+		rule:  dnatRule,
+	}
+
+	if err := r.iptablesClient.Append(info.table, info.chain, info.rule...); err != nil {
+		return fmt.Errorf("add inbound DNAT rule: %w", err)
+	}
+	r.rules[ruleID] = info.rule
+
+	r.updateState()
+	return nil
+}
+
+// RemoveInboundDNAT removes an inbound DNAT rule.
+func (r *family) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	ruleID := firewall.RuleID(fmt.Sprintf("inbound-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	if dnatRule, exists := r.rules[ruleID]; exists {
+		if err := r.iptablesClient.Delete(tableNat, chainRTRdr, dnatRule...); err != nil {
+			return fmt.Errorf("delete inbound DNAT rule: %w", err)
+		}
+		delete(r.rules, ruleID)
+	}
+
+	r.updateState()
+	return nil
+}
+
+// ensureNATOutputChain lazily creates the OUTPUT NAT chain and jump rule on first use.
+func (r *family) ensureNATOutputChain() error {
+	if _, exists := r.rules[jumpNATOutput]; exists {
+		return nil
+	}
+
+	chainExists, err := r.iptablesClient.ChainExists(tableNat, chainNATOutput)
+	if err != nil {
+		return fmt.Errorf("check chain %s: %w", chainNATOutput, err)
+	}
+	if !chainExists {
+		if err := r.iptablesClient.NewChain(tableNat, chainNATOutput); err != nil {
+			return fmt.Errorf("create chain %s: %w", chainNATOutput, err)
+		}
+	}
+
+	jumpRule := jumpRuleSpec(chainNATOutput)
+	if err := r.iptablesClient.Insert(tableNat, chainOutput, 1, jumpRule...); err != nil {
+		if !chainExists {
+			if delErr := r.iptablesClient.ClearAndDeleteChain(tableNat, chainNATOutput); delErr != nil {
+				log.Warnf("failed to rollback chain %s: %v", chainNATOutput, delErr)
+			}
+		}
+		return fmt.Errorf("add OUTPUT jump rule: %w", err)
+	}
+	r.rules[jumpNATOutput] = jumpRule
+
+	r.updateState()
+	return nil
+}
+
+// AddOutputDNAT adds an OUTPUT chain DNAT rule for locally-generated traffic.
+func (r *family) AddOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	ruleID := firewall.RuleID(fmt.Sprintf("output-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	if _, exists := r.rules[ruleID]; exists {
+		return nil
+	}
+
+	if err := r.ensureNATOutputChain(); err != nil {
+		return err
+	}
+
+	dnatRule := []string{
+		"-p", strings.ToLower(protoForFamily(protocol, localAddr.Is6())),
+		"--dport", strconv.Itoa(int(originalPort)),
+		"-d", localAddr.String(),
+		"-j", "DNAT",
+		"--to-destination", ":" + strconv.Itoa(int(translatedPort)),
+	}
+
+	if err := r.iptablesClient.Append(tableNat, chainNATOutput, dnatRule...); err != nil {
+		return fmt.Errorf("add output DNAT rule: %w", err)
+	}
+	r.rules[ruleID] = dnatRule
+
+	r.updateState()
+	return nil
+}
+
+// RemoveOutputDNAT removes an OUTPUT chain DNAT rule.
+func (r *family) RemoveOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	ruleID := firewall.RuleID(fmt.Sprintf("output-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	if dnatRule, exists := r.rules[ruleID]; exists {
+		if err := r.iptablesClient.Delete(tableNat, chainNATOutput, dnatRule...); err != nil {
+			return fmt.Errorf("delete output DNAT rule: %w", err)
+		}
+		delete(r.rules, ruleID)
+	}
+
+	r.updateState()
+	return nil
+}

client/firewall/iptables/dnat_refcount_linux_test.go

@@ -0,0 +1,240 @@
+//go:build privileged
+
+package iptables
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	fw "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+func iptRefcountIfaceV4() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("10.20.0.1"),
+				Network: netip.MustParsePrefix("10.20.0.0/24"),
+			}
+		},
+	}
+}
+
+func iptRefcountIfaceDual() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("10.20.0.1"),
+				Network: netip.MustParsePrefix("10.20.0.0/24"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+}
+
+func newIptRefcountManager(t *testing.T, dual bool) *Manager {
+	t.Helper()
+	var ifMock *iFaceMock
+	if dual {
+		ifMock = iptRefcountIfaceDual()
+	} else {
+		ifMock = iptRefcountIfaceV4()
+	}
+	m, err := Create(ifMock, iface.DefaultMTU)
+	require.NoError(t, err, "create manager")
+	require.NoError(t, m.Init(nil), "init manager")
+	t.Cleanup(func() {
+		require.NoError(t, m.Close(nil), "close manager")
+	})
+	return m
+}
+
+func iptDnatV4(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("10.20.0.2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+func iptDnatV6(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("fd00::2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+// TestIptablesRouting_RepeatedEnableSingleReference verifies that EnableRouting
+// (called on every network-map update) holds at most one reference per family
+// and a single DisableRouting drops both back to zero.
+func TestIptablesRouting_RepeatedEnableSingleReference(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	require.NoError(t, m.EnableRouting(), "first enable")
+	require.NoError(t, m.EnableRouting(), "second enable")
+	require.NoError(t, m.EnableRouting(), "third enable")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "repeated enable holds a single v4 reference")
+	assert.Equal(t, 1, v6, "repeated enable holds a single v6 reference")
+
+	require.NoError(t, m.DisableRouting(), "disable")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "single disable releases the v4 reference")
+	assert.Equal(t, 0, v6, "single disable releases the v6 reference")
+}
+
+// TestIptablesRouting_DisableKeepsDNATReference verifies that an unpaired
+// DisableRouting does not release references held by active DNAT rules.
+func TestIptablesRouting_DisableKeepsDNATReference(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV6(9095))
+	require.NoError(t, err, "add v6 dnat")
+
+	require.NoError(t, m.DisableRouting(), "unpaired disable")
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6, "DNAT-held reference survives unpaired DisableRouting")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "delete releases the DNAT reference")
+}
+
+// TestIptablesDNAT_RefcountBalancedV4 covers a Balanced Add/Delete pair on v4.
+func TestIptablesDNAT_RefcountBalancedV4(t *testing.T) {
+	m := newIptRefcountManager(t, false)
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV4(7081))
+	require.NoError(t, err, "add v4 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	r2, err := m.AddDNATRule(iptDnatV4(7082))
+	require.NoError(t, err, "add v4 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 2, v4, "v4 refcount after second add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r1))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r2))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount after second delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+}
+
+// TestIptablesDNAT_RefcountBalancedV6 checks the v6 path increments v6 only and
+// decrements back to zero.
+func TestIptablesDNAT_RefcountBalancedV6(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	require.NotNil(t, m.family6, "v6 family")
+	require.Same(t, m.family4.ipFwdState, m.family6.ipFwdState, "shared state")
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV6(9081))
+	require.NoError(t, err, "add v6 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 1, v6, "v6 refcount after first add")
+
+	r2, err := m.AddDNATRule(iptDnatV6(9082))
+	require.NoError(t, err, "add v6 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 2, v6, "v6 refcount after second add")
+
+	require.NoError(t, m.DeleteDNATRule(r1))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 1, v6, "v6 refcount after first delete")
+
+	require.NoError(t, m.DeleteDNATRule(r2))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6, "v6 refcount after second delete")
+}
+
+// TestIptablesDNAT_DuplicateAddNoLeak verifies the duplicate-rule path returns
+// without bumping the refcount.
+func TestIptablesDNAT_DuplicateAddNoLeak(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	rule := iptDnatV4(7083)
+	r1, err := m.AddDNATRule(rule)
+	require.NoError(t, err)
+	v4, _ := state.Counts()
+	assert.Equal(t, 1, v4)
+
+	_, err = m.AddDNATRule(rule)
+	require.NoError(t, err, "duplicate add")
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "duplicate add must not increment")
+
+	require.NoError(t, m.DeleteDNATRule(r1))
+	v4, _ = state.Counts()
+	assert.Equal(t, 0, v4, "single delete must drop to zero")
+}
+
+// TestIptablesDNAT_DeleteMissingNoUnderflow verifies Delete on an unknown rule
+// neither errors nor releases the refcount.
+func TestIptablesDNAT_DeleteMissingNoUnderflow(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	phantom := iptDnatV4(7099)
+	require.NoError(t, m.DeleteDNATRule(&phantom), "delete missing v4")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6)
+
+	phantom6 := iptDnatV6(9099)
+	require.NoError(t, m.DeleteDNATRule(&phantom6), "delete missing v6")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6)
+
+	r1, err := m.AddDNATRule(iptDnatV4(7100))
+	require.NoError(t, err)
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "real add still increments after phantom delete")
+	require.NoError(t, m.DeleteDNATRule(r1))
+}
+
+// TestIptablesDNAT_DoubleDeleteNoUnderflow verifies a second Delete on the same
+// rule is a no-op.
+func TestIptablesDNAT_DoubleDeleteNoUnderflow(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV6(9083))
+	require.NoError(t, err)
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "first delete")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "second delete must be no-op")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "double delete must not underflow")
+}

client/firewall/iptables/family_linux.go

@@ -0,0 +1,258 @@
+//go:build !android
+
+package iptables
+
+import (
+	"fmt"
+	"maps"
+	"net/netip"
+
+	"github.com/coreos/go-iptables/iptables"
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+	nbid "github.com/netbirdio/netbird/client/internal/acl/id"
+	"github.com/netbirdio/netbird/client/internal/routemanager/ipfwdstate"
+	"github.com/netbirdio/netbird/client/internal/routemanager/refcounter"
+	"github.com/netbirdio/netbird/client/internal/statemanager"
+)
+
+// constants needed to manage and create iptable rules
+const (
+	tableFilter = "filter"
+	tableNat    = "nat"
+	tableMangle = "mangle"
+
+	// chainACLInput is the peer ACL chain that holds installed
+	// peer-filtering rules.
+	chainACLInput = "NETBIRD-ACL-INPUT"
+
+	// mangleForwardKey is the entries map key for mangle FORWARD guard
+	// rules that prevent external DNAT from bypassing ACL rules.
+	mangleForwardKey chainKey = "MANGLE-FORWARD"
+
+	chainInput       = "INPUT"
+	chainPostrouting = "POSTROUTING"
+	chainPrerouting  = "PREROUTING"
+	chainForward     = "FORWARD"
+	chainRTNAT       = "NETBIRD-RT-NAT"
+	chainRTFwdIn     = "NETBIRD-RT-FWD-IN"
+	chainRTFwdOut    = "NETBIRD-RT-FWD-OUT"
+	chainRTPre       = "NETBIRD-RT-PRE"
+	chainRTRdr       = "NETBIRD-RT-RDR"
+	chainNATOutput   = "NETBIRD-NAT-OUTPUT"
+	chainRTMSSClamp  = "NETBIRD-RT-MSSCLAMP"
+
+	jumpManglePre  = "jump-mangle-pre"
+	jumpNATPre     = "jump-nat-pre"
+	jumpNATPost    = "jump-nat-post"
+	jumpNATOutput  = "jump-nat-output"
+	jumpMSSClamp   = "jump-mss-clamp"
+	markManglePre  = "mark-mangle-pre"
+	markManglePost = "mark-mangle-post"
+	matchSet       = "--match-set"
+
+	dnatSuffix firewall.RuleID = "_dnat"
+	snatSuffix firewall.RuleID = "_snat"
+	fwdSuffix  firewall.RuleID = "_fwd"
+
+	// ipv4TCPHeaderSize is the minimum IPv4 (20) + TCP (20) header size for MSS calculation.
+	ipv4TCPHeaderSize = 40
+	// ipv6TCPHeaderSize is the minimum IPv6 (40) + TCP (20) header size for MSS calculation.
+	ipv6TCPHeaderSize = 60
+)
+
+type ruleInfo struct {
+	chain string
+	table string
+	rule  []string
+}
+
+type routeRules map[firewall.RuleID][]string
+
+// ruleSpec is a single iptables rule expressed as its argument list
+// (e.g. {"-i", "wg0", "-j", "DROP"}).
+type ruleSpec []string
+
+// chainKey identifies the chain a seeded entry belongs to. It holds
+// built-in chain names ("INPUT", "FORWARD", "PREROUTING") plus the
+// synthetic mangleForwardKey bucket for the mangle FORWARD guard rules.
+type chainKey string
+
+// aclEntries maps a chain to the rules seeded into it to jump into or
+// guard the netbird ACL chains.
+type aclEntries map[chainKey][]ruleSpec
+
+type entry struct {
+	spec     ruleSpec
+	position int
+}
+
+// ipsetCounter is the shared hash:net refcounter used by peer and
+// route ACLs alike. The ipset library does not support comments, so
+// the key is just the set name (string).
+type ipsetCounter = refcounter.Counter[string, []netip.Prefix, struct{}]
+
+// family holds the per-address-family iptables state. One instance
+// handles route ACLs, peer ACLs, NAT, DNAT, and MSS clamping for a
+// single family; the top-level Manager owns one for v4 and another
+// for v6.
+type family struct {
+	iptablesClient *iptables.IPTables
+	wgIface        iFaceMapper
+	v6             bool
+
+	// Peer ACL chain bookkeeping.
+	entries         aclEntries
+	optionalEntries map[chainKey][]entry
+
+	// filters holds peer + route filter rules keyed by content hash.
+	// AddFilterRule writes here; DeleteFilterRule looks up by id.
+	filters      map[nbid.RuleID]*Rule
+	ipsetCounter *ipsetCounter
+	// ipsetSupported records whether the kernel can create the hash:net
+	// sets the source matches rely on; probed once at init. When false,
+	// multi-source rules expand to one rule per source prefix.
+	ipsetSupported bool
+
+	// rules holds NAT, jump, and MSS-clamping rules (auxiliary
+	// plumbing that isn't a filter rule).
+	rules routeRules
+
+	// Routing / NAT.
+	legacyManagement bool
+	mtu              uint16
+	ipFwdState       *ipfwdstate.IPForwardingState
+
+	stateManager *statemanager.Manager
+}
+
+func newFamily(iptablesClient *iptables.IPTables, wgIface iFaceMapper, mtu uint16) (*family, error) {
+	r := &family{
+		iptablesClient:  iptablesClient,
+		wgIface:         wgIface,
+		v6:              iptablesClient.Proto() == iptables.ProtocolIPv6,
+		entries:         make(aclEntries),
+		optionalEntries: make(map[chainKey][]entry),
+		filters:         make(map[nbid.RuleID]*Rule),
+		rules:           make(routeRules),
+		mtu:             mtu,
+		ipFwdState:      ipfwdstate.NewIPForwardingState(wgIface.Name()),
+	}
+
+	r.ipsetCounter = refcounter.New(
+		func(name string, sources []netip.Prefix) (struct{}, error) {
+			return struct{}{}, r.createIpSet(name, sources)
+		},
+		func(name string, _ struct{}) error {
+			return r.deleteIpSet(name)
+		},
+	)
+
+	return r, nil
+}
+
+// init wires the family to the state manager and installs both the
+// route ACL containers and the peer ACL chain skeleton.
+func (r *family) init(stateManager *statemanager.Manager) error {
+	r.stateManager = stateManager
+
+	r.ipsetSupported = r.probeIPSetSupport()
+
+	if err := r.cleanUpDefaultForwardRules(); err != nil {
+		log.Errorf("failed to clean up rules from FORWARD chain: %s", err)
+	}
+
+	if err := r.createContainers(); err != nil {
+		return fmt.Errorf("create containers: %w", err)
+	}
+
+	if err := r.setupDataPlaneMark(); err != nil {
+		log.Errorf("failed to set up data plane mark: %v", err)
+	}
+
+	r.seedInitialEntries()
+	r.seedInitialOptionalEntries()
+
+	if err := r.cleanAclChains(); err != nil {
+		return fmt.Errorf("clean acl chains: %w", err)
+	}
+	if err := r.createDefaultChains(); err != nil {
+		return fmt.Errorf("create default chains: %w", err)
+	}
+
+	r.updateState()
+
+	return nil
+}
+
+// Reset tears down all firewall state owned by this family. ACL
+// chain cleanup runs before route-chain cleanup because the route
+// chains are still referenced by FORWARD jumps installed during
+// seedInitialEntries; deleting them first would trip EBUSY.
+func (r *family) Reset() error {
+	var merr *multierror.Error
+
+	if err := r.cleanAclChains(); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	if err := r.cleanUpDefaultForwardRules(); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	if err := r.ipsetCounter.Flush(); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	if err := r.cleanupDataPlaneMark(); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	clear(r.rules)
+	clear(r.filters)
+	r.updateState()
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) updateState() {
+	if r.stateManager == nil {
+		return
+	}
+
+	var currentState *ShutdownState
+	if existing := r.stateManager.GetState(currentState); existing != nil {
+		if existingState, ok := existing.(*ShutdownState); ok {
+			currentState = existingState
+		}
+	}
+	if currentState == nil {
+		currentState = &ShutdownState{}
+	}
+
+	currentState.Lock()
+	defer currentState.Unlock()
+
+	// Clone the rule maps so the persisted state holds a private snapshot.
+	// The live maps keep being mutated by subsequent rule operations while
+	// the state manager marshals the state from its periodic-save goroutine.
+	// Sharing the maps by reference races the two and aborts the process with
+	// a concurrent map iteration and write. The ipset counter guards itself
+	// during marshaling, so it can be shared directly.
+	if r.v6 {
+		currentState.RouteRules6 = maps.Clone(r.rules)
+		currentState.RouteIPsetCounter6 = r.ipsetCounter
+		currentState.ACLEntries6 = maps.Clone(r.entries)
+	} else {
+		currentState.RouteRules = maps.Clone(r.rules)
+		currentState.RouteIPsetCounter = r.ipsetCounter
+		currentState.ACLEntries = maps.Clone(r.entries)
+	}
+
+	if err := r.stateManager.UpdateState(currentState); err != nil {
+		log.Errorf("failed to update state: %v", err)
+	}
+}

client/firewall/iptables/filter_linux.go

@@ -0,0 +1,430 @@
+//go:build !android
+
+package iptables
+
+import (
+	"fmt"
+	"net/netip"
+	"slices"
+	"strconv"
+	"strings"
+
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+	nbid "github.com/netbirdio/netbird/client/internal/acl/id"
+	nbnet "github.com/netbirdio/netbird/client/net"
+)
+
+// AddFilterRule installs a packet-filtering rule. With destination
+// empty, the rule goes to the peer ACL input chain plus a paired
+// mangle PREROUTING rule for the redirect mark. With destination set
+// (prefix or named set), it goes to the route ACL forward chain.
+// Multi-source rules collapse to one iptables rule via the shared
+// hash:net ipset.
+func (r *family) AddFilterRule(
+	id []byte,
+	sources []netip.Prefix,
+	destination firewall.Network,
+	proto firewall.Protocol,
+	sPort *firewall.Port,
+	dPort *firewall.Port,
+	action firewall.Action,
+) (firewall.Rule, error) {
+	ruleID := nbid.GenerateRuleID(sources, destination, proto, sPort, dPort, action)
+	if existing, ok := r.filters[ruleID]; ok {
+		return existing, nil
+	}
+
+	rule, err := r.installFilterRules(ruleID, sources, destination, proto, sPort, dPort, action, r.ipsetSupported)
+	if err != nil {
+		return nil, err
+	}
+
+	r.filters[ruleID] = rule
+	r.updateState()
+	return rule, nil
+}
+
+// installFilterRules resolves the source matches and installs one
+// iptables rule per match. It is more than one rule only when useIPSet
+// is false and a multi-source rule has to be expanded per prefix.
+func (r *family) installFilterRules(
+	ruleID nbid.RuleID,
+	sources []netip.Prefix,
+	destination firewall.Network,
+	proto firewall.Protocol,
+	sPort *firewall.Port,
+	dPort *firewall.Port,
+	action firewall.Action,
+	useIPSet bool,
+) (*Rule, error) {
+	srcMatches, err := r.applySourceMatches(sources, useIPSet)
+	if err != nil {
+		return nil, fmt.Errorf("apply source match: %w", err)
+	}
+
+	rule, err := r.installFilterRule(ruleID, srcMatches, destination, proto, sPort, dPort, action)
+	if err != nil {
+		for _, srcMatch := range srcMatches {
+			r.dropSourceMatch(srcMatch)
+		}
+		return nil, err
+	}
+	return rule, nil
+}
+
+func (r *family) hasRule(id nbid.RuleID) bool {
+	_, ok := r.filters[id]
+	return ok
+}
+
+// hasDNATRule reports whether this family owns the DNAT rule set for
+// the given user id. DNAT rules live in r.rules under the well-known
+// "<id>_dnat" key; the lookup here is used by Manager.DeleteDNATRule
+// to pick the right family.
+func (r *family) hasDNATRule(id firewall.RuleID) bool {
+	_, ok := r.rules[id+dnatSuffix]
+	return ok
+}
+
+// DeleteFilterRule removes a previously installed filter rule. The
+// rule's stored chain/table identify where to delete from; source set
+// references are recovered from the spec via findSets and dropped
+// from the shared ipset counter.
+func (r *family) DeleteFilterRule(rule firewall.Rule) error {
+	ruleID := rule.ID()
+	pr, ok := r.filters[ruleID]
+	if !ok {
+		log.Debugf("filter rule %s not found", ruleID)
+		return nil
+	}
+
+	// DeleteIfExists keeps the deletes idempotent so a retry after a
+	// partial failure does not error on the parts already removed.
+	var merr *multierror.Error
+	for _, fs := range pr.allSpecs() {
+		if err := r.iptablesClient.DeleteIfExists(tableFilter, pr.chain, fs.specs...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete rule from %s: %w", pr.chain, err))
+		}
+		if fs.mangleSpecs != nil {
+			if err := r.iptablesClient.DeleteIfExists(tableMangle, chainRTPre, fs.mangleSpecs...); err != nil {
+				merr = multierror.Append(merr, fmt.Errorf("delete mangle rule: %w", err))
+			}
+		}
+	}
+	if merr != nil {
+		// Leave the rule tracked so the caller retries the remaining part.
+		return nberrors.FormatErrorOrNil(merr)
+	}
+
+	// The rule is gone from iptables, so untrack it regardless of how the
+	// refcount decrement goes, but surface decrement failures so callers
+	// see the ipset desync. Only the primary spec can reference sets: the
+	// per-prefix expansion never uses them.
+	delete(r.filters, ruleID)
+	r.updateState()
+	if err := r.decrementSetCounter(pr.specs); err != nil {
+		return fmt.Errorf("drop source set references: %w", err)
+	}
+	return nil
+}
+
+// findSets scans an iptables rule spec for "-m set --match-set <name>
+// <dir>" fragments and returns the named sets in occurrence order.
+// Used at delete time to drop ipsetCounter references.
+func findSets(rule []string) []string {
+	var sets []string
+	for i, arg := range rule {
+		if arg == "-m" && i+3 < len(rule) && rule[i+1] == "set" && rule[i+2] == matchSet {
+			sets = append(sets, rule[i+3])
+		}
+	}
+	return sets
+}
+
+// sourceNetwork classifies a source-prefix list into the firewall.Network
+// shape the rest of the spec-builder consumes: empty for match-any, a
+// single prefix inline, or an ipset for multiple sources.
+func sourceNetwork(sources []netip.Prefix) firewall.Network {
+	switch {
+	case len(sources) == 0:
+		return firewall.Network{}
+	case len(sources) == 1 && sources[0].Bits() == 0:
+		return firewall.Network{}
+	case len(sources) == 1:
+		return firewall.Network{Prefix: sources[0]}
+	default:
+		return firewall.Network{Set: firewall.NewPrefixSet(sources)}
+	}
+}
+
+// applySourceMatches returns one source match fragment per iptables
+// rule needed for the sources: normally a single fragment (a set match,
+// a direct -s match, or nil for match-any), and one -s fragment per
+// prefix when a multi-source rule cannot use ipset. Per-prefix rules
+// are the only form a kernel without the ipset modules can express.
+func (r *family) applySourceMatches(sources []netip.Prefix, useIPSet bool) ([][]string, error) {
+	network := sourceNetwork(sources)
+	if !network.IsSet() || useIPSet {
+		match, err := r.applySourceMatch(network, sources)
+		if err != nil {
+			return nil, err
+		}
+		return [][]string{match}, nil
+	}
+
+	matches := make([][]string, 0, len(sources))
+	for _, source := range sources {
+		matches = append(matches, []string{"-s", source.String()})
+	}
+	return matches, nil
+}
+
+// applySourceMatch returns the iptables match fragment for the rule's
+// source. For a Set it increments the shared ipset's refcount; for a
+// Prefix it emits a direct -s match; for the wildcard it returns nil.
+func (r *family) applySourceMatch(network firewall.Network, prefixes []netip.Prefix) ([]string, error) {
+	switch {
+	case network.IsSet():
+		if r.ipsetCounter == nil {
+			return nil, fmt.Errorf("multi-source peer rule requires shared ipset counter")
+		}
+		name := r.ipsetName(network.Set.HashedName())
+		if _, err := r.ipsetCounter.Increment(name, prefixes); err != nil {
+			return nil, fmt.Errorf("ipset increment %s: %w", name, err)
+		}
+		return []string{"-m", "set", matchSet, name, "src"}, nil
+	case network.IsPrefix():
+		return []string{"-s", network.Prefix.String()}, nil
+	default:
+		return nil, nil
+	}
+}
+
+// dropSourceMatch undoes whatever applySourceMatch reserved when
+// installing a rule fails. Safe to call when the spec is empty or holds
+// only inline matchers. Decrement errors are logged but not returned:
+// the install error is what the caller needs to see.
+func (r *family) dropSourceMatch(srcMatch []string) {
+	if r.ipsetCounter == nil {
+		return
+	}
+	for _, name := range findSets(srcMatch) {
+		if _, err := r.ipsetCounter.Decrement(name); err != nil {
+			log.Errorf("rollback ipset decrement %s: %v", name, err)
+		}
+	}
+}
+
+// decrementSetCounter drops ipset references owned by a raw rule spec
+// stored in r.rules (NAT / legacy route entries). It returns an error
+// aggregate so the caller surfaces decrement failures.
+func (r *family) decrementSetCounter(rule []string) error {
+	if r.ipsetCounter == nil {
+		return nil
+	}
+	var merr *multierror.Error
+	for _, name := range findSets(rule) {
+		if _, err := r.ipsetCounter.Decrement(name); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("decrement counter: %w", err))
+		}
+	}
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// installFilterRule assembles and writes the iptables filter-chain
+// rules for one filter rule, one per source match fragment. With
+// destination empty the rules land in the peer ACL input chain and each
+// gets a paired mangle PREROUTING rule for the redirect mark. With
+// destination set the rules land in the route ACL forward chain and
+// there is no mangle pairing.
+func (r *family) installFilterRule(
+	ruleID nbid.RuleID,
+	srcMatches [][]string,
+	destination firewall.Network,
+	protocol firewall.Protocol,
+	sPort, dPort *firewall.Port,
+	action firewall.Action,
+) (*Rule, error) {
+	isRoute := !destination.IsZero()
+
+	proto := protoForFamily(protocol, r.v6)
+
+	var destExp []string
+	if isRoute {
+		var err error
+		destExp, err = r.applyNetwork("-d", destination, nil)
+		if err != nil {
+			return nil, fmt.Errorf("apply network -d: %w", err)
+		}
+	}
+	matchSpecs := filterMatchSpecs(proto, sPort, dPort)
+
+	chain := chainACLInput
+	if isRoute {
+		chain = chainRTFwdIn
+	}
+
+	var installed []filterSpecs
+	for _, srcMatch := range srcMatches {
+		specs := slices.Clone(srcMatch)
+		specs = append(specs, destExp...)
+		specs = append(specs, matchSpecs...)
+
+		var mangleSpecs []string
+		if !isRoute {
+			mangleSpecs = slices.Clone(specs)
+			mangleSpecs = append(mangleSpecs,
+				"-i", r.wgIface.Name(),
+				"-m", "addrtype", "--dst-type", "LOCAL",
+				"-j", "MARK", "--set-xmark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkRedirected),
+			)
+		}
+
+		specs = append(specs, "-j", actionToStr(action))
+
+		if err := r.insertFilterRule(chain, action, specs); err != nil {
+			// Leave nothing half-installed: the caller sees an error, so a
+			// partial rule would silently keep matching without being tracked.
+			r.removeFilterSpecs(chain, installed)
+			r.dropSourceMatch(destExp)
+			return nil, fmt.Errorf("install filter rule on %s: %w", chain, err)
+		}
+
+		// The mangle redirect-mark rule is best effort: the filter rule itself
+		// is what enforces the ACL, so a mangle failure must not undo it. Drop
+		// the spec so teardown does not try to remove a rule that was not added.
+		if mangleSpecs != nil {
+			if err := r.iptablesClient.Append(tableMangle, chainRTPre, mangleSpecs...); err != nil {
+				log.Errorf("add mangle rule: %v", err)
+				mangleSpecs = nil
+			}
+		}
+
+		installed = append(installed, filterSpecs{specs: specs, mangleSpecs: mangleSpecs})
+	}
+
+	return &Rule{
+		id:          ruleID,
+		specs:       installed[0].specs,
+		mangleSpecs: installed[0].mangleSpecs,
+		extraRules:  installed[1:],
+		chain:       chain,
+		v6:          r.v6,
+	}, nil
+}
+
+// insertFilterRule writes one assembled rule spec into the given ACL
+// chain. Peer ACL drops are inserted at position 1 so they precede the
+// chain's catch-all; route ACL drops are inserted at position 2 to sit
+// immediately after the established/related accept rule.
+func (r *family) insertFilterRule(chain string, action firewall.Action, specs []string) error {
+	if action == firewall.ActionDrop {
+		pos := 1
+		if chain == chainRTFwdIn {
+			pos = 2
+		}
+		return r.iptablesClient.Insert(tableFilter, chain, pos, specs...)
+	}
+	return r.iptablesClient.Append(tableFilter, chain, specs...)
+}
+
+// removeFilterSpecs deletes the already-installed rules of a partially
+// applied filter rule.
+func (r *family) removeFilterSpecs(chain string, installed []filterSpecs) {
+	for _, fs := range installed {
+		if err := r.iptablesClient.DeleteIfExists(tableFilter, chain, fs.specs...); err != nil {
+			log.Debugf("delete partial filter rule: %v", err)
+		}
+		if fs.mangleSpecs != nil {
+			if err := r.iptablesClient.DeleteIfExists(tableMangle, chainRTPre, fs.mangleSpecs...); err != nil {
+				log.Debugf("delete partial mangle rule: %v", err)
+			}
+		}
+	}
+}
+
+// applyNetwork resolves a firewall.Network into the iptables match
+// fragment for the given direction flag (-s or -d). Set networks
+// increment the shared ipset refcount; prefixes emit a direct match;
+// an empty network returns no spec ("match any").
+func (r *family) applyNetwork(flag string, network firewall.Network, prefixes []netip.Prefix) ([]string, error) {
+	direction := "src"
+	if flag == "-d" {
+		direction = "dst"
+	}
+
+	if network.IsSet() {
+		// A destination set is populated later from DNS results, so unlike a
+		// source set it cannot be expanded into per-prefix rules. Without
+		// ipset such a rule is not expressible; report it instead of
+		// installing something broader than the policy allows.
+		if flag == "-d" && !r.ipsetSupported {
+			return nil, fmt.Errorf("destination set %s requires ipset (ip_set_hash_net and xt_set)", network.Set.HashedName())
+		}
+
+		name := r.ipsetName(network.Set.HashedName())
+		if _, err := r.ipsetCounter.Increment(name, prefixes); err != nil {
+			return nil, fmt.Errorf("create or get ipset: %w", err)
+		}
+
+		return []string{"-m", "set", matchSet, name, direction}, nil
+	}
+	if network.IsPrefix() {
+		return []string{flag, network.Prefix.String()}, nil
+	}
+
+	// nolint:nilnil
+	return nil, nil
+}
+
+// protoForFamily translates ICMP to ICMPv6 for ip6tables.
+// ip6tables requires "ipv6-icmp" (or "icmpv6") instead of "icmp".
+func protoForFamily(protocol firewall.Protocol, v6 bool) string {
+	if v6 && protocol == firewall.ProtocolICMP {
+		return "ipv6-icmp"
+	}
+	return string(protocol)
+}
+
+// filterMatchSpecs returns the proto/port match fragment for a
+// filtering rule. The source match (-s or -m set) is built by the
+// caller and prepended.
+func filterMatchSpecs(protocol string, sPort, dPort *firewall.Port) (specs []string) {
+	if protocol != "all" {
+		specs = append(specs, "-p", protocol)
+	}
+	specs = append(specs, applyPort("--sport", sPort)...)
+	specs = append(specs, applyPort("--dport", dPort)...)
+	return specs
+}
+
+func actionToStr(action firewall.Action) string {
+	if action == firewall.ActionAccept {
+		return "ACCEPT"
+	}
+	return "DROP"
+}
+
+func applyPort(flag string, port *firewall.Port) []string {
+	if port == nil {
+		return nil
+	}
+
+	if port.IsRange && len(port.Values) == 2 {
+		return []string{flag, fmt.Sprintf("%d:%d", port.Values[0], port.Values[1])}
+	}
+
+	if len(port.Values) > 1 {
+		portList := make([]string, len(port.Values))
+		for i, p := range port.Values {
+			portList[i] = strconv.Itoa(int(p))
+		}
+		return []string{"-m", "multiport", flag, strings.Join(portList, ",")}
+	}
+
+	return []string{flag, strconv.Itoa(int(port.Values[0]))}
+}

client/firewall/iptables/interface_allower_linux.go

@@ -0,0 +1,93 @@
+package iptables
+
+import (
+	"fmt"
+
+	"github.com/coreos/go-iptables/iptables"
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+)
+
+// InterfaceAllower opens the NetBird interface on the iptables filter INPUT
+// chain so the host firewall doesn't drop traffic the userspace firewall
+// handles. It is the fallback used when nftables is unavailable (an
+// iptables-legacy host).
+//
+// It opens INPUT only: the userspace router never forwards in the kernel.
+// firewalld trust is handled by the uspfilter manager, not here.
+type InterfaceAllower struct {
+	ifaceName string
+	ipt4      *iptables.IPTables
+	// ipt6 is nil when the interface has no IPv6 overlay address.
+	ipt6 *iptables.IPTables
+}
+
+// NewInterfaceAllower builds an iptables allower for the interface. It returns
+// an error when iptables is unavailable, so the caller can fall back to
+// firewalld trust.
+func NewInterfaceAllower(wgIface iFaceMapper) (*InterfaceAllower, error) {
+	ipt4, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
+	if err != nil {
+		return nil, fmt.Errorf("iptables not available: %w", err)
+	}
+	if _, err := ipt4.ListChains(tableFilter); err != nil {
+		return nil, fmt.Errorf("iptables filter table not available: %w", err)
+	}
+
+	a := &InterfaceAllower{ifaceName: wgIface.Name(), ipt4: ipt4}
+
+	// Missing v6 must not break the v4 path: open v4 only and continue.
+	if wgIface.Address().HasIPv6() {
+		ipt6, err := iptables.NewWithProtocol(iptables.ProtocolIPv6)
+		if err != nil {
+			log.Warnf("ip6tables not available, opening interface on v4 only: %v", err)
+		} else if _, err := ipt6.ListChains(tableFilter); err != nil {
+			log.Warnf("ip6tables filter table not available, opening interface on v4 only: %v", err)
+		} else {
+			a.ipt6 = ipt6
+		}
+	}
+
+	return a, nil
+}
+
+// Apply inserts the interface accept rule on the filter INPUT chain. It removes
+// any stale rule first so an unclean exit (e.g. SIGKILL, where Close never ran)
+// is recovered deterministically rather than accumulating duplicates.
+func (a *InterfaceAllower) Apply() error {
+	var merr *multierror.Error
+	for _, ipt := range a.clients() {
+		if err := ipt.DeleteIfExists(tableFilter, chainInput, a.inputRule()...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("clean stale interface accept rule: %w", err))
+		}
+		if err := ipt.Insert(tableFilter, chainInput, 1, a.inputRule()...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("add interface accept rule: %w", err))
+		}
+	}
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// Close removes the interface accept rule.
+func (a *InterfaceAllower) Close() error {
+	var merr *multierror.Error
+	for _, ipt := range a.clients() {
+		if err := ipt.DeleteIfExists(tableFilter, chainInput, a.inputRule()...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove interface accept rule: %w", err))
+		}
+	}
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (a *InterfaceAllower) inputRule() []string {
+	return []string{"-i", a.ifaceName, "-j", "ACCEPT"}
+}
+
+func (a *InterfaceAllower) clients() []*iptables.IPTables {
+	clients := []*iptables.IPTables{a.ipt4}
+	if a.ipt6 != nil {
+		clients = append(clients, a.ipt6)
+	}
+	return clients
+}

client/firewall/iptables/ipset_linux.go

@@ -0,0 +1,131 @@
+//go:build !android
+
+package iptables
+
+import (
+	"fmt"
+	"net/netip"
+
+	"github.com/google/uuid"
+	"github.com/hashicorp/go-multierror"
+	"github.com/lrh3321/ipset-go"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+)
+
+// probeIPSetSupport checks whether the kernel can create the ipset type
+// used for source and destination matches. On kernels lacking the
+// required ipset hash module, set creation fails (e.g. "invalid
+// argument"), which would otherwise fail every multi-source rule and
+// leave traffic the policy permits blocked by the catch-all drop. When
+// unsupported, multi-source rules fall back to one rule per prefix.
+func (r *family) probeIPSetSupport() bool {
+	// Use a unique name so concurrent processes don't collide and we only ever
+	// destroy the set we created ourselves. ipset names are limited to 31 chars,
+	// so use a short random suffix.
+	probeName := "nb-probe-" + uuid.New().String()[:8]
+
+	if err := r.createIPSet(probeName); err != nil {
+		log.Warnf("ipset is not available (failed to create probe set: %v); "+
+			"falling back to per-IP iptables ACL rules. Ensure the kernel provides "+
+			"the ipset hash:net module (ip_set_hash_net) for better performance with large rule sets", err)
+		return false
+	}
+
+	if err := r.destroyIPSet(probeName); err != nil {
+		log.Debugf("destroy ipset probe set %q: %v", probeName, err)
+	}
+
+	return true
+}
+
+func (r *family) createIpSet(setName string, sources []netip.Prefix) error {
+	if err := r.createIPSet(setName); err != nil {
+		return fmt.Errorf("create set %s: %w", setName, err)
+	}
+
+	for _, prefix := range sources {
+		if err := r.addPrefixToIPSet(setName, prefix); err != nil {
+			// The refcounter records nothing when this callback errors,
+			// so destroy the set or it leaks in the kernel. A partial
+			// source set would also fail-open for deny rules, so the
+			// rule must fail rather than install with a missing source.
+			if derr := r.destroyIPSet(setName); derr != nil {
+				log.Warnf("rollback ipset %s after add failure: %v", setName, derr)
+			}
+			return fmt.Errorf("add element to set %s: %w", setName, err)
+		}
+	}
+
+	return nil
+}
+
+func (r *family) deleteIpSet(setName string) error {
+	if err := r.destroyIPSet(setName); err != nil {
+		return fmt.Errorf("destroy set %s: %w", setName, err)
+	}
+
+	log.Debugf("deleted unused ipset %s", setName)
+	return nil
+}
+
+func (r *family) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
+	name := r.ipsetName(set.HashedName())
+	var merr *multierror.Error
+	for _, prefix := range prefixes {
+		if err := r.addPrefixToIPSet(name, prefix); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("add prefix to ipset: %w", err))
+		}
+	}
+	if merr == nil {
+		log.Debugf("updated set %s with prefixes %v", name, prefixes)
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) ipsetName(name string) string {
+	if r.v6 {
+		return name + "-v6"
+	}
+	return name
+}
+
+func (r *family) createIPSet(name string) error {
+	opts := ipset.CreateOptions{
+		Replace: true,
+	}
+	if r.v6 {
+		opts.Family = ipset.FamilyIPV6
+	}
+
+	if err := ipset.Create(name, ipset.TypeHashNet, opts); err != nil {
+		return fmt.Errorf("create ipset %s: %w", name, err)
+	}
+
+	log.Debugf("created ipset %s with type hash:net", name)
+	return nil
+}
+
+func (r *family) addPrefixToIPSet(name string, prefix netip.Prefix) error {
+	addr := prefix.Addr()
+	ip := addr.AsSlice()
+
+	entry := &ipset.Entry{
+		IP:      ip,
+		CIDR:    uint8(prefix.Bits()),
+		Replace: true,
+	}
+
+	if err := ipset.Add(name, entry); err != nil {
+		return fmt.Errorf("add prefix to ipset %s: %w", name, err)
+	}
+
+	return nil
+}
+
+func (r *family) destroyIPSet(name string) error {
+	return ipset.Destroy(name)
+}

client/firewall/iptables/manager_linux.go

@@ -3,7 +3,6 @@ package iptables
 import (
 	"context"
 	"fmt"
-	"net"
 	"net/netip"
 	"sync"
 
@@ -18,16 +17,21 @@ import (
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 )
 
-// Manager of iptables firewall
+// Manager of iptables firewall. Per-family state (peer ACLs, route
+// ACLs, NAT, DNAT, MSS clamping) lives on family; Manager dispatches
+// by family and provides the public firewall.Manager surface.
 type Manager struct {
 	mutex sync.Mutex
 
 	wgIface iFaceMapper
 
 	ipv4Client   *iptables.IPTables
-	aclMgr       *aclManager
-	router       *router
+	family4      *family
 	rawSupported bool
+
+	// IPv6 counterparts, nil when no v6 overlay
+	ipv6Client *iptables.IPTables
+	family6    *family
 }
 
 // iFaceMapper defines subset methods of interface required for manager
@@ -48,25 +52,51 @@ func Create(wgIface iFaceMapper, mtu uint16) (*Manager, error) {
 		ipv4Client: iptablesClient,
 	}
 
-	m.router, err = newRouter(iptablesClient, wgIface, mtu)
+	m.family4, err = newFamily(iptablesClient, wgIface, mtu)
 	if err != nil {
-		return nil, fmt.Errorf("create router: %w", err)
+		return nil, fmt.Errorf("create family: %w", err)
 	}
 
-	m.aclMgr, err = newAclManager(iptablesClient, wgIface)
-	if err != nil {
-		return nil, fmt.Errorf("create acl manager: %w", err)
+	if wgIface.Address().HasIPv6() {
+		if err := m.createIPv6Components(wgIface, mtu); err != nil {
+			return nil, fmt.Errorf("create IPv6 firewall: %w", err)
+		}
 	}
 
 	return m, nil
 }
 
+func (m *Manager) createIPv6Components(wgIface iFaceMapper, mtu uint16) error {
+	ip6Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv6)
+	if err != nil {
+		return fmt.Errorf("init ip6tables: %w", err)
+	}
+
+	family6, err := newFamily(ip6Client, wgIface, mtu)
+	if err != nil {
+		return fmt.Errorf("create v6 family: %w", err)
+	}
+
+	// Share the same IP forwarding state with the v4 family, since the
+	// forwarding refcounter is per-family but shared between both families.
+	family6.ipFwdState = m.family4.ipFwdState
+
+	m.ipv6Client = ip6Client
+	m.family6 = family6
+
+	return nil
+}
+
+func (m *Manager) hasIPv6() bool {
+	return m.ipv6Client != nil
+}
+
 func (m *Manager) Init(stateManager *statemanager.Manager) error {
 	state := &ShutdownState{
 		InterfaceState: &InterfaceState{
 			NameStr:   m.wgIface.Name(),
 			WGAddress: m.wgIface.Address(),
-			MTU:       m.router.mtu,
+			MTU:       m.family4.mtu,
 		},
 	}
 	stateManager.RegisterState(state)
@@ -74,13 +104,8 @@ func (m *Manager) Init(stateManager *statemanager.Manager) error {
 		log.Errorf("failed to update state: %v", err)
 	}
 
-	if err := m.router.init(stateManager); err != nil {
-		return fmt.Errorf("router init: %w", err)
-	}
-
-	if err := m.aclMgr.init(stateManager); err != nil {
-		// TODO: cleanup router
-		return fmt.Errorf("acl manager init: %w", err)
+	if err := m.initChains(stateManager); err != nil {
+		return err
 	}
 
 	if err := m.initNoTrackChain(); err != nil {
@@ -103,55 +128,78 @@ func (m *Manager) Init(stateManager *statemanager.Manager) error {
 	return nil
 }
 
-// AddPeerFiltering adds a rule to the firewall
-//
-// Comment will be ignored because some system this feature is not supported
-func (m *Manager) AddPeerFiltering(
-	id []byte,
-	ip net.IP,
-	proto firewall.Protocol,
-	sPort *firewall.Port,
-	dPort *firewall.Port,
-	action firewall.Action,
-	ipsetName string,
-) ([]firewall.Rule, error) {
-	m.mutex.Lock()
-	defer m.mutex.Unlock()
+// initChains initializes the per-family firewall state for both
+// address families, rolling back on failure.
+func (m *Manager) initChains(stateManager *statemanager.Manager) error {
+	type initStep struct {
+		name string
+		r    *family
+	}
 
-	return m.aclMgr.AddPeerFiltering(id, ip, proto, sPort, dPort, action, ipsetName)
+	steps := []initStep{{"v4", m.family4}}
+	if m.hasIPv6() {
+		steps = append(steps, initStep{"v6", m.family6})
+	}
+
+	var initialized []initStep
+	for _, s := range steps {
+		if err := s.r.init(stateManager); err != nil {
+			for i := len(initialized) - 1; i >= 0; i-- {
+				if rerr := initialized[i].r.Reset(); rerr != nil {
+					log.Warnf("rollback %s: %v", initialized[i].name, rerr)
+				}
+			}
+			return fmt.Errorf("%s init: %w", s.name, err)
+		}
+		initialized = append(initialized, s)
+	}
+	return nil
 }
 
-func (m *Manager) AddRouteFiltering(
+// AddFilterRule installs a packet-filtering rule. See firewall.Manager
+// docs for destination semantics. Sources are a single address family;
+// the rule is dispatched to the matching v4 / v6 backend.
+func (m *Manager) AddFilterRule(
 	id []byte,
 	sources []netip.Prefix,
 	destination firewall.Network,
 	proto firewall.Protocol,
-	sPort, dPort *firewall.Port,
+	sPort *firewall.Port,
+	dPort *firewall.Port,
 	action firewall.Action,
 ) (firewall.Rule, error) {
-	m.mutex.Lock()
-	defer m.mutex.Unlock()
-
-	if destination.IsPrefix() && !destination.Prefix.Addr().Is4() {
-		return nil, fmt.Errorf("unsupported IP version: %s", destination.Prefix.Addr().String())
+	if len(sources) == 0 {
+		return nil, firewall.ErrNoSources
 	}
 
-	return m.router.AddRouteFiltering(id, sources, destination, proto, sPort, dPort, action)
-}
-
-// DeletePeerRule from the firewall by rule definition
-func (m *Manager) DeletePeerRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.aclMgr.DeletePeerRule(rule)
+	fam := m.family4
+	if isIPv6Rule(sources, destination) {
+		if !m.hasIPv6() {
+			return nil, fmt.Errorf("add filtering: %w", firewall.ErrIPv6NotInitialized)
+		}
+		fam = m.family6
+	}
+	return fam.AddFilterRule(id, sources, destination, proto, sPort, dPort, action)
 }
 
-func (m *Manager) DeleteRouteRule(rule firewall.Rule) error {
+// DeleteFilterRule removes a rule previously added via AddFilterRule.
+// The rule is looked up by id in each family's filter cache.
+func (m *Manager) DeleteFilterRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.DeleteRouteRule(rule)
+	id := rule.ID()
+	if m.family4.hasRule(id) {
+		return m.family4.DeleteFilterRule(rule)
+	}
+	if m.hasIPv6() && m.family6.hasRule(id) {
+		return m.family6.DeleteFilterRule(rule)
+	}
+	log.Debugf("filter rule %s not found in any family", id)
+	return nil
 }
 
 func (m *Manager) IsServerRouteSupported() bool {
@@ -166,18 +214,68 @@ func (m *Manager) AddNatRule(pair firewall.RouterPair) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.AddNatRule(pair)
+	if pair.Destination.IsPrefix() && pair.Destination.Prefix.Addr().Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("add NAT rule: %w", firewall.ErrIPv6NotInitialized)
+		}
+		return m.family6.AddNatRule(pair)
+	}
+
+	if err := m.family4.AddNatRule(pair); err != nil {
+		return err
+	}
+
+	// Dynamic routes need NAT in both tables since resolved IPs can be
+	// either v4 or v6. This covers both DomainSet (modern) and the legacy
+	// wildcard 0.0.0.0/0 destination where the client resolves DNS.
+	if m.hasIPv6() && pair.Dynamic {
+		v6Pair := firewall.ToV6NatPair(pair)
+		if err := m.family6.AddNatRule(v6Pair); err != nil {
+			return fmt.Errorf("add v6 NAT rule: %w", err)
+		}
+	}
+
+	return nil
 }
 
 func (m *Manager) RemoveNatRule(pair firewall.RouterPair) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.RemoveNatRule(pair)
+	if pair.Destination.IsPrefix() && pair.Destination.Prefix.Addr().Is6() {
+		if !m.hasIPv6() {
+			return nil
+		}
+		return m.family6.RemoveNatRule(pair)
+	}
+
+	var merr *multierror.Error
+
+	if err := m.family4.RemoveNatRule(pair); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("remove v4 NAT rule: %w", err))
+	}
+
+	if m.hasIPv6() && pair.Dynamic {
+		v6Pair := firewall.ToV6NatPair(pair)
+		if err := m.family6.RemoveNatRule(v6Pair); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove v6 NAT rule: %w", err))
+		}
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
 }
 
 func (m *Manager) SetLegacyManagement(isLegacy bool) error {
-	return firewall.SetLegacyManagement(m.router, isLegacy)
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	if err := firewall.SetLegacyManagement(m.family4, isLegacy); err != nil {
+		return err
+	}
+	if m.hasIPv6() {
+		return firewall.SetLegacyManagement(m.family6, isLegacy)
+	}
+	return nil
 }
 
 // Reset firewall to the default state
@@ -191,11 +289,14 @@ func (m *Manager) Close(stateManager *statemanager.Manager) error {
 		merr = multierror.Append(merr, fmt.Errorf("cleanup notrack chain: %w", err))
 	}
 
-	if err := m.aclMgr.Reset(); err != nil {
-		merr = multierror.Append(merr, fmt.Errorf("reset acl manager: %w", err))
+	if m.hasIPv6() {
+		if err := m.family6.Reset(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("reset v6 family: %w", err))
+		}
 	}
-	if err := m.router.Reset(); err != nil {
-		merr = multierror.Append(merr, fmt.Errorf("reset router: %w", err))
+
+	if err := m.family4.Reset(); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("reset family: %w", err))
 	}
 
 	// Appending to merr intentionally blocks DeleteState below so ShutdownState
@@ -214,30 +315,6 @@ func (m *Manager) Close(stateManager *statemanager.Manager) error {
 	return nberrors.FormatErrorOrNil(merr)
 }
 
-// AllowNetbird allows netbird interface traffic.
-// This is called when USPFilter wraps the native firewall, adding blanket accept
-// rules so that packet filtering is handled in userspace instead of by netfilter.
-func (m *Manager) AllowNetbird() error {
-	_, err := m.AddPeerFiltering(
-		nil,
-		net.IP{0, 0, 0, 0},
-		firewall.ProtocolALL,
-		nil,
-		nil,
-		firewall.ActionAccept,
-		"",
-	)
-	if err != nil {
-		return fmt.Errorf("allow netbird interface traffic: %w", err)
-	}
-
-	if err := firewalld.TrustInterface(m.wgIface.Name()); err != nil {
-		log.Warnf("failed to trust interface in firewalld: %v", err)
-	}
-
-	return nil
-}
-
 // Flush doesn't need to be implemented for this manager
 func (m *Manager) Flush() error { return nil }
 
@@ -247,17 +324,12 @@ func (m *Manager) SetLogLevel(log.Level) {
 }
 
 func (m *Manager) EnableRouting() error {
-	if err := m.router.ipFwdState.RequestForwarding(); err != nil {
-		return fmt.Errorf("enable IP forwarding: %w", err)
-	}
-	return nil
+	// v6 only when the overlay actually has v6.
+	return m.family4.ipFwdState.RequestRouting(m.hasIPv6())
 }
 
 func (m *Manager) DisableRouting() error {
-	if err := m.router.ipFwdState.ReleaseForwarding(); err != nil {
-		return fmt.Errorf("disable IP forwarding: %w", err)
-	}
-	return nil
+	return m.family4.ipFwdState.ReleaseRouting()
 }
 
 // AddDNATRule adds a DNAT rule
@@ -265,7 +337,13 @@ func (m *Manager) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error)
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.AddDNATRule(rule)
+	if rule.TranslatedAddress.Is6() {
+		if !m.hasIPv6() {
+			return nil, fmt.Errorf("add DNAT rule: %w", firewall.ErrIPv6NotInitialized)
+		}
+		return m.family6.AddDNATRule(rule)
+	}
+	return m.family4.AddDNATRule(rule)
 }
 
 // DeleteDNATRule deletes a DNAT rule
@@ -273,7 +351,10 @@ func (m *Manager) DeleteDNATRule(rule firewall.Rule) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.DeleteDNATRule(rule)
+	if m.hasIPv6() && !m.family4.hasDNATRule(rule.ID()) {
+		return m.family6.DeleteDNATRule(rule)
+	}
+	return m.family4.DeleteDNATRule(rule)
 }
 
 // UpdateSet updates the set with the given prefixes
@@ -281,44 +362,87 @@ func (m *Manager) UpdateSet(set firewall.Set, prefixes []netip.Prefix) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.UpdateSet(set, prefixes)
+	var v4Prefixes, v6Prefixes []netip.Prefix
+	for _, p := range prefixes {
+		if p.Addr().Is6() {
+			v6Prefixes = append(v6Prefixes, p)
+		} else {
+			v4Prefixes = append(v4Prefixes, p)
+		}
+	}
+
+	if err := m.family4.UpdateSet(set, v4Prefixes); err != nil {
+		return err
+	}
+
+	if m.hasIPv6() && len(v6Prefixes) > 0 {
+		if err := m.family6.UpdateSet(set, v6Prefixes); err != nil {
+			return fmt.Errorf("update v6 set: %w", err)
+		}
+	}
+
+	return nil
 }
 
 // AddInboundDNAT adds an inbound DNAT rule redirecting traffic from NetBird peers to local services.
-func (m *Manager) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, sourcePort, targetPort uint16) error {
+func (m *Manager) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.AddInboundDNAT(localAddr, protocol, sourcePort, targetPort)
+	if localAddr.Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("add inbound DNAT: %w", firewall.ErrIPv6NotInitialized)
+		}
+		return m.family6.AddInboundDNAT(localAddr, protocol, originalPort, translatedPort)
+	}
+	return m.family4.AddInboundDNAT(localAddr, protocol, originalPort, translatedPort)
 }
 
 // RemoveInboundDNAT removes an inbound DNAT rule.
-func (m *Manager) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, sourcePort, targetPort uint16) error {
+func (m *Manager) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.RemoveInboundDNAT(localAddr, protocol, sourcePort, targetPort)
+	if localAddr.Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("remove inbound DNAT: %w", firewall.ErrIPv6NotInitialized)
+		}
+		return m.family6.RemoveInboundDNAT(localAddr, protocol, originalPort, translatedPort)
+	}
+	return m.family4.RemoveInboundDNAT(localAddr, protocol, originalPort, translatedPort)
 }
 
 // AddOutputDNAT adds an OUTPUT chain DNAT rule for locally-generated traffic.
-func (m *Manager) AddOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, sourcePort, targetPort uint16) error {
+func (m *Manager) AddOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.AddOutputDNAT(localAddr, protocol, sourcePort, targetPort)
+	if localAddr.Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("add output DNAT: %w", firewall.ErrIPv6NotInitialized)
+		}
+		return m.family6.AddOutputDNAT(localAddr, protocol, originalPort, translatedPort)
+	}
+	return m.family4.AddOutputDNAT(localAddr, protocol, originalPort, translatedPort)
 }
 
 // RemoveOutputDNAT removes an OUTPUT chain DNAT rule.
-func (m *Manager) RemoveOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, sourcePort, targetPort uint16) error {
+func (m *Manager) RemoveOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
 	m.mutex.Lock()
 	defer m.mutex.Unlock()
 
-	return m.router.RemoveOutputDNAT(localAddr, protocol, sourcePort, targetPort)
+	if localAddr.Is6() {
+		if !m.hasIPv6() {
+			return fmt.Errorf("remove output DNAT: %w", firewall.ErrIPv6NotInitialized)
+		}
+		return m.family6.RemoveOutputDNAT(localAddr, protocol, originalPort, translatedPort)
+	}
+	return m.family4.RemoveOutputDNAT(localAddr, protocol, originalPort, translatedPort)
 }
 
 const (
 	chainNameRaw = "NETBIRD-RAW"
-	chainOUTPUT  = "OUTPUT"
+	chainOutput  = "OUTPUT"
 	tableRaw     = "raw"
 )
 
@@ -393,15 +517,15 @@ func (m *Manager) initNoTrackChain() error {
 
 	jumpRule := []string{"-j", chainNameRaw}
 
-	if err := m.ipv4Client.InsertUnique(tableRaw, chainOUTPUT, 1, jumpRule...); err != nil {
+	if err := m.ipv4Client.InsertUnique(tableRaw, chainOutput, 1, jumpRule...); err != nil {
 		if delErr := m.ipv4Client.DeleteChain(tableRaw, chainNameRaw); delErr != nil {
 			log.Debugf("delete orphan chain: %v", delErr)
 		}
 		return fmt.Errorf("add output jump rule: %w", err)
 	}
 
-	if err := m.ipv4Client.InsertUnique(tableRaw, chainPREROUTING, 1, jumpRule...); err != nil {
-		if delErr := m.ipv4Client.DeleteIfExists(tableRaw, chainOUTPUT, jumpRule...); delErr != nil {
+	if err := m.ipv4Client.InsertUnique(tableRaw, chainPrerouting, 1, jumpRule...); err != nil {
+		if delErr := m.ipv4Client.DeleteIfExists(tableRaw, chainOutput, jumpRule...); delErr != nil {
 			log.Debugf("delete output jump rule: %v", delErr)
 		}
 		if delErr := m.ipv4Client.DeleteChain(tableRaw, chainNameRaw); delErr != nil {
@@ -428,11 +552,11 @@ func (m *Manager) cleanupNoTrackChain() error {
 
 	jumpRule := []string{"-j", chainNameRaw}
 
-	if err := m.ipv4Client.DeleteIfExists(tableRaw, chainOUTPUT, jumpRule...); err != nil {
+	if err := m.ipv4Client.DeleteIfExists(tableRaw, chainOutput, jumpRule...); err != nil {
 		return fmt.Errorf("remove output jump rule: %w", err)
 	}
 
-	if err := m.ipv4Client.DeleteIfExists(tableRaw, chainPREROUTING, jumpRule...); err != nil {
+	if err := m.ipv4Client.DeleteIfExists(tableRaw, chainPrerouting, jumpRule...); err != nil {
 		return fmt.Errorf("remove prerouting jump rule: %w", err)
 	}
 
@@ -447,3 +571,13 @@ func (m *Manager) cleanupNoTrackChain() error {
 func getConntrackEstablished() []string {
 	return []string{"-m", "conntrack", "--ctstate", "RELATED,ESTABLISHED", "-j", "ACCEPT"}
 }
+
+// isIPv6Rule reports whether the rule belongs to the IPv6 family, from
+// the destination prefix when set, otherwise from the (single-family)
+// sources.
+func isIPv6Rule(sources []netip.Prefix, destination firewall.Network) bool {
+	if destination.IsPrefix() {
+		return destination.Prefix.Addr().Is6()
+	}
+	return len(sources) > 0 && sources[0].Addr().Is6()
+}

client/firewall/iptables/manager_linux_test.go

@@ -1,18 +1,23 @@
+//go:build privileged
+
 package iptables
 
 import (
 	"fmt"
 	"net/netip"
+	"slices"
 	"strings"
 	"testing"
 	"time"
 
 	"github.com/coreos/go-iptables/iptables"
+	"github.com/lrh3321/ipset-go"
 	"github.com/stretchr/testify/require"
 
 	fw "github.com/netbirdio/netbird/client/firewall/manager"
 	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/iface/wgaddr"
+	"github.com/netbirdio/netbird/shared/management/domain"
 )
 
 var ifaceMock = &iFaceMock{
@@ -65,47 +70,37 @@ func TestIptablesManager(t *testing.T) {
 		time.Sleep(time.Second)
 	}()
 
-	var rule2 []fw.Rule
+	var rule2 fw.Rule
 	t.Run("add second rule", func(t *testing.T) {
 		ip := netip.MustParseAddr("10.20.0.3")
 		port := &fw.Port{
 			IsRange: true,
 			Values:  []uint16{8043, 8046},
 		}
-		rule2, err = manager.AddPeerFiltering(nil, ip.AsSlice(), "tcp", port, nil, fw.ActionAccept, "")
+		rule2, err = manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "tcp", port, nil, fw.ActionAccept)
 		require.NoError(t, err, "failed to add rule")
 
-		for _, r := range rule2 {
-			rr := r.(*Rule)
-			checkRuleSpecs(t, ipv4Client, rr.chain, true, rr.specs...)
-		}
+		rr := rule2.(*Rule)
+		checkRuleSpecs(t, ipv4Client, rr.chain, true, rr.specs...)
 	})
 
 	t.Run("delete second rule", func(t *testing.T) {
-		for _, r := range rule2 {
-			err := manager.DeletePeerRule(r)
-			require.NoError(t, err, "failed to delete rule")
-		}
-
-		require.Empty(t, manager.aclMgr.ipsetStore.ipsets, "rulesets index after removed second rule must be empty")
+		require.NoError(t, manager.DeleteFilterRule(rule2), "failed to delete rule")
 	})
 
 	t.Run("reset check", func(t *testing.T) {
 		// add second rule
 		ip := netip.MustParseAddr("10.20.0.3")
 		port := &fw.Port{Values: []uint16{5353}}
-		_, err = manager.AddPeerFiltering(nil, ip.AsSlice(), "udp", nil, port, fw.ActionAccept, "")
+		_, err = manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "udp", nil, port, fw.ActionAccept)
 		require.NoError(t, err, "failed to add rule")
 
 		err = manager.Close(nil)
 		require.NoError(t, err, "failed to reset")
 
-		ok, err := ipv4Client.ChainExists("filter", chainNameInputRules)
+		ok, err := ipv4Client.ChainExists("filter", chainACLInput)
 		require.NoError(t, err, "failed check chain exists")
-
-		if ok {
-			require.NoErrorf(t, err, "chain '%v' still exists after Close", chainNameInputRules)
-		}
+		require.Falsef(t, ok, "chain %q still exists after Close", chainACLInput)
 	})
 }
 
@@ -126,15 +121,13 @@ func TestIptablesManagerDenyRules(t *testing.T) {
 		ip := netip.MustParseAddr("10.20.0.3")
 		port := &fw.Port{Values: []uint16{22}}
 
-		rule, err := manager.AddPeerFiltering(nil, ip.AsSlice(), "tcp", nil, port, fw.ActionDrop, "deny-ssh")
+		rule, err := manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "tcp", nil, port, fw.ActionDrop)
 		require.NoError(t, err, "failed to add deny rule")
-		require.NotEmpty(t, rule, "deny rule should not be empty")
+		require.NotNil(t, rule, "deny rule should not be nil")
 
 		// Verify the rule was added by checking iptables
-		for _, r := range rule {
-			rr := r.(*Rule)
-			checkRuleSpecs(t, ipv4Client, rr.chain, true, rr.specs...)
-		}
+		rr := rule.(*Rule)
+		checkRuleSpecs(t, ipv4Client, rr.chain, true, rr.specs...)
 	})
 
 	t.Run("deny rule precedence test", func(t *testing.T) {
@@ -142,36 +135,40 @@ func TestIptablesManagerDenyRules(t *testing.T) {
 		port := &fw.Port{Values: []uint16{80}}
 
 		// Add accept rule first
-		_, err := manager.AddPeerFiltering(nil, ip.AsSlice(), "tcp", nil, port, fw.ActionAccept, "accept-http")
+		_, err := manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "tcp", nil, port, fw.ActionAccept)
 		require.NoError(t, err, "failed to add accept rule")
 
 		// Add deny rule second for same IP/port - this should take precedence
-		_, err = manager.AddPeerFiltering(nil, ip.AsSlice(), "tcp", nil, port, fw.ActionDrop, "deny-http")
+		_, err = manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "tcp", nil, port, fw.ActionDrop)
 		require.NoError(t, err, "failed to add deny rule")
 
 		// Inspect the actual iptables rules to verify deny rule comes before accept rule
-		rules, err := ipv4Client.List("filter", chainNameInputRules)
+		rules, err := ipv4Client.List("filter", chainACLInput)
 		require.NoError(t, err, "failed to list iptables rules")
 
 		// Debug: print all rules
-		t.Logf("All iptables rules in chain %s:", chainNameInputRules)
+		t.Logf("All iptables rules in chain %s:", chainACLInput)
 		for i, rule := range rules {
 			t.Logf("  [%d] %s", i, rule)
 		}
 
+		// Single-source rules emit a direct `-s <ip>/32 ... --dport 80`
+		// match. Match on that shape instead of the legacy
+		// per-(action,port) ipset names ("deny-http"/"accept-http")
+		// that this test predates.
+		srcMatch := fmt.Sprintf("-s %s/32", ip)
 		var denyRuleIndex, acceptRuleIndex = -1, -1
 		for i, rule := range rules {
-			if strings.Contains(rule, "DROP") {
-				t.Logf("Found DROP rule at index %d: %s", i, rule)
-				if strings.Contains(rule, "deny-http") && strings.Contains(rule, "80") {
-					denyRuleIndex = i
-				}
+			if !strings.Contains(rule, srcMatch) || !strings.Contains(rule, "--dport 80") {
+				continue
 			}
-			if strings.Contains(rule, "ACCEPT") {
+			if strings.Contains(rule, "-j DROP") {
+				t.Logf("Found DROP rule at index %d: %s", i, rule)
+				denyRuleIndex = i
+			}
+			if strings.Contains(rule, "-j ACCEPT") {
 				t.Logf("Found ACCEPT rule at index %d: %s", i, rule)
-				if strings.Contains(rule, "accept-http") && strings.Contains(rule, "80") {
-					acceptRuleIndex = i
-				}
+				acceptRuleIndex = i
 			}
 		}
 
@@ -196,7 +193,6 @@ func TestIptablesManagerIPSet(t *testing.T) {
 		},
 	}
 
-	// just check on the local interface
 	manager, err := Create(mock, iface.DefaultMTU)
 	require.NoError(t, err)
 	require.NoError(t, manager.Init(nil))
@@ -210,27 +206,39 @@ func TestIptablesManagerIPSet(t *testing.T) {
 		time.Sleep(time.Second)
 	}()
 
-	var rule2 []fw.Rule
-	t.Run("add second rule", func(t *testing.T) {
+	var rule2 fw.Rule
+	t.Run("single source uses direct -s match (no ipset)", func(t *testing.T) {
 		ip := netip.MustParseAddr("10.20.0.3")
 		port := &fw.Port{
 			Values: []uint16{443},
 		}
-		rule2, err = manager.AddPeerFiltering(nil, ip.AsSlice(), "tcp", port, nil, fw.ActionAccept, "default")
-		for _, r := range rule2 {
-			require.NoError(t, err, "failed to add rule")
-			require.Equal(t, r.(*Rule).ipsetName, "default-sport", "ipset name must be set")
-			require.Equal(t, r.(*Rule).ip, "10.20.0.3", "ipset IP must be set")
-		}
+		rule2, err = manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "tcp", port, nil, fw.ActionAccept)
+		require.NoError(t, err, "failed to add rule")
+		require.NotNil(t, rule2)
+		require.Contains(t, rule2.(*Rule).specs, "-s",
+			"single-source rule should use direct -s match, not an ipset")
+		require.Empty(t, findSets(rule2.(*Rule).specs),
+			"single-source rule should not allocate a shared ipset")
 	})
 
-	t.Run("delete second rule", func(t *testing.T) {
-		for _, r := range rule2 {
-			err := manager.DeletePeerRule(r)
-			require.NoError(t, err, "failed to delete rule")
+	t.Run("delete single-source rule", func(t *testing.T) {
+		require.NoError(t, manager.DeleteFilterRule(rule2), "failed to delete rule")
+	})
 
-			require.Empty(t, manager.aclMgr.ipsetStore.ipsets, "rulesets index after removed second rule must be empty")
+	t.Run("multi-source uses shared ipset", func(t *testing.T) {
+		sources := []netip.Prefix{
+			netip.PrefixFrom(netip.MustParseAddr("10.20.0.3"), 32),
+			netip.PrefixFrom(netip.MustParseAddr("10.20.0.4"), 32),
+			netip.PrefixFrom(netip.MustParseAddr("10.20.0.5"), 32),
 		}
+		port := &fw.Port{Values: []uint16{8080}}
+		multi, err := manager.AddFilterRule(nil, sources, fw.Network{}, "tcp", nil, port, fw.ActionAccept)
+		require.NoError(t, err, "failed to add multi-source rule")
+		require.NotNil(t, multi, "multi-source rule must produce one iptables rule")
+		sets := findSets(multi.(*Rule).specs)
+		require.Len(t, sets, 1, "multi-source rule must reference exactly one ipset")
+
+		require.NoError(t, manager.DeleteFilterRule(multi))
 	})
 
 	t.Run("reset check", func(t *testing.T) {
@@ -239,9 +247,324 @@ func TestIptablesManagerIPSet(t *testing.T) {
 	})
 }
 
+// TestIptablesFilterIPSetFallback verifies that when the kernel lacks
+// ipset support, a multi-source rule falls back to one iptables rule
+// per source prefix instead of silently leaving the chain empty. See
+// discussion #6125.
+func TestIptablesFilterIPSetFallback(t *testing.T) {
+	ipv4Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
+	require.NoError(t, err)
+
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	defer func() {
+		require.NoError(t, manager.Close(nil))
+	}()
+
+	// Simulate a kernel without the ipset hash module.
+	manager.family4.ipsetSupported = false
+
+	sources := []netip.Prefix{
+		netip.MustParsePrefix("10.20.0.42/32"),
+		netip.MustParsePrefix("10.20.0.43/32"),
+	}
+	port := &fw.Port{Values: []uint16{22}}
+
+	rule, err := manager.AddFilterRule(nil, sources, fw.Network{}, "tcp", nil, port, fw.ActionAccept)
+	require.NoError(t, err, "AddFilterRule should succeed via fallback")
+
+	rr := rule.(*Rule)
+	all := rr.allSpecs()
+	require.Len(t, all, len(sources), "each source prefix needs its own rule")
+	for i, fs := range all {
+		joined := strings.Join(fs.specs, " ")
+		require.Contains(t, joined, "-s "+sources[i].String(), "fallback rule must match by source prefix")
+		require.NotContains(t, joined, matchSet, "fallback rule must not use ipset matching")
+
+		// The rule must actually be present in the ACL chain (not silently dropped).
+		checkRuleSpecs(t, ipv4Client, rr.chain, true, fs.specs...)
+
+		// Every expanded peer rule keeps its own redirect-mark pairing.
+		require.NotNil(t, fs.mangleSpecs, "peer rule must carry a mangle pairing")
+		checkTableRuleSpecs(t, ipv4Client, tableMangle, chainRTPre, true, fs.mangleSpecs...)
+	}
+
+	require.NoError(t, manager.DeleteFilterRule(rule), "failed to delete fallback rule")
+	for _, fs := range all {
+		checkRuleSpecs(t, ipv4Client, rr.chain, false, fs.specs...)
+		checkTableRuleSpecs(t, ipv4Client, tableMangle, chainRTPre, false, fs.mangleSpecs...)
+	}
+}
+
+// TestIptablesFilterDestinationSetRequiresIPSet documents that a dynamic
+// (domain) destination cannot be expressed without ipset: its prefixes are only
+// known after DNS resolution, so there is nothing to expand into per-prefix
+// rules. The call must report that rather than install a broader rule than the
+// policy allows.
+func TestIptablesFilterDestinationSetRequiresIPSet(t *testing.T) {
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	defer func() {
+		require.NoError(t, manager.Close(nil))
+	}()
+
+	manager.family4.ipsetSupported = false
+
+	destination := fw.Network{Set: fw.NewDomainSet(domain.List{"example.com"})}
+
+	_, err = manager.AddFilterRule(nil, []netip.Prefix{netip.MustParsePrefix("172.16.0.0/16")},
+		destination, fw.ProtocolALL, nil, nil, fw.ActionAccept)
+	require.Error(t, err, "a domain destination is not expressible without ipset")
+	require.ErrorContains(t, err, "requires ipset")
+}
+
+// TestIptablesNatRuleDropsSourceSetOnDestinationFailure covers a marking rule
+// whose source set is created but whose destination set is not: the source
+// reference has to go back, or the set it created stays in the kernel with a
+// count nothing will ever drop.
+func TestIptablesNatRuleDropsSourceSetOnDestinationFailure(t *testing.T) {
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	defer func() {
+		require.NoError(t, manager.Close(nil))
+	}()
+
+	sourceSet := fw.NewPrefixSet([]netip.Prefix{
+		netip.MustParsePrefix("100.0.0.0/16"),
+		netip.MustParsePrefix("10.10.0.0/16"),
+	})
+	destSet := fw.NewDomainSet(domain.List{"example.org"})
+
+	// Poison the destination set's name so its hash:net creation fails after
+	// the source set has already been created.
+	poisoned := manager.family4.ipsetName(destSet.HashedName())
+	require.NoError(t, ipset.Create(poisoned, ipset.TypeHashIP, ipset.CreateOptions{}))
+	t.Cleanup(func() {
+		if err := ipset.Destroy(poisoned); err != nil {
+			t.Logf("destroy poisoned set %s: %v", poisoned, err)
+		}
+	})
+
+	pair := fw.RouterPair{
+		ID:          "nat-source-set-test",
+		Source:      fw.Network{Set: sourceSet},
+		Destination: fw.Network{Set: destSet},
+		Masquerade:  true,
+		Dynamic:     true,
+	}
+
+	require.Error(t, manager.AddNatRule(pair), "the destination set must fail to be created")
+
+	_, ok := manager.family4.ipsetCounter.Get(manager.family4.ipsetName(sourceSet.HashedName()))
+	require.False(t, ok, "the source set reference must be released")
+}
+
+// TestIptablesNatRuleReAddKeepsSetReferences re-adds the same NAT rule the way
+// a repeated network-map update does. The marking rule's set references must not
+// grow, or RemoveNatRule can never drop the count to zero and the set stays in
+// the kernel for the rest of the process lifetime.
+func TestIptablesNatRuleReAddKeepsSetReferences(t *testing.T) {
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	defer func() {
+		require.NoError(t, manager.Close(nil))
+	}()
+
+	set := fw.NewDomainSet(domain.List{"example.com"})
+	pair := fw.RouterPair{
+		ID:          "nat-reference-test",
+		Source:      fw.Network{Prefix: netip.MustParsePrefix("100.0.0.0/16")},
+		Destination: fw.Network{Set: set},
+		Masquerade:  true,
+		Dynamic:     true,
+	}
+
+	require.NoError(t, manager.AddNatRule(pair), "add nat rule")
+	name := manager.family4.ipsetName(set.HashedName())
+	first, ok := manager.family4.ipsetCounter.Get(name)
+	require.True(t, ok, "the marking rule must hold a reference to its set")
+
+	require.NoError(t, manager.AddNatRule(pair), "re-add nat rule")
+	second, ok := manager.family4.ipsetCounter.Get(name)
+	require.True(t, ok, "the set must still be referenced")
+	require.Equal(t, first.Count, second.Count, "re-adding the same rule must not add references")
+
+	require.NoError(t, manager.RemoveNatRule(pair), "remove nat rule")
+	_, ok = manager.family4.ipsetCounter.Get(name)
+	require.False(t, ok, "removing the rule must drop the last reference")
+}
+
+// TestIptablesRouteFilterIPSetFallback covers the route ACL side of the
+// fallback: with a destination set, the expanded per-source rules land
+// in the route forward chain and are all removed on delete.
+func TestIptablesRouteFilterIPSetFallback(t *testing.T) {
+	ipv4Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
+	require.NoError(t, err)
+
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	defer func() {
+		require.NoError(t, manager.Close(nil))
+	}()
+
+	manager.family4.ipsetSupported = false
+
+	sources := []netip.Prefix{
+		netip.MustParsePrefix("172.16.0.0/16"),
+		netip.MustParsePrefix("192.168.0.0/16"),
+	}
+	destination := fw.Network{Prefix: netip.MustParsePrefix("10.0.0.0/8")}
+	port := &fw.Port{Values: []uint16{443}}
+
+	rule, err := manager.AddFilterRule(nil, sources, destination, "tcp", nil, port, fw.ActionAccept)
+	require.NoError(t, err, "route ACL must install without ipset")
+
+	rr := rule.(*Rule)
+	require.Equal(t, chainRTFwdIn, rr.chain, "route rule must land in the forward chain")
+
+	all := rr.allSpecs()
+	require.Len(t, all, len(sources), "each source prefix needs its own rule")
+	for i, fs := range all {
+		joined := strings.Join(fs.specs, " ")
+		require.Contains(t, joined, "-s "+sources[i].String(), "fallback rule must match by source prefix")
+		require.NotContains(t, joined, matchSet, "fallback rule must not use ipset matching")
+		require.Nil(t, fs.mangleSpecs, "route rules have no mangle pairing")
+
+		checkRuleSpecs(t, ipv4Client, rr.chain, true, fs.specs...)
+	}
+
+	require.NoError(t, manager.DeleteFilterRule(rule), "failed to delete fallback rule")
+	for _, fs := range all {
+		checkRuleSpecs(t, ipv4Client, rr.chain, false, fs.specs...)
+	}
+}
+
+// TestIptablesCloseRemovesAllState exercises a spread of rule kinds and then
+// asserts Close puts every table it touches back exactly as it found it. A
+// leaked chain, jump, or ipset survives the daemon and nothing can remove it
+// afterwards, since the tracking that knew about it is gone.
+func TestIptablesCloseRemovesAllState(t *testing.T) {
+	ipv4Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
+	require.NoError(t, err)
+
+	before := snapshotIptables(t, ipv4Client)
+
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	// A failed assertion below returns before the Close under test, which would
+	// leave this test's chains and sets in the kernel for the next one.
+	t.Cleanup(func() {
+		if err := manager.Close(nil); err != nil {
+			t.Logf("close after failure: %v", err)
+		}
+	})
+
+	sources := []netip.Prefix{
+		netip.MustParsePrefix("10.20.0.42/32"),
+		netip.MustParsePrefix("10.20.0.43/32"),
+	}
+
+	// A multi-source peer rule: shared ipset plus the mangle redirect pairing.
+	_, err = manager.AddFilterRule(nil, sources, fw.Network{}, "tcp",
+		nil, &fw.Port{Values: []uint16{22}}, fw.ActionAccept)
+	require.NoError(t, err, "add peer rule")
+
+	// A route rule with a dynamic destination: a second set, in the forward chain.
+	_, err = manager.AddFilterRule(nil, sources,
+		fw.Network{Set: fw.NewDomainSet(domain.List{"example.com"})},
+		fw.ProtocolALL, nil, nil, fw.ActionDrop)
+	require.NoError(t, err, "add route rule")
+
+	// NAT marking for a routed destination, both directions.
+	pair := fw.RouterPair{
+		ID:          "cleanup-test",
+		Source:      fw.Network{Prefix: netip.MustParsePrefix("100.0.0.0/16")},
+		Destination: fw.Network{Prefix: netip.MustParsePrefix("192.168.55.0/24")},
+		Masquerade:  true,
+	}
+	require.NoError(t, manager.AddNatRule(pair), "add nat rule")
+	require.NoError(t, manager.EnableRouting(), "enable routing")
+
+	// A DNAT redirect, which also holds a forwarding reference.
+	dnat := fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{8080}},
+		TranslatedAddress: netip.MustParseAddr("10.20.0.44"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+	_, err = manager.AddDNATRule(dnat)
+	require.NoError(t, err, "add dnat rule")
+
+	require.NotEqual(t, before, snapshotIptables(t, ipv4Client), "the manager must have installed state")
+
+	// Everything above stays in place, so Close is what has to remove it.
+	require.NoError(t, manager.Close(nil), "close")
+
+	after := snapshotIptables(t, ipv4Client)
+	require.Equal(t, before.chains, after.chains, "Close must remove every chain it created")
+	require.Equal(t, before.rules, after.rules, "Close must remove every rule it created")
+	require.Equal(t, before.sets, after.sets, "Close must destroy every ipset it created")
+}
+
+// iptablesState is a snapshot of the tables the manager writes to, used to
+// compare the kernel before and after a manager lifetime.
+type iptablesState struct {
+	chains map[string][]string
+	rules  map[string][]string
+	sets   []string
+}
+
+func snapshotIptables(t *testing.T, client *iptables.IPTables) iptablesState {
+	t.Helper()
+
+	state := iptablesState{
+		chains: map[string][]string{},
+		rules:  map[string][]string{},
+	}
+
+	for _, table := range []string{tableFilter, tableNat, tableMangle, tableRaw} {
+		chains, err := client.ListChains(table)
+		require.NoErrorf(t, err, "list chains in %s", table)
+		slices.Sort(chains)
+		state.chains[table] = chains
+
+		for _, chain := range chains {
+			rules, err := client.List(table, chain)
+			require.NoErrorf(t, err, "list rules in %s/%s", table, chain)
+			state.rules[table+"/"+chain] = rules
+		}
+	}
+
+	sets, err := ipset.ListAll()
+	require.NoError(t, err, "list ipsets")
+	for _, set := range sets {
+		state.sets = append(state.sets, set.SetName)
+	}
+	slices.Sort(state.sets)
+
+	return state
+}
+
 func checkRuleSpecs(t *testing.T, ipv4Client *iptables.IPTables, chainName string, mustExists bool, rulespec ...string) {
 	t.Helper()
-	exists, err := ipv4Client.Exists("filter", chainName, rulespec...)
+	checkTableRuleSpecs(t, ipv4Client, tableFilter, chainName, mustExists, rulespec...)
+}
+
+func checkTableRuleSpecs(t *testing.T, ipv4Client *iptables.IPTables, table, chainName string, mustExists bool, rulespec ...string) {
+	t.Helper()
+	exists, err := ipv4Client.Exists(table, chainName, rulespec...)
 	require.NoError(t, err, "failed to check rule")
 	require.Falsef(t, !exists && mustExists, "rule '%v' does not exist", rulespec)
 	require.Falsef(t, exists && !mustExists, "rule '%v' exist", rulespec)
@@ -281,7 +604,7 @@ func TestIptablesCreatePerformance(t *testing.T) {
 			start := time.Now()
 			for i := 0; i < testMax; i++ {
 				port := &fw.Port{Values: []uint16{uint16(1000 + i)}}
-				_, err = manager.AddPeerFiltering(nil, ip.AsSlice(), "tcp", nil, port, fw.ActionAccept, "")
+				_, err = manager.AddFilterRule(nil, pfx(ip.AsSlice()), fw.Network{}, "tcp", nil, port, fw.ActionAccept)
 
 				require.NoError(t, err, "failed to add rule")
 			}

client/firewall/iptables/router_linux_test.go

@@ -1,4 +1,4 @@
-//go:build !android
+//go:build !android && privileged
 
 package iptables
 
@@ -31,7 +31,7 @@ func TestIptablesManager_RestoreOrCreateContainers(t *testing.T) {
 	iptablesClient, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
 	require.NoError(t, err, "failed to init iptables client")
 
-	manager, err := newRouter(iptablesClient, ifaceMock, iface.DefaultMTU)
+	manager, err := newFamily(iptablesClient, ifaceMock, iface.DefaultMTU)
 	require.NoError(t, err, "should return a valid iptables manager")
 	require.NoError(t, manager.init(nil))
 
@@ -52,12 +52,12 @@ func TestIptablesManager_RestoreOrCreateContainers(t *testing.T) {
 	// 11. MSS clamping rule for outbound traffic
 	require.Len(t, manager.rules, 11, "should have created rules map")
 
-	exists, err := manager.iptablesClient.Exists(tableNat, chainPOSTROUTING, "-j", chainRTNAT)
-	require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableNat, chainPOSTROUTING)
+	exists, err := manager.iptablesClient.Exists(tableNat, chainPostrouting, "-j", chainRTNAT)
+	require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableNat, chainPostrouting)
 	require.True(t, exists, "postrouting jump rule should exist")
 
-	exists, err = manager.iptablesClient.Exists(tableMangle, chainPREROUTING, "-j", chainRTPRE)
-	require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainPREROUTING)
+	exists, err = manager.iptablesClient.Exists(tableMangle, chainPrerouting, "-j", chainRTPre)
+	require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainPrerouting)
 	require.True(t, exists, "prerouting jump rule should exist")
 
 	pair := firewall.RouterPair{
@@ -84,7 +84,7 @@ func TestIptablesManager_AddNatRule(t *testing.T) {
 			iptablesClient, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
 			require.NoError(t, err, "failed to init iptables client")
 
-			manager, err := newRouter(iptablesClient, ifaceMock, iface.DefaultMTU)
+			manager, err := newFamily(iptablesClient, ifaceMock, iface.DefaultMTU)
 			require.NoError(t, err, "shouldn't return error")
 			require.NoError(t, manager.init(nil))
 
@@ -95,7 +95,7 @@ func TestIptablesManager_AddNatRule(t *testing.T) {
 			err = manager.AddNatRule(testCase.InputPair)
 			require.NoError(t, err, "marking rule should be inserted")
 
-			natRuleKey := firewall.GenKey(firewall.NatFormat, testCase.InputPair)
+			natRuleKey := testCase.InputPair.GenKey(firewall.NatFormat)
 			markingRule := []string{
 				"-i", ifaceMock.Name(),
 				"-m", "conntrack",
@@ -106,8 +106,8 @@ func TestIptablesManager_AddNatRule(t *testing.T) {
 				fmt.Sprintf("%#x", nbnet.PreroutingFwmarkMasquerade),
 			}
 
-			exists, err := iptablesClient.Exists(tableMangle, chainRTPRE, markingRule...)
-			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPRE)
+			exists, err := iptablesClient.Exists(tableMangle, chainRTPre, markingRule...)
+			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPre)
 			if testCase.InputPair.Masquerade {
 				require.True(t, exists, "marking rule should be created")
 				foundRule, found := manager.rules[natRuleKey]
@@ -121,7 +121,7 @@ func TestIptablesManager_AddNatRule(t *testing.T) {
 
 			// Check inverse rule
 			inversePair := firewall.GetInversePair(testCase.InputPair)
-			inverseRuleKey := firewall.GenKey(firewall.NatFormat, inversePair)
+			inverseRuleKey := inversePair.GenKey(firewall.NatFormat)
 			inverseMarkingRule := []string{
 				"!", "-i", ifaceMock.Name(),
 				"-m", "conntrack",
@@ -132,8 +132,8 @@ func TestIptablesManager_AddNatRule(t *testing.T) {
 				fmt.Sprintf("%#x", nbnet.PreroutingFwmarkMasqueradeReturn),
 			}
 
-			exists, err = iptablesClient.Exists(tableMangle, chainRTPRE, inverseMarkingRule...)
-			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPRE)
+			exists, err = iptablesClient.Exists(tableMangle, chainRTPre, inverseMarkingRule...)
+			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPre)
 			if testCase.InputPair.Masquerade {
 				require.True(t, exists, "inverse marking rule should be created")
 				foundRule, found := manager.rules[inverseRuleKey]
@@ -157,7 +157,7 @@ func TestIptablesManager_RemoveNatRule(t *testing.T) {
 		t.Run(testCase.Name, func(t *testing.T) {
 			iptablesClient, _ := iptables.NewWithProtocol(iptables.ProtocolIPv4)
 
-			manager, err := newRouter(iptablesClient, ifaceMock, iface.DefaultMTU)
+			manager, err := newFamily(iptablesClient, ifaceMock, iface.DefaultMTU)
 			require.NoError(t, err, "shouldn't return error")
 			require.NoError(t, manager.init(nil))
 			defer func() {
@@ -170,7 +170,7 @@ func TestIptablesManager_RemoveNatRule(t *testing.T) {
 			err = manager.RemoveNatRule(testCase.InputPair)
 			require.NoError(t, err, "shouldn't return error")
 
-			natRuleKey := firewall.GenKey(firewall.NatFormat, testCase.InputPair)
+			natRuleKey := testCase.InputPair.GenKey(firewall.NatFormat)
 			markingRule := []string{
 				"-i", ifaceMock.Name(),
 				"-m", "conntrack",
@@ -181,8 +181,8 @@ func TestIptablesManager_RemoveNatRule(t *testing.T) {
 				fmt.Sprintf("%#x", nbnet.PreroutingFwmarkMasquerade),
 			}
 
-			exists, err := iptablesClient.Exists(tableMangle, chainRTPRE, markingRule...)
-			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPRE)
+			exists, err := iptablesClient.Exists(tableMangle, chainRTPre, markingRule...)
+			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPre)
 			require.False(t, exists, "marking rule should not exist")
 
 			_, found := manager.rules[natRuleKey]
@@ -190,7 +190,7 @@ func TestIptablesManager_RemoveNatRule(t *testing.T) {
 
 			// Check inverse rule removal
 			inversePair := firewall.GetInversePair(testCase.InputPair)
-			inverseRuleKey := firewall.GenKey(firewall.NatFormat, inversePair)
+			inverseRuleKey := inversePair.GenKey(firewall.NatFormat)
 			inverseMarkingRule := []string{
 				"!", "-i", ifaceMock.Name(),
 				"-m", "conntrack",
@@ -201,8 +201,8 @@ func TestIptablesManager_RemoveNatRule(t *testing.T) {
 				fmt.Sprintf("%#x", nbnet.PreroutingFwmarkMasqueradeReturn),
 			}
 
-			exists, err = iptablesClient.Exists(tableMangle, chainRTPRE, inverseMarkingRule...)
-			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPRE)
+			exists, err = iptablesClient.Exists(tableMangle, chainRTPre, inverseMarkingRule...)
+			require.NoError(t, err, "should be able to query the iptables %s table and %s chain", tableMangle, chainRTPre)
 			require.False(t, exists, "inverse marking rule should not exist")
 
 			_, found = manager.rules[inverseRuleKey]
@@ -219,13 +219,13 @@ func TestRouter_AddRouteFiltering(t *testing.T) {
 	iptablesClient, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
 	require.NoError(t, err, "Failed to create iptables client")
 
-	r, err := newRouter(iptablesClient, ifaceMock, iface.DefaultMTU)
-	require.NoError(t, err, "Failed to create router manager")
+	r, err := newFamily(iptablesClient, ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err, "Failed to create family manager")
 	require.NoError(t, r.init(nil))
 
 	defer func() {
 		err := r.Reset()
-		require.NoError(t, err, "Failed to reset router")
+		require.NoError(t, err, "Failed to reset family")
 	}()
 
 	tests := []struct {
@@ -334,62 +334,30 @@ func TestRouter_AddRouteFiltering(t *testing.T) {
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			ruleKey, err := r.AddRouteFiltering(nil, tt.sources, firewall.Network{Prefix: tt.destination}, tt.proto, tt.sPort, tt.dPort, tt.action)
-			require.NoError(t, err, "AddRouteFiltering failed")
+			ruleKey, err := r.AddFilterRule(nil, tt.sources, firewall.Network{Prefix: tt.destination}, tt.proto, tt.sPort, tt.dPort, tt.action)
+			require.NoError(t, err, "AddFilterRule failed")
 
-			// Check if the rule is in the internal map
-			rule, ok := r.rules[ruleKey.ID()]
-			assert.True(t, ok, "Rule not found in internal map")
+			stored, ok := r.filters[ruleKey.ID()]
+			require.True(t, ok, "rule not stored in filters")
+			t.Logf("Internal rule: %v", stored.specs)
 
-			// Log the internal rule
-			t.Logf("Internal rule: %v", rule)
-
-			// Check if the rule exists in iptables
-			exists, err := iptablesClient.Exists(tableFilter, chainRTFWDIN, rule...)
+			exists, err := iptablesClient.Exists(tableFilter, chainRTFwdIn, stored.specs...)
 			assert.NoError(t, err, "Failed to check rule existence")
 			assert.True(t, exists, "Rule not found in iptables")
 
-			var source firewall.Network
-			if len(tt.sources) > 1 {
-				source.Set = firewall.NewPrefixSet(tt.sources)
-			} else if len(tt.sources) > 0 {
-				source.Prefix = tt.sources[0]
-			}
-			// Verify rule content
-			params := routeFilteringRuleParams{
-				Source:      source,
-				Destination: firewall.Network{Prefix: tt.destination},
-				Proto:       tt.proto,
-				SPort:       tt.sPort,
-				DPort:       tt.dPort,
-				Action:      tt.action,
-			}
-
-			expectedRule, err := r.genRouteRuleSpec(params, nil)
-			require.NoError(t, err, "Failed to generate expected rule spec")
-
 			if tt.expectSet {
 				setName := firewall.NewPrefixSet(tt.sources).HashedName()
-				expectedRule, err = r.genRouteRuleSpec(params, nil)
-				require.NoError(t, err, "Failed to generate expected rule spec with set")
-
-				// Check if the set was created
 				_, exists := r.ipsetCounter.Get(setName)
 				assert.True(t, exists, "IPSet not created")
+				assert.NotEmpty(t, findSets(stored.specs), "Rule should reference an ipset")
 			}
 
-			assert.Equal(t, expectedRule, rule, "Rule content mismatch")
-
-			// Clean up
-			err = r.DeleteRouteRule(ruleKey)
-			require.NoError(t, err, "Failed to delete rule")
+			require.NoError(t, r.DeleteFilterRule(ruleKey), "Failed to delete rule")
 		})
 	}
 }
 
 func TestFindSetNameInRule(t *testing.T) {
-	r := &router{}
-
 	testCases := []struct {
 		name     string
 		rule     []string
@@ -430,7 +398,7 @@ func TestFindSetNameInRule(t *testing.T) {
 
 	for _, tc := range testCases {
 		t.Run(tc.name, func(t *testing.T) {
-			result := r.findSets(tc.rule)
+			result := findSets(tc.rule)
 
 			if len(result) != len(tc.expected) {
 				t.Errorf("Expected %d sets, got %d. Sets found: %v", len(tc.expected), len(result), result)

client/firewall/iptables/routing_linux.go

@@ -0,0 +1,273 @@
+//go:build !android
+
+package iptables
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+	nbnet "github.com/netbirdio/netbird/client/net"
+)
+
+func (r *family) AddNatRule(pair firewall.RouterPair) error {
+	if r.legacyManagement {
+		log.Warnf("This peer is connected to a NetBird Management service with an older version. Allowing all traffic for %s", pair.Destination)
+		if err := r.addLegacyRouteRule(pair); err != nil {
+			return fmt.Errorf("add legacy routing rule: %w", err)
+		}
+	}
+
+	if pair.Masquerade {
+		if err := r.addNatRule(pair); err != nil {
+			return fmt.Errorf("add nat rule: %w", err)
+		}
+
+		if err := r.addNatRule(firewall.GetInversePair(pair)); err != nil {
+			return fmt.Errorf("add inverse nat rule: %w", err)
+		}
+	}
+
+	r.updateState()
+
+	return nil
+}
+
+// RemoveNatRule removes an iptables rule pair from forwarding and nat chains
+func (r *family) RemoveNatRule(pair firewall.RouterPair) error {
+	if pair.Masquerade {
+		if err := r.removeNatRule(pair); err != nil {
+			return fmt.Errorf("remove nat rule: %w", err)
+		}
+
+		if err := r.removeNatRule(firewall.GetInversePair(pair)); err != nil {
+			return fmt.Errorf("remove inverse nat rule: %w", err)
+		}
+	}
+
+	if err := r.removeLegacyRouteRule(pair); err != nil {
+		return fmt.Errorf("remove legacy routing rule: %w", err)
+	}
+
+	r.updateState()
+
+	return nil
+}
+
+// addLegacyRouteRule adds a legacy routing rule for mgmt servers pre route acls
+func (r *family) addLegacyRouteRule(pair firewall.RouterPair) error {
+	ruleID := pair.GenKey(firewall.ForwardingFormat)
+
+	if err := r.removeLegacyRouteRule(pair); err != nil {
+		return err
+	}
+
+	rule := []string{"-s", pair.Source.String(), "-d", pair.Destination.String(), "-j", "ACCEPT"}
+	if err := r.iptablesClient.Append(tableFilter, chainRTFwdIn, rule...); err != nil {
+		return fmt.Errorf("add legacy forwarding rule %s -> %s: %w", pair.Source, pair.Destination, err)
+	}
+
+	r.rules[ruleID] = rule
+
+	return nil
+}
+
+func (r *family) removeLegacyRouteRule(pair firewall.RouterPair) error {
+	ruleID := pair.GenKey(firewall.ForwardingFormat)
+
+	if rule, exists := r.rules[ruleID]; exists {
+		if err := r.iptablesClient.DeleteIfExists(tableFilter, chainRTFwdIn, rule...); err != nil {
+			return fmt.Errorf("remove legacy forwarding rule %s -> %s: %w", pair.Source, pair.Destination, err)
+		}
+		delete(r.rules, ruleID)
+
+		if err := r.decrementSetCounter(rule); err != nil {
+			return fmt.Errorf("decrement ipset counter: %w", err)
+		}
+	}
+
+	return nil
+}
+
+// GetLegacyManagement returns the current legacy management mode
+func (r *family) GetLegacyManagement() bool {
+	return r.legacyManagement
+}
+
+// SetLegacyManagement sets the route manager to use legacy management mode
+func (r *family) SetLegacyManagement(isLegacy bool) {
+	r.legacyManagement = isLegacy
+}
+
+// RemoveAllLegacyRouteRules removes all legacy routing rules for mgmt servers pre route acls
+func (r *family) RemoveAllLegacyRouteRules() error {
+	var merr *multierror.Error
+	for k, rule := range r.rules {
+		if !strings.HasPrefix(string(k), firewall.ForwardingFormatPrefix) {
+			continue
+		}
+		if err := r.iptablesClient.DeleteIfExists(tableFilter, chainRTFwdIn, rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove legacy forwarding rule: %w", err))
+		} else {
+			delete(r.rules, k)
+		}
+	}
+
+	r.updateState()
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) addPostroutingRules() error {
+	// First rule for outbound masquerade
+	rule1 := []string{
+		"-m", "mark", "--mark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkMasquerade),
+		"!", "-o", "lo",
+		"-j", "MASQUERADE",
+	}
+	if err := r.iptablesClient.Append(tableNat, chainRTNAT, rule1...); err != nil {
+		return fmt.Errorf("add outbound masquerade rule: %w", err)
+	}
+	r.rules["static-nat-outbound"] = rule1
+
+	// Second rule for return traffic masquerade
+	rule2 := []string{
+		"-m", "mark", "--mark", fmt.Sprintf("%#x", nbnet.PreroutingFwmarkMasqueradeReturn),
+		"-o", r.wgIface.Name(),
+		"-j", "MASQUERADE",
+	}
+	if err := r.iptablesClient.Append(tableNat, chainRTNAT, rule2...); err != nil {
+		return fmt.Errorf("add return masquerade rule: %w", err)
+	}
+	r.rules["static-nat-return"] = rule2
+
+	return nil
+}
+
+// addMSSClampingRules adds MSS clamping rules to prevent fragmentation for forwarded traffic.
+func (r *family) addMSSClampingRules() error {
+	overhead := uint16(ipv4TCPHeaderSize)
+	if r.v6 {
+		overhead = ipv6TCPHeaderSize
+	}
+	mss := r.mtu - overhead
+
+	// Add jump rule from FORWARD chain in mangle table to our custom chain
+	jumpRule := jumpRuleSpec(chainRTMSSClamp)
+	if err := r.iptablesClient.Insert(tableMangle, chainForward, 1, jumpRule...); err != nil {
+		return fmt.Errorf("add jump to MSS clamp chain: %w", err)
+	}
+	r.rules[jumpMSSClamp] = jumpRule
+
+	ruleOut := []string{
+		"-o", r.wgIface.Name(),
+		"-p", "tcp",
+		"--tcp-flags", "SYN,RST", "SYN",
+		"-j", "TCPMSS",
+		"--set-mss", fmt.Sprintf("%d", mss),
+	}
+	if err := r.iptablesClient.Append(tableMangle, chainRTMSSClamp, ruleOut...); err != nil {
+		return fmt.Errorf("add outbound MSS clamp rule: %w", err)
+	}
+	r.rules["mss-clamp-out"] = ruleOut
+
+	return nil
+}
+
+func (r *family) insertEstablishedRule(chain string) error {
+	establishedRule := getConntrackEstablished()
+
+	err := r.iptablesClient.Insert(tableFilter, chain, 1, establishedRule...)
+	if err != nil {
+		return fmt.Errorf("insert established rule: %w", err)
+	}
+
+	ruleID := firewall.RuleID("established-" + chain)
+	r.rules[ruleID] = establishedRule
+
+	return nil
+}
+
+func (r *family) addNatRule(pair firewall.RouterPair) (err error) {
+	ruleID := pair.GenKey(firewall.NatFormat)
+
+	if rule, exists := r.rules[ruleID]; exists {
+		if derr := r.iptablesClient.DeleteIfExists(tableMangle, chainRTPre, rule...); derr != nil {
+			return fmt.Errorf("remove existing marking rule for %s: %w", pair.Destination, derr)
+		}
+		delete(r.rules, ruleID)
+
+		// Drop the replaced spec's set references only once the new spec has
+		// taken its own, so a set both specs share is not destroyed and
+		// recreated, which would lose the prefixes UpdateSet put in it.
+		defer func() {
+			if derr := r.decrementSetCounter(rule); derr != nil && err == nil {
+				err = fmt.Errorf("decrement ipset counter: %w", derr)
+			}
+		}()
+	}
+
+	markValue := nbnet.PreroutingFwmarkMasquerade
+	if pair.Inverse {
+		markValue = nbnet.PreroutingFwmarkMasqueradeReturn
+	}
+
+	rule := []string{"-i", r.wgIface.Name()}
+	if pair.Inverse {
+		rule = []string{"!", "-i", r.wgIface.Name()}
+	}
+
+	rule = append(rule,
+		"-m", "conntrack",
+		"--ctstate", "NEW",
+	)
+	sourceExp, err := r.applyNetwork("-s", pair.Source, nil)
+	if err != nil {
+		return fmt.Errorf("apply network -s: %w", err)
+	}
+	destExp, err := r.applyNetwork("-d", pair.Destination, nil)
+	if err != nil {
+		r.dropSourceMatch(sourceExp)
+		return fmt.Errorf("apply network -d: %w", err)
+	}
+
+	rule = append(rule, sourceExp...)
+	rule = append(rule, destExp...)
+	rule = append(rule,
+		"-j", "MARK", "--set-mark", fmt.Sprintf("%#x", markValue),
+	)
+
+	// Ensure nat rules come first, so the mark can be overwritten.
+	// Currently overwritten by the dst-type LOCAL rules for redirected traffic.
+	if err := r.iptablesClient.Insert(tableMangle, chainRTPre, 1, rule...); err != nil {
+		r.dropSourceMatch(rule)
+		return fmt.Errorf("add marking rule for %s: %w", pair.Destination, err)
+	}
+
+	r.rules[ruleID] = rule
+
+	return nil
+}
+
+func (r *family) removeNatRule(pair firewall.RouterPair) error {
+	ruleID := pair.GenKey(firewall.NatFormat)
+
+	if rule, exists := r.rules[ruleID]; exists {
+		if err := r.iptablesClient.DeleteIfExists(tableMangle, chainRTPre, rule...); err != nil {
+			return fmt.Errorf("remove marking rule for %s: %w", pair.Destination, err)
+		}
+		delete(r.rules, ruleID)
+
+		if err := r.decrementSetCounter(rule); err != nil {
+			return fmt.Errorf("decrement ipset counter: %w", err)
+		}
+	} else {
+		log.Debugf("marking rule %s not found", ruleID)
+	}
+
+	return nil
+}

client/firewall/iptables/rule.go

@@ -1,17 +1,37 @@
 package iptables
 
-// Rule to handle management of rules
-type Rule struct {
-	ruleID    string
-	ipsetName string
+import "github.com/netbirdio/netbird/client/firewall/manager"
 
+// Rule to handle management of rules. Source set membership (when the
+// rule was built against a shared hash:net ipset) is encoded in specs;
+// DeleteFilterRule recovers it via findSets so the refcounter can drop
+// the right reference.
+type Rule struct {
+	id          manager.RuleID
 	specs       []string
 	mangleSpecs []string
-	ip          string
-	chain       string
+	// extraRules holds the rules beyond the first when the ipset
+	// fallback expands a multi-source rule into one rule per prefix.
+	extraRules []filterSpecs
+	chain      string
+	v6         bool
 }
 
-// GetRuleID returns the rule id
-func (r *Rule) ID() string {
-	return r.ruleID
+// filterSpecs is one installed iptables rule: its filter-table spec and
+// the paired mangle redirect-mark spec (nil for route rules or when the
+// mangle rule could not be added).
+type filterSpecs struct {
+	specs       []string
+	mangleSpecs []string
+}
+
+// allSpecs returns the spec pairs of every iptables rule backing this
+// Rule, the primary one first.
+func (r *Rule) allSpecs() []filterSpecs {
+	return append([]filterSpecs{{specs: r.specs, mangleSpecs: r.mangleSpecs}}, r.extraRules...)
+}
+
+// ID returns the rule id
+func (r *Rule) ID() manager.RuleID {
+	return r.id
 }

client/firewall/iptables/rulestore_linux.go

@@ -1,103 +0,0 @@
-package iptables
-
-import "encoding/json"
-
-type ipList struct {
-	ips map[string]struct{}
-}
-
-func newIpList(ip string) *ipList {
-	ips := make(map[string]struct{})
-	ips[ip] = struct{}{}
-
-	return &ipList{
-		ips: ips,
-	}
-}
-
-func (s *ipList) addIP(ip string) {
-	s.ips[ip] = struct{}{}
-}
-
-// MarshalJSON implements json.Marshaler
-func (s *ipList) MarshalJSON() ([]byte, error) {
-	return json.Marshal(struct {
-		IPs map[string]struct{} `json:"ips"`
-	}{
-		IPs: s.ips,
-	})
-}
-
-// UnmarshalJSON implements json.Unmarshaler
-func (s *ipList) UnmarshalJSON(data []byte) error {
-	temp := struct {
-		IPs map[string]struct{} `json:"ips"`
-	}{}
-	if err := json.Unmarshal(data, &temp); err != nil {
-		return err
-	}
-	s.ips = temp.IPs
-
-	if temp.IPs == nil {
-		temp.IPs = make(map[string]struct{})
-	}
-
-	return nil
-}
-
-type ipsetStore struct {
-	ipsets map[string]*ipList
-}
-
-func newIpsetStore() *ipsetStore {
-	return &ipsetStore{
-		ipsets: make(map[string]*ipList),
-	}
-}
-
-func (s *ipsetStore) ipset(ipsetName string) (*ipList, bool) {
-	r, ok := s.ipsets[ipsetName]
-	return r, ok
-}
-
-func (s *ipsetStore) addIpList(ipsetName string, list *ipList) {
-	s.ipsets[ipsetName] = list
-}
-
-func (s *ipsetStore) deleteIpset(ipsetName string) {
-	delete(s.ipsets, ipsetName)
-}
-
-func (s *ipsetStore) ipsetNames() []string {
-	names := make([]string, 0, len(s.ipsets))
-	for name := range s.ipsets {
-		names = append(names, name)
-	}
-	return names
-}
-
-// MarshalJSON implements json.Marshaler
-func (s *ipsetStore) MarshalJSON() ([]byte, error) {
-	return json.Marshal(struct {
-		IPSets map[string]*ipList `json:"ipsets"`
-	}{
-		IPSets: s.ipsets,
-	})
-}
-
-// UnmarshalJSON implements json.Unmarshaler
-func (s *ipsetStore) UnmarshalJSON(data []byte) error {
-	temp := struct {
-		IPSets map[string]*ipList `json:"ipsets"`
-	}{}
-	if err := json.Unmarshal(data, &temp); err != nil {
-		return err
-	}
-	s.ipsets = temp.IPSets
-
-	if temp.IPSets == nil {
-		temp.IPSets = make(map[string]*ipList)
-	}
-
-	return nil
-}

client/firewall/iptables/state_linux.go

@@ -4,6 +4,8 @@ import (
 	"fmt"
 	"sync"
 
+	log "github.com/sirupsen/logrus"
+
 	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/iface/wgaddr"
 )
@@ -27,11 +29,13 @@ type ShutdownState struct {
 
 	InterfaceState *InterfaceState `json:"interface_state,omitempty"`
 
-	RouteRules        routeRules    `json:"route_rules,omitempty"`
-	RouteIPsetCounter *ipsetCounter `json:"route_ipset_counter,omitempty"`
+	RouteRules         routeRules    `json:"route_rules,omitempty"`
+	RouteRules6        routeRules    `json:"route_rules_v6,omitempty"`
+	RouteIPsetCounter  *ipsetCounter `json:"route_ipset_counter,omitempty"`
+	RouteIPsetCounter6 *ipsetCounter `json:"route_ipset_counter_v6,omitempty"`
 
-	ACLEntries    aclEntries  `json:"acl_entries,omitempty"`
-	ACLIPsetStore *ipsetStore `json:"acl_ipset_store,omitempty"`
+	ACLEntries  aclEntries `json:"acl_entries,omitempty"`
+	ACLEntries6 aclEntries `json:"acl_entries_v6,omitempty"`
 }
 
 func (s *ShutdownState) Name() string {
@@ -49,17 +53,33 @@ func (s *ShutdownState) Cleanup() error {
 	}
 
 	if s.RouteRules != nil {
-		ipt.router.rules = s.RouteRules
+		ipt.family4.rules = s.RouteRules
 	}
 	if s.RouteIPsetCounter != nil {
-		ipt.router.ipsetCounter.LoadData(s.RouteIPsetCounter)
+		ipt.family4.ipsetCounter.LoadData(s.RouteIPsetCounter)
 	}
 
 	if s.ACLEntries != nil {
-		ipt.aclMgr.entries = s.ACLEntries
+		ipt.family4.entries = s.ACLEntries
 	}
-	if s.ACLIPsetStore != nil {
-		ipt.aclMgr.ipsetStore = s.ACLIPsetStore
+
+	// Clean up v6 state even if the current run has no IPv6.
+	// The previous run may have left ip6tables rules behind.
+	if !ipt.hasIPv6() {
+		if err := ipt.createIPv6Components(s.InterfaceState, mtu); err != nil {
+			log.Warnf("failed to create v6 components for cleanup: %v", err)
+		}
+	}
+	if ipt.hasIPv6() {
+		if s.RouteRules6 != nil {
+			ipt.family6.rules = s.RouteRules6
+		}
+		if s.RouteIPsetCounter6 != nil {
+			ipt.family6.ipsetCounter.LoadData(s.RouteIPsetCounter6)
+		}
+		if s.ACLEntries6 != nil {
+			ipt.family6.entries = s.ACLEntries6
+		}
 	}
 
 	if err := ipt.Close(nil); err != nil {

client/firewall/iptables/testhelpers_linux_test.go

@@ -0,0 +1,27 @@
+//go:build privileged
+
+package iptables
+
+import (
+	"fmt"
+	"net"
+	"net/netip"
+)
+
+func pfx(ip net.IP) []netip.Prefix {
+	if ip == nil {
+		return []netip.Prefix{netip.PrefixFrom(netip.IPv4Unspecified(), 0)}
+	}
+	if ip.IsUnspecified() {
+		if ip.To4() != nil {
+			return []netip.Prefix{netip.PrefixFrom(netip.IPv4Unspecified(), 0)}
+		}
+		return []netip.Prefix{netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+	}
+	a, ok := netip.AddrFromSlice(ip)
+	if !ok {
+		panic(fmt.Sprintf("invalid IP length: %d", len(ip)))
+	}
+	a = a.Unmap()
+	return []netip.Prefix{netip.PrefixFrom(a, a.BitLen())}
+}

client/firewall/manager/firewall.go

@@ -1,8 +1,8 @@
 package manager
 
 import (
+	"errors"
 	"fmt"
-	"net"
 	"net/netip"
 	"sort"
 
@@ -11,6 +11,16 @@ import (
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 )
 
+// ErrIPv6NotInitialized is returned when an IPv6 address is passed to a firewall
+// method but the IPv6 firewall components were not initialized.
+var ErrIPv6NotInitialized = errors.New("IPv6 firewall not initialized")
+
+// ErrNoSources is returned when AddFilterRule is called with an empty
+// source list. "Match any source" must be expressed explicitly with a
+// /0 prefix; an empty list is a caller error and is rejected rather
+// than silently widening the rule to every source.
+var ErrNoSources = errors.New("rule has no sources")
+
 const (
 	ForwardingFormatPrefix = "netbird-fwd-"
 	ForwardingFormat       = "netbird-fwd-%s-%t"
@@ -18,13 +28,18 @@ const (
 	NatFormat              = "netbird-nat-%s-%t"
 )
 
+// RuleID identifies a firewall rule. It is a typed string so the
+// compiler catches accidental mixing with arbitrary string keys. It is
+// only an identifier and does not implement Rule.
+type RuleID string
+
 // Rule abstraction should be implemented by each firewall manager
 //
 // Each firewall type for different OS can use different type
 // of the properties to hold data of the created rule
 type Rule interface {
 	// ID returns the rule id
-	ID() string
+	ID() RuleID
 }
 
 // RuleDirection is the traffic direction which a rule is applied
@@ -86,6 +101,13 @@ func (d Network) IsPrefix() bool {
 	return d.Prefix.IsValid()
 }
 
+// IsZero returns true if the network designates no destination, i.e. it
+// is the zero value. A zero Network is the peer-rule sentinel; a non-zero
+// one carries a prefix or set destination.
+func (d Network) IsZero() bool {
+	return !d.IsPrefix() && !d.IsSet()
+}
+
 // Manager is the high level abstraction of a firewall manager
 //
 // It declares methods which handle actions required by the
@@ -93,46 +115,42 @@ func (d Network) IsPrefix() bool {
 type Manager interface {
 	Init(stateManager *statemanager.Manager) error
 
-	// AllowNetbird allows netbird interface traffic
-	AllowNetbird() error
-
-	// AddPeerFiltering adds a rule to the firewall
+	// AddFilterRule adds a packet-filtering rule to the firewall.
 	//
-	// If comment argument is empty firewall manager should set
-	// rule ID as comment for the rule
+	// If destination is the zero Network, the rule applies to traffic
+	// inbound to this node, i.e. peer ACL semantics, installed in
+	// the kernel's input chain. If destination is set (prefix or
+	// set), the rule applies to forwarded traffic with that
+	// destination, route ACL semantics, installed in the forward
+	// chain.
 	//
-	// Note: Callers should call Flush() after adding rules to ensure
-	// they are applied to the kernel and rule handles are refreshed.
-	AddPeerFiltering(
+	// sources must be a single address family; the caller splits mixed
+	// families and calls once per family. "Match any source" must be
+	// expressed with an explicit /0 prefix; an empty sources list is
+	// rejected with ErrNoSources so a zeroed list can never widen a
+	// rule to every source.
+	//
+	// Note: callers should call Flush() after adding rules.
+	AddFilterRule(
 		id []byte,
-		ip net.IP,
+		sources []netip.Prefix,
+		destination Network,
 		proto Protocol,
 		sPort *Port,
 		dPort *Port,
 		action Action,
-		ipsetName string,
-	) ([]Rule, error)
+	) (Rule, error)
 
-	// DeletePeerRule from the firewall by rule definition
-	DeletePeerRule(rule Rule) error
+	// DeleteFilterRule removes a filtering rule previously added via
+	// AddFilterRule. The rule's own type identifies whether it lives
+	// in the peer (input) or route (forward) path.
+	DeleteFilterRule(rule Rule) error
 
 	// IsServerRouteSupported returns true if the firewall supports server side routing operations
 	IsServerRouteSupported() bool
 
 	IsStateful() bool
 
-	AddRouteFiltering(
-		id []byte,
-		sources []netip.Prefix,
-		destination Network,
-		proto Protocol,
-		sPort, dPort *Port,
-		action Action,
-	) (Rule, error)
-
-	// DeleteRouteRule deletes a routing rule
-	DeleteRouteRule(rule Rule) error
-
 	// AddNatRule inserts a routing NAT rule
 	AddNatRule(pair RouterPair) error
 
@@ -164,26 +182,25 @@ type Manager interface {
 	UpdateSet(hash Set, prefixes []netip.Prefix) error
 
 	// AddInboundDNAT adds an inbound DNAT rule redirecting traffic from NetBird peers to local services
-	AddInboundDNAT(localAddr netip.Addr, protocol Protocol, sourcePort, targetPort uint16) error
+	AddInboundDNAT(localAddr netip.Addr, protocol Protocol, originalPort, translatedPort uint16) error
 
 	// RemoveInboundDNAT removes inbound DNAT rule
-	RemoveInboundDNAT(localAddr netip.Addr, protocol Protocol, sourcePort, targetPort uint16) error
+	RemoveInboundDNAT(localAddr netip.Addr, protocol Protocol, originalPort, translatedPort uint16) error
 
 	// AddOutputDNAT adds an OUTPUT chain DNAT rule for locally-generated traffic.
-	// localAddr must be IPv4; the underlying iptables/nftables backends are IPv4-only.
-	AddOutputDNAT(localAddr netip.Addr, protocol Protocol, sourcePort, targetPort uint16) error
+	AddOutputDNAT(localAddr netip.Addr, protocol Protocol, originalPort, translatedPort uint16) error
 
 	// RemoveOutputDNAT removes an OUTPUT chain DNAT rule.
-	// localAddr must be IPv4; the underlying iptables/nftables backends are IPv4-only.
-	RemoveOutputDNAT(localAddr netip.Addr, protocol Protocol, sourcePort, targetPort uint16) error
+	RemoveOutputDNAT(localAddr netip.Addr, protocol Protocol, originalPort, translatedPort uint16) error
 
 	// SetupEBPFProxyNoTrack creates static notrack rules for eBPF proxy loopback traffic.
 	// This prevents conntrack from interfering with WireGuard proxy communication.
 	SetupEBPFProxyNoTrack(proxyPort, wgPort uint16) error
 }
 
-func GenKey(format string, pair RouterPair) string {
-	return fmt.Sprintf(format, pair.ID, pair.Inverse)
+// GenKey builds the rule id for this pair from the given format.
+func (p RouterPair) GenKey(format string) RuleID {
+	return RuleID(fmt.Sprintf(format, p.ID, p.Inverse))
 }
 
 // LegacyManager defines the interface for legacy management operations
@@ -239,6 +256,20 @@ func MergeIPRanges(prefixes []netip.Prefix) []netip.Prefix {
 	return merged
 }
 
+// UnmapPrefix normalizes a v4-mapped v6 prefix (::ffff:a.b.c.d) to its
+// plain v4 form, shifting the prefix length out of the 96-bit mapped
+// range. Other prefixes are returned unchanged. Keeping prefixes
+// unmapped ensures v4 rules match consistently and the match builders
+// read the correct address length.
+func UnmapPrefix(p netip.Prefix) netip.Prefix {
+	addr := p.Addr()
+	if !addr.Is4In6() {
+		return p
+	}
+	bits := max(p.Bits()-96, 0)
+	return netip.PrefixFrom(addr.Unmap(), bits)
+}
+
 // SortPrefixes sorts the given slice of netip.Prefix in place.
 // It sorts first by IP address, then by prefix length (most specific to least specific).
 func SortPrefixes(prefixes []netip.Prefix) {

client/firewall/manager/forward_rule.go

@@ -13,13 +13,13 @@ type ForwardRule struct {
 	TranslatedPort    Port
 }
 
-func (r ForwardRule) ID() string {
+func (r ForwardRule) ID() RuleID {
 	id := fmt.Sprintf("%s;%s;%s;%s",
 		r.Protocol,
 		r.DestinationPort.String(),
 		r.TranslatedAddress.String(),
 		r.TranslatedPort.String())
-	return id
+	return RuleID(id)
 }
 
 func (r ForwardRule) String() string {

client/firewall/manager/routerpair.go

@@ -1,6 +1,8 @@
 package manager
 
 import (
+	"net/netip"
+
 	"github.com/netbirdio/netbird/route"
 )
 
@@ -10,6 +12,10 @@ type RouterPair struct {
 	Destination Network
 	Masquerade  bool
 	Inverse     bool
+	// Dynamic indicates the route is domain-based. NAT rules for dynamic
+	// routes are duplicated to the v6 table so that resolved AAAA records
+	// are masqueraded correctly.
+	Dynamic bool
 }
 
 func GetInversePair(pair RouterPair) RouterPair {
@@ -20,5 +26,17 @@ func GetInversePair(pair RouterPair) RouterPair {
 		Destination: pair.Source,
 		Masquerade:  pair.Masquerade,
 		Inverse:     true,
+		Dynamic:     pair.Dynamic,
 	}
 }
+
+// ToV6NatPair creates a v6 counterpart of a v4 NAT pair with `::/0` source
+// and, for prefix destinations, `::/0` destination.
+func ToV6NatPair(pair RouterPair) RouterPair {
+	v6 := pair
+	v6.Source = Network{Prefix: netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+	if v6.Destination.IsPrefix() {
+		v6.Destination = Network{Prefix: netip.PrefixFrom(netip.IPv6Unspecified(), 0)}
+	}
+	return v6
+}

client/firewall/manager/set.go

@@ -40,7 +40,7 @@ func (h Set) Comment() string {
 
 // NewPrefixSet generates a unique name for an ipset based on the given prefixes.
 func NewPrefixSet(prefixes []netip.Prefix) Set {
-	// sort for consistent naming
+	prefixes = slices.Clone(prefixes)
 	SortPrefixes(prefixes)
 
 	hash := sha256.New()

client/firewall/nftables/acl_linux.go

@@ -1,721 +0,0 @@
-package nftables
-
-import (
-	"bytes"
-	"fmt"
-	"net"
-	"slices"
-	"strconv"
-	"strings"
-	"time"
-
-	"github.com/google/nftables"
-	"github.com/google/nftables/binaryutil"
-	"github.com/google/nftables/expr"
-	log "github.com/sirupsen/logrus"
-	"golang.org/x/sys/unix"
-
-	firewall "github.com/netbirdio/netbird/client/firewall/manager"
-	nbnet "github.com/netbirdio/netbird/client/net"
-)
-
-const (
-
-	// rules chains contains the effective ACL rules
-	chainNameInputRules = "netbird-acl-input-rules"
-
-	// filter chains contains the rules that jump to the rules chains
-	chainNameInputFilter       = "netbird-acl-input-filter"
-	chainNameForwardFilter     = "netbird-acl-forward-filter"
-	chainNameManglePrerouting  = "netbird-mangle-prerouting"
-	chainNameManglePostrouting = "netbird-mangle-postrouting"
-)
-
-const flushError = "flush: %w"
-
-var (
-	anyIP = []byte{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
-)
-
-type AclManager struct {
-	rConn              *nftables.Conn
-	sConn              *nftables.Conn
-	wgIface            iFaceMapper
-	routingFwChainName string
-
-	workTable       *nftables.Table
-	chainInputRules *nftables.Chain
-	chainPrerouting *nftables.Chain
-
-	ipsetStore *ipsetStore
-	rules      map[string]*Rule
-}
-
-func newAclManager(table *nftables.Table, wgIface iFaceMapper, routingFwChainName string) (*AclManager, error) {
-	// sConn is used for creating sets and adding/removing elements from them
-	// it's differ then rConn (which does create new conn for each flush operation)
-	// and is permanent. Using same connection for both type of operations
-	// overloads netlink with high amount of rules ( > 10000)
-	sConn, err := nftables.New(nftables.AsLasting())
-	if err != nil {
-		return nil, fmt.Errorf("create nf conn: %w", err)
-	}
-
-	return &AclManager{
-		rConn:              &nftables.Conn{},
-		sConn:              sConn,
-		wgIface:            wgIface,
-		workTable:          table,
-		routingFwChainName: routingFwChainName,
-
-		ipsetStore: newIpsetStore(),
-		rules:      make(map[string]*Rule),
-	}, nil
-}
-
-func (m *AclManager) init(workTable *nftables.Table) error {
-	m.workTable = workTable
-	return m.createDefaultChains()
-}
-
-// AddPeerFiltering rule to the firewall
-//
-// If comment argument is empty firewall manager should set
-// rule ID as comment for the rule
-func (m *AclManager) AddPeerFiltering(
-	id []byte,
-	ip net.IP,
-	proto firewall.Protocol,
-	sPort *firewall.Port,
-	dPort *firewall.Port,
-	action firewall.Action,
-	ipsetName string,
-) ([]firewall.Rule, error) {
-	var ipset *nftables.Set
-	if ipsetName != "" {
-		var err error
-		ipset, err = m.addIpToSet(ipsetName, ip)
-		if err != nil {
-			return nil, err
-		}
-	}
-
-	newRules := make([]firewall.Rule, 0, 2)
-	ioRule, err := m.addIOFiltering(ip, proto, sPort, dPort, action, ipset)
-	if err != nil {
-		return nil, err
-	}
-
-	newRules = append(newRules, ioRule)
-	return newRules, nil
-}
-
-// DeletePeerRule from the firewall by rule definition
-func (m *AclManager) DeletePeerRule(rule firewall.Rule) error {
-	r, ok := rule.(*Rule)
-	if !ok {
-		return fmt.Errorf("invalid rule type")
-	}
-
-	if r.nftSet == nil {
-		if err := m.rConn.DelRule(r.nftRule); err != nil {
-			log.Errorf("failed to delete rule: %v", err)
-		}
-		if r.mangleRule != nil {
-			if err := m.rConn.DelRule(r.mangleRule); err != nil {
-				log.Errorf("failed to delete mangle rule: %v", err)
-			}
-		}
-		delete(m.rules, r.ID())
-		return m.rConn.Flush()
-	}
-
-	ips, ok := m.ipsetStore.ips(r.nftSet.Name)
-	if !ok {
-		if err := m.rConn.DelRule(r.nftRule); err != nil {
-			log.Errorf("failed to delete rule: %v", err)
-		}
-		if r.mangleRule != nil {
-			if err := m.rConn.DelRule(r.mangleRule); err != nil {
-				log.Errorf("failed to delete mangle rule: %v", err)
-			}
-		}
-		delete(m.rules, r.ID())
-		return m.rConn.Flush()
-	}
-
-	if _, ok := ips[r.ip.String()]; ok {
-		err := m.sConn.SetDeleteElements(r.nftSet, []nftables.SetElement{{Key: r.ip.To4()}})
-		if err != nil {
-			log.Errorf("delete elements for set %q: %v", r.nftSet.Name, err)
-		}
-		if err := m.sConn.Flush(); err != nil {
-			log.Debugf("flush error of set delete element, %s", r.nftSet.Name)
-			return err
-		}
-		m.ipsetStore.DeleteIpFromSet(r.nftSet.Name, r.ip)
-	}
-
-	// if after delete, set still contains other IPs,
-	// no need to delete firewall rule and we should exit here
-	if len(ips) > 0 {
-		return nil
-	}
-
-	if err := m.rConn.DelRule(r.nftRule); err != nil {
-		log.Errorf("failed to delete rule: %v", err)
-	}
-	if r.mangleRule != nil {
-		if err := m.rConn.DelRule(r.mangleRule); err != nil {
-			log.Errorf("failed to delete mangle rule: %v", err)
-		}
-	}
-
-	if err := m.rConn.Flush(); err != nil {
-		return err
-	}
-
-	delete(m.rules, r.ID())
-	m.ipsetStore.DeleteReferenceFromIpSet(r.nftSet.Name)
-
-	if m.ipsetStore.HasReferenceToSet(r.nftSet.Name) {
-		return nil
-	}
-
-	// we delete last IP from the set, that means we need to delete
-	// set itself and associated firewall rule too
-	m.rConn.FlushSet(r.nftSet)
-	m.rConn.DelSet(r.nftSet)
-	m.ipsetStore.deleteIpset(r.nftSet.Name)
-	return nil
-}
-
-// createDefaultAllowRules creates default allow rules for the input and output chains
-func (m *AclManager) createDefaultAllowRules() error {
-	expIn := []expr.Any{
-		&expr.Verdict{
-			Kind: expr.VerdictAccept,
-		},
-	}
-
-	_ = m.rConn.InsertRule(&nftables.Rule{
-		Table:    m.workTable,
-		Chain:    m.chainInputRules,
-		Position: 0,
-		Exprs:    expIn,
-	})
-
-	if err := m.rConn.Flush(); err != nil {
-		return fmt.Errorf(flushError, err)
-	}
-	return nil
-}
-
-// Flush rule/chain/set operations from the buffer
-//
-// Method also get all rules after flush and refreshes handle values in the rulesets
-func (m *AclManager) Flush() error {
-	if err := m.flushWithBackoff(); err != nil {
-		return err
-	}
-
-	if err := m.refreshRuleHandles(m.chainInputRules, false); err != nil {
-		log.Errorf("failed to refresh rule handles ipv4 input chain: %v", err)
-	}
-	if err := m.refreshRuleHandles(m.chainPrerouting, true); err != nil {
-		log.Errorf("failed to refresh rule handles prerouting chain: %v", err)
-	}
-
-	return nil
-}
-
-func (m *AclManager) addIOFiltering(
-	ip net.IP,
-	proto firewall.Protocol,
-	sPort *firewall.Port,
-	dPort *firewall.Port,
-	action firewall.Action,
-	ipset *nftables.Set,
-) (*Rule, error) {
-	ruleId := generatePeerRuleId(ip, proto, sPort, dPort, action, ipset)
-	if r, ok := m.rules[ruleId]; ok {
-		return &Rule{
-			nftRule:    r.nftRule,
-			mangleRule: r.mangleRule,
-			nftSet:     r.nftSet,
-			ruleID:     r.ruleID,
-			ip:         ip,
-		}, nil
-	}
-
-	var expressions []expr.Any
-
-	if proto != firewall.ProtocolALL {
-		expressions = append(expressions, &expr.Payload{
-			DestRegister: 1,
-			Base:         expr.PayloadBaseNetworkHeader,
-			Offset:       uint32(9),
-			Len:          uint32(1),
-		})
-
-		protoData, err := protoToInt(proto)
-		if err != nil {
-			return nil, fmt.Errorf("convert protocol to number: %v", err)
-		}
-
-		expressions = append(expressions, &expr.Cmp{
-			Register: 1,
-			Op:       expr.CmpOpEq,
-			Data:     []byte{protoData},
-		})
-	}
-
-	rawIP := ip.To4()
-	// check if rawIP contains zeroed IPv4 0.0.0.0 value
-	// in that case not add IP match expression into the rule definition
-	if !bytes.HasPrefix(anyIP, rawIP) {
-		// source address position
-		addrOffset := uint32(12)
-
-		expressions = append(expressions,
-			&expr.Payload{
-				DestRegister: 1,
-				Base:         expr.PayloadBaseNetworkHeader,
-				Offset:       addrOffset,
-				Len:          4,
-			},
-		)
-		// add individual IP for match if no ipset defined
-		if ipset == nil {
-			expressions = append(expressions,
-				&expr.Cmp{
-					Op:       expr.CmpOpEq,
-					Register: 1,
-					Data:     rawIP,
-				},
-			)
-		} else {
-			expressions = append(expressions,
-				&expr.Lookup{
-					SourceRegister: 1,
-					SetName:        ipset.Name,
-					SetID:          ipset.ID,
-				},
-			)
-		}
-	}
-
-	expressions = append(expressions, applyPort(sPort, true)...)
-	expressions = append(expressions, applyPort(dPort, false)...)
-
-	mainExpressions := slices.Clone(expressions)
-
-	switch action {
-	case firewall.ActionAccept:
-		mainExpressions = append(mainExpressions, &expr.Verdict{Kind: expr.VerdictAccept})
-	case firewall.ActionDrop:
-		mainExpressions = append(mainExpressions, &expr.Verdict{Kind: expr.VerdictDrop})
-	}
-
-	userData := []byte(ruleId)
-
-	chain := m.chainInputRules
-	rule := &nftables.Rule{
-		Table:    m.workTable,
-		Chain:    chain,
-		Exprs:    mainExpressions,
-		UserData: userData,
-	}
-
-	// Insert DROP rules at the beginning, append ACCEPT rules at the end
-	var nftRule *nftables.Rule
-	if action == firewall.ActionDrop {
-		nftRule = m.rConn.InsertRule(rule)
-	} else {
-		nftRule = m.rConn.AddRule(rule)
-	}
-
-	if err := m.rConn.Flush(); err != nil {
-		return nil, fmt.Errorf("flush input rule %s: %v", ruleId, err)
-	}
-
-	ruleStruct := &Rule{
-		nftRule: nftRule,
-		// best effort mangle rule
-		mangleRule: m.createPreroutingRule(expressions, userData),
-		nftSet:     ipset,
-		ruleID:     ruleId,
-		ip:         ip,
-	}
-	m.rules[ruleId] = ruleStruct
-	if ipset != nil {
-		m.ipsetStore.AddReferenceToIpset(ipset.Name)
-	}
-
-	return ruleStruct, nil
-}
-
-func (m *AclManager) createPreroutingRule(expressions []expr.Any, userData []byte) *nftables.Rule {
-	if m.chainPrerouting == nil {
-		log.Warn("prerouting chain is not created")
-		return nil
-	}
-
-	preroutingExprs := slices.Clone(expressions)
-
-	// interface
-	preroutingExprs = append([]expr.Any{
-		&expr.Meta{
-			Key:      expr.MetaKeyIIFNAME,
-			Register: 1,
-		},
-		&expr.Cmp{
-			Op:       expr.CmpOpEq,
-			Register: 1,
-			Data:     ifname(m.wgIface.Name()),
-		},
-	}, preroutingExprs...)
-
-	// local destination and mark
-	preroutingExprs = append(preroutingExprs,
-		&expr.Fib{
-			Register:       1,
-			ResultADDRTYPE: true,
-			FlagDADDR:      true,
-		},
-		&expr.Cmp{
-			Op:       expr.CmpOpEq,
-			Register: 1,
-			Data:     binaryutil.NativeEndian.PutUint32(unix.RTN_LOCAL),
-		},
-
-		&expr.Immediate{
-			Register: 1,
-			Data:     binaryutil.NativeEndian.PutUint32(nbnet.PreroutingFwmarkRedirected),
-		},
-		&expr.Meta{
-			Key:            expr.MetaKeyMARK,
-			Register:       1,
-			SourceRegister: true,
-		},
-	)
-
-	nfRule := m.rConn.AddRule(&nftables.Rule{
-		Table:    m.workTable,
-		Chain:    m.chainPrerouting,
-		Exprs:    preroutingExprs,
-		UserData: userData,
-	})
-
-	if err := m.rConn.Flush(); err != nil {
-		log.Errorf("failed to flush mangle rule %s: %v", string(userData), err)
-		return nil
-	}
-
-	return nfRule
-}
-
-func (m *AclManager) createDefaultChains() (err error) {
-	// chainNameInputRules
-	chain := m.createChain(chainNameInputRules)
-	err = m.rConn.Flush()
-	if err != nil {
-		log.Debugf("failed to create chain (%s): %s", chain.Name, err)
-		return fmt.Errorf(flushError, err)
-	}
-	m.chainInputRules = chain
-
-	// netbird-acl-input-filter
-	// type filter hook input priority filter; policy accept;
-	chain = m.createFilterChainWithHook(chainNameInputFilter, nftables.ChainHookInput)
-	m.addJumpRule(chain, m.chainInputRules.Name, expr.MetaKeyIIFNAME) // to netbird-acl-input-rules
-	m.addDropExpressions(chain, expr.MetaKeyIIFNAME)
-	err = m.rConn.Flush()
-	if err != nil {
-		log.Debugf("failed to create chain (%s): %s", chain.Name, err)
-		return err
-	}
-
-	// netbird-acl-forward-filter
-	chainFwFilter := m.createFilterChainWithHook(chainNameForwardFilter, nftables.ChainHookForward)
-	m.addJumpRulesToRtForward(chainFwFilter) // to netbird-rt-fwd
-	m.addDropExpressions(chainFwFilter, expr.MetaKeyIIFNAME)
-
-	err = m.rConn.Flush()
-	if err != nil {
-		log.Debugf("failed to create chain (%s): %s", chainNameForwardFilter, err)
-		return fmt.Errorf(flushError, err)
-	}
-
-	if err := m.allowRedirectedTraffic(chainFwFilter); err != nil {
-		log.Errorf("failed to allow redirected traffic: %s", err)
-	}
-
-	return nil
-}
-
-// Makes redirected traffic originally destined for the host itself (now subject to the forward filter)
-// go through the input filter as well. This will enable e.g. Docker services to keep working by accessing the
-// netbird peer IP.
-func (m *AclManager) allowRedirectedTraffic(chainFwFilter *nftables.Chain) error {
-	// Chain is created by route manager
-	// TODO: move creation to a common place
-	m.chainPrerouting = &nftables.Chain{
-		Name:     chainNameManglePrerouting,
-		Table:    m.workTable,
-		Type:     nftables.ChainTypeFilter,
-		Hooknum:  nftables.ChainHookPrerouting,
-		Priority: nftables.ChainPriorityMangle,
-	}
-
-	m.addFwmarkToForward(chainFwFilter)
-
-	if err := m.rConn.Flush(); err != nil {
-		return fmt.Errorf(flushError, err)
-	}
-
-	return nil
-}
-
-func (m *AclManager) addFwmarkToForward(chainFwFilter *nftables.Chain) {
-	m.rConn.InsertRule(&nftables.Rule{
-		Table: m.workTable,
-		Chain: chainFwFilter,
-		Exprs: []expr.Any{
-			&expr.Meta{
-				Key:      expr.MetaKeyMARK,
-				Register: 1,
-			},
-			&expr.Cmp{
-				Op:       expr.CmpOpEq,
-				Register: 1,
-				Data:     binaryutil.NativeEndian.PutUint32(nbnet.PreroutingFwmarkRedirected),
-			},
-			&expr.Verdict{
-				Kind: expr.VerdictAccept,
-			},
-		},
-	})
-}
-
-func (m *AclManager) addJumpRulesToRtForward(chainFwFilter *nftables.Chain) {
-	expressions := []expr.Any{
-		&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
-		&expr.Cmp{
-			Op:       expr.CmpOpEq,
-			Register: 1,
-			Data:     ifname(m.wgIface.Name()),
-		},
-		&expr.Verdict{
-			Kind:  expr.VerdictJump,
-			Chain: m.routingFwChainName,
-		},
-	}
-
-	_ = m.rConn.AddRule(&nftables.Rule{
-		Table: m.workTable,
-		Chain: chainFwFilter,
-		Exprs: expressions,
-	})
-}
-
-func (m *AclManager) createChain(name string) *nftables.Chain {
-	chain := &nftables.Chain{
-		Name:  name,
-		Table: m.workTable,
-	}
-
-	chain = m.rConn.AddChain(chain)
-
-	insertReturnTrafficRule(m.rConn, m.workTable, chain)
-
-	return chain
-}
-
-func (m *AclManager) createFilterChainWithHook(name string, hookNum *nftables.ChainHook) *nftables.Chain {
-	polAccept := nftables.ChainPolicyAccept
-	chain := &nftables.Chain{
-		Name:     name,
-		Table:    m.workTable,
-		Hooknum:  hookNum,
-		Priority: nftables.ChainPriorityFilter,
-		Type:     nftables.ChainTypeFilter,
-		Policy:   &polAccept,
-	}
-
-	return m.rConn.AddChain(chain)
-}
-
-func (m *AclManager) addDropExpressions(chain *nftables.Chain, ifaceKey expr.MetaKey) []expr.Any {
-	expressions := []expr.Any{
-		&expr.Meta{Key: ifaceKey, Register: 1},
-		&expr.Cmp{
-			Op:       expr.CmpOpEq,
-			Register: 1,
-			Data:     ifname(m.wgIface.Name()),
-		},
-		&expr.Verdict{Kind: expr.VerdictDrop},
-	}
-	_ = m.rConn.AddRule(&nftables.Rule{
-		Table: m.workTable,
-		Chain: chain,
-		Exprs: expressions,
-	})
-	return nil
-}
-
-func (m *AclManager) addJumpRule(chain *nftables.Chain, to string, ifaceKey expr.MetaKey) {
-	expressions := []expr.Any{
-		&expr.Meta{Key: ifaceKey, Register: 1},
-		&expr.Cmp{
-			Op:       expr.CmpOpEq,
-			Register: 1,
-			Data:     ifname(m.wgIface.Name()),
-		},
-		&expr.Verdict{
-			Kind:  expr.VerdictJump,
-			Chain: to,
-		},
-	}
-
-	_ = m.rConn.AddRule(&nftables.Rule{
-		Table: chain.Table,
-		Chain: chain,
-		Exprs: expressions,
-	})
-}
-
-func (m *AclManager) addIpToSet(ipsetName string, ip net.IP) (*nftables.Set, error) {
-	ipset, err := m.rConn.GetSetByName(m.workTable, ipsetName)
-	rawIP := ip.To4()
-	if err != nil {
-		if ipset, err = m.createSet(m.workTable, ipsetName); err != nil {
-			return nil, fmt.Errorf("get set name: %v", err)
-		}
-
-		m.ipsetStore.newIpset(ipset.Name)
-	}
-
-	if m.ipsetStore.IsIpInSet(ipset.Name, ip) {
-		return ipset, nil
-	}
-
-	if err := m.sConn.SetAddElements(ipset, []nftables.SetElement{{Key: rawIP}}); err != nil {
-		return nil, fmt.Errorf("add set element for the first time: %v", err)
-	}
-
-	m.ipsetStore.AddIpToSet(ipset.Name, ip)
-
-	if err := m.sConn.Flush(); err != nil {
-		return nil, fmt.Errorf("flush add elements: %v", err)
-	}
-
-	return ipset, nil
-}
-
-// createSet in given table by name
-func (m *AclManager) createSet(table *nftables.Table, name string) (*nftables.Set, error) {
-	ipset := &nftables.Set{
-		Name:    name,
-		Table:   table,
-		Dynamic: true,
-		KeyType: nftables.TypeIPAddr,
-	}
-
-	if err := m.rConn.AddSet(ipset, nil); err != nil {
-		return nil, fmt.Errorf("create set: %v", err)
-	}
-
-	if err := m.rConn.Flush(); err != nil {
-		return nil, fmt.Errorf("flush created set: %v", err)
-	}
-
-	return ipset, nil
-}
-
-func (m *AclManager) flushWithBackoff() (err error) {
-	backoff := 4
-	backoffTime := 1000 * time.Millisecond
-	for i := 0; ; i++ {
-		err = m.rConn.Flush()
-		if err != nil {
-			log.Debugf("failed to flush nftables: %v", err)
-			if !strings.Contains(err.Error(), "busy") {
-				return
-			}
-			log.Error("failed to flush nftables, retrying...")
-			if i == backoff-1 {
-				return err
-			}
-			time.Sleep(backoffTime)
-			backoffTime *= 2
-			continue
-		}
-		break
-	}
-	return
-}
-
-func (m *AclManager) refreshRuleHandles(chain *nftables.Chain, mangle bool) error {
-	if m.workTable == nil || chain == nil {
-		return nil
-	}
-
-	list, err := m.rConn.GetRules(m.workTable, chain)
-	if err != nil {
-		return err
-	}
-
-	for _, rule := range list {
-		if len(rule.UserData) == 0 {
-			continue
-		}
-		split := bytes.Split(rule.UserData, []byte(" "))
-		r, ok := m.rules[string(split[0])]
-		if ok {
-			if mangle {
-				*r.mangleRule = *rule
-			} else {
-				*r.nftRule = *rule
-			}
-		}
-	}
-
-	return nil
-}
-
-func generatePeerRuleId(ip net.IP, proto firewall.Protocol, sPort *firewall.Port, dPort *firewall.Port, action firewall.Action, ipset *nftables.Set) string {
-	rulesetID := ":" + string(proto) + ":"
-	if sPort != nil {
-		rulesetID += sPort.String()
-	}
-	rulesetID += ":"
-	if dPort != nil {
-		rulesetID += dPort.String()
-	}
-	rulesetID += ":"
-	rulesetID += strconv.Itoa(int(action))
-	if ipset == nil {
-		return "ip:" + ip.String() + rulesetID
-	}
-	return "set:" + ipset.Name + rulesetID
-}
-
-func ifname(n string) []byte {
-	b := make([]byte, 16)
-	copy(b, n+"\x00")
-	return b
-}
-
-func protoToInt(protocol firewall.Protocol) (uint8, error) {
-	switch protocol {
-	case firewall.ProtocolTCP:
-		return unix.IPPROTO_TCP, nil
-	case firewall.ProtocolUDP:
-		return unix.IPPROTO_UDP, nil
-	case firewall.ProtocolICMP:
-		return unix.IPPROTO_ICMP, nil
-	}
-
-	return 0, fmt.Errorf("unsupported protocol: %s", protocol)
-}

client/firewall/nftables/addr_family_linux.go

@@ -0,0 +1,81 @@
+package nftables
+
+import (
+	"fmt"
+	"net"
+
+	"github.com/google/nftables"
+	"golang.org/x/sys/unix"
+
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+)
+
+var (
+	// afIPv4 defines IPv4 header layout and nftables types.
+	afIPv4 = addrFamily{
+		protoOffset:   9,
+		srcAddrOffset: 12,
+		dstAddrOffset: 16,
+		addrLen:       net.IPv4len,
+		totalBits:     8 * net.IPv4len,
+		setKeyType:    nftables.TypeIPAddr,
+		tableFamily:   nftables.TableFamilyIPv4,
+		icmpProto:     unix.IPPROTO_ICMP,
+	}
+	// afIPv6 defines IPv6 header layout and nftables types.
+	afIPv6 = addrFamily{
+		protoOffset:   6,
+		srcAddrOffset: 8,
+		dstAddrOffset: 24,
+		addrLen:       net.IPv6len,
+		totalBits:     8 * net.IPv6len,
+		setKeyType:    nftables.TypeIP6Addr,
+		tableFamily:   nftables.TableFamilyIPv6,
+		icmpProto:     unix.IPPROTO_ICMPV6,
+	}
+)
+
+// addrFamily holds protocol-specific constants for nftables expression building.
+type addrFamily struct {
+	// protoOffset is the IP header offset for the protocol/next-header field (9 for v4, 6 for v6)
+	protoOffset uint32
+	// srcAddrOffset is the IP header offset for the source address (12 for v4, 8 for v6)
+	srcAddrOffset uint32
+	// dstAddrOffset is the IP header offset for the destination address (16 for v4, 24 for v6)
+	dstAddrOffset uint32
+	// addrLen is the byte length of addresses (4 for v4, 16 for v6)
+	addrLen uint32
+	// totalBits is the address size in bits (32 for v4, 128 for v6)
+	totalBits int
+	// setKeyType is the nftables set data type for addresses
+	setKeyType nftables.SetDatatype
+	// tableFamily is the nftables table family
+	tableFamily nftables.TableFamily
+	// icmpProto is the ICMP protocol number for this family (1 for v4, 58 for v6)
+	icmpProto uint8
+}
+
+// familyForAddr returns the address family for the given IP.
+func familyForAddr(is4 bool) addrFamily {
+	if is4 {
+		return afIPv4
+	}
+	return afIPv6
+}
+
+// protoNum converts a firewall protocol to the IP protocol number,
+// using the correct ICMP variant for the address family.
+func (af addrFamily) protoNum(protocol firewall.Protocol) (uint8, error) {
+	switch protocol {
+	case firewall.ProtocolTCP:
+		return unix.IPPROTO_TCP, nil
+	case firewall.ProtocolUDP:
+		return unix.IPPROTO_UDP, nil
+	case firewall.ProtocolICMP:
+		return af.icmpProto, nil
+	case firewall.ProtocolALL:
+		return 0, nil
+	default:
+		return 0, fmt.Errorf("unsupported protocol: %s", protocol)
+	}
+}

client/firewall/nftables/chains_linux.go

@@ -0,0 +1,880 @@
+//go:build !android
+
+package nftables
+
+import (
+	"bytes"
+	"errors"
+	"fmt"
+	"slices"
+	"strings"
+	"time"
+
+	"github.com/coreos/go-iptables/iptables"
+	"github.com/google/nftables"
+	"github.com/google/nftables/binaryutil"
+	"github.com/google/nftables/expr"
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+	"golang.org/x/sys/unix"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	"github.com/netbirdio/netbird/client/firewall/firewalld"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+	nbnet "github.com/netbirdio/netbird/client/net"
+)
+
+func (r *family) createContainers() error {
+	r.chains[chainNameRoutingFw] = r.conn.AddChain(&nftables.Chain{
+		Name:  chainNameRoutingFw,
+		Table: r.workTable,
+	})
+
+	prio := *nftables.ChainPriorityNATSource - 1
+	r.chains[chainNameRoutingNat] = r.conn.AddChain(&nftables.Chain{
+		Name:     chainNameRoutingNat,
+		Table:    r.workTable,
+		Hooknum:  nftables.ChainHookPostrouting,
+		Priority: &prio,
+		Type:     nftables.ChainTypeNAT,
+	})
+
+	r.chains[chainNameRoutingRdr] = r.conn.AddChain(&nftables.Chain{
+		Name:     chainNameRoutingRdr,
+		Table:    r.workTable,
+		Hooknum:  nftables.ChainHookPrerouting,
+		Priority: nftables.ChainPriorityNATDest,
+		Type:     nftables.ChainTypeNAT,
+	})
+
+	r.chains[chainNameManglePostrouting] = r.conn.AddChain(&nftables.Chain{
+		Name:     chainNameManglePostrouting,
+		Table:    r.workTable,
+		Hooknum:  nftables.ChainHookPostrouting,
+		Priority: nftables.ChainPriorityMangle,
+		Type:     nftables.ChainTypeFilter,
+	})
+
+	r.chains[chainNameManglePrerouting] = r.conn.AddChain(&nftables.Chain{
+		Name:     chainNameManglePrerouting,
+		Table:    r.workTable,
+		Hooknum:  nftables.ChainHookPrerouting,
+		Priority: nftables.ChainPriorityMangle,
+		Type:     nftables.ChainTypeFilter,
+	})
+
+	r.chains[chainNameMangleForward] = r.conn.AddChain(&nftables.Chain{
+		Name:     chainNameMangleForward,
+		Table:    r.workTable,
+		Hooknum:  nftables.ChainHookForward,
+		Priority: nftables.ChainPriorityMangle,
+		Type:     nftables.ChainTypeFilter,
+	})
+
+	insertReturnTrafficRule(r.conn, r.workTable, r.chains[chainNameRoutingFw])
+
+	r.addPostroutingRules()
+
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("initialize tables: %v", err)
+	}
+
+	if err := r.addMSSClampingRules(); err != nil {
+		log.Errorf("failed to add MSS clamping rules: %s", err)
+	}
+
+	// Kernel routing opens both INPUT and FORWARD.
+	if err := r.openInterface(true); err != nil {
+		log.Errorf("failed to open interface in foreign chains: %s", err)
+	}
+
+	if err := firewalld.TrustInterface(r.wgIface.Name()); err != nil {
+		log.Warnf("failed to trust interface in firewalld: %v", err)
+	}
+
+	if err := r.refreshRulesMap(); err != nil {
+		log.Errorf("failed to refresh rules: %s", err)
+	}
+
+	return nil
+}
+
+// setupDataPlaneMark configures the fwmark for the data plane
+func (r *family) setupDataPlaneMark() error {
+	if r.chains[chainNameManglePrerouting] == nil || r.chains[chainNameManglePostrouting] == nil {
+		return errors.New("no mangle chains found")
+	}
+
+	ctNew := getCtNewExprs()
+	preExprs := []expr.Any{
+		&expr.Meta{
+			Key:      expr.MetaKeyIIFNAME,
+			Register: 1,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+	}
+	preExprs = append(preExprs, ctNew...)
+	preExprs = append(preExprs,
+		&expr.Immediate{
+			Register: 1,
+			Data:     binaryutil.NativeEndian.PutUint32(nbnet.DataPlaneMarkIn),
+		},
+		&expr.Ct{
+			Key:            expr.CtKeyMARK,
+			Register:       1,
+			SourceRegister: true,
+		},
+	)
+
+	preNftRule := &nftables.Rule{
+		Table: r.workTable,
+		Chain: r.chains[chainNameManglePrerouting],
+		Exprs: preExprs,
+	}
+	r.conn.AddRule(preNftRule)
+
+	postExprs := []expr.Any{
+		&expr.Meta{
+			Key:      expr.MetaKeyOIFNAME,
+			Register: 1,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+	}
+	postExprs = append(postExprs, ctNew...)
+	postExprs = append(postExprs,
+		&expr.Immediate{
+			Register: 1,
+			Data:     binaryutil.NativeEndian.PutUint32(nbnet.DataPlaneMarkOut),
+		},
+		&expr.Ct{
+			Key:            expr.CtKeyMARK,
+			Register:       1,
+			SourceRegister: true,
+		},
+	)
+
+	postNftRule := &nftables.Rule{
+		Table: r.workTable,
+		Chain: r.chains[chainNameManglePostrouting],
+		Exprs: postExprs,
+	}
+	r.conn.AddRule(postNftRule)
+
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("flush: %w", err)
+	}
+
+	return nil
+}
+
+// openInterface adds passthrough accept rules for the NetBird interface to the
+// kernel's filter table and external chains so they don't drop our traffic.
+// includeForward also opens the FORWARD chains (kernel routing); when false only
+// INPUT is opened, which is all the userspace router needs since it never
+// forwards in the kernel.
+func (r *family) openInterface(includeForward bool) error {
+	var merr *multierror.Error
+
+	if err := r.acceptFilterTableRules(includeForward); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	if err := r.acceptExternalChainsRules(includeForward); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("add accept rules to external chains: %w", err))
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) acceptFilterTableRules(includeForward bool) error {
+	if r.filterTable == nil {
+		return nil
+	}
+
+	fw := "iptables"
+
+	defer func() {
+		log.Debugf("Used %s to add accept input/forward rules", fw)
+	}()
+
+	// Try iptables first and fallback to nftables if iptables is not available.
+	// Use the correct protocol (iptables vs ip6tables) for the address family.
+	ipt, err := iptables.NewWithProtocol(r.iptablesProto())
+	if err != nil {
+		log.Warnf("Will use nftables to manipulate the filter table because iptables is not available: %v", err)
+
+		fw = "nftables"
+		return r.acceptFilterRulesNftables(r.filterTable, includeForward)
+	}
+
+	if err := r.acceptFilterRulesIptables(ipt, includeForward); err != nil {
+		log.Warnf("iptables failed (table may be incompatible), falling back to nftables: %v", err)
+		fw = "nftables"
+		return r.acceptFilterRulesNftables(r.filterTable, includeForward)
+	}
+	return nil
+}
+
+func (r *family) acceptFilterRulesIptables(ipt *iptables.IPTables, includeForward bool) error {
+	var merr *multierror.Error
+
+	if includeForward {
+		for _, rule := range r.getAcceptForwardRules() {
+			if err := ipt.Insert("filter", chainNameForward, 1, rule...); err != nil {
+				merr = multierror.Append(merr, fmt.Errorf("add iptables forward rule: %v", err))
+			} else {
+				log.Debugf("added iptables forward rule: %v", rule)
+			}
+		}
+	}
+
+	inputRule := r.getAcceptInputRule()
+	if err := ipt.Insert("filter", chainNameInput, 1, inputRule...); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("add iptables input rule: %v", err))
+	} else {
+		log.Debugf("added iptables input rule: %v", inputRule)
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) getAcceptForwardRules() [][]string {
+	intf := r.wgIface.Name()
+	return [][]string{
+		{"-i", intf, "-j", "ACCEPT"},
+		{"-o", intf, "-m", "conntrack", "--ctstate", "RELATED,ESTABLISHED", "-j", "ACCEPT"},
+	}
+}
+
+func (r *family) getAcceptInputRule() []string {
+	return []string{"-i", r.wgIface.Name(), "-j", "ACCEPT"}
+}
+
+// acceptFilterRulesNftables adds accept rules to the ip filter table using nftables.
+// This is used when iptables is not available.
+func (r *family) acceptFilterRulesNftables(table *nftables.Table, includeForward bool) error {
+	intf := ifname(r.wgIface.Name())
+
+	if includeForward {
+		forwardChain := &nftables.Chain{
+			Name:     chainNameForward,
+			Table:    table,
+			Type:     nftables.ChainTypeFilter,
+			Hooknum:  nftables.ChainHookForward,
+			Priority: nftables.ChainPriorityFilter,
+		}
+		r.insertForwardAcceptRules(forwardChain, intf)
+	}
+
+	inputChain := &nftables.Chain{
+		Name:     chainNameInput,
+		Table:    table,
+		Type:     nftables.ChainTypeFilter,
+		Hooknum:  nftables.ChainHookInput,
+		Priority: nftables.ChainPriorityFilter,
+	}
+	r.insertInputAcceptRule(inputChain, intf)
+
+	return r.conn.Flush()
+}
+
+// acceptExternalChainsRules adds accept rules to external chains (non-netbird, non-iptables tables).
+// It dynamically finds chains at call time to handle chains that may have been created after startup.
+func (r *family) acceptExternalChainsRules(includeForward bool) error {
+	chains := r.findExternalChains()
+	if len(chains) == 0 {
+		return nil
+	}
+
+	intf := ifname(r.wgIface.Name())
+	for _, chain := range chains {
+		r.applyExternalChainAccept(chain, intf, includeForward)
+	}
+
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("flush external chain rules: %w", err)
+	}
+	return nil
+}
+
+func (r *family) applyExternalChainAccept(chain *nftables.Chain, intf []byte, includeForward bool) {
+	if chain.Hooknum == nil {
+		log.Debugf("skipping external chain %s/%s: hooknum is nil", chain.Table.Name, chain.Name)
+		return
+	}
+
+	log.Debugf("adding accept rules to external %s chain: %s %s/%s",
+		hookName(chain.Hooknum), familyName(chain.Table.Family), chain.Table.Name, chain.Name)
+
+	switch *chain.Hooknum {
+	case *nftables.ChainHookForward:
+		if includeForward {
+			r.insertForwardAcceptRules(chain, intf)
+		}
+	case *nftables.ChainHookInput:
+		r.insertInputAcceptRule(chain, intf)
+	}
+}
+
+func (r *family) insertForwardAcceptRules(chain *nftables.Chain, intf []byte) {
+	existing, err := r.existingNetbirdRulesInChain(chain)
+	if err != nil {
+		log.Warnf("skip forward accept rules in %s/%s: %v", chain.Table.Name, chain.Name, err)
+		return
+	}
+	r.insertForwardIifRule(chain, intf, existing)
+	r.insertForwardOifEstablishedRule(chain, intf, existing)
+}
+
+func (r *family) insertForwardIifRule(chain *nftables.Chain, intf []byte, existing map[string]bool) {
+	if existing[userDataAcceptForwardRuleIif] {
+		return
+	}
+	r.conn.InsertRule(&nftables.Rule{
+		Table: chain.Table,
+		Chain: chain,
+		Exprs: []expr.Any{
+			&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: intf},
+			&expr.Counter{},
+			&expr.Verdict{Kind: expr.VerdictAccept},
+		},
+		UserData: []byte(userDataAcceptForwardRuleIif),
+	})
+}
+
+func (r *family) insertForwardOifEstablishedRule(chain *nftables.Chain, intf []byte, existing map[string]bool) {
+	if existing[userDataAcceptForwardRuleOif] {
+		return
+	}
+	exprs := []expr.Any{
+		&expr.Meta{Key: expr.MetaKeyOIFNAME, Register: 1},
+		&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: intf},
+	}
+	r.conn.InsertRule(&nftables.Rule{
+		Table:    chain.Table,
+		Chain:    chain,
+		Exprs:    append(exprs, getEstablishedExprs(2)...),
+		UserData: []byte(userDataAcceptForwardRuleOif),
+	})
+}
+
+func (r *family) insertInputAcceptRule(chain *nftables.Chain, intf []byte) {
+	existing, err := r.existingNetbirdRulesInChain(chain)
+	if err != nil {
+		log.Warnf("skip input accept rule in %s/%s: %v", chain.Table.Name, chain.Name, err)
+		return
+	}
+	if existing[userDataAcceptInputRule] {
+		return
+	}
+	r.conn.InsertRule(&nftables.Rule{
+		Table: chain.Table,
+		Chain: chain,
+		Exprs: []expr.Any{
+			&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+			&expr.Cmp{Op: expr.CmpOpEq, Register: 1, Data: intf},
+			&expr.Counter{},
+			&expr.Verdict{Kind: expr.VerdictAccept},
+		},
+		UserData: []byte(userDataAcceptInputRule),
+	})
+}
+
+// existingNetbirdRulesInChain returns the set of netbird-owned UserData tags present in a chain; callers must bail on error since InsertRule is additive.
+func (r *family) existingNetbirdRulesInChain(chain *nftables.Chain) (map[string]bool, error) {
+	rules, err := r.conn.GetRules(chain.Table, chain)
+	if err != nil {
+		return nil, fmt.Errorf("list rules: %w", err)
+	}
+	present := map[string]bool{}
+	for _, rule := range rules {
+		if !isNetbirdAcceptRuleTag(rule.UserData) {
+			continue
+		}
+		present[string(rule.UserData)] = true
+	}
+	return present, nil
+}
+
+func isNetbirdAcceptRuleTag(userData []byte) bool {
+	switch string(userData) {
+	case userDataAcceptForwardRuleIif,
+		userDataAcceptForwardRuleOif,
+		userDataAcceptInputRule:
+		return true
+	}
+	return false
+}
+
+func (r *family) removeAcceptFilterRules() error {
+	var merr *multierror.Error
+
+	if err := r.removeFilterTableRules(); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	if err := r.removeExternalChainsRules(); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("remove external chain rules: %w", err))
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) removeFilterTableRules() error {
+	if r.filterTable == nil {
+		return nil
+	}
+
+	ipt, err := iptables.NewWithProtocol(r.iptablesProto())
+	if err != nil {
+		log.Debugf("iptables not available, using nftables to remove filter rules: %v", err)
+		return r.removeAcceptRulesFromTable(r.filterTable)
+	}
+
+	if err := r.removeAcceptFilterRulesIptables(ipt); err != nil {
+		log.Debugf("iptables removal failed (table may be incompatible), falling back to nftables: %v", err)
+		return r.removeAcceptRulesFromTable(r.filterTable)
+	}
+	return nil
+}
+
+func (r *family) removeAcceptRulesFromTable(table *nftables.Table) error {
+	chains, err := r.conn.ListChainsOfTableFamily(table.Family)
+	if err != nil {
+		return fmt.Errorf("list chains: %v", err)
+	}
+
+	for _, chain := range chains {
+		if chain.Table.Name != table.Name {
+			continue
+		}
+
+		if chain.Name != chainNameForward && chain.Name != chainNameInput {
+			continue
+		}
+
+		if err := r.removeAcceptRulesFromChain(table, chain); err != nil {
+			return err
+		}
+	}
+
+	return r.conn.Flush()
+}
+
+func (r *family) removeAcceptRulesFromChain(table *nftables.Table, chain *nftables.Chain) error {
+	rules, err := r.conn.GetRules(table, chain)
+	if err != nil {
+		return fmt.Errorf("get rules from %s/%s: %v", table.Name, chain.Name, err)
+	}
+
+	for _, rule := range rules {
+		if bytes.Equal(rule.UserData, []byte(userDataAcceptForwardRuleIif)) ||
+			bytes.Equal(rule.UserData, []byte(userDataAcceptForwardRuleOif)) ||
+			bytes.Equal(rule.UserData, []byte(userDataAcceptInputRule)) {
+			if err := r.conn.DelRule(rule); err != nil {
+				return fmt.Errorf("delete rule from %s/%s: %v", table.Name, chain.Name, err)
+			}
+		}
+	}
+	return nil
+}
+
+// removeExternalChainsRules removes our accept rules from all external chains.
+// This is deterministic - it scans for chains at removal time rather than relying on saved state,
+// ensuring cleanup works even after a crash or if chains changed.
+func (r *family) removeExternalChainsRules() error {
+	chains := r.findExternalChains()
+	if len(chains) == 0 {
+		return nil
+	}
+
+	var merr *multierror.Error
+	for _, chain := range chains {
+		if err := r.removeAcceptRulesFromChain(chain.Table, chain); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove rules from external chain %s/%s: %w", chain.Table.Name, chain.Name, err))
+			continue
+		}
+		if err := r.conn.Flush(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("flush external chain %s/%s: %w", chain.Table.Name, chain.Name, err))
+		}
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// findExternalChains scans for chains from non-netbird tables that have FORWARD or INPUT hooks.
+// This is used both at startup (to know where to add rules) and at cleanup (to ensure deterministic removal).
+func (r *family) findExternalChains() []*nftables.Chain {
+	var chains []*nftables.Chain
+
+	families := []nftables.TableFamily{r.af.tableFamily, nftables.TableFamilyINet}
+
+	for _, family := range families {
+		allChains, err := r.conn.ListChainsOfTableFamily(family)
+		if err != nil {
+			log.Debugf("list chains for family %d: %v", family, err)
+			continue
+		}
+
+		for _, chain := range allChains {
+			if r.isExternalChain(chain) {
+				chains = append(chains, chain)
+			}
+		}
+	}
+
+	return chains
+}
+
+func (r *family) isExternalChain(chain *nftables.Chain) bool {
+	if r.workTable != nil && chain.Table.Name == r.workTable.Name {
+		return false
+	}
+
+	// Skip firewalld-owned chains. Firewalld creates its chains with the
+	// NFT_CHAIN_OWNER flag, so inserting rules into them returns EPERM.
+	// We delegate acceptance to firewalld by trusting the interface instead.
+	if chain.Table.Name == firewalldTableName {
+		return false
+	}
+
+	// Skip iptables/ip6tables-managed tables (adding nft-native rules breaks iptables-save compat)
+	if (chain.Table.Family == nftables.TableFamilyIPv4 || chain.Table.Family == nftables.TableFamilyIPv6) && isIptablesTable(chain.Table.Name) {
+		return false
+	}
+
+	if chain.Type != nftables.ChainTypeFilter {
+		return false
+	}
+
+	if chain.Hooknum == nil {
+		return false
+	}
+
+	return *chain.Hooknum == *nftables.ChainHookForward || *chain.Hooknum == *nftables.ChainHookInput
+}
+
+func isIptablesTable(name string) bool {
+	switch name {
+	case tableNameFilter, tableNat, tableMangle, tableRaw, tableSecurity:
+		return true
+	}
+	return false
+}
+
+func (r *family) removeAcceptFilterRulesIptables(ipt *iptables.IPTables) error {
+	var merr *multierror.Error
+
+	for _, rule := range r.getAcceptForwardRules() {
+		if err := ipt.DeleteIfExists("filter", chainNameForward, rule...); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove iptables forward rule: %v", err))
+		}
+	}
+
+	inputRule := r.getAcceptInputRule()
+	if err := ipt.DeleteIfExists("filter", chainNameInput, inputRule...); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("remove iptables input rule: %v", err))
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// Flush rule/chain/set operations from the buffer
+//
+// Method also get all rules after flush and refreshes handle values in the rulesets
+func (r *family) Flush() error {
+	if err := r.flushWithBackoff(); err != nil {
+		return err
+	}
+
+	if err := r.refreshRuleHandles(r.chainInputRules, false); err != nil {
+		log.Errorf("failed to refresh rule handles ipv4 input chain: %v", err)
+	}
+	if err := r.refreshRuleHandles(r.chainPrerouting, true); err != nil {
+		log.Errorf("failed to refresh rule handles prerouting chain: %v", err)
+	}
+
+	return nil
+}
+
+// queuePreroutingRule builds the prerouting mangle rule that marks
+// redirected traffic and queues it on the connection without flushing,
+// so the caller can commit it in the same transaction as the rule it
+// pairs with. Returns nil when the prerouting chain is absent, in which
+// case nothing is queued.
+func (r *family) queuePreroutingRule(expressions []expr.Any, userData []byte) *nftables.Rule {
+	if r.chainPrerouting == nil {
+		log.Warn("prerouting chain is not created")
+		return nil
+	}
+
+	preroutingExprs := slices.Clone(expressions)
+
+	// interface
+	preroutingExprs = append([]expr.Any{
+		&expr.Meta{
+			Key:      expr.MetaKeyIIFNAME,
+			Register: 1,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+	}, preroutingExprs...)
+
+	// local destination and mark
+	preroutingExprs = append(preroutingExprs,
+		&expr.Fib{
+			Register:       1,
+			ResultADDRTYPE: true,
+			FlagDADDR:      true,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     binaryutil.NativeEndian.PutUint32(unix.RTN_LOCAL),
+		},
+
+		&expr.Immediate{
+			Register: 1,
+			Data:     binaryutil.NativeEndian.PutUint32(nbnet.PreroutingFwmarkRedirected),
+		},
+		&expr.Meta{
+			Key:            expr.MetaKeyMARK,
+			Register:       1,
+			SourceRegister: true,
+		},
+	)
+
+	return r.conn.AddRule(&nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chainPrerouting,
+		Exprs:    preroutingExprs,
+		UserData: userData,
+	})
+}
+
+func (r *family) createDefaultChains() (err error) {
+	// chainNameInputRules
+	chain := r.createChain(chainNameInputRules)
+	err = r.conn.Flush()
+	if err != nil {
+		log.Debugf("failed to create chain (%s): %s", chain.Name, err)
+		return fmt.Errorf(flushError, err)
+	}
+	r.chainInputRules = chain
+
+	// netbird-acl-input-filter
+	// type filter hook input priority filter; policy accept;
+	chain = r.createFilterChainWithHook(chainNameInputFilter, nftables.ChainHookInput)
+	r.addJumpRule(chain, r.chainInputRules.Name, expr.MetaKeyIIFNAME) // to netbird-acl-input-rules
+	r.addDropExpressions(chain, expr.MetaKeyIIFNAME)
+	err = r.conn.Flush()
+	if err != nil {
+		log.Debugf("failed to create chain (%s): %s", chain.Name, err)
+		return err
+	}
+
+	// netbird-acl-forward-filter
+	chainFwFilter := r.createFilterChainWithHook(chainNameForwardFilter, nftables.ChainHookForward)
+	r.addJumpRulesToRtForward(chainFwFilter) // to netbird-rt-fwd
+	r.addDropExpressions(chainFwFilter, expr.MetaKeyIIFNAME)
+
+	err = r.conn.Flush()
+	if err != nil {
+		log.Debugf("failed to create chain (%s): %s", chainNameForwardFilter, err)
+		return fmt.Errorf(flushError, err)
+	}
+
+	if err := r.allowRedirectedTraffic(chainFwFilter); err != nil {
+		log.Errorf("failed to allow redirected traffic: %s", err)
+	}
+
+	return nil
+}
+
+// Makes redirected traffic originally destined for the host itself (now subject to the forward filter)
+// go through the input filter as well. This will enable e.g. Docker services to keep working by accessing the
+// netbird peer IP.
+func (r *family) allowRedirectedTraffic(chainFwFilter *nftables.Chain) error {
+	r.chainPrerouting = r.chains[chainNameManglePrerouting]
+
+	r.addFwmarkToForward(chainFwFilter)
+
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf(flushError, err)
+	}
+
+	return nil
+}
+
+func (r *family) addFwmarkToForward(chainFwFilter *nftables.Chain) {
+	r.conn.InsertRule(&nftables.Rule{
+		Table: r.workTable,
+		Chain: chainFwFilter,
+		Exprs: []expr.Any{
+			&expr.Meta{
+				Key:      expr.MetaKeyMARK,
+				Register: 1,
+			},
+			&expr.Cmp{
+				Op:       expr.CmpOpEq,
+				Register: 1,
+				Data:     binaryutil.NativeEndian.PutUint32(nbnet.PreroutingFwmarkRedirected),
+			},
+			&expr.Verdict{
+				Kind: expr.VerdictAccept,
+			},
+		},
+	})
+}
+
+func (r *family) addJumpRulesToRtForward(chainFwFilter *nftables.Chain) {
+	expressions := []expr.Any{
+		&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+		&expr.Verdict{
+			Kind:  expr.VerdictJump,
+			Chain: r.routingFwChainName,
+		},
+	}
+
+	_ = r.conn.AddRule(&nftables.Rule{
+		Table: r.workTable,
+		Chain: chainFwFilter,
+		Exprs: expressions,
+	})
+}
+
+func (r *family) createChain(name string) *nftables.Chain {
+	chain := &nftables.Chain{
+		Name:  name,
+		Table: r.workTable,
+	}
+
+	chain = r.conn.AddChain(chain)
+
+	insertReturnTrafficRule(r.conn, r.workTable, chain)
+
+	return chain
+}
+
+func (r *family) createFilterChainWithHook(name string, hookNum *nftables.ChainHook) *nftables.Chain {
+	polAccept := nftables.ChainPolicyAccept
+	chain := &nftables.Chain{
+		Name:     name,
+		Table:    r.workTable,
+		Hooknum:  hookNum,
+		Priority: nftables.ChainPriorityFilter,
+		Type:     nftables.ChainTypeFilter,
+		Policy:   &polAccept,
+	}
+
+	return r.conn.AddChain(chain)
+}
+
+func (r *family) addDropExpressions(chain *nftables.Chain, ifaceKey expr.MetaKey) []expr.Any {
+	expressions := []expr.Any{
+		&expr.Meta{Key: ifaceKey, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+		&expr.Verdict{Kind: expr.VerdictDrop},
+	}
+	_ = r.conn.AddRule(&nftables.Rule{
+		Table: r.workTable,
+		Chain: chain,
+		Exprs: expressions,
+	})
+	return nil
+}
+
+func (r *family) addJumpRule(chain *nftables.Chain, to string, ifaceKey expr.MetaKey) {
+	expressions := []expr.Any{
+		&expr.Meta{Key: ifaceKey, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+		&expr.Verdict{
+			Kind:  expr.VerdictJump,
+			Chain: to,
+		},
+	}
+
+	_ = r.conn.AddRule(&nftables.Rule{
+		Table: chain.Table,
+		Chain: chain,
+		Exprs: expressions,
+	})
+}
+
+func (r *family) flushWithBackoff() (err error) {
+	backoff := 4
+	backoffTime := 1000 * time.Millisecond
+	for i := 0; ; i++ {
+		err = r.conn.Flush()
+		if err != nil {
+			log.Debugf("failed to flush nftables: %v", err)
+			if !strings.Contains(err.Error(), "busy") {
+				return
+			}
+			log.Error("failed to flush nftables, retrying...")
+			if i == backoff-1 {
+				return err
+			}
+			time.Sleep(backoffTime)
+			backoffTime *= 2
+			continue
+		}
+		break
+	}
+	return
+}
+
+func (r *family) refreshRuleHandles(chain *nftables.Chain, mangle bool) error {
+	if r.workTable == nil || chain == nil {
+		return nil
+	}
+
+	list, err := r.conn.GetRules(r.workTable, chain)
+	if err != nil {
+		return err
+	}
+
+	for _, rule := range list {
+		if len(rule.UserData) == 0 {
+			continue
+		}
+		pr, ok := r.filters[firewall.RuleID(rule.UserData)]
+		if !ok {
+			continue
+		}
+		if mangle {
+			if pr.mangleRule != nil {
+				*pr.mangleRule = *rule
+			}
+		} else {
+			*pr.nftRule = *rule
+		}
+	}
+
+	return nil
+}

client/firewall/nftables/dnat_linux.go

@@ -0,0 +1,573 @@
+//go:build !android
+
+package nftables
+
+import (
+	"fmt"
+	"net/netip"
+
+	"github.com/google/nftables"
+	"github.com/google/nftables/binaryutil"
+	"github.com/google/nftables/expr"
+	"github.com/google/nftables/xt"
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+)
+
+func (r *family) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
+	ruleID := rule.ID()
+	if _, exists := r.rules[ruleID+dnatSuffix]; exists {
+		return rule, nil
+	}
+
+	protoNum, err := r.af.protoNum(rule.Protocol)
+	if err != nil {
+		return nil, fmt.Errorf("convert protocol to number: %w", err)
+	}
+
+	// Request forwarding before queueing rules: addDnatRedirect/addDnatMasq
+	// buffer netlink messages on r.conn that the next caller's Flush would
+	// commit if we returned without flushing them ourselves.
+	if err := r.ipFwdState.RequestForwarding(r.isV6()); err != nil {
+		return nil, fmt.Errorf("enable forwarding: %w", err)
+	}
+
+	if err := r.addDnatRedirect(rule, protoNum, ruleID); err != nil {
+		r.releaseForwarding()
+		return nil, err
+	}
+
+	if err := r.addDnatMasq(rule, protoNum, ruleID); err != nil {
+		r.releaseForwarding()
+		delete(r.rules, ruleID+dnatSuffix)
+		return nil, err
+	}
+
+	// Unlike iptables, there's no point in adding "out" rules in the forward chain here as our policy is ACCEPT.
+	// To overcome DROP policies in other chains, we'd have to add rules to the chains there.
+	// We also cannot just add "oif <iface> accept" there and filter in our own table as we don't know what is supposed to be allowed.
+	// TODO: find chains with drop policies and add rules there
+
+	if err := r.conn.Flush(); err != nil {
+		r.releaseForwarding()
+		delete(r.rules, ruleID+dnatSuffix)
+		delete(r.rules, ruleID+snatSuffix)
+		return nil, fmt.Errorf("flush rules: %w", err)
+	}
+
+	return &rule, nil
+}
+
+func (r *family) addDnatRedirect(rule firewall.ForwardRule, protoNum uint8, ruleID firewall.RuleID) error {
+	dnatExprs := []expr.Any{
+		&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpNeq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+		&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     []byte{protoNum},
+		},
+		&expr.Payload{
+			DestRegister: 1,
+			Base:         expr.PayloadBaseTransportHeader,
+			Offset:       2,
+			Len:          2,
+		},
+	}
+	portExprs, err := r.applyPort(&rule.DestinationPort, false)
+	if err != nil {
+		return fmt.Errorf("apply destination port: %w", err)
+	}
+	dnatExprs = append(dnatExprs, portExprs...)
+
+	// shifted translated port is not supported in nftables, so we hand this over to xtables
+	if rule.TranslatedPort.IsRange && len(rule.TranslatedPort.Values) == 2 {
+		if rule.TranslatedPort.Values[0] != rule.DestinationPort.Values[0] ||
+			rule.TranslatedPort.Values[1] != rule.DestinationPort.Values[1] {
+			return r.addXTablesRedirect(dnatExprs, ruleID, rule)
+		}
+	}
+
+	additionalExprs, regProtoMin, regProtoMax, err := r.handleTranslatedPort(rule)
+	if err != nil {
+		return err
+	}
+	dnatExprs = append(dnatExprs, additionalExprs...)
+
+	dnatExprs = append(dnatExprs,
+		&expr.NAT{
+			Type:        expr.NATTypeDestNAT,
+			Family:      uint32(r.af.tableFamily),
+			RegAddrMin:  1,
+			RegProtoMin: regProtoMin,
+			RegProtoMax: regProtoMax,
+		},
+	)
+
+	dnatRule := &nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chains[chainNameRoutingRdr],
+		Exprs:    dnatExprs,
+		UserData: []byte(ruleID + dnatSuffix),
+	}
+	r.conn.AddRule(dnatRule)
+	r.rules[ruleID+dnatSuffix] = dnatRule
+
+	return nil
+}
+
+func (r *family) handleTranslatedPort(rule firewall.ForwardRule) ([]expr.Any, uint32, uint32, error) {
+	switch {
+	case rule.TranslatedPort.IsRange && len(rule.TranslatedPort.Values) == 2:
+		return r.handlePortRange(rule)
+	case len(rule.TranslatedPort.Values) == 0:
+		return r.handleAddressOnly(rule)
+	case len(rule.TranslatedPort.Values) == 1:
+		return r.handleSinglePort(rule)
+	default:
+		return nil, 0, 0, fmt.Errorf("invalid translated port: %v", rule.TranslatedPort)
+	}
+}
+
+func (r *family) handlePortRange(rule firewall.ForwardRule) ([]expr.Any, uint32, uint32, error) {
+	exprs := []expr.Any{
+		&expr.Immediate{
+			Register: 1,
+			Data:     rule.TranslatedAddress.AsSlice(),
+		},
+		&expr.Immediate{
+			Register: 2,
+			Data:     binaryutil.BigEndian.PutUint16(rule.TranslatedPort.Values[0]),
+		},
+		&expr.Immediate{
+			Register: 3,
+			Data:     binaryutil.BigEndian.PutUint16(rule.TranslatedPort.Values[1]),
+		},
+	}
+	return exprs, 2, 3, nil
+}
+
+func (r *family) handleAddressOnly(rule firewall.ForwardRule) ([]expr.Any, uint32, uint32, error) {
+	exprs := []expr.Any{
+		&expr.Immediate{
+			Register: 1,
+			Data:     rule.TranslatedAddress.AsSlice(),
+		},
+	}
+	return exprs, 0, 0, nil
+}
+
+func (r *family) handleSinglePort(rule firewall.ForwardRule) ([]expr.Any, uint32, uint32, error) {
+	exprs := []expr.Any{
+		&expr.Immediate{
+			Register: 1,
+			Data:     rule.TranslatedAddress.AsSlice(),
+		},
+		&expr.Immediate{
+			Register: 2,
+			Data:     binaryutil.BigEndian.PutUint16(rule.TranslatedPort.Values[0]),
+		},
+	}
+	return exprs, 2, 0, nil
+}
+
+func (r *family) addXTablesRedirect(dnatExprs []expr.Any, ruleID firewall.RuleID, rule firewall.ForwardRule) error {
+	dnatExprs = append(dnatExprs,
+		&expr.Counter{},
+		&expr.Target{
+			Name: "DNAT",
+			Rev:  2,
+			Info: &xt.NatRange2{
+				NatRange: xt.NatRange{
+					Flags:   uint(xt.NatRangeMapIPs | xt.NatRangeProtoSpecified | xt.NatRangeProtoOffset),
+					MinIP:   rule.TranslatedAddress.AsSlice(),
+					MaxIP:   rule.TranslatedAddress.AsSlice(),
+					MinPort: rule.TranslatedPort.Values[0],
+					MaxPort: rule.TranslatedPort.Values[1],
+				},
+				BasePort: rule.DestinationPort.Values[0],
+			},
+		},
+	)
+
+	natTable := &nftables.Table{
+		Name:   tableNat,
+		Family: r.af.tableFamily,
+	}
+	dnatRule := &nftables.Rule{
+		Table: natTable,
+		Chain: &nftables.Chain{
+			Name:     chainNameNatPrerouting,
+			Table:    natTable,
+			Type:     nftables.ChainTypeNAT,
+			Hooknum:  nftables.ChainHookPrerouting,
+			Priority: nftables.ChainPriorityNATDest,
+		},
+		Exprs:    dnatExprs,
+		UserData: []byte(ruleID + dnatSuffix),
+	}
+	r.conn.AddRule(dnatRule)
+	r.rules[ruleID+dnatSuffix] = dnatRule
+
+	return nil
+}
+
+func (r *family) addDnatMasq(rule firewall.ForwardRule, protoNum uint8, ruleID firewall.RuleID) error {
+	portExprs, err := r.applyPort(&rule.TranslatedPort, false)
+	if err != nil {
+		return fmt.Errorf("apply translated port: %w", err)
+	}
+
+	masqExprs := []expr.Any{
+		&expr.Meta{Key: expr.MetaKeyOIFNAME, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+		&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     []byte{protoNum},
+		},
+		&expr.Payload{
+			DestRegister: 1,
+			Base:         expr.PayloadBaseNetworkHeader,
+			Offset:       r.af.dstAddrOffset,
+			Len:          r.af.addrLen,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     rule.TranslatedAddress.AsSlice(),
+		},
+	}
+
+	masqExprs = append(masqExprs, portExprs...)
+	masqExprs = append(masqExprs, &expr.Masq{})
+
+	masqRule := &nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chains[chainNameRoutingNat],
+		Exprs:    masqExprs,
+		UserData: []byte(ruleID + snatSuffix),
+	}
+	r.conn.AddRule(masqRule)
+	r.rules[ruleID+snatSuffix] = masqRule
+
+	return nil
+}
+
+func (r *family) DeleteDNATRule(rule firewall.Rule) error {
+	ruleID := rule.ID()
+
+	if err := r.refreshRulesMap(); err != nil {
+		return fmt.Errorf(refreshRulesMapError, err)
+	}
+
+	var merr *multierror.Error
+	var needsFlush bool
+	var found bool
+
+	if dnatRule, exists := r.rules[ruleID+dnatSuffix]; exists {
+		found = true
+		if dnatRule.Handle == 0 {
+			log.Warnf("dnat rule %s has no handle, removing stale entry", ruleID+dnatSuffix)
+			delete(r.rules, ruleID+dnatSuffix)
+		} else if err := r.conn.DelRule(dnatRule); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete dnat rule: %w", err))
+		} else {
+			needsFlush = true
+		}
+	}
+
+	if masqRule, exists := r.rules[ruleID+snatSuffix]; exists {
+		found = true
+		if masqRule.Handle == 0 {
+			log.Warnf("snat rule %s has no handle, removing stale entry", ruleID+snatSuffix)
+			delete(r.rules, ruleID+snatSuffix)
+		} else if err := r.conn.DelRule(masqRule); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("delete snat rule: %w", err))
+		} else {
+			needsFlush = true
+		}
+	}
+
+	if needsFlush {
+		if err := r.conn.Flush(); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf(flushError, err))
+		}
+	}
+
+	if merr != nil {
+		return nberrors.FormatErrorOrNil(merr)
+	}
+
+	delete(r.rules, ruleID+dnatSuffix)
+	delete(r.rules, ruleID+snatSuffix)
+
+	// Release once, only if the rule was present and removed.
+	if found {
+		r.releaseForwarding()
+	}
+
+	return nil
+}
+
+// releaseForwarding drops one IP forwarding reference, logging any error.
+func (r *family) releaseForwarding() {
+	if err := r.ipFwdState.ReleaseForwarding(r.isV6()); err != nil {
+		log.Errorf("release IP forwarding: %v", err)
+	}
+}
+
+// isV6 reports whether this family handles the IPv6 table.
+func (r *family) isV6() bool {
+	return r.af.tableFamily == nftables.TableFamilyIPv6
+}
+
+func (r *family) AddInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	ruleID := firewall.RuleID(fmt.Sprintf("inbound-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	if _, exists := r.rules[ruleID]; exists {
+		return nil
+	}
+
+	protoNum, err := r.af.protoNum(protocol)
+	if err != nil {
+		return fmt.Errorf("convert protocol to number: %w", err)
+	}
+
+	exprs := []expr.Any{
+		&expr.Meta{Key: expr.MetaKeyIIFNAME, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     ifname(r.wgIface.Name()),
+		},
+		&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 2},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 2,
+			Data:     []byte{protoNum},
+		},
+		&expr.Payload{
+			DestRegister: 3,
+			Base:         expr.PayloadBaseTransportHeader,
+			Offset:       2,
+			Len:          2,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 3,
+			Data:     binaryutil.BigEndian.PutUint16(originalPort),
+		},
+	}
+
+	bits := 32
+	if localAddr.Is6() {
+		bits = 128
+	}
+	exprs = append(exprs, prefixMatchExprs(r.af, netip.PrefixFrom(localAddr, bits), false)...)
+
+	exprs = append(exprs,
+		&expr.Immediate{
+			Register: 1,
+			Data:     localAddr.AsSlice(),
+		},
+		&expr.Immediate{
+			Register: 2,
+			Data:     binaryutil.BigEndian.PutUint16(translatedPort),
+		},
+		&expr.NAT{
+			Type:        expr.NATTypeDestNAT,
+			Family:      uint32(r.af.tableFamily),
+			RegAddrMin:  1,
+			RegProtoMin: 2,
+			RegProtoMax: 0,
+		},
+	)
+
+	dnatRule := &nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chains[chainNameRoutingRdr],
+		Exprs:    exprs,
+		UserData: []byte(ruleID),
+	}
+	r.conn.AddRule(dnatRule)
+
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("add inbound DNAT rule: %w", err)
+	}
+
+	r.rules[ruleID] = dnatRule
+
+	return nil
+}
+
+// RemoveInboundDNAT removes an inbound DNAT rule.
+func (r *family) RemoveInboundDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	if err := r.refreshRulesMap(); err != nil {
+		return fmt.Errorf(refreshRulesMapError, err)
+	}
+
+	ruleID := firewall.RuleID(fmt.Sprintf("inbound-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	rule, exists := r.rules[ruleID]
+	if !exists {
+		return nil
+	}
+
+	if rule.Handle == 0 {
+		log.Warnf("inbound DNAT rule %s has no handle, removing stale entry", ruleID)
+		delete(r.rules, ruleID)
+		return nil
+	}
+
+	if err := r.conn.DelRule(rule); err != nil {
+		return fmt.Errorf("delete inbound DNAT rule %s: %w", ruleID, err)
+	}
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("flush delete inbound DNAT rule: %w", err)
+	}
+	delete(r.rules, ruleID)
+
+	return nil
+}
+
+// ensureNATOutputChain lazily creates the OUTPUT NAT chain on first use.
+func (r *family) ensureNATOutputChain() error {
+	if _, exists := r.chains[chainNameNATOutput]; exists {
+		return nil
+	}
+
+	r.chains[chainNameNATOutput] = r.conn.AddChain(&nftables.Chain{
+		Name:     chainNameNATOutput,
+		Table:    r.workTable,
+		Hooknum:  nftables.ChainHookOutput,
+		Priority: nftables.ChainPriorityNATDest,
+		Type:     nftables.ChainTypeNAT,
+	})
+
+	if err := r.conn.Flush(); err != nil {
+		delete(r.chains, chainNameNATOutput)
+		return fmt.Errorf("create NAT output chain: %w", err)
+	}
+	return nil
+}
+
+// AddOutputDNAT adds an OUTPUT chain DNAT rule for locally-generated traffic.
+func (r *family) AddOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	ruleID := firewall.RuleID(fmt.Sprintf("output-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	if _, exists := r.rules[ruleID]; exists {
+		return nil
+	}
+
+	if err := r.ensureNATOutputChain(); err != nil {
+		return err
+	}
+
+	protoNum, err := r.af.protoNum(protocol)
+	if err != nil {
+		return fmt.Errorf("convert protocol to number: %w", err)
+	}
+
+	exprs := []expr.Any{
+		&expr.Meta{Key: expr.MetaKeyL4PROTO, Register: 1},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 1,
+			Data:     []byte{protoNum},
+		},
+		&expr.Payload{
+			DestRegister: 2,
+			Base:         expr.PayloadBaseTransportHeader,
+			Offset:       2,
+			Len:          2,
+		},
+		&expr.Cmp{
+			Op:       expr.CmpOpEq,
+			Register: 2,
+			Data:     binaryutil.BigEndian.PutUint16(originalPort),
+		},
+	}
+
+	bits := 32
+	if localAddr.Is6() {
+		bits = 128
+	}
+	exprs = append(exprs, prefixMatchExprs(r.af, netip.PrefixFrom(localAddr, bits), false)...)
+
+	exprs = append(exprs,
+		&expr.Immediate{
+			Register: 1,
+			Data:     localAddr.AsSlice(),
+		},
+		&expr.Immediate{
+			Register: 2,
+			Data:     binaryutil.BigEndian.PutUint16(translatedPort),
+		},
+		&expr.NAT{
+			Type:        expr.NATTypeDestNAT,
+			Family:      uint32(r.af.tableFamily),
+			RegAddrMin:  1,
+			RegProtoMin: 2,
+		},
+	)
+
+	dnatRule := &nftables.Rule{
+		Table:    r.workTable,
+		Chain:    r.chains[chainNameNATOutput],
+		Exprs:    exprs,
+		UserData: []byte(ruleID),
+	}
+	r.conn.AddRule(dnatRule)
+
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("add output DNAT rule: %w", err)
+	}
+
+	r.rules[ruleID] = dnatRule
+
+	return nil
+}
+
+// RemoveOutputDNAT removes an OUTPUT chain DNAT rule.
+func (r *family) RemoveOutputDNAT(localAddr netip.Addr, protocol firewall.Protocol, originalPort, translatedPort uint16) error {
+	if err := r.refreshRulesMap(); err != nil {
+		return fmt.Errorf(refreshRulesMapError, err)
+	}
+
+	ruleID := firewall.RuleID(fmt.Sprintf("output-dnat-%s-%s-%d-%d", localAddr.String(), protocol, originalPort, translatedPort))
+
+	rule, exists := r.rules[ruleID]
+	if !exists {
+		return nil
+	}
+
+	if rule.Handle == 0 {
+		log.Warnf("output DNAT rule %s has no handle, removing stale entry", ruleID)
+		delete(r.rules, ruleID)
+		return nil
+	}
+
+	if err := r.conn.DelRule(rule); err != nil {
+		return fmt.Errorf("delete output DNAT rule %s: %w", ruleID, err)
+	}
+	if err := r.conn.Flush(); err != nil {
+		return fmt.Errorf("flush delete output DNAT rule: %w", err)
+	}
+	delete(r.rules, ruleID)
+
+	return nil
+}

client/firewall/nftables/dnat_refcount_linux_test.go

@@ -0,0 +1,249 @@
+//go:build privileged
+
+package nftables
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	fw "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+func nftRefcountIfaceV4() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("100.96.0.1"),
+				Network: netip.MustParsePrefix("100.96.0.0/16"),
+			}
+		},
+	}
+}
+
+func nftRefcountIfaceDual() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("100.96.0.1"),
+				Network: netip.MustParsePrefix("100.96.0.0/16"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+}
+
+func newNftRefcountManager(t *testing.T, dual bool) *Manager {
+	t.Helper()
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+	var ifMock *iFaceMock
+	if dual {
+		ifMock = nftRefcountIfaceDual()
+	} else {
+		ifMock = nftRefcountIfaceV4()
+	}
+	m, err := Create(ifMock, iface.DefaultMTU)
+	require.NoError(t, err, "create manager")
+	require.NoError(t, m.Init(nil), "init manager")
+	t.Cleanup(func() {
+		require.NoError(t, m.Close(nil), "close manager")
+	})
+	return m
+}
+
+func dnatV4(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("100.96.0.2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+func dnatV6(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("fd00::2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+// TestNftablesDNAT_RefcountBalancedV4 verifies that Add/Delete pairs leave the
+// v4 refcount at zero.
+func TestNftablesDNAT_RefcountBalancedV4(t *testing.T) {
+	m := newNftRefcountManager(t, false)
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV4(8081))
+	require.NoError(t, err, "add v4 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	r2, err := m.AddDNATRule(dnatV4(8082))
+	require.NoError(t, err, "add v4 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 2, v4, "v4 refcount after second add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v4 dnat 1")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r2), "delete v4 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount after second delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+}
+
+// TestNftablesDNAT_RefcountBalancedV6 verifies the v6 path increments v6 only
+// and decrements back to zero on Delete.
+func TestNftablesDNAT_RefcountBalancedV6(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	require.NotNil(t, m.family6, "v6 family")
+	require.Same(t, m.family4.ipFwdState, m.family6.ipFwdState, "shared state")
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV6(9091))
+	require.NoError(t, err, "add v6 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 1, v6, "v6 refcount after first add")
+
+	r2, err := m.AddDNATRule(dnatV6(9092))
+	require.NoError(t, err, "add v6 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 2, v6, "v6 refcount after second add")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat 1")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 1, v6, "v6 refcount after first delete")
+
+	require.NoError(t, m.DeleteDNATRule(r2), "delete v6 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6, "v6 refcount after second delete")
+}
+
+// TestNftablesDNAT_DuplicateAddNoLeak verifies that a duplicate Add (same
+// ForwardRule) does not double-increment the refcount.
+func TestNftablesDNAT_DuplicateAddNoLeak(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	rule := dnatV4(8083)
+	r1, err := m.AddDNATRule(rule)
+	require.NoError(t, err, "add v4 dnat")
+	v4, _ := state.Counts()
+	assert.Equal(t, 1, v4)
+
+	// duplicate add: same rule ID, must be a no-op for the refcount.
+	_, err = m.AddDNATRule(rule)
+	require.NoError(t, err, "duplicate add")
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "duplicate add must not increment")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v4 dnat")
+	v4, _ = state.Counts()
+	assert.Equal(t, 0, v4, "single delete must drop to zero")
+}
+
+// TestNftablesDNAT_DeleteMissingNoUnderflow verifies deleting a rule that was
+// never added does not underflow the refcount.
+func TestNftablesDNAT_DeleteMissingNoUnderflow(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	// Construct a Rule reference for something never added. The router stores
+	// rules by ID(), and DeleteDNATRule looks them up in r.rules; a missing
+	// entry must be a no-op rather than calling Release.
+	phantom := dnatV4(8099)
+	require.NoError(t, m.DeleteDNATRule(&phantom), "delete missing v4 dnat")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unaffected by missing delete")
+	assert.Equal(t, 0, v6, "v6 refcount unaffected")
+
+	phantom6 := dnatV6(9099)
+	require.NoError(t, m.DeleteDNATRule(&phantom6), "delete missing v6 dnat")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6, "v6 refcount unaffected by missing delete")
+
+	// And after a phantom delete, a real add still results in count=1.
+	r1, err := m.AddDNATRule(dnatV4(8100))
+	require.NoError(t, err, "add v4 dnat after phantom delete")
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "real add still increments after phantom delete")
+	require.NoError(t, m.DeleteDNATRule(r1))
+}
+
+// TestNftablesRouting_RepeatedEnableSingleReference verifies that EnableRouting
+// (called on every network-map update) holds at most one reference per family
+// and a single DisableRouting drops both back to zero.
+func TestNftablesRouting_RepeatedEnableSingleReference(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	require.NoError(t, m.EnableRouting(), "first enable")
+	require.NoError(t, m.EnableRouting(), "second enable")
+	require.NoError(t, m.EnableRouting(), "third enable")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "repeated enable holds a single v4 reference")
+	assert.Equal(t, 1, v6, "repeated enable holds a single v6 reference")
+
+	require.NoError(t, m.DisableRouting(), "disable")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "single disable releases the v4 reference")
+	assert.Equal(t, 0, v6, "single disable releases the v6 reference")
+}
+
+// TestNftablesRouting_DisableKeepsDNATReference verifies that an unpaired
+// DisableRouting does not release references held by active DNAT rules.
+func TestNftablesRouting_DisableKeepsDNATReference(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV6(9095))
+	require.NoError(t, err, "add v6 dnat")
+
+	require.NoError(t, m.DisableRouting(), "unpaired disable")
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6, "DNAT-held reference survives unpaired DisableRouting")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "delete releases the DNAT reference")
+}
+
+// TestNftablesDNAT_DoubleDeleteNoUnderflow verifies that deleting the same rule
+// twice does not underflow the refcount (the second delete is a no-op).
+func TestNftablesDNAT_DoubleDeleteNoUnderflow(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.family4.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV6(9093))
+	require.NoError(t, err)
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "first delete")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "second delete must be no-op")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "double delete must not underflow")
+}

client/firewall/nftables/external_chain_monitor_integration_linux_test.go

@@ -0,0 +1,76 @@
+//go:build linux
+
+package nftables
+
+import (
+	"os"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/google/nftables"
+	"github.com/stretchr/testify/require"
+)
+
+// TestExternalChainMonitorRootIntegration verifies that adding a new chain
+// in an external (non-netbird) filter table triggers the reconciler.
+// Requires CAP_NET_ADMIN; skip otherwise.
+func TestExternalChainMonitorRootIntegration(t *testing.T) {
+	if os.Geteuid() != 0 {
+		t.Skip("root required")
+	}
+
+	calls := make(chan struct{}, 8)
+	var count atomic.Int32
+	rec := &countingReconciler{calls: calls, count: &count}
+
+	m := newExternalChainMonitor(rec)
+	m.start()
+	t.Cleanup(m.stop)
+
+	// Give the netlink subscription a moment to register.
+	time.Sleep(200 * time.Millisecond)
+
+	conn := &nftables.Conn{}
+	table := conn.AddTable(&nftables.Table{
+		Name:   "nbmon_integration_test",
+		Family: nftables.TableFamilyINet,
+	})
+	t.Cleanup(func() {
+		cleanup := &nftables.Conn{}
+		cleanup.DelTable(table)
+		_ = cleanup.Flush()
+	})
+
+	chain := conn.AddChain(&nftables.Chain{
+		Name:     "filter_INPUT",
+		Table:    table,
+		Hooknum:  nftables.ChainHookInput,
+		Priority: nftables.ChainPriorityFilter,
+		Type:     nftables.ChainTypeFilter,
+	})
+	_ = chain
+	require.NoError(t, conn.Flush(), "create external test chain")
+
+	select {
+	case <-calls:
+		// success
+	case <-time.After(3 * time.Second):
+		t.Fatalf("reconcile was not invoked after creating an external chain")
+	}
+	require.GreaterOrEqual(t, count.Load(), int32(1))
+}
+
+type countingReconciler struct {
+	calls chan struct{}
+	count *atomic.Int32
+}
+
+func (c *countingReconciler) reconcileExternalChains() error {
+	c.count.Add(1)
+	select {
+	case c.calls <- struct{}{}:
+	default:
+	}
+	return nil
+}

client/firewall/nftables/external_chain_monitor_linux.go

@@ -0,0 +1,200 @@
+package nftables
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"sync"
+	"time"
+
+	"github.com/cenkalti/backoff/v4"
+	"github.com/google/nftables"
+	log "github.com/sirupsen/logrus"
+)
+
+const (
+	externalMonitorReconcileDelay = 500 * time.Millisecond
+	externalMonitorInitInterval   = 5 * time.Second
+	externalMonitorMaxInterval    = 5 * time.Minute
+	externalMonitorRandomization  = 0.5
+)
+
+// externalChainReconciler re-applies passthrough accept rules to external
+// nftables chains. Implementations must be safe to call from the monitor
+// goroutine; the Manager locks its mutex internally.
+type externalChainReconciler interface {
+	reconcileExternalChains() error
+}
+
+// externalChainMonitor watches nftables netlink events and triggers a
+// reconcile when a new table or chain appears (e.g. after
+// `firewall-cmd --reload`). Netlink errors trigger exponential-backoff
+// reconnect.
+type externalChainMonitor struct {
+	reconciler externalChainReconciler
+
+	mu     sync.Mutex
+	cancel context.CancelFunc
+	done   chan struct{}
+}
+
+func newExternalChainMonitor(r externalChainReconciler) *externalChainMonitor {
+	return &externalChainMonitor{reconciler: r}
+}
+
+func (m *externalChainMonitor) start() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.cancel != nil {
+		return
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	m.cancel = cancel
+	done := make(chan struct{})
+	m.done = done
+
+	go m.run(ctx, done)
+}
+
+func (m *externalChainMonitor) stop() {
+	m.mu.Lock()
+	cancel := m.cancel
+	done := m.done
+	m.cancel = nil
+	m.done = nil
+	m.mu.Unlock()
+
+	if cancel == nil {
+		return
+	}
+	cancel()
+	<-done
+}
+
+func (m *externalChainMonitor) run(ctx context.Context, done chan struct{}) {
+	defer close(done)
+
+	bo := &backoff.ExponentialBackOff{
+		InitialInterval:     externalMonitorInitInterval,
+		RandomizationFactor: externalMonitorRandomization,
+		Multiplier:          backoff.DefaultMultiplier,
+		MaxInterval:         externalMonitorMaxInterval,
+		MaxElapsedTime:      0,
+		Clock:               backoff.SystemClock,
+	}
+	bo.Reset()
+
+	for ctx.Err() == nil {
+		err := m.watch(ctx)
+		if ctx.Err() != nil {
+			return
+		}
+
+		delay := bo.NextBackOff()
+		log.Warnf("external chain monitor: %v, reconnecting in %s", err, delay)
+		select {
+		case <-ctx.Done():
+			return
+		case <-time.After(delay):
+		}
+	}
+}
+
+func (m *externalChainMonitor) watch(ctx context.Context) error {
+	events, closeMon, err := m.subscribe()
+	if err != nil {
+		return err
+	}
+	defer closeMon()
+
+	debounce := time.NewTimer(time.Hour)
+	if !debounce.Stop() {
+		<-debounce.C
+	}
+	defer debounce.Stop()
+
+	pending := false
+	for {
+		select {
+		case <-ctx.Done():
+			return nil
+		case <-debounce.C:
+			pending = false
+			m.reconcile()
+		case ev, ok := <-events:
+			if !ok {
+				return errors.New("monitor channel closed")
+			}
+			if ev.Error != nil {
+				return fmt.Errorf("monitor event: %w", ev.Error)
+			}
+			if !isRelevantMonitorEvent(ev) {
+				continue
+			}
+			resetDebounce(debounce, pending)
+			pending = true
+		}
+	}
+}
+
+func (m *externalChainMonitor) subscribe() (chan *nftables.MonitorEvent, func(), error) {
+	conn := &nftables.Conn{}
+	mon := nftables.NewMonitor(
+		nftables.WithMonitorAction(nftables.MonitorActionNew),
+		nftables.WithMonitorObject(nftables.MonitorObjectChains|nftables.MonitorObjectTables),
+	)
+	events, err := conn.AddMonitor(mon)
+	if err != nil {
+		return nil, nil, fmt.Errorf("add netlink monitor: %w", err)
+	}
+	return events, func() { _ = mon.Close() }, nil
+}
+
+// resetDebounce reschedules a pending debounce timer without leaking a stale
+// fire on its channel. pending must reflect whether the timer is armed.
+func resetDebounce(t *time.Timer, pending bool) {
+	if pending && !t.Stop() {
+		select {
+		case <-t.C:
+		default:
+		}
+	}
+	t.Reset(externalMonitorReconcileDelay)
+}
+
+func (m *externalChainMonitor) reconcile() {
+	if err := m.reconciler.reconcileExternalChains(); err != nil {
+		log.Warnf("reconcile external chain rules: %v", err)
+	}
+}
+
+// isRelevantMonitorEvent returns true for table/chain creation events on
+// families we care about. The reconciler filters to actual external filter
+// chains.
+func isRelevantMonitorEvent(ev *nftables.MonitorEvent) bool {
+	switch ev.Type {
+	case nftables.MonitorEventTypeNewChain:
+		chain, ok := ev.Data.(*nftables.Chain)
+		if !ok || chain == nil || chain.Table == nil {
+			return false
+		}
+		return isMonitoredFamily(chain.Table.Family)
+	case nftables.MonitorEventTypeNewTable:
+		table, ok := ev.Data.(*nftables.Table)
+		if !ok || table == nil {
+			return false
+		}
+		return isMonitoredFamily(table.Family)
+	}
+	return false
+}
+
+func isMonitoredFamily(family nftables.TableFamily) bool {
+	switch family {
+	case nftables.TableFamilyIPv4, nftables.TableFamilyIPv6, nftables.TableFamilyINet:
+		return true
+	}
+	return false
+}

client/firewall/nftables/external_chain_monitor_linux_test.go

@@ -0,0 +1,137 @@
+package nftables
+
+import (
+	"testing"
+
+	"github.com/google/nftables"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestIsMonitoredFamily(t *testing.T) {
+	tests := []struct {
+		family nftables.TableFamily
+		want   bool
+	}{
+		{nftables.TableFamilyIPv4, true},
+		{nftables.TableFamilyIPv6, true},
+		{nftables.TableFamilyINet, true},
+		{nftables.TableFamilyARP, false},
+		{nftables.TableFamilyBridge, false},
+		{nftables.TableFamilyNetdev, false},
+		{nftables.TableFamilyUnspecified, false},
+	}
+	for _, tc := range tests {
+		assert.Equal(t, tc.want, isMonitoredFamily(tc.family), "family=%d", tc.family)
+	}
+}
+
+func TestIsRelevantMonitorEvent(t *testing.T) {
+	inetTable := &nftables.Table{Name: "firewalld", Family: nftables.TableFamilyINet}
+	ipTable := &nftables.Table{Name: "filter", Family: nftables.TableFamilyIPv4}
+	arpTable := &nftables.Table{Name: "arp", Family: nftables.TableFamilyARP}
+
+	tests := []struct {
+		name string
+		ev   *nftables.MonitorEvent
+		want bool
+	}{
+		{
+			name: "new chain in inet firewalld",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeNewChain,
+				Data: &nftables.Chain{Name: "filter_INPUT", Table: inetTable},
+			},
+			want: true,
+		},
+		{
+			name: "new chain in ip filter",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeNewChain,
+				Data: &nftables.Chain{Name: "INPUT", Table: ipTable},
+			},
+			want: true,
+		},
+		{
+			name: "new chain in unwatched arp family",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeNewChain,
+				Data: &nftables.Chain{Name: "x", Table: arpTable},
+			},
+			want: false,
+		},
+		{
+			name: "new table inet",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeNewTable,
+				Data: inetTable,
+			},
+			want: true,
+		},
+		{
+			name: "del chain (we only act on new)",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeDelChain,
+				Data: &nftables.Chain{Name: "filter_INPUT", Table: inetTable},
+			},
+			want: false,
+		},
+		{
+			name: "chain with nil table",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeNewChain,
+				Data: &nftables.Chain{Name: "x"},
+			},
+			want: false,
+		},
+		{
+			name: "nil data",
+			ev: &nftables.MonitorEvent{
+				Type: nftables.MonitorEventTypeNewChain,
+				Data: (*nftables.Chain)(nil),
+			},
+			want: false,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			assert.Equal(t, tc.want, isRelevantMonitorEvent(tc.ev))
+		})
+	}
+}
+
+// fakeReconciler records reconcile invocations for debounce tests.
+type fakeReconciler struct {
+	calls chan struct{}
+}
+
+func (f *fakeReconciler) reconcileExternalChains() error {
+	f.calls <- struct{}{}
+	return nil
+}
+
+func TestExternalChainMonitorStopWithoutStart(t *testing.T) {
+	m := newExternalChainMonitor(&fakeReconciler{calls: make(chan struct{}, 1)})
+	// Must not panic or block.
+	m.stop()
+}
+
+func TestExternalChainMonitorDoubleStart(t *testing.T) {
+	// start() twice should be a no-op; stop() cleans up once.
+	// We avoid exercising the netlink watch loop here because it needs root.
+	m := newExternalChainMonitor(&fakeReconciler{calls: make(chan struct{}, 1)})
+
+	// Replace run with a stub that just waits for cancel, so start() stays
+	// deterministic without opening a netlink socket.
+	origDone := make(chan struct{})
+	m.done = origDone
+	m.cancel = func() { close(origDone) }
+
+	// Second start should be a no-op (cancel already set).
+	m.start()
+	assert.NotNil(t, m.cancel)
+
+	m.stop()
+	assert.Nil(t, m.cancel)
+	assert.Nil(t, m.done)
+}

client/firewall/nftables/family_linux.go

@@ -0,0 +1,249 @@
+//go:build !android
+
+package nftables
+
+import (
+	"fmt"
+	"net/netip"
+
+	"github.com/coreos/go-iptables/iptables"
+	"github.com/google/nftables"
+	"github.com/hashicorp/go-multierror"
+	log "github.com/sirupsen/logrus"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+	"github.com/netbirdio/netbird/client/firewall/firewalld"
+	firewall "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/internal/routemanager/ipfwdstate"
+	"github.com/netbirdio/netbird/client/internal/routemanager/refcounter"
+)
+
+const (
+	tableNat      = "nat"
+	tableMangle   = "mangle"
+	tableRaw      = "raw"
+	tableSecurity = "security"
+
+	chainNameNatPrerouting = "PREROUTING"
+	chainNameRoutingFw     = "netbird-rt-fwd"
+	chainNameRoutingNat    = "netbird-rt-postrouting"
+	chainNameRoutingRdr    = "netbird-rt-redirect"
+	chainNameNATOutput     = "netbird-nat-output"
+	chainNameForward       = "FORWARD"
+	chainNameMangleForward = "netbird-mangle-forward"
+
+	// Peer ACL chain names.
+	chainNameInputRules        = "netbird-acl-input-rules"
+	chainNameInputFilter       = "netbird-acl-input-filter"
+	chainNameForwardFilter     = "netbird-acl-forward-filter"
+	chainNameManglePrerouting  = "netbird-mangle-prerouting"
+	chainNameManglePostrouting = "netbird-mangle-postrouting"
+
+	flushError = "flush: %w"
+
+	firewalldTableName = "firewalld"
+
+	userDataAcceptForwardRuleIif = "frwacceptiif"
+	userDataAcceptForwardRuleOif = "frwacceptoif"
+	userDataAcceptInputRule      = "inputaccept"
+
+	dnatSuffix firewall.RuleID = "_dnat"
+	snatSuffix firewall.RuleID = "_snat"
+
+	// ipv4TCPHeaderSize is the minimum IPv4 (20) + TCP (20) header size for MSS calculation.
+	ipv4TCPHeaderSize = 40
+	// ipv6TCPHeaderSize is the minimum IPv6 (40) + TCP (20) header size for MSS calculation.
+	ipv6TCPHeaderSize = 60
+
+	// maxPrefixesSet 1638 prefixes start to fail, taking some margin
+	maxPrefixesSet       = 1500
+	refreshRulesMapError = "refresh rules map: %w"
+)
+
+var (
+	errFilterTableNotFound = fmt.Errorf("'filter' table not found")
+)
+
+type setInput struct {
+	set      firewall.Set
+	prefixes []netip.Prefix
+}
+
+// family holds the per-address-family nftables state. One instance
+// handles route ACLs, peer ACLs, NAT, DNAT, and MSS clamping for a
+// single family; the top-level Manager owns one for v4 and another
+// for v6. The name predates the peer-ACL absorption; it's effectively
+// the per-family backend now.
+type family struct {
+	conn        *nftables.Conn
+	workTable   *nftables.Table
+	filterTable *nftables.Table
+	chains      map[string]*nftables.Chain
+
+	// filters holds peer + route filter rules keyed by content hash.
+	// AddFilterRule writes here; DeleteFilterRule looks up by id.
+	filters map[firewall.RuleID]*Rule
+
+	// rules holds NAT, DNAT, and external accept rules (auxiliary
+	// plumbing that isn't a filter rule).
+	rules map[firewall.RuleID]*nftables.Rule
+
+	// Peer ACL chain handles.
+	chainInputRules    *nftables.Chain
+	chainPrerouting    *nftables.Chain
+	routingFwChainName string
+
+	ipsetCounter *refcounter.Counter[string, setInput, *nftables.Set]
+
+	af               addrFamily
+	wgIface          iFaceMapper
+	ipFwdState       *ipfwdstate.IPForwardingState
+	legacyManagement bool
+	mtu              uint16
+}
+
+func newFamily(workTable *nftables.Table, wgIface iFaceMapper, mtu uint16) *family {
+	r := &family{
+		conn:               &nftables.Conn{},
+		workTable:          workTable,
+		chains:             make(map[string]*nftables.Chain),
+		filters:            make(map[firewall.RuleID]*Rule),
+		rules:              make(map[firewall.RuleID]*nftables.Rule),
+		routingFwChainName: chainNameRoutingFw,
+		af:                 familyForAddr(workTable.Family == nftables.TableFamilyIPv4),
+		wgIface:            wgIface,
+		ipFwdState:         ipfwdstate.NewIPForwardingState(wgIface.Name()),
+		mtu:                mtu,
+	}
+
+	r.ipsetCounter = refcounter.New(
+		r.createIpSet,
+		r.deleteIpSet,
+	)
+
+	var err error
+	r.filterTable, err = r.loadFilterTable()
+	if err != nil {
+		log.Debugf("ip filter table not found: %v", err)
+	}
+
+	return r
+}
+
+func (r *family) init(workTable *nftables.Table) error {
+	r.workTable = workTable
+
+	if err := r.removeAcceptFilterRules(); err != nil {
+		log.Errorf("failed to clean up rules from filter table: %s", err)
+	}
+
+	if err := r.createContainers(); err != nil {
+		return fmt.Errorf("create containers: %w", err)
+	}
+
+	if err := r.setupDataPlaneMark(); err != nil {
+		log.Errorf("failed to set up data plane mark: %v", err)
+	}
+
+	if err := r.createDefaultChains(); err != nil {
+		return fmt.Errorf("create default acl chains: %w", err)
+	}
+
+	return nil
+}
+
+// Reset cleans existing nftables filter table rules from the system
+func (r *family) Reset() error {
+	// clear without deleting the ipsets, the nf table will be deleted by the caller
+	r.ipsetCounter.Clear()
+
+	var merr *multierror.Error
+
+	if err := r.removeAcceptFilterRules(); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("remove accept filter rules: %w", err))
+	}
+
+	if err := firewalld.UntrustInterface(r.wgIface.Name()); err != nil {
+		merr = multierror.Append(merr, err)
+	}
+
+	if err := r.removeNatPreroutingRules(); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("remove filter prerouting rules: %w", err))
+	}
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+func (r *family) loadFilterTable() (*nftables.Table, error) {
+	tables, err := r.conn.ListTablesOfFamily(r.af.tableFamily)
+	if err != nil {
+		return nil, fmt.Errorf("list tables: %w", err)
+	}
+
+	for _, table := range tables {
+		if table.Name == "filter" {
+			return table, nil
+		}
+	}
+
+	return nil, errFilterTableNotFound
+}
+
+func hookName(hook *nftables.ChainHook) string {
+	if hook == nil {
+		return "unknown"
+	}
+	switch *hook {
+	case *nftables.ChainHookForward:
+		return chainNameForward
+	case *nftables.ChainHookInput:
+		return chainNameInput
+	default:
+		return fmt.Sprintf("hook(%d)", *hook)
+	}
+}
+
+func familyName(family nftables.TableFamily) string {
+	switch family {
+	case nftables.TableFamilyIPv4:
+		return "ip"
+	case nftables.TableFamilyIPv6:
+		return "ip6"
+	case nftables.TableFamilyINet:
+		return "inet"
+	default:
+		return fmt.Sprintf("family(%d)", family)
+	}
+}
+
+func (r *family) iptablesProto() iptables.Protocol {
+	if r.af.tableFamily == nftables.TableFamilyIPv6 {
+		return iptables.ProtocolIPv6
+	}
+	return iptables.ProtocolIPv4
+}
+
+func (r *family) refreshRulesMap() error {
+	var merr *multierror.Error
+	newRules := make(map[firewall.RuleID]*nftables.Rule)
+	for _, chain := range r.chains {
+		rules, err := r.conn.GetRules(chain.Table, chain)
+		if err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("list rules for chain %s: %w", chain.Name, err))
+			// preserve existing entries for this chain since we can't verify their state
+			for k, v := range r.rules {
+				if v.Chain != nil && v.Chain.Name == chain.Name {
+					newRules[k] = v
+				}
+			}
+			continue
+		}
+		for _, rule := range rules {
+			if len(rule.UserData) > 0 {
+				newRules[firewall.RuleID(rule.UserData)] = rule
+			}
+		}
+	}
+	r.rules = newRules
+	return nberrors.FormatErrorOrNil(merr)
+}