[client] make AllowedIPs swap self-healing on WireGuard errors

Address CodeRabbit review on #6799.

Decrement now decides whether to reprogram based on whether the currently
installed peer still holds references (e.peers[e.active] > 0) instead of
whether the released peer was the active one. If a prior swap's remove/add
failed, e.active was left pointing at a peer with zero references and every
later Decrement returned early, permanently stranding the prefix and leaking
the entry. The active peer is now detached only when e.active != "", and a
failed hand-off is retried on the next Decrement/Increment.

Also clear currentPeerKey unconditionally in the static handler's
RemoveAllowedIPs (matching dynamic/dnsinterceptor) and assert Increment
errors in the tests. Added self-heal tests for the failed remove/add paths.

agent-network/README.md

@@ -1,16 +1,47 @@
 # NetBird Agent Network
 
-Agent Network is NetBird's access control layer for AI agents and the people who run
-them. It gives every agent a real identity, tied to your identity provider (IdP), and
-governs what it can reach — the LLM APIs and AI gateways it can call, and the internal
-resources it can access. Traffic flows only over the encrypted NetBird tunnel, scoped by
-policy, with no API keys to leak.
+Agent Network is NetBird's access control layer for AI agents and the people who run them. 
+It gives every agent a real identity, tied to an identity provider (IdP), and governs what it can reach: LLM APIs and 
+AI gateways it can call, and the internal resources it can access. Traffic flows only over the encrypted NetBird tunnel, 
+scoped by policy, with no API keys or other credentials to leak. It also gives you control over cost and token usage.
 
-> **Beta.** Agent Network is open source and can be self-hosted on your own
-> infrastructure.
+Because every LLM request passes through an
+identity-aware proxy, you can:
+
+- **Set spending and rate limits** per agent, per user, or per team — with hard caps
+  that stop requests once a budget is reached.
+- **Restrict models and providers** so agents can only call approved (and cost-appropriate)
+  endpoints, keeping expensive models off-limits unless explicitly allowed.
+- **Attribute usage** by tracking token consumption and cost per identity, group, or cost center so every
+  request is tied back to the agent and person responsible.
+- **Reuse your existing AI gateway** — point the proxy at a gateway you already run,
+  keeping its routing and config in place while it adds identity on top, so you skip
+  API key distribution.
+
+https://github.com/user-attachments/assets/44d18286-d8ab-49f8-a457-98ccd66f3268
+
+> **Beta.** Agent Network is in beta, but it's stable and already running in
+> production environments. It's fully open source and can be self-hosted on your own
+> infrastructure, with no vendor lock-in and no data leaving your environment.
 
 ## How it works
 
+Say you have a simple use case: your Engineering or IT team needs access to Claude Code or Codex, and you want visibility into usage plus the ability to enforce budgets.
+How can you do that without creating a dedicated API key for every team?
+
+With Agent Network you get a private endpoint inside your network, for example: https://mirror.netbird.ai
+Teams configure their agents to point to that endpoint instead of using individual API keys directly. 
+
+This endpoint is only reachable when users are connected to your NetBird network and authenticated through your IdP. Otherwise, it is not accessible from the public internet.
+You can then use this private endpoint to configure your AI agents, whether that is Claude Code, Codex, or another tool.
+
+## Quickstart
+
+Full step-by-step setup:
+**https://docs.netbird.io/agent-network/quickstart**
+
+## Architecture 
+
 Agent Network is built on two existing NetBird capabilities:
 
 - **Overlay network** — the encrypted WireGuard mesh between peers.
@@ -22,6 +53,9 @@ LLM traffic is routed through the proxy's identity-aware pipeline, while interna
 resources (databases, internal APIs, self-hosted models) are reached directly over
 peer-to-peer WireGuard tunnels, governed by the same identities and access policies.
 
+<img width="4720" height="2218" alt="image" src="https://github.com/user-attachments/assets/1afa5da1-4b82-4f8a-a7a8-f417efadf1eb" />
+
+
 ## Where the code lives
 
 There is no separate "agent-network" service — it reuses the reverse-proxy and management

client/cmd/login.go

@@ -17,7 +17,9 @@ import (
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	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"
 )
@@ -331,6 +333,14 @@ func doForegroundLogin(ctx context.Context, cmd *cobra.Command, setupKey string,
 		return fmt.Errorf("read config file %s: %v", configFilePath, err)
 	}
 
+	// 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)

client/cmd/service.go

@@ -5,6 +5,7 @@ package cmd
 import (
 	"context"
 	"fmt"
+	"net/http"
 	"runtime"
 	"strings"
 	"sync"
@@ -22,15 +23,21 @@ 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
+	jsonServ         *http.Server
+	jsonServMu       sync.Mutex
 	serverInstance   *server.Server
 	serverInstanceMu sync.Mutex
 }
@@ -46,6 +53,8 @@ func init() {
 	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 +5,6 @@ package cmd
 import (
 	"context"
 	"fmt"
-	"net"
-	"os"
-	"strings"
 	"time"
 
 	"github.com/kardianos/service"
@@ -22,41 +19,56 @@ import (
 	"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
+}
+
 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])
-	}
-
-	listen, err := net.Listen(split[0], split[1])
+	daemonListener, err := listenOnAddress(daemonAddr)
 	if err != nil {
 		return fmt.Errorf("listen daemon interface: %w", 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])
+	var jsonListener *socketListener
+	if enableJSONSocket {
+		jsonListener, err = listenOnAddress(jsonSocket)
+		if err != nil {
+			_ = daemonListener.Close()
+			return fmt.Errorf("listen daemon JSON interface: %w", err)
+		}
+	} else {
+		removeStaleUnixSocketForAddress(jsonSocket)
+	}
+
+	go func() {
+		defer daemonListener.Close()
+		if jsonListener != nil {
+			defer jsonListener.Close()
+		}
+
+		if err := daemonListener.chmodUnixSocket("daemon"); err != nil {
+			log.Error(err)
+			return
+		}
+		if jsonListener != nil {
+			if err := jsonListener.chmodUnixSocket("daemon JSON"); err != nil {
+				log.Error(err)
 				return
 			}
 		}
@@ -71,8 +83,16 @@ func (p *program) Start(svc service.Service) error {
 		p.serverInstance = serverInstance
 		p.serverInstanceMu.Unlock()
 
-		log.Printf("started daemon server: %v", split[1])
-		if err := p.serv.Serve(listen); err != nil {
+		if jsonListener != nil {
+			if err := p.startJSONGateway(jsonListener, daemonAddr); err != nil {
+				log.Fatalf("failed to start daemon JSON server: %v", err)
+			}
+		} else {
+			log.Debug("daemon JSON socket disabled")
+		}
+
+		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)
 		}
 	}()
@@ -92,6 +112,20 @@ func (p *program) Stop(srv service.Service) error {
 
 	p.cancel()
 
+	p.jsonServMu.Lock()
+	jsonServ := p.jsonServ
+	p.jsonServMu.Unlock()
+	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()
 	}
@@ -148,6 +182,9 @@ var runCmd = &cobra.Command{
 		if err != nil {
 			return err
 		}
+		if err := validateJSONSocketFlags(); err != nil {
+			return err
+		}
 
 		return s.Run()
 	},
@@ -162,6 +199,9 @@ var startCmd = &cobra.Command{
 		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)
@@ -198,6 +238,9 @@ var restartCmd = &cobra.Command{
 		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)

client/cmd/service_installer.go

@@ -67,6 +67,10 @@ func buildServiceArguments() []string {
 		args = append(args, "--disable-networks")
 	}
 
+	if enableJSONSocket {
+		args = append(args, "--enable-json-socket", "--json-socket", jsonSocket)
+	}
+
 	return args
 }
 
@@ -106,6 +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 +1,52 @@
+//go:build !ios && !android
+
+package cmd
+
+import (
+	"context"
+	"errors"
+	"net"
+	"net/http"
+	"strings"
+	"time"
+
+	"github.com/grpc-ecosystem/grpc-gateway/v2/runtime"
+	log "github.com/sirupsen/logrus"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/credentials/insecure"
+
+	"github.com/netbirdio/netbird/client/proto"
+)
+
+func grpcGatewayEndpoint(addr string) string {
+	return strings.TrimPrefix(addr, "tcp://")
+}
+
+func (p *program) startJSONGateway(jsonListener *socketListener, daemonEndpoint string) error {
+	mux := runtime.NewServeMux()
+	opts := []grpc.DialOption{grpc.WithTransportCredentials(insecure.NewCredentials())}
+	if err := proto.RegisterDaemonServiceHandlerFromEndpoint(p.ctx, mux, grpcGatewayEndpoint(daemonEndpoint), opts); err != nil {
+		return err
+	}
+
+	jsonServer := &http.Server{
+		Handler:           mux,
+		ReadHeaderTimeout: 5 * time.Second,
+		BaseContext: func(net.Listener) context.Context {
+			return p.ctx
+		},
+	}
+
+	p.jsonServMu.Lock()
+	p.jsonServ = jsonServer
+	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_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

@@ -23,6 +23,7 @@ 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"`
@@ -30,6 +31,7 @@ type serviceParams struct {
 	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"`
 }
 
@@ -75,6 +77,7 @@ func currentServiceParams() *serviceParams {
 	params := &serviceParams{
 		LogLevel:              logLevel,
 		DaemonAddr:            daemonAddr,
+		JSONSocket:            jsonSocket,
 		ManagementURL:         managementURL,
 		ConfigPath:            configPath,
 		LogFiles:              logFiles,
@@ -82,6 +85,7 @@ func currentServiceParams() *serviceParams {
 		DisableUpdateSettings: updateSettingsDisabled,
 		EnableCapture:         captureEnabled,
 		DisableNetworks:       networksDisabled,
+		EnableJSONSocket:      enableJSONSocket,
 	}
 
 	if len(serviceEnvVars) > 0 {
@@ -113,9 +117,8 @@ 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
 	}
@@ -124,6 +127,14 @@ func applyServiceParams(cmd *cobra.Command, params *serviceParams) {
 		daemonAddr = params.DaemonAddr
 	}
 
+	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
 	// that an explicit clear (--management-url "") persists across reinstalls.
 	if !rootCmd.PersistentFlags().Changed("management-url") {

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) {
@@ -530,6 +556,7 @@ func fieldToGlobalVar(field string) string {
 	m := map[string]string{
 		"LogLevel":              "logLevel",
 		"DaemonAddr":            "daemonAddr",
+		"JSONSocket":            "jsonSocket",
 		"ManagementURL":         "managementURL",
 		"ConfigPath":            "configPath",
 		"LogFiles":              "logFiles",
@@ -537,6 +564,7 @@ func fieldToGlobalVar(field string) string {
 		"DisableUpdateSettings": "updateSettingsDisabled",
 		"EnableCapture":         "captureEnabled",
 		"DisableNetworks":       "networksDisabled",
+		"EnableJSONSocket":      "enableJSONSocket",
 		"ServiceEnvVars":        "serviceEnvVars",
 	}
 	if v, ok := m[field]; ok {

client/cmd/service_socket.go

@@ -0,0 +1,111 @@
+//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 == "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]://[path|host:port] format: %q", addr)
+	}
+
+	switch network {
+	case "unix", "tcp":
+		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/up.go

@@ -22,6 +22,8 @@ import (
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/proto"
+	nbnet "github.com/netbirdio/netbird/client/net"
+	"github.com/netbirdio/netbird/client/server"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/shared/management/domain"
 	"github.com/netbirdio/netbird/util"
@@ -229,6 +231,24 @@ func runInForegroundMode(ctx context.Context, cmd *cobra.Command, activeProf *pr
 
 	_, _ = profilemanager.UpdateOldManagementURL(ctx, config, configFilePath)
 
+	// 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)

client/internal/auth/device_flow.go

@@ -259,12 +259,18 @@ func (d *DeviceAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowIn
 	ticker := time.NewTicker(interval)
 	defer ticker.Stop()
 
+	log.Infof("device flow: waiting for user authorization, polling token endpoint every %s, code expires in %s", interval, timeout)
+
+	start := time.Now()
+	polls := 0
+
 	for {
 		select {
 		case <-waitCtx.Done():
 			return TokenInfo{}, waitCtx.Err()
 		case <-ticker.C:
 
+			polls++
 			tokenResponse, err := d.requestToken(info)
 			if err != nil {
 				return TokenInfo{}, fmt.Errorf("parsing token response failed with error: %v", err)
@@ -272,10 +278,12 @@ func (d *DeviceAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowIn
 
 			if tokenResponse.Error != "" {
 				if tokenResponse.Error == "authorization_pending" {
+					log.Tracef("device flow: authorization still pending after poll %d", polls)
 					continue
 				} else if tokenResponse.Error == "slow_down" {
 					interval += (3 * time.Second)
 					ticker.Reset(interval)
+					log.Infof("device flow: IdP requested slow_down, polling interval increased to %s", interval)
 					continue
 				}
 
@@ -296,6 +304,7 @@ func (d *DeviceAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowIn
 				return TokenInfo{}, fmt.Errorf("validate access token failed with error: %v", err)
 			}
 
+			log.Infof("device flow: user authorization confirmed after %d polls in %s", polls, time.Since(start).Round(time.Second))
 			return tokenInfo, err
 		}
 	}

client/internal/auth/pkce_flow.go

@@ -188,6 +188,8 @@ func (p *PKCEAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowInfo
 	waitCtx, cancel := context.WithTimeout(ctx, timeout)
 	defer cancel()
 
+	log.Infof("pkce flow: waiting for authorization callback on %s, timeout %s", p.oAuthConfig.RedirectURL, timeout)
+
 	tokenChan := make(chan *oauth2.Token, 1)
 	errChan := make(chan error, 1)
 
@@ -221,6 +223,7 @@ func (p *PKCEAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowInfo
 func (p *PKCEAuthorizationFlow) startServer(server *http.Server, tokenChan chan<- *oauth2.Token, errChan chan<- error) {
 	mux := http.NewServeMux()
 	mux.HandleFunc("/", func(w http.ResponseWriter, req *http.Request) {
+		log.Infof("pkce flow: received authorization callback from IdP")
 		cert := p.providerConfig.ClientCertPair
 		if cert != nil {
 			tr := &http.Transport{
@@ -271,11 +274,18 @@ func (p *PKCEAuthorizationFlow) handleRequest(req *http.Request) (*oauth2.Token,
 		return nil, fmt.Errorf("authentication failed: missing code")
 	}
 
-	return p.oAuthConfig.Exchange(
+	exchangeStart := time.Now()
+	token, err := p.oAuthConfig.Exchange(
 		req.Context(),
 		code,
 		oauth2.SetAuthURLParam("code_verifier", p.codeVerifier),
 	)
+	if err != nil {
+		return nil, err
+	}
+
+	log.Infof("pkce flow: authorization code exchanged for token in %s", time.Since(exchangeStart).Round(time.Millisecond))
+	return token, nil
 }
 
 func (p *PKCEAuthorizationFlow) parseOAuthToken(token *oauth2.Token) (TokenInfo, error) {

client/internal/conn_mgr.go

@@ -109,7 +109,7 @@ func (e *ConnMgr) UpdatedRemoteFeatureFlag(ctx context.Context, enabled bool) er
 			return nil
 		}
 
-		log.Warnf("lazy connection manager is enabled by management feature flag")
+		log.Infof("lazy connection manager is enabled by the management feature flag")
 		e.initLazyManager(ctx)
 		e.statusRecorder.UpdateLazyConnection(true)
 		return e.addPeersToLazyConnManager()

client/internal/debug/debug.go

@@ -480,7 +480,6 @@ func (g *BundleGenerator) addStatus() error {
 
 		fullStatus := g.statusRecorder.GetFullStatus()
 		protoFullStatus := nbstatus.ToProtoFullStatus(fullStatus)
-		protoFullStatus.Events = g.statusRecorder.GetEventHistory()
 		overview := nbstatus.ConvertToStatusOutputOverview(protoFullStatus, nbstatus.ConvertOptions{
 			Anonymize:     g.anonymize,
 			ProfileName:   profName,

client/internal/engine.go

@@ -551,7 +551,7 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 		} else {
 			log.Infof("running rosenpass in strict mode")
 		}
-		e.rpManager, err = rosenpass.NewManager(e.config.PreSharedKey, e.config.WgIfaceName)
+		e.rpManager, err = rosenpass.NewManager(e.config.PreSharedKey, e.config.WgIfaceName, publicKey)
 		if err != nil {
 			return fmt.Errorf("create rosenpass manager: %w", err)
 		}
@@ -2605,13 +2605,14 @@ func (e *Engine) updateForwardRules(rules []*mgmProto.ForwardingRule) ([]firewal
 
 func (e *Engine) toExcludedLazyPeers(rules []firewallManager.ForwardRule, peers []*mgmProto.RemotePeerConfig) map[string]bool {
 	excludedPeers := make(map[string]bool)
+
+	// Ingress forward targets: inbound forwarded traffic is initiated remotely and
+	// cannot wake a lazy connection, so the peer routing the target must stay
+	// permanently connected. AllowedIPs are already parsed on the peer conn, so
+	// reuse those typed prefixes instead of re-parsing the network map strings.
 	for _, r := range rules {
-		ip := r.TranslatedAddress
 		for _, p := range peers {
-			for _, allowedIP := range p.GetAllowedIps() {
-				if allowedIP != ip.String() {
-					continue
-				}
+			if e.peerRoutesAddr(p, r.TranslatedAddress) {
 				log.Infof("exclude forwarder peer from lazy connection: %s", p.GetWgPubKey())
 				excludedPeers[p.GetWgPubKey()] = true
 			}
@@ -2621,6 +2622,27 @@ func (e *Engine) toExcludedLazyPeers(rules []firewallManager.ForwardRule, peers
 	return excludedPeers
 }
 
+// peerRoutesAddr reports whether the peer is a router for addr, matched against
+// the peer's already-parsed AllowedIPs from the store (the same typed value the
+// lazy manager consumes) rather than re-parsing the network map strings.
+func (e *Engine) peerRoutesAddr(p *mgmProto.RemotePeerConfig, addr netip.Addr) bool {
+	prefixes, ok := e.peerStore.AllowedIPs(p.GetWgPubKey())
+	if !ok {
+		return false
+	}
+	return prefixesContain(prefixes, addr)
+}
+
+// prefixesContain reports whether addr falls within any of the prefixes.
+func prefixesContain(prefixes []netip.Prefix, addr netip.Addr) bool {
+	for _, prefix := range prefixes {
+		if prefix.Contains(addr) {
+			return true
+		}
+	}
+	return false
+}
+
 // isChecksEqual checks if two slices of checks are equal.
 func isChecksEqual(checks1, checks2 []*mgmProto.Checks) bool {
 	normalize := func(checks []*mgmProto.Checks) []string {

client/internal/engine_lazy_exclude_test.go

@@ -0,0 +1,87 @@
+package internal
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+
+	firewallManager "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/internal/peer"
+	"github.com/netbirdio/netbird/client/internal/peerstore"
+	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
+)
+
+func TestPrefixesContain(t *testing.T) {
+	tests := []struct {
+		name     string
+		prefixes []string
+		addr     string
+		want     bool
+	}{
+		{name: "own overlay /32 matches", prefixes: []string{"100.110.8.145/32"}, addr: "100.110.8.145", want: true},
+		{name: "addr inside routed subnet", prefixes: []string{"10.121.0.0/16"}, addr: "10.121.208.4", want: true},
+		{name: "addr outside subnet", prefixes: []string{"10.121.0.0/16"}, addr: "10.122.0.1", want: false},
+		{name: "different /32", prefixes: []string{"100.110.8.145/32"}, addr: "100.110.8.146", want: false},
+		{name: "ipv6 /128 matches", prefixes: []string{"fd00::1/128"}, addr: "fd00::1", want: true},
+		{name: "no prefixes", prefixes: nil, addr: "10.121.208.4", want: false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			prefixes := make([]netip.Prefix, 0, len(tt.prefixes))
+			for _, p := range tt.prefixes {
+				prefixes = append(prefixes, netip.MustParsePrefix(p))
+			}
+			require.Equal(t, tt.want, prefixesContain(prefixes, netip.MustParseAddr(tt.addr)))
+		})
+	}
+}
+
+// TestToExcludedLazyPeers_ForwardTarget guards a regression: the forward-target
+// peer (the peer routing a ForwardRule.TranslatedAddress) must be excluded from
+// lazy connections, matched via the peer's already-parsed AllowedIPs.
+func TestToExcludedLazyPeers_ForwardTarget(t *testing.T) {
+	const targetPeerKey = "cccccccccccccccccccccccccccccccccccccccccc0="
+	const otherPeerKey = "dddddddddddddddddddddddddddddddddddddddddd0="
+
+	store := peerstore.NewConnStore()
+	store.AddPeerConn(targetPeerKey, newTestConn(t, targetPeerKey, "100.110.8.145/32"))
+	store.AddPeerConn(otherPeerKey, newTestConn(t, otherPeerKey, "100.110.9.10/32"))
+
+	e := &Engine{peerStore: store}
+
+	peers := []*mgmProto.RemotePeerConfig{
+		{WgPubKey: targetPeerKey, AllowedIps: []string{"100.110.8.145/32"}},
+		{WgPubKey: otherPeerKey, AllowedIps: []string{"100.110.9.10/32"}},
+	}
+	rules := []firewallManager.ForwardRule{
+		{TranslatedAddress: netip.MustParseAddr("100.110.8.145")},
+	}
+
+	excluded := e.toExcludedLazyPeers(rules, peers)
+
+	require.True(t, excluded[targetPeerKey], "forward-target peer must be excluded from lazy connections")
+	require.False(t, excluded[otherPeerKey], "non-target peer must not be excluded")
+	require.Len(t, excluded, 1)
+}
+
+func TestToExcludedLazyPeers_NoRules(t *testing.T) {
+	e := &Engine{peerStore: peerstore.NewConnStore()}
+
+	peers := []*mgmProto.RemotePeerConfig{
+		{WgPubKey: "peer-a", AllowedIps: []string{"100.110.8.145/32"}},
+	}
+
+	require.Empty(t, e.toExcludedLazyPeers(nil, peers))
+}
+
+func newTestConn(t *testing.T, key, allowedIP string) *peer.Conn {
+	t.Helper()
+	conn, err := peer.NewConn(peer.ConnConfig{
+		Key:      key,
+		WgConfig: peer.WgConfig{AllowedIps: []netip.Prefix{netip.MustParsePrefix(allowedIP)}},
+	}, peer.ServiceDependencies{})
+	require.NoError(t, err)
+	return conn
+}

client/internal/netflow/logger/logger.go

@@ -27,7 +27,7 @@ type Logger struct {
 	wgIfaceNetV6       netip.Prefix
 	dnsCollection      atomic.Bool
 	exitNodeCollection atomic.Bool
-	Store              types.Store
+	Store              types.AggregatingStore
 }
 
 func New(statusRecorder *peer.Status, wgIfaceIPNet, wgIfaceIPNetV6 netip.Prefix) *Logger {
@@ -35,7 +35,7 @@ func New(statusRecorder *peer.Status, wgIfaceIPNet, wgIfaceIPNetV6 netip.Prefix)
 		statusRecorder: statusRecorder,
 		wgIfaceNet:     wgIfaceIPNet,
 		wgIfaceNetV6:   wgIfaceIPNetV6,
-		Store:          store.NewMemoryStore(),
+		Store:          store.NewAggregatingMemoryStore(),
 	}
 }
 
@@ -125,6 +125,10 @@ func (l *Logger) stop() {
 	l.mux.Unlock()
 }
 
+func (l *Logger) ResetAggregationWindow() types.FlowEventAggregator {
+	return l.Store.ResetAggregationWindow()
+}
+
 func (l *Logger) GetEvents() []*types.Event {
 	return l.Store.GetEvents()
 }

client/internal/netflow/manager.go

@@ -9,12 +9,14 @@ import (
 	"sync"
 	"time"
 
+	"github.com/cenkalti/backoff/v4"
 	"github.com/google/uuid"
 	log "github.com/sirupsen/logrus"
 	"google.golang.org/protobuf/types/known/timestamppb"
 
 	"github.com/netbirdio/netbird/client/internal/netflow/conntrack"
 	"github.com/netbirdio/netbird/client/internal/netflow/logger"
+	"github.com/netbirdio/netbird/client/internal/netflow/store"
 	nftypes "github.com/netbirdio/netbird/client/internal/netflow/types"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/flow/client"
@@ -23,14 +25,16 @@ import (
 
 // Manager handles netflow tracking and logging
 type Manager struct {
-	mux            sync.Mutex
-	shutdownWg     sync.WaitGroup
-	logger         nftypes.FlowLogger
-	flowConfig     *nftypes.FlowConfig
-	conntrack      nftypes.ConnTracker
-	receiverClient *client.GRPCClient
-	publicKey      []byte
-	cancel         context.CancelFunc
+	mux               sync.Mutex
+	shutdownWg        sync.WaitGroup
+	logger            nftypes.FlowLogger
+	flowConfig        *nftypes.FlowConfig
+	conntrack         nftypes.ConnTracker
+	receiverClient    *client.GRPCClient
+	eventsWithoutAcks nftypes.Store
+	publicKey         []byte
+	cancel            context.CancelFunc
+	retryInterval     time.Duration
 }
 
 // NewManager creates a new netflow manager
@@ -48,9 +52,11 @@ func NewManager(iface nftypes.IFaceMapper, publicKey []byte, statusRecorder *pee
 	}
 
 	return &Manager{
-		logger:    flowLogger,
-		conntrack: ct,
-		publicKey: publicKey,
+		logger:            flowLogger,
+		conntrack:         ct,
+		publicKey:         publicKey,
+		retryInterval:     time.Second,
+		eventsWithoutAcks: store.NewMemoryStore(),
 	}
 }
 
@@ -66,6 +72,7 @@ func (m *Manager) needsNewClient(previous *nftypes.FlowConfig) bool {
 }
 
 // enableFlow starts components for flow tracking
+// must be called under m.mux lock
 func (m *Manager) enableFlow(previous *nftypes.FlowConfig) error {
 	// first make sender ready so events don't pile up
 	if m.needsNewClient(previous) {
@@ -85,6 +92,7 @@ func (m *Manager) enableFlow(previous *nftypes.FlowConfig) error {
 	return nil
 }
 
+// must be called under m.mux lock
 func (m *Manager) resetClient() error {
 	if m.receiverClient != nil {
 		if err := m.receiverClient.Close(); err != nil {
@@ -107,14 +115,19 @@ func (m *Manager) resetClient() error {
 	ctx, cancel := context.WithCancel(context.Background())
 	m.cancel = cancel
 
-	m.shutdownWg.Add(2)
+	m.shutdownWg.Add(3)
+	flowConfigInterval := m.flowConfig.Interval
 	go func() {
 		defer m.shutdownWg.Done()
-		m.receiveACKs(ctx, flowClient)
+		m.receiveACKs(ctx, flowClient, flowConfigInterval)
 	}()
 	go func() {
 		defer m.shutdownWg.Done()
-		m.startSender(ctx)
+		m.startSender(ctx, flowConfigInterval)
+	}()
+	go func() {
+		defer m.shutdownWg.Done()
+		m.startRetries(ctx, flowConfigInterval)
 	}()
 
 	return nil
@@ -198,8 +211,8 @@ func (m *Manager) GetLogger() nftypes.FlowLogger {
 	return m.logger
 }
 
-func (m *Manager) startSender(ctx context.Context) {
-	ticker := time.NewTicker(m.flowConfig.Interval)
+func (m *Manager) startSender(ctx context.Context, flowConfigInterval time.Duration) {
+	ticker := time.NewTicker(flowConfigInterval)
 	defer ticker.Stop()
 
 	for {
@@ -207,27 +220,29 @@ func (m *Manager) startSender(ctx context.Context) {
 		case <-ctx.Done():
 			return
 		case <-ticker.C:
-			events := m.logger.GetEvents()
+			collectedEvents := m.logger.ResetAggregationWindow()
+			events := collectedEvents.GetAggregatedEvents()
 			for _, event := range events {
+				m.eventsWithoutAcks.StoreEvent(event)
 				if err := m.send(event); err != nil {
 					log.Errorf("failed to send flow event to server: %v", err)
-					continue
+				} else {
+					log.Tracef("sent flow event: %s", event.ID)
 				}
-				log.Tracef("sent flow event: %s", event.ID)
 			}
 		}
 	}
 }
 
-func (m *Manager) receiveACKs(ctx context.Context, client *client.GRPCClient) {
-	err := client.Receive(ctx, m.flowConfig.Interval, func(ack *proto.FlowEventAck) error {
+func (m *Manager) receiveACKs(ctx context.Context, client *client.GRPCClient, flowConfigInterval time.Duration) {
+	err := client.Receive(ctx, flowConfigInterval, func(ack *proto.FlowEventAck) error {
 		id, err := uuid.FromBytes(ack.EventId)
 		if err != nil {
 			log.Warnf("failed to convert ack event id to uuid: %v", err)
 			return nil
 		}
 		log.Tracef("received flow event ack: %s", id)
-		m.logger.DeleteEvents([]uuid.UUID{id})
+		m.eventsWithoutAcks.DeleteEvents([]uuid.UUID{id})
 		return nil
 	})
 
@@ -236,6 +251,51 @@ func (m *Manager) receiveACKs(ctx context.Context, client *client.GRPCClient) {
 	}
 }
 
+// We effectively never drop events (see MaxInterval), which makes eventsWithoutAcks unbounded.
+// We may want to limit the max size of the store, and start dropping oldest events when the threshold is reached.
+func (m *Manager) startRetries(ctx context.Context, flowConfigInterval time.Duration) {
+	timer := time.NewTimer(m.retryInterval)
+	retryBackoff := backoff.WithContext(&backoff.ExponentialBackOff{
+		InitialInterval:     1 * time.Second,
+		RandomizationFactor: 0.5,
+		Multiplier:          1.7,
+		MaxInterval:         flowConfigInterval / 2,
+		MaxElapsedTime:      3 * 30 * 24 * time.Hour, // 3 months
+		Stop:                backoff.Stop,
+		Clock:               backoff.SystemClock,
+	}, ctx)
+	defer timer.Stop()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-timer.C:
+			resetBackoff := true
+			for _, e := range m.eventsWithoutAcks.GetEvents() {
+				if e.Timestamp.Add(time.Second).After(time.Now()) {
+					// grace period on retries to avoid early retries
+					// do not retry if the event is less than 1 sec old
+					continue
+				}
+				if err := m.send(e); err != nil {
+					if nextBackoff := retryBackoff.NextBackOff(); nextBackoff != backoff.Stop {
+						timer = time.NewTimer(nextBackoff)
+						resetBackoff = false
+					} else {
+						resetBackoff = true // we exhausted retries, reset retry loop
+					}
+					break
+				}
+			}
+			if resetBackoff { // use regular retry interval in absence of network errors
+				retryBackoff.Reset()
+				timer = time.NewTimer(m.retryInterval)
+			}
+		}
+	}
+}
+
 func (m *Manager) send(event *nftypes.Event) error {
 	m.mux.Lock()
 	client := m.receiverClient
@@ -250,9 +310,11 @@ func (m *Manager) send(event *nftypes.Event) error {
 
 func toProtoEvent(publicKey []byte, event *nftypes.Event) *proto.FlowEvent {
 	protoEvent := &proto.FlowEvent{
-		EventId:   event.ID[:],
-		Timestamp: timestamppb.New(event.Timestamp),
-		PublicKey: publicKey,
+		EventId:     event.ID[:],
+		Timestamp:   timestamppb.New(event.Timestamp),
+		PublicKey:   publicKey,
+		WindowStart: timestamppb.New(event.WindowStart),
+		WindowEnd:   timestamppb.New(event.WindowEnd),
 		FlowFields: &proto.FlowFields{
 			FlowId:           event.FlowID[:],
 			RuleId:           event.RuleID,
@@ -267,6 +329,9 @@ func toProtoEvent(publicKey []byte, event *nftypes.Event) *proto.FlowEvent {
 			TxBytes:          event.TxBytes,
 			SourceResourceId: event.SourceResourceID,
 			DestResourceId:   event.DestResourceID,
+			NumOfStarts:      event.NumOfStarts,
+			NumOfEnds:        event.NumOfEnds,
+			NumOfDrops:       event.NumOfDrops,
 		},
 	}
 

client/internal/netflow/manager_integration_test.go

@@ -0,0 +1,291 @@
+package netflow
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"net"
+	"net/netip"
+	"slices"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+	"github.com/netbirdio/netbird/client/internal/netflow/types"
+	"github.com/netbirdio/netbird/flow/proto"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"google.golang.org/grpc"
+)
+
+type testServer struct {
+	proto.UnimplementedFlowServiceServer
+	events         chan *proto.FlowEvent
+	acks           chan *proto.FlowEventAck
+	grpcSrv        *grpc.Server
+	addr           string
+	handlerDone    chan struct{} // signaled each time Events() exits
+	handlerStarted chan struct{} // signaled each time Events() begins
+}
+
+func newTestServer(t *testing.T) *testServer {
+	listener, err := net.Listen("tcp", "127.0.0.1:0")
+	require.NoError(t, err)
+
+	s := &testServer{
+		events:         make(chan *proto.FlowEvent, 100),
+		acks:           make(chan *proto.FlowEventAck, 100),
+		grpcSrv:        grpc.NewServer(),
+		addr:           listener.Addr().String(),
+		handlerDone:    make(chan struct{}, 10),
+		handlerStarted: make(chan struct{}, 10),
+	}
+
+	proto.RegisterFlowServiceServer(s.grpcSrv, s)
+
+	go func() {
+		if err := s.grpcSrv.Serve(listener); err != nil && !errors.Is(err, grpc.ErrServerStopped) {
+			t.Logf("server error: %v", err)
+		}
+	}()
+
+	t.Cleanup(func() {
+		s.grpcSrv.Stop()
+	})
+
+	return s
+}
+
+func (s *testServer) Events(stream proto.FlowService_EventsServer) error {
+	defer func() {
+		select {
+		case s.handlerDone <- struct{}{}:
+		default:
+		}
+	}()
+
+	err := stream.Send(&proto.FlowEventAck{IsInitiator: true})
+	if err != nil {
+		return err
+	}
+
+	select {
+	case s.handlerStarted <- struct{}{}:
+	default:
+	}
+
+	ctx, cancel := context.WithCancel(stream.Context())
+	defer cancel()
+
+	go func() {
+		defer cancel()
+		for {
+			event, err := stream.Recv()
+			if err != nil {
+				return
+			}
+
+			if !event.IsInitiator {
+				select {
+				case s.events <- event:
+				case <-ctx.Done():
+					return
+				}
+			}
+		}
+	}()
+
+	for {
+		select {
+		case ack := <-s.acks:
+			if err := stream.Send(ack); err != nil {
+				return err
+			}
+		case <-ctx.Done():
+			return ctx.Err()
+		}
+	}
+}
+
+func TestSendEventReceiveAck(t *testing.T) {
+	_, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	t.Cleanup(cancel)
+
+	server := newTestServer(t)
+	manager := createManager(t, server.addr, 60*time.Second) // set high to prevent retries in this test
+	defer manager.Close()
+
+	assert.Eventually(t, func() bool {
+		select {
+		case <-server.handlerStarted:
+			return true
+		default:
+			return false
+		}
+	}, 3*time.Second, 100*time.Millisecond)
+
+	event1 := types.EventFields{
+		FlowID:    uuid.New(),
+		Type:      types.TypeStart,
+		Direction: types.Ingress,
+		DestIP:    ipAddr("172.16.1.2"),
+		DestPort:  2345,
+		Protocol:  6,
+	}
+	manager.logger.StoreEvent(event1)
+	event2 := types.EventFields{
+		FlowID:    uuid.New(),
+		Type:      types.TypeStart,
+		Direction: types.Ingress,
+		DestIP:    ipAddr("172.16.1.1"),
+		DestPort:  1234,
+		Protocol:  6,
+	}
+	manager.logger.StoreEvent(event2)
+
+	// verify the server received logged events
+	serverSideEvents := make([]*proto.FlowEvent, 0)
+	assert.Eventually(t, func() bool {
+		select {
+		case event := <-server.events:
+			serverSideEvents = append(serverSideEvents, event)
+			if len(serverSideEvents) == 2 {
+				return true
+			}
+		default:
+			if len(serverSideEvents) == 2 {
+				return true
+			}
+		}
+		return false
+	}, 5*time.Second, 100*time.Millisecond)
+
+	serverSideFlowIds := make([]uuid.UUID, 0, 2)
+	slices.Values(serverSideEvents)(func(e *proto.FlowEvent) bool {
+		id, err := uuid.FromBytes(e.FlowFields.FlowId)
+		assert.NoError(t, err)
+		serverSideFlowIds = append(serverSideFlowIds, id)
+		return true
+	})
+	assert.ElementsMatch(t, []uuid.UUID{event1.FlowID, event2.FlowID}, serverSideFlowIds)
+
+	// verify the manager tracks un-acked events
+	unackedEvents := manager.eventsWithoutAcks.GetEvents()
+	assert.Len(t, unackedEvents, 2)
+	flowIds := make([]uuid.UUID, 0)
+	slices.Values(unackedEvents)(func(e *types.Event) bool {
+		flowIds = append(flowIds, e.FlowID)
+		return true
+	})
+	assert.ElementsMatch(t, flowIds, []uuid.UUID{event1.FlowID, event2.FlowID})
+}
+
+// verify handling of retries:
+//   - unacked events are retried
+//   - when acks arrive, events are removed from the un-acked event tracker
+func TestRetryEvents(t *testing.T) {
+	_, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	t.Cleanup(cancel)
+
+	server := newTestServer(t)
+	manager := createManager(t, server.addr, time.Second) // set low to start retries sooner
+	defer manager.Close()
+
+	assert.Eventually(t, func() bool {
+		select {
+		case <-server.handlerStarted:
+			return true
+		default:
+			return false
+		}
+	}, 3*time.Second, 100*time.Millisecond)
+
+	event1 := types.EventFields{
+		FlowID:    uuid.New(),
+		Type:      types.TypeStart,
+		Direction: types.Ingress,
+		DestIP:    ipAddr("172.16.1.2"),
+		DestPort:  2345,
+		Protocol:  6,
+	}
+	manager.logger.StoreEvent(event1)
+	event2 := types.EventFields{
+		FlowID:    uuid.New(),
+		Type:      types.TypeStart,
+		Direction: types.Ingress,
+		DestIP:    ipAddr("172.16.1.1"),
+		DestPort:  1234,
+		Protocol:  6,
+	}
+	manager.logger.StoreEvent(event2)
+
+	// verify the server received retries of logged events
+	serverSideEvents := make([]*proto.FlowEvent, 0)
+	func() {
+		c := time.After(2500 * time.Millisecond)
+		for {
+			select {
+			case event := <-server.events:
+				serverSideEvents = append(serverSideEvents, event)
+			case <-c:
+				return
+			}
+		}
+	}()
+	assert.True(t, len(serverSideEvents) > 2) // must see retries
+
+	uniqueServerSideEvents := make(map[uuid.UUID]*proto.FlowEvent)
+	slices.Values(serverSideEvents)(func(e *proto.FlowEvent) bool {
+		id, err := uuid.FromBytes(e.FlowFields.FlowId)
+		assert.NoError(t, err)
+		uniqueServerSideEvents[id] = e
+		return true
+	})
+	assert.Contains(t, uniqueServerSideEvents, event1.FlowID)
+	assert.Contains(t, uniqueServerSideEvents, event2.FlowID)
+
+	// ack events
+	server.acks <- &proto.FlowEventAck{EventId: uniqueServerSideEvents[event1.FlowID].EventId}
+	server.acks <- &proto.FlowEventAck{EventId: uniqueServerSideEvents[event2.FlowID].EventId}
+
+	assert.EventuallyWithT(t, func(c *assert.CollectT) {
+		unackedEvents := manager.eventsWithoutAcks.GetEvents()
+		assert.Empty(c, unackedEvents)
+
+	}, 3*time.Second, 100*time.Millisecond)
+}
+
+func createManager(t *testing.T, serverAddr string, retryInterval time.Duration) *Manager {
+	t.Helper()
+
+	mockIFace := &mockIFaceMapper{
+		address: wgaddr.Address{
+			Network: netip.MustParsePrefix("192.168.1.1/32"),
+		},
+		isUserspaceBind: true,
+	}
+
+	publicKey := []byte("test-public-key")
+	manager := NewManager(mockIFace, publicKey, nil)
+	manager.retryInterval = retryInterval
+
+	initialConfig := &types.FlowConfig{
+		Enabled:        true,
+		URL:            fmt.Sprintf("http://%s", serverAddr),
+		TokenPayload:   "initial-payload",
+		TokenSignature: "initial-signature",
+		Interval:       500 * time.Millisecond,
+	}
+
+	err := manager.Update(initialConfig)
+	require.NoError(t, err)
+
+	return manager
+}
+
+func ipAddr(a string) netip.Addr {
+	addr, _ := netip.ParseAddr(a)
+	return addr
+}

client/internal/netflow/store/event_aggregation_test.go

@@ -0,0 +1,365 @@
+package store
+
+import (
+	"math/rand"
+	"net/netip"
+	"testing"
+	"time"
+
+	"github.com/google/uuid"
+	"github.com/netbirdio/netbird/client/internal/netflow/types"
+	"github.com/stretchr/testify/assert"
+)
+
+var random = rand.New(rand.NewSource(time.Now().UnixNano()))
+
+func TestFlowAggregation(t *testing.T) {
+	var protocols = []types.Protocol{types.ICMP, types.ICMPv6, types.TCP, types.UDP}
+	var tests = []struct {
+		description string
+		addresses   [][]netip.Addr
+		dstPort     uint16
+		eventTypes  []types.Type
+	}{
+		{
+			description: "start and stop",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeStart, types.TypeEnd},
+		},
+		{
+			description: "start and drop",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeStart, types.TypeDrop},
+		},
+		{
+			description: "start only",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeStart},
+		},
+		{
+			description: "drop only",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeDrop},
+		}}
+
+	for _, protocol := range protocols {
+		for _, tt := range tests {
+			t.Run(tt.description+" "+protocol.String(), func(t *testing.T) {
+				store := NewAggregatingMemoryStore()
+				store.WindowEnd = time.Now().Add(5 * time.Second)
+
+				allExpected := make([]*types.Event, 0)
+
+				for _, srcAndDst := range tt.addresses {
+					inEvents, expected := generateEvents(srcAndDst[0], srcAndDst[1], tt.dstPort, tt.eventTypes, protocol, types.Ingress, 0, store.WindowStart, store.WindowEnd)
+					for _, e := range inEvents {
+						store.StoreEvent(e)
+					}
+					allExpected = append(allExpected, expected)
+				}
+
+				events := store.GetAggregatedEvents()
+				assert.ElementsMatch(t, events, allExpected)
+			})
+		}
+	}
+}
+
+func TestIcmpEventAggregation(t *testing.T) {
+	var protocols = []types.Protocol{types.ICMP, types.ICMPv6}
+	var icmpTypes = []uint8{1, 2, 3}
+
+	var tests = []struct {
+		description string
+		addresses   [][]netip.Addr
+		eventTypes  []types.Type
+	}{
+		{
+			description: "start and stop",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}},
+			eventTypes:  []types.Type{types.TypeStart, types.TypeEnd},
+		},
+		{
+			description: "start and drop",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}},
+			eventTypes:  []types.Type{types.TypeStart, types.TypeDrop},
+		},
+		{
+			description: "start only",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}},
+			eventTypes:  []types.Type{types.TypeStart},
+		},
+		{
+			description: "drop only",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}},
+			eventTypes:  []types.Type{types.TypeDrop},
+		}}
+
+	for _, protocol := range protocols {
+		for _, tt := range tests {
+			t.Run(tt.description+" "+protocol.String(), func(t *testing.T) {
+				store := NewAggregatingMemoryStore()
+				store.WindowEnd = time.Now().Add(5 * time.Second)
+
+				allExpected := make([]*types.Event, 0)
+				for _, icmpType := range icmpTypes {
+					events, expected := generateEvents(tt.addresses[0][0], tt.addresses[0][1], 0, tt.eventTypes, protocol, types.Ingress, icmpType, store.WindowStart, store.WindowEnd)
+					for _, e := range events {
+						store.StoreEvent(e)
+					}
+					allExpected = append(allExpected, expected)
+				}
+				aggregatedEvents := store.GetAggregatedEvents()
+				assert.Len(t, aggregatedEvents, len(allExpected))
+				assert.ElementsMatch(t, aggregatedEvents, allExpected)
+			})
+		}
+	}
+}
+
+func TestFlowAggregationOfUnknownProtocols(t *testing.T) {
+	var tests = []struct {
+		description string
+		addresses   [][]netip.Addr
+		dstPort     uint16
+		eventTypes  []types.Type
+	}{
+		{
+			description: "start and stop",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeStart, types.TypeEnd},
+		},
+		{
+			description: "start and drop",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeStart, types.TypeDrop},
+		},
+		{
+			description: "start only",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeStart},
+		},
+		{
+			description: "drop only",
+			addresses:   [][]netip.Addr{{netip.MustParseAddr("1.1.1.1"), netip.MustParseAddr("2.2.2.2")}, {netip.MustParseAddr("3.3.3.3"), netip.MustParseAddr("2.2.2.2")}},
+			dstPort:     uint16(random.Uint32() >> 16),
+			eventTypes:  []types.Type{types.TypeDrop},
+		}}
+
+	for _, tt := range tests {
+		t.Run(tt.description+" "+types.ProtocolUnknown.String(), func(t *testing.T) {
+			store := NewAggregatingMemoryStore()
+			store.WindowEnd = time.Now().Add(5 * time.Second)
+
+			allExpected := make([]*types.Event, 0)
+
+			for _, srcAndDst := range tt.addresses {
+				inEvents, expected := generateEventsForUnknownProtocol(srcAndDst[0], srcAndDst[1], tt.dstPort, tt.eventTypes, types.ProtocolUnknown, types.Ingress, store.WindowStart, store.WindowEnd)
+				for _, e := range inEvents {
+					store.StoreEvent(e)
+				}
+				allExpected = append(allExpected, expected...)
+			}
+
+			events := store.GetAggregatedEvents()
+			assert.ElementsMatch(t, events, allExpected)
+		})
+	}
+}
+
+func TestResetAggregationWindow(t *testing.T) {
+	now := time.Now()
+	nowFunc := func() time.Time { return now }
+	store := NewAggregatingMemoryStoreWithTimeFunc(nowFunc)
+	store.StoreEvent(&types.Event{
+		ID:        uuid.New(),
+		Timestamp: time.Now(),
+		EventFields: types.EventFields{
+			FlowID:           uuid.New(),
+			Type:             types.TypeStart,
+			Protocol:         types.TCP,
+			RuleID:           []byte("rule-id-1"),
+			Direction:        types.Ingress,
+			SourceIP:         netip.MustParseAddr("1.1.1.1"),
+			SourcePort:       1234,
+			DestIP:           netip.MustParseAddr("2.2.2.2"),
+			DestPort:         5678,
+			SourceResourceID: []byte("source-resource-id"),
+			DestResourceID:   []byte("dest-resource-id"),
+			RxPackets:        random.Uint64(),
+			TxPackets:        random.Uint64(),
+			RxBytes:          random.Uint64(),
+			TxBytes:          random.Uint64(),
+		},
+	})
+
+	now = now.Add(1 * time.Second)
+	reset := store.ResetAggregationWindow()
+	previousEvents, ok := reset.(*AggregatingMemory)
+	assert.True(t, ok)
+	assert.NotEqual(t, previousEvents.WindowStart, store.WindowStart)
+	assert.Equal(t, previousEvents.WindowEnd, store.WindowStart)
+	assert.NotEmpty(t, previousEvents.events)
+	assert.Empty(t, store.events)
+}
+
+func generateEvents(srcIp, dstIp netip.Addr, dstPort uint16, eventTypes []types.Type, protocol types.Protocol,
+	direction types.Direction, icmpType uint8, windowStart, windowEnd time.Time) ([]*types.Event, *types.Event) {
+	var rxPackets, txPackets, rxBytes, txBytes uint64
+	inEvents := make([]*types.Event, 0)
+	ts := time.Now()
+	flowId := uuid.New()
+	srcPort := uint16(random.Uint32() >> 16)
+
+	for idx, eventType := range eventTypes {
+		e := &types.Event{
+			ID:        uuid.New(),
+			Timestamp: ts.Add(time.Duration(idx) * time.Second),
+			EventFields: types.EventFields{
+				FlowID:           flowId,
+				Type:             eventType,
+				Protocol:         protocol,
+				RuleID:           []byte("rule-id-1"),
+				Direction:        direction,
+				SourceIP:         srcIp,
+				SourcePort:       srcPort,
+				DestIP:           dstIp,
+				DestPort:         dstPort,
+				SourceResourceID: []byte("source-resource-id"),
+				DestResourceID:   []byte("dest-resource-id"),
+				RxPackets:        random.Uint64(),
+				TxPackets:        random.Uint64(),
+				RxBytes:          random.Uint64(),
+				TxBytes:          random.Uint64(),
+			}}
+		rxBytes += e.RxBytes
+		txBytes += e.TxBytes
+		rxPackets += e.RxPackets
+		txPackets += e.TxPackets
+		inEvents = append(inEvents, e)
+		if protocol == types.ICMP || protocol == types.ICMPv6 {
+			e.ICMPType = icmpType
+		}
+	}
+
+	var start, end, drop uint64
+	for _, eventType := range eventTypes {
+		switch eventType {
+		case types.TypeStart:
+			start += 1
+		case types.TypeDrop:
+			drop += 1
+		case types.TypeEnd:
+			end += 1
+		}
+	}
+	aggregatedEvent := &types.Event{
+		ID:          inEvents[0].ID,
+		Timestamp:   inEvents[0].Timestamp,
+		WindowStart: windowStart,
+		WindowEnd:   windowEnd,
+		EventFields: types.EventFields{
+			FlowID:           flowId,
+			Type:             types.TypeUnknown,
+			Protocol:         inEvents[0].Protocol,
+			RuleID:           []byte("rule-id-1"),
+			Direction:        inEvents[0].Direction,
+			SourceIP:         srcIp,
+			SourcePort:       srcPort,
+			DestIP:           dstIp,
+			DestPort:         dstPort,
+			SourceResourceID: []byte("source-resource-id"),
+			DestResourceID:   []byte("dest-resource-id"),
+			RxPackets:        rxPackets,
+			TxPackets:        txPackets,
+			RxBytes:          rxBytes,
+			TxBytes:          txBytes,
+			NumOfStarts:      start,
+			NumOfEnds:        end,
+			NumOfDrops:       drop,
+		}}
+	if protocol == types.ICMP || protocol == types.ICMPv6 {
+		aggregatedEvent.ICMPType = icmpType
+	}
+
+	return inEvents, aggregatedEvent
+}
+
+func generateEventsForUnknownProtocol(srcIp, dstIp netip.Addr, dstPort uint16, eventTypes []types.Type, protocol types.Protocol,
+	direction types.Direction, windowStart, windowEnd time.Time) ([]*types.Event, []*types.Event) {
+	inEvents := make([]*types.Event, 0)
+	expectedEvents := make([]*types.Event, 0)
+
+	ts := time.Now()
+	flowId := uuid.New()
+	srcPort := uint16(random.Uint32() >> 16)
+
+	for idx, eventType := range eventTypes {
+		e := &types.Event{
+			ID:        uuid.New(),
+			Timestamp: ts.Add(time.Duration(idx) * time.Second),
+			EventFields: types.EventFields{
+				FlowID:           flowId,
+				Type:             eventType,
+				Protocol:         protocol,
+				RuleID:           []byte("rule-id-1"),
+				Direction:        direction,
+				SourceIP:         srcIp,
+				SourcePort:       srcPort,
+				DestIP:           dstIp,
+				DestPort:         dstPort,
+				SourceResourceID: []byte("source-resource-id"),
+				DestResourceID:   []byte("dest-resource-id"),
+				RxPackets:        random.Uint64(),
+				TxPackets:        random.Uint64(),
+				RxBytes:          random.Uint64(),
+				TxBytes:          random.Uint64(),
+			}}
+		inEvents = append(inEvents, e)
+
+		var start, end, drop uint64
+		switch eventType {
+		case types.TypeStart:
+			start = 1
+		case types.TypeDrop:
+			drop = 1
+		case types.TypeEnd:
+			end = 1
+		}
+
+		expectedEvents = append(expectedEvents, &types.Event{
+			ID:          e.ID,
+			Timestamp:   e.Timestamp,
+			WindowStart: windowStart,
+			WindowEnd:   windowEnd,
+			EventFields: types.EventFields{
+				FlowID:           flowId,
+				Type:             types.TypeUnknown,
+				Protocol:         e.Protocol,
+				RuleID:           []byte("rule-id-1"),
+				Direction:        e.Direction,
+				SourceIP:         srcIp,
+				SourcePort:       srcPort,
+				DestIP:           dstIp,
+				DestPort:         dstPort,
+				SourceResourceID: []byte("source-resource-id"),
+				DestResourceID:   []byte("dest-resource-id"),
+				RxPackets:        e.RxPackets,
+				TxPackets:        e.TxPackets,
+				RxBytes:          e.RxBytes,
+				TxBytes:          e.TxBytes,
+				NumOfStarts:      start,
+				NumOfEnds:        end,
+				NumOfDrops:       drop,
+			}})
+	}
+
+	return inEvents, expectedEvents
+}

client/internal/netflow/store/memory.go

@@ -1,10 +1,15 @@
 package store
 
 import (
+	"maps"
+	"math/rand"
+	v2 "math/rand/v2"
+	"net/netip"
+	"slices"
 	"sync"
+	"time"
 
 	"github.com/google/uuid"
-
 	"github.com/netbirdio/netbird/client/internal/netflow/types"
 )
 
@@ -19,6 +24,14 @@ type Memory struct {
 	events map[uuid.UUID]*types.Event
 }
 
+type AggregatingMemory struct {
+	Memory
+	WindowStart time.Time
+	WindowEnd   time.Time
+	rnd         *v2.PCG
+	nowFunc     func() time.Time
+}
+
 func (m *Memory) StoreEvent(event *types.Event) {
 	m.mux.Lock()
 	defer m.mux.Unlock()
@@ -48,3 +61,104 @@ func (m *Memory) DeleteEvents(ids []uuid.UUID) {
 		delete(m.events, id)
 	}
 }
+
+func NewAggregatingMemoryStore() *AggregatingMemory {
+	return NewAggregatingMemoryStoreWithTimeFunc(defaultNowFunc)
+}
+
+// used in tests when deterministic (less random) time intervals are required
+func NewAggregatingMemoryStoreWithTimeFunc(nowFunc func() time.Time) *AggregatingMemory {
+	return &AggregatingMemory{WindowStart: nowFunc(), Memory: Memory{events: make(map[uuid.UUID]*types.Event)}, nowFunc: nowFunc, rnd: v2.NewPCG(rand.Uint64(), rand.Uint64())}
+}
+
+func (am *AggregatingMemory) ResetAggregationWindow() types.FlowEventAggregator {
+	am.mux.Lock()
+	defer am.mux.Unlock()
+
+	now := am.nowFunc()
+	toret := AggregatingMemory{WindowStart: am.WindowStart, WindowEnd: now, Memory: Memory{events: am.events}, rnd: v2.NewPCG(rand.Uint64(), rand.Uint64())}
+
+	am.events = make(map[uuid.UUID]*types.Event)
+	am.WindowStart = now
+
+	return &toret
+}
+
+type aggregationKey struct {
+	srcAddr   netip.Addr
+	destAddr  netip.Addr
+	destPort  uint16
+	direction int
+	protocol  uint8
+	icmpType  uint8
+	unique    uint64 // used to prevent aggregation on non icmp/udp/tcp events
+}
+
+func (am *AggregatingMemory) GetAggregatedEvents() []*types.Event {
+	am.mux.Lock()
+	defer am.mux.Unlock()
+
+	aggregated := make(map[aggregationKey]*types.Event)
+	for _, v := range am.events {
+		lookupKey := aggregationKey{srcAddr: v.SourceIP, destAddr: v.DestIP, destPort: v.DestPort, direction: int(v.Direction), protocol: uint8(v.Protocol), icmpType: v.ICMPType}
+		if _, ok := aggregated[lookupKey]; !ok {
+			event := v.Clone()
+
+			switch event.Type {
+			case types.TypeStart:
+				event.NumOfStarts += 1
+			case types.TypeDrop:
+				event.NumOfDrops += 1
+			case types.TypeEnd:
+				event.NumOfEnds += 1
+			}
+			event.Type = types.TypeUnknown
+
+			// Please note that ICMPCode field isn't propagated by the manager (see flow/proto/flow.pb.go, FlowFields struct)
+			// so the field value in an icmp event in the "aggregated" doesn't matter
+
+			event.WindowStart = am.WindowStart
+			event.WindowEnd = am.WindowEnd
+
+			if event.Protocol != types.ICMP && event.Protocol != types.ICMPv6 && event.Protocol != types.UDP && event.Protocol != types.TCP {
+				lookupKey.unique = am.rnd.Uint64() // to make the lookup key unique so we don't aggregate on it
+			}
+
+			aggregated[lookupKey] = event
+			continue
+		}
+
+		aggregatedEvent := aggregated[lookupKey]
+		if aggregatedEvent.Protocol != types.ICMP && aggregatedEvent.Protocol != types.ICMPv6 && aggregatedEvent.Protocol != types.UDP && aggregatedEvent.Protocol != types.TCP {
+			continue // we don't aggregate this type of events; shouldn't ever get here
+		}
+
+		// track the number of connections, duration?, open and close events?
+		aggregatedEvent.RxBytes += v.RxBytes
+		aggregatedEvent.RxPackets += v.RxPackets
+		aggregatedEvent.TxBytes += v.TxBytes
+		aggregatedEvent.TxPackets += v.TxPackets
+		switch v.Type {
+		case types.TypeStart:
+			aggregatedEvent.NumOfStarts += 1
+		case types.TypeDrop:
+			aggregatedEvent.NumOfDrops += 1
+		case types.TypeEnd:
+			aggregatedEvent.NumOfEnds += 1
+		}
+		if aggregatedEvent.Timestamp.Compare(v.Timestamp) > 0 {
+			aggregatedEvent.Timestamp = v.Timestamp
+			aggregatedEvent.ID = v.ID
+			aggregatedEvent.SourcePort = v.SourcePort
+		}
+		if len(aggregatedEvent.RuleID) == 0 && len(v.RuleID) != 0 {
+			aggregatedEvent.RuleID = slices.Clone(v.RuleID)
+		}
+	}
+
+	return slices.Collect(maps.Values(aggregated)) // could return an iterator instead here
+}
+
+func defaultNowFunc() time.Time {
+	return time.Now()
+}

client/internal/netflow/types/types.go

@@ -2,6 +2,7 @@ package types
 
 import (
 	"net/netip"
+	"slices"
 	"strconv"
 	"time"
 
@@ -69,8 +70,10 @@ const (
 )
 
 type Event struct {
-	ID        uuid.UUID
-	Timestamp time.Time
+	ID          uuid.UUID
+	Timestamp   time.Time
+	WindowStart time.Time
+	WindowEnd   time.Time
 	EventFields
 }
 
@@ -92,6 +95,17 @@ type EventFields struct {
 	TxPackets        uint64
 	RxBytes          uint64
 	TxBytes          uint64
+	NumOfStarts      uint64
+	NumOfEnds        uint64
+	NumOfDrops       uint64
+}
+
+func (e *Event) Clone() *Event {
+	toret := *e
+	toret.RuleID = slices.Clone(e.RuleID)
+	toret.SourceResourceID = slices.Clone(e.SourceResourceID)
+	toret.DestResourceID = slices.Clone(e.DestResourceID)
+	return &toret
 }
 
 type FlowConfig struct {
@@ -114,13 +128,15 @@ type FlowManager interface {
 	GetLogger() FlowLogger
 }
 
+type FlowEventAggregator interface {
+	ResetAggregationWindow() FlowEventAggregator
+	GetAggregatedEvents() []*Event
+}
+
 type FlowLogger interface {
+	ResetAggregationWindow() FlowEventAggregator
 	// StoreEvent stores a flow event
 	StoreEvent(flowEvent EventFields)
-	// GetEvents returns all stored events
-	GetEvents() []*Event
-	// DeleteEvents deletes events from the store
-	DeleteEvents([]uuid.UUID)
 	// Close closes the logger
 	Close()
 	// Enable enables the flow logger receiver
@@ -140,6 +156,11 @@ type Store interface {
 	Close()
 }
 
+type AggregatingStore interface {
+	FlowEventAggregator
+	Store
+}
+
 // ConnTracker defines the interface for connection tracking functionality
 type ConnTracker interface {
 	// Start begins tracking connections by listening for conntrack events.

client/internal/peer/conn.go

@@ -30,6 +30,11 @@ import (
 	relayClient "github.com/netbirdio/netbird/shared/relay/client"
 )
 
+// wgTimeoutEscalationThreshold is the number of consecutive WireGuard
+// handshake timeouts after which the rosenpass state for the peer is
+// considered desynced and gets reset.
+const wgTimeoutEscalationThreshold = 3
+
 // MetricsRecorder is an interface for recording peer connection metrics
 type MetricsRecorder interface {
 	RecordConnectionStages(
@@ -118,6 +123,9 @@ type Conn struct {
 	wgWatcher       *WGWatcher
 	wgWatcherWg     sync.WaitGroup
 	wgWatcherCancel context.CancelFunc
+	// wgTimeouts counts consecutive WireGuard handshake timeouts without a
+	// successful handshake in between. Guarded by mu.
+	wgTimeouts int
 
 	// used to store the remote Rosenpass key for Relayed connection in case of connection update from ice
 	rosenpassRemoteKey []byte
@@ -195,7 +203,6 @@ func NewConn(config ConnConfig, services ServiceDependencies) (*Conn, error) {
 		statusICE:          worker.NewAtomicStatus(),
 		dumpState:          dumpState,
 		endpointUpdater:    NewEndpointUpdater(connLog, config.WgConfig, isController(config)),
-		wgWatcher:          NewWGWatcher(connLog, config.WgConfig.WgInterface, config.Key, dumpState),
 		metricsRecorder:    services.MetricsRecorder,
 	}
 
@@ -663,11 +670,12 @@ func (conn *Conn) onGuardEvent() {
 	}
 }
 
-func (conn *Conn) onWGDisconnected() {
+func (conn *Conn) onWGDisconnected(watcherCtx context.Context) {
 	conn.mu.Lock()
 	defer conn.mu.Unlock()
 
-	if conn.ctx.Err() != nil {
+	// watcherCtx guards against a stale watcher tearing down a connection that already superseded it.
+	if conn.ctx.Err() != nil || watcherCtx.Err() != nil {
 		return
 	}
 
@@ -683,6 +691,29 @@ func (conn *Conn) onWGDisconnected() {
 	default:
 		conn.Log.Debugf("No active connection to close on WG timeout")
 	}
+
+	conn.escalateWGTimeoutLocked()
+}
+
+// escalateWGTimeoutLocked resets the peer's rosenpass state after repeated
+// handshake timeouts. With rosenpass enabled, persistent timeouts mean the
+// preshared keys have desynced; the renewal exchange runs over the dead
+// tunnel and cannot resync them. Reporting the peer disconnected drops its
+// rosenpass state, so the next connection configuration programs the
+// rendezvous key and the tunnel can bootstrap again. Callers must hold mu.
+func (conn *Conn) escalateWGTimeoutLocked() {
+	if conn.config.RosenpassConfig.PubKey == nil {
+		return
+	}
+
+	conn.wgTimeouts++
+	if conn.wgTimeouts < wgTimeoutEscalationThreshold || conn.onDisconnected == nil {
+		return
+	}
+	conn.wgTimeouts = 0
+
+	conn.Log.Warnf("%d consecutive WireGuard handshake timeouts, resetting rosenpass state for peer", wgTimeoutEscalationThreshold)
+	conn.onDisconnected(conn.config.WgConfig.RemoteKey)
 }
 
 func (conn *Conn) updateRelayStatus(relayServerAddr string, rosenpassPubKey []byte, updateTime time.Time) {
@@ -802,25 +833,39 @@ func (conn *Conn) isConnectedOnAllWay() (status guard.ConnStatus) {
 	})
 }
 
+// enableWgWatcherIfNeeded starts a fresh watcher instance per connection attempt, so its
+// lifecycle stays bound to conn.mu and enable/disable can't race an old goroutine's shutdown.
+// Caller must hold conn.mu.
 func (conn *Conn) enableWgWatcherIfNeeded(enabledTime time.Time) {
-	if !conn.wgWatcher.PrepareInitialHandshake() {
+	if conn.wgWatcher != nil {
 		return
 	}
 
+	watcher := NewWGWatcher(conn.Log, conn.config.WgConfig.WgInterface, conn.config.Key, conn.dumpState)
+	watcher.PrepareInitialHandshake()
+
 	wgWatcherCtx, wgWatcherCancel := context.WithCancel(conn.ctx)
+	conn.wgWatcher = watcher
 	conn.wgWatcherCancel = wgWatcherCancel
+
 	conn.wgWatcherWg.Add(1)
 	go func() {
 		defer conn.wgWatcherWg.Done()
-		conn.wgWatcher.EnableWgWatcher(wgWatcherCtx, enabledTime, conn.onWGDisconnected, conn.onWGHandshakeSuccess)
+		onDisconnected := func() { conn.onWGDisconnected(wgWatcherCtx) }
+		watcher.EnableWgWatcher(wgWatcherCtx, enabledTime, onDisconnected, conn.onWGHandshakeSuccess, conn.onWGCheckSuccess)
 	}()
 }
 
+// disableWgWatcherIfNeeded cancels and drops the watcher once no transport is active. It never
+// waits for the goroutine: the timeout path reentrantly calls back here under conn.mu, so
+// blocking would deadlock. Caller must hold conn.mu.
 func (conn *Conn) disableWgWatcherIfNeeded() {
-	if conn.currentConnPriority == conntype.None && conn.wgWatcherCancel != nil {
-		conn.wgWatcherCancel()
-		conn.wgWatcherCancel = nil
+	if conn.currentConnPriority != conntype.None || conn.wgWatcher == nil {
+		return
 	}
+	conn.wgWatcherCancel()
+	conn.wgWatcher = nil
+	conn.wgWatcherCancel = nil
 }
 
 func (conn *Conn) newProxy(remoteConn net.Conn) (wgproxy.Proxy, error) {
@@ -843,7 +888,9 @@ func (conn *Conn) resetEndpoint() {
 		return
 	}
 	conn.Log.Infof("reset wg endpoint")
-	conn.wgWatcher.Reset()
+	if conn.wgWatcher != nil {
+		conn.wgWatcher.Reset()
+	}
 	if err := conn.endpointUpdater.RemoveEndpointAddress(); err != nil {
 		conn.Log.Warnf("failed to remove endpoint address before update: %v", err)
 	}
@@ -892,6 +939,15 @@ func (conn *Conn) onWGHandshakeSuccess(when time.Time) {
 	conn.recordConnectionMetrics()
 }
 
+// onWGCheckSuccess is called for every watcher check that observed a fresh
+// handshake, including handshakes of connections that were already up when
+// the watcher started.
+func (conn *Conn) onWGCheckSuccess() {
+	conn.mu.Lock()
+	conn.wgTimeouts = 0
+	conn.mu.Unlock()
+}
+
 // recordConnectionMetrics records connection stage timestamps as metrics
 func (conn *Conn) recordConnectionMetrics() {
 	if conn.metricsRecorder == nil {

client/internal/peer/conn_test.go

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

client/internal/peer/wg_watcher.go

@@ -3,7 +3,6 @@ package peer
 import (
 	"context"
 	"fmt"
-	"sync"
 	"time"
 
 	log "github.com/sirupsen/logrus"
@@ -24,14 +23,14 @@ type WGInterfaceStater interface {
 	GetStats() (map[string]configurer.WGStats, error)
 }
 
+// WGWatcher is single-shot: one instance per connection attempt, run once, then discarded.
+// Lifecycle is owned by Conn under conn.mu, so it keeps no "enabled" state to go stale.
 type WGWatcher struct {
 	log           *log.Entry
 	wgIfaceStater WGInterfaceStater
 	peerKey       string
 	stateDump     *stateDump
 
-	enabled   bool
-	muEnabled sync.Mutex
 	// initialHandshake is not thread-safe; never call PrepareInitialHandshake and EnableWgWatcher concurrently.
 	initialHandshake time.Time
 
@@ -48,36 +47,23 @@ func NewWGWatcher(log *log.Entry, wgIfaceStater WGInterfaceStater, peerKey strin
 	}
 }
 
-// PrepareInitialHandshake reserves the watcher and reads the peer's current WireGuard
-// handshake time. It must be called before the peer is (re)configured on the WireGuard
-// interface, so the captured baseline reflects the state prior to this connection attempt
-// instead of racing with that configuration. Returns ok=false if the watcher is already
-// running, in which case EnableWgWatcher must not be called.
-func (w *WGWatcher) PrepareInitialHandshake() (ok bool) {
-	w.muEnabled.Lock()
-	if w.enabled {
-		w.muEnabled.Unlock()
-		return false
-	}
-
+// PrepareInitialHandshake reads the peer's current WireGuard handshake time. It must be
+// called before the peer is (re)configured on the WireGuard interface, so the captured
+// baseline reflects the state prior to this connection attempt instead of racing with
+// that configuration.
+func (w *WGWatcher) PrepareInitialHandshake() {
 	w.log.Debugf("enable WireGuard watcher")
-	w.enabled = true
-	w.muEnabled.Unlock()
-
 	handshake, _ := w.wgState()
 	w.initialHandshake = handshake
-	return true
 }
 
 // EnableWgWatcher runs the WireGuard watcher loop using the handshake baseline captured by
 // PrepareInitialHandshake. The watcher runs until ctx is cancelled. Caller is responsible
-// for context lifecycle management.
-func (w *WGWatcher) EnableWgWatcher(ctx context.Context, enabledTime time.Time, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time)) {
-	w.periodicHandshakeCheck(ctx, onDisconnectedFn, onHandshakeSuccessFn, enabledTime, w.initialHandshake)
-
-	w.muEnabled.Lock()
-	w.enabled = false
-	w.muEnabled.Unlock()
+// for context lifecycle management. onHandshakeSuccessFn is called only for the first
+// handshake observed by this run, onCheckSuccessFn for every check that observed a fresh
+// handshake, including the first.
+func (w *WGWatcher) EnableWgWatcher(ctx context.Context, enabledTime time.Time, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time), onCheckSuccessFn func()) {
+	w.periodicHandshakeCheck(ctx, onDisconnectedFn, onHandshakeSuccessFn, onCheckSuccessFn, enabledTime, w.initialHandshake)
 }
 
 // Reset signals the watcher that the WireGuard peer has been reset and a new
@@ -90,7 +76,7 @@ func (w *WGWatcher) Reset() {
 }
 
 // wgStateCheck help to check the state of the WireGuard handshake and relay connection
-func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time), enabledTime time.Time, initialHandshake time.Time) {
+func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn func(), onHandshakeSuccessFn func(when time.Time), onCheckSuccessFn func(), enabledTime time.Time, initialHandshake time.Time) {
 	w.log.Infof("WireGuard watcher started")
 
 	timer := time.NewTimer(wgHandshakeOvertime)
@@ -103,6 +89,7 @@ func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn
 		case <-timer.C:
 			handshake, ok := w.handshakeCheck(lastHandshake)
 			if !ok {
+				// early ctx cancel check return
 				if ctx.Err() != nil {
 					return
 				}
@@ -117,6 +104,10 @@ func (w *WGWatcher) periodicHandshakeCheck(ctx context.Context, onDisconnectedFn
 				}
 			}
 
+			if onCheckSuccessFn != nil && ctx.Err() == nil {
+				onCheckSuccessFn()
+			}
+
 			lastHandshake = *handshake
 
 			resetTime := time.Until(handshake.Add(checkPeriod))
@@ -147,9 +138,9 @@ func (w *WGWatcher) handshakeCheck(lastHandshake time.Time) (*time.Time, bool) {
 
 	w.log.Tracef("previous handshake, handshake: %v, %v", lastHandshake, handshake)
 
-	// the current know handshake did not change
+	// the current known handshake did not change
 	if handshake.Equal(lastHandshake) {
-		w.log.Warnf("WireGuard handshake timed out: %v", handshake)
+		w.log.Warnf("WireGuard handshake not updated: %v", handshake)
 		return nil, false
 	}
 

client/internal/peer/wg_watcher_test.go

@@ -7,7 +7,6 @@ import (
 	"time"
 
 	log "github.com/sirupsen/logrus"
-	"github.com/stretchr/testify/require"
 
 	"github.com/netbirdio/netbird/client/iface/configurer"
 )
@@ -24,6 +23,72 @@ func (m *MocWgIface) disconnect() {
 	m.stop = true
 }
 
+type mockHandshakeStats struct {
+	mu        sync.Mutex
+	handshake time.Time
+}
+
+func (m *mockHandshakeStats) GetStats() (map[string]configurer.WGStats, error) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return map[string]configurer.WGStats{"": {LastHandshake: m.handshake}}, nil
+}
+
+func (m *mockHandshakeStats) advance() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.handshake = time.Now()
+}
+
+// TestWGWatcher_CheckSuccessCallback: onCheckSuccessFn must fire for a fresh
+// handshake even when the watcher started with an existing handshake baseline,
+// the case where onHandshakeSuccessFn stays silent.
+func TestWGWatcher_CheckSuccessCallback(t *testing.T) {
+	// checkPeriod bounds how stale a handshake may be before the watcher treats it
+	// as a suspended-machine timeout. The first check fires after wgHandshakeOvertime,
+	// so keep checkPeriod well above any scheduling jitter to avoid a false timeout
+	// converting the expected success into a disconnect on a loaded runner.
+	checkPeriod = 1 * time.Minute
+	wgHandshakeOvertime = 1 * time.Second
+
+	mlog := log.WithField("peer", "tet")
+	// Use an old baseline so advance() yields a strictly newer handshake even on
+	// platforms with coarse clock resolution (Windows), where two time.Now() calls
+	// microseconds apart can return the same instant and read as a timed-out handshake.
+	stats := &mockHandshakeStats{handshake: time.Now().Add(-time.Hour)}
+	watcher := NewWGWatcher(mlog, stats, "", newStateDump("peer", mlog, &Status{}))
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	watcher.PrepareInitialHandshake()
+
+	firstHandshake := make(chan struct{}, 1)
+	checkSuccess := make(chan struct{}, 1)
+	go watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {
+		firstHandshake <- struct{}{}
+	}, func() {
+		select {
+		case checkSuccess <- struct{}{}:
+		default:
+		}
+	})
+
+	stats.advance()
+
+	select {
+	case <-checkSuccess:
+	case <-time.After(10 * time.Second):
+		t.Errorf("timeout waiting for check success callback")
+	}
+
+	select {
+	case <-firstHandshake:
+		t.Errorf("first-handshake callback must not fire for a non-zero baseline")
+	default:
+	}
+}
+
 func TestWGWatcher_EnableWgWatcher(t *testing.T) {
 	checkPeriod = 5 * time.Second
 	wgHandshakeOvertime = 1 * time.Second
@@ -35,8 +100,7 @@ func TestWGWatcher_EnableWgWatcher(t *testing.T) {
 	ctx, cancel := context.WithCancel(context.Background())
 	defer cancel()
 
-	ok := watcher.PrepareInitialHandshake()
-	require.True(t, ok, "watcher should not be enabled yet")
+	watcher.PrepareInitialHandshake()
 
 	onDisconnected := make(chan struct{}, 1)
 	go watcher.EnableWgWatcher(ctx, time.Now(), func() {
@@ -44,7 +108,7 @@ func TestWGWatcher_EnableWgWatcher(t *testing.T) {
 		onDisconnected <- struct{}{}
 	}, func(when time.Time) {
 		mlog.Infof("onHandshakeSuccess: %v", when)
-	})
+	}, nil)
 
 	// wait for initial reading
 	time.Sleep(2 * time.Second)
@@ -66,14 +130,13 @@ func TestWGWatcher_ReEnable(t *testing.T) {
 	watcher := NewWGWatcher(mlog, mocWgIface, "", newStateDump("peer", mlog, &Status{}))
 
 	ctx, cancel := context.WithCancel(context.Background())
-	ok := watcher.PrepareInitialHandshake()
-	require.True(t, ok, "watcher should not be enabled yet")
+	watcher.PrepareInitialHandshake()
 
 	wg := &sync.WaitGroup{}
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
-		watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {})
+		watcher.EnableWgWatcher(ctx, time.Now(), func() {}, func(when time.Time) {}, nil)
 	}()
 	cancel()
 
@@ -83,13 +146,12 @@ func TestWGWatcher_ReEnable(t *testing.T) {
 	ctx, cancel = context.WithCancel(context.Background())
 	defer cancel()
 
-	ok = watcher.PrepareInitialHandshake()
-	require.True(t, ok, "watcher should be re-enabled after the previous run stopped")
+	watcher.PrepareInitialHandshake()
 
 	onDisconnected := make(chan struct{}, 1)
 	go watcher.EnableWgWatcher(ctx, time.Now(), func() {
 		onDisconnected <- struct{}{}
-	}, func(when time.Time) {})
+	}, func(when time.Time) {}, nil)
 
 	time.Sleep(2 * time.Second)
 	mocWgIface.disconnect()

client/internal/profilemanager/service.go

@@ -11,6 +11,7 @@ import (
 	"runtime"
 	"sort"
 	"strings"
+	"syscall"
 
 	log "github.com/sirupsen/logrus"
 
@@ -439,7 +440,11 @@ func (s *ServiceManager) GetStatePath() string {
 
 	activeProf, err := s.GetActiveProfileState()
 	if err != nil {
-		log.Warnf("failed to get active profile state: %v", err)
+		if errors.Is(err, syscall.ENOSYS) {
+			log.Debugf("active profile state unavailable on this platform: %v", err)
+		} else {
+			log.Warnf("failed to get active profile state: %v", err)
+		}
 		return defaultStatePath
 	}
 

client/internal/rosenpass/manager.go

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

client/internal/rosenpass/manager_test.go

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

client/internal/rosenpass/netbird_handler.go

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

client/internal/rosenpass/netbird_handler_test.go

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

client/internal/rosenpass/seed.go

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

client/internal/routemanager/dnsinterceptor/handler.go

@@ -95,7 +95,7 @@ func (d *DnsInterceptor) RemoveRoute() error {
 
 			// AllowedIPs should use real IPs
 			if d.currentPeerKey != "" {
-				if _, err := d.allowedIPsRefcounter.Decrement(prefix); err != nil {
+				if _, err := d.allowedIPsRefcounter.Decrement(prefix, d.currentPeerKey); err != nil {
 					merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %s: %v", prefix, err))
 				}
 			}
@@ -172,7 +172,7 @@ func (d *DnsInterceptor) removeAllowedIP(realPrefix netip.Prefix) error {
 	}
 
 	// AllowedIPs use real IPs
-	if _, err := d.allowedIPsRefcounter.Decrement(realPrefix); err != nil {
+	if _, err := d.allowedIPsRefcounter.Decrement(realPrefix, d.currentPeerKey); err != nil {
 		return fmt.Errorf("remove allowed IP %s: %v", realPrefix, err)
 	}
 
@@ -205,7 +205,7 @@ func (d *DnsInterceptor) RemoveAllowedIPs() error {
 	for _, prefixes := range d.interceptedDomains {
 		for _, prefix := range prefixes {
 			// AllowedIPs use real IPs
-			if _, err := d.allowedIPsRefcounter.Decrement(prefix); err != nil {
+			if _, err := d.allowedIPsRefcounter.Decrement(prefix, d.currentPeerKey); err != nil {
 				merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %s: %v", prefix, err))
 			}
 		}

client/internal/routemanager/dynamic/route.go

@@ -135,7 +135,7 @@ func (r *Route) RemoveAllowedIPs() error {
 	var merr *multierror.Error
 	for _, domainPrefixes := range r.dynamicDomains {
 		for _, prefix := range domainPrefixes {
-			if _, err := r.allowedIPsRefcounter.Decrement(prefix); err != nil {
+			if _, err := r.allowedIPsRefcounter.Decrement(prefix, r.currentPeerKey); err != nil {
 				merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %s: %w", prefix, err))
 			}
 		}
@@ -320,7 +320,7 @@ func (r *Route) removeRoutes(prefixes []netip.Prefix) ([]netip.Prefix, error) {
 			merr = multierror.Append(merr, fmt.Errorf("remove dynamic route for IP %s: %w", prefix, err))
 		}
 		if r.currentPeerKey != "" {
-			if _, err := r.allowedIPsRefcounter.Decrement(prefix); err != nil {
+			if _, err := r.allowedIPsRefcounter.Decrement(prefix, r.currentPeerKey); err != nil {
 				merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %s: %w", prefix, err))
 			}
 		}

client/internal/routemanager/manager.go

@@ -12,6 +12,7 @@ import (
 	"strings"
 	"sync"
 	"sync/atomic"
+	"syscall"
 	"time"
 
 	"github.com/google/uuid"
@@ -214,7 +215,7 @@ func (m *DefaultManager) setupRefCounters(useNoop bool) {
 		)
 	}
 
-	m.allowedIPsRefCounter = refcounter.New(
+	m.allowedIPsRefCounter = refcounter.NewAllowedIPs(
 		func(prefix netip.Prefix, peerKey string) (string, error) {
 			// save peerKey to use it in the remove function
 			return peerKey, m.wgInterface.AddAllowedIP(peerKey, prefix)
@@ -264,7 +265,11 @@ func (m *DefaultManager) initSelector() *routeselector.RouteSelector {
 
 	// restore selector state if it exists
 	if err := m.stateManager.LoadState(state); err != nil {
-		log.Warnf("failed to load state: %v", err)
+		if errors.Is(err, syscall.ENOSYS) {
+			log.Debugf("route selector state unavailable on this platform: %v", err)
+		} else {
+			log.Warnf("failed to load state: %v", err)
+		}
 		return routeselector.NewRouteSelector()
 	}
 

client/internal/routemanager/refcounter/allowedips.go

@@ -0,0 +1,185 @@
+package refcounter
+
+import (
+	"errors"
+	"fmt"
+	"net/netip"
+	"sort"
+	"sync"
+
+	"github.com/hashicorp/go-multierror"
+
+	nberrors "github.com/netbirdio/netbird/client/errors"
+)
+
+// allowedIPsEntry holds the per-peer reference counts for a single prefix and which peer is
+// currently installed in WireGuard. WireGuard allows a prefix on exactly one peer, so at most
+// one peer is active at a time even when several peers reference the prefix.
+type allowedIPsEntry struct {
+	// peers maps a peerKey to the number of references holding the prefix for that peer.
+	peers map[string]int
+	// active is the peerKey currently installed in WireGuard for this prefix ("" if none).
+	active string
+	// total is the sum of all per-peer reference counts (kept in sync with peers).
+	total int
+}
+
+// AllowedIPsRefCounter is a peer-aware reference counter for WireGuard AllowedIPs.
+//
+// The generic Counter keys only by prefix and remembers a single Out value set by the first
+// caller, which it never changes. That is wrong for AllowedIPs: two independent watchers (or
+// multiple resolved domains) can reference the same prefix through different peers, and when the
+// peer currently installed in WireGuard releases its last reference the prefix must be handed over
+// to a surviving peer instead of being left pointing at the released one.
+//
+// It calls add/remove (which program WireGuard) only on the transitions that matter:
+//   - add on the first reference for a prefix, or when swapping the active peer;
+//   - remove on the last reference for a prefix, or on the old peer during a swap.
+type AllowedIPsRefCounter struct {
+	mu      sync.Mutex
+	entries map[netip.Prefix]*allowedIPsEntry
+	add     AddFunc[netip.Prefix, string, string]
+	remove  RemoveFunc[netip.Prefix, string]
+}
+
+// NewAllowedIPs creates a new peer-aware AllowedIPs reference counter.
+// add programs a prefix on a peer in WireGuard and returns the peerKey to store as the active peer.
+// remove unprograms the prefix from the given peer.
+func NewAllowedIPs(add AddFunc[netip.Prefix, string, string], remove RemoveFunc[netip.Prefix, string]) *AllowedIPsRefCounter {
+	return &AllowedIPsRefCounter{
+		entries: map[netip.Prefix]*allowedIPsEntry{},
+		add:     add,
+		remove:  remove,
+	}
+}
+
+// Increment adds a reference to prefix for peerKey. WireGuard is programmed only for the first
+// reference to a prefix; while a different peer is already installed the prefix is left with it
+// (first peer wins, HA at the WireGuard layer is not possible) and only the reference count is kept.
+func (rm *AllowedIPsRefCounter) Increment(prefix netip.Prefix, peerKey string) (Ref[string], error) {
+	rm.mu.Lock()
+	defer rm.mu.Unlock()
+
+	e, ok := rm.entries[prefix]
+	if !ok {
+		e = &allowedIPsEntry{peers: map[string]int{}}
+		rm.entries[prefix] = e
+	}
+
+	logCallerF("Increasing allowed IP ref count for prefix %v peer %s [peer %d -> %d, total %d -> %d, active %q]",
+		prefix, peerKey, e.peers[peerKey], e.peers[peerKey]+1, e.total, e.total+1, e.active)
+
+	// Program WireGuard only when nothing is installed yet for this prefix.
+	if e.active == "" {
+		out, err := rm.add(prefix, peerKey)
+		if errors.Is(err, ErrIgnore) {
+			if e.total == 0 {
+				delete(rm.entries, prefix)
+			}
+			return Ref[string]{Count: e.total, Out: e.active}, nil
+		}
+		if err != nil {
+			if e.total == 0 {
+				delete(rm.entries, prefix)
+			}
+			return Ref[string]{}, fmt.Errorf("failed to add allowed IP %v for peer %s: %w", prefix, peerKey, err)
+		}
+		e.active = out
+	}
+
+	e.peers[peerKey]++
+	e.total++
+
+	return Ref[string]{Count: e.total, Out: e.active}, nil
+}
+
+// Decrement removes a reference to prefix for peerKey. When the peer currently installed in
+// WireGuard releases its last reference, the prefix is swapped to a surviving peer if one exists,
+// otherwise it is removed from WireGuard.
+func (rm *AllowedIPsRefCounter) Decrement(prefix netip.Prefix, peerKey string) (Ref[string], error) {
+	rm.mu.Lock()
+	defer rm.mu.Unlock()
+
+	e, ok := rm.entries[prefix]
+	if !ok {
+		logCallerF("No allowed IP reference found for prefix %v", prefix)
+		return Ref[string]{}, nil
+	}
+
+	if e.peers[peerKey] > 0 {
+		logCallerF("Decreasing allowed IP ref count for prefix %v peer %s [peer %d -> %d, total %d -> %d, active %q]",
+			prefix, peerKey, e.peers[peerKey], e.peers[peerKey]-1, e.total, e.total-1, e.active)
+		e.peers[peerKey]--
+		e.total--
+		if e.peers[peerKey] == 0 {
+			delete(e.peers, peerKey)
+		}
+	} else {
+		logCallerF("No allowed IP reference found for prefix %v peer %s", prefix, peerKey)
+	}
+
+	// If the peer currently installed in WireGuard still holds references, nothing to reprogram.
+	// Keying the check on the active peer (not the one just released) makes this self-healing:
+	// a prior swap whose remove/add failed leaves e.active pointing at a peer with no references,
+	// and this retries the hand-off on the next Decrement instead of getting stuck.
+	if e.active != "" && e.peers[e.active] > 0 {
+		return Ref[string]{Count: e.total, Out: e.active}, nil
+	}
+
+	// Detach the stale/gone active peer from WireGuard before reprogramming.
+	if e.active != "" {
+		if err := rm.remove(prefix, e.active); err != nil {
+			return Ref[string]{Count: e.total, Out: e.active}, fmt.Errorf("remove allowed IP %v for peer %s: %w", prefix, e.active, err)
+		}
+		e.active = ""
+	}
+
+	// Hand the prefix over to a surviving peer, or drop the entry when none remain.
+	if survivor, ok := pickSurvivor(e.peers); ok {
+		out, err := rm.add(prefix, survivor)
+		if err != nil {
+			return Ref[string]{Count: e.total, Out: ""}, fmt.Errorf("swap allowed IP %v to peer %s: %w", prefix, survivor, err)
+		}
+		e.active = out
+		return Ref[string]{Count: e.total, Out: e.active}, nil
+	}
+
+	delete(rm.entries, prefix)
+	return Ref[string]{Count: 0, Out: ""}, nil
+}
+
+// Flush removes all prefixes from WireGuard and clears the counter.
+func (rm *AllowedIPsRefCounter) Flush() error {
+	rm.mu.Lock()
+	defer rm.mu.Unlock()
+
+	var merr *multierror.Error
+	for prefix, e := range rm.entries {
+		if e.active == "" {
+			continue
+		}
+		logCallerF("Flushing allowed IP for prefix %v peer %s", prefix, e.active)
+		if err := rm.remove(prefix, e.active); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove allowed IP %v for peer %s: %w", prefix, e.active, err))
+		}
+	}
+
+	clear(rm.entries)
+
+	return nberrors.FormatErrorOrNil(merr)
+}
+
+// pickSurvivor deterministically selects a peer still referencing the prefix. WireGuard cannot do
+// multipath for a single prefix, so any surviving peer is a valid winner; the choice is made stable
+// (lowest peerKey) for predictable behavior and testability.
+func pickSurvivor(peers map[string]int) (string, bool) {
+	if len(peers) == 0 {
+		return "", false
+	}
+	keys := make([]string, 0, len(peers))
+	for k := range peers {
+		keys = append(keys, k)
+	}
+	sort.Strings(keys)
+	return keys[0], true
+}

client/internal/routemanager/refcounter/allowedips_test.go

@@ -0,0 +1,241 @@
+package refcounter
+
+import (
+	"errors"
+	"net/netip"
+	"testing"
+)
+
+// fakeWG models WireGuard's cryptokey routing: a prefix can be installed on exactly one peer.
+// failAdd/failRemove make the next add/remove fail once, to exercise the self-healing error paths.
+type fakeWG struct {
+	installed  map[netip.Prefix]string
+	adds       int
+	removes    int
+	failAdd    bool
+	failRemove bool
+}
+
+func newFakeWG() *fakeWG {
+	return &fakeWG{installed: map[netip.Prefix]string{}}
+}
+
+func (f *fakeWG) counter() *AllowedIPsRefCounter {
+	return NewAllowedIPs(
+		func(prefix netip.Prefix, peerKey string) (string, error) {
+			if f.failAdd {
+				f.failAdd = false
+				return "", errors.New("add failed")
+			}
+			f.adds++
+			f.installed[prefix] = peerKey
+			return peerKey, nil
+		},
+		func(prefix netip.Prefix, peerKey string) error {
+			if f.failRemove {
+				f.failRemove = false
+				return errors.New("remove failed")
+			}
+			f.removes++
+			// only clear if this peer is the one installed, mirroring wg semantics
+			if f.installed[prefix] == peerKey {
+				delete(f.installed, prefix)
+			}
+			return nil
+		},
+	)
+}
+
+func mustPrefix(t *testing.T, s string) netip.Prefix {
+	t.Helper()
+	p, err := netip.ParsePrefix(s)
+	if err != nil {
+		t.Fatalf("parse prefix %q: %v", s, err)
+	}
+	return p
+}
+
+func mustIncrement(t *testing.T, c *AllowedIPsRefCounter, p netip.Prefix, peer string) Ref[string] {
+	t.Helper()
+	ref, err := c.Increment(p, peer)
+	if err != nil {
+		t.Fatalf("Increment(%v, %s): %v", p, peer, err)
+	}
+	return ref
+}
+
+func mustDecrement(t *testing.T, c *AllowedIPsRefCounter, p netip.Prefix, peer string) Ref[string] {
+	t.Helper()
+	ref, err := c.Decrement(p, peer)
+	if err != nil {
+		t.Fatalf("Decrement(%v, %s): %v", p, peer, err)
+	}
+	return ref
+}
+
+// TestAllowedIPs_SwapOnActivePeerRemoval reproduces the reported bug: two networks with the same
+// prefix routed by different peers. Removing the network whose peer is installed must hand the
+// prefix over to the surviving peer instead of leaving it on the removed one.
+func TestAllowedIPs_SwapOnActivePeerRemoval(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p := mustPrefix(t, "10.44.8.0/24")
+
+	mustIncrement(t, c, p, "peerA")
+	mustIncrement(t, c, p, "peerB")
+	// First peer wins while both are present.
+	if got := f.installed[p]; got != "peerA" {
+		t.Fatalf("expected peerA installed, got %q", got)
+	}
+
+	// Remove the active peer's network -> must swap to peerB.
+	mustDecrement(t, c, p, "peerA")
+	if got := f.installed[p]; got != "peerB" {
+		t.Fatalf("BUG: prefix stuck on removed peer, want peerB got %q", got)
+	}
+
+	// Remove the last one -> prefix gone.
+	mustDecrement(t, c, p, "peerB")
+	if _, ok := f.installed[p]; ok {
+		t.Fatalf("expected prefix removed, still installed on %q", f.installed[p])
+	}
+}
+
+// TestAllowedIPs_RemoveNonActivePeer removing a non-installed peer must not touch WireGuard.
+func TestAllowedIPs_RemoveNonActivePeer(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p := mustPrefix(t, "10.44.8.0/24")
+
+	mustIncrement(t, c, p, "peerA")
+	mustIncrement(t, c, p, "peerB")
+	removesBefore := f.removes
+
+	mustDecrement(t, c, p, "peerB")
+	if f.installed[p] != "peerA" {
+		t.Fatalf("active peer must stay peerA, got %q", f.installed[p])
+	}
+	if f.removes != removesBefore {
+		t.Fatalf("removing a non-active peer must not call wg remove")
+	}
+}
+
+// TestAllowedIPs_SamePeerMultipleRefs two references via the same peer must keep the prefix until
+// the last reference is released (the reason the per-peer count must be an int, not a set).
+func TestAllowedIPs_SamePeerMultipleRefs(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p := mustPrefix(t, "10.44.8.0/24")
+
+	mustIncrement(t, c, p, "peerA")
+	mustIncrement(t, c, p, "peerA")
+	if f.adds != 1 {
+		t.Fatalf("expected a single wg add for the same peer, got %d", f.adds)
+	}
+
+	mustDecrement(t, c, p, "peerA")
+	if f.installed[p] != "peerA" {
+		t.Fatalf("prefix must stay while a reference remains, got %q", f.installed[p])
+	}
+	if f.removes != 0 {
+		t.Fatalf("no wg remove expected while a reference remains, got %d", f.removes)
+	}
+
+	mustDecrement(t, c, p, "peerA")
+	if _, ok := f.installed[p]; ok {
+		t.Fatalf("prefix must be removed after last reference")
+	}
+}
+
+// TestAllowedIPs_RefCountAndActive checks the Ref returned to callers (used for the HA-disabled log).
+func TestAllowedIPs_RefCountAndActive(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p := mustPrefix(t, "10.44.8.0/24")
+
+	ref := mustIncrement(t, c, p, "peerA")
+	if ref.Count != 1 || ref.Out != "peerA" {
+		t.Fatalf("want {1, peerA}, got {%d, %q}", ref.Count, ref.Out)
+	}
+	ref = mustIncrement(t, c, p, "peerB")
+	if ref.Count != 2 || ref.Out != "peerA" {
+		t.Fatalf("want {2, peerA}, got {%d, %q}", ref.Count, ref.Out)
+	}
+}
+
+// TestAllowedIPs_Flush removes everything installed and clears the counter.
+func TestAllowedIPs_Flush(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p1 := mustPrefix(t, "10.44.8.0/24")
+	p2 := mustPrefix(t, "10.44.9.0/24")
+
+	mustIncrement(t, c, p1, "peerA")
+	mustIncrement(t, c, p2, "peerB")
+
+	if err := c.Flush(); err != nil {
+		t.Fatal(err)
+	}
+	if len(f.installed) != 0 {
+		t.Fatalf("expected all prefixes removed, got %v", f.installed)
+	}
+	// After flush, a fresh increment must add again.
+	mustIncrement(t, c, p1, "peerC")
+	if f.installed[p1] != "peerC" {
+		t.Fatalf("counter not reset after flush")
+	}
+}
+
+// TestAllowedIPs_SelfHealAfterSwapAddError ensures a failed add during a swap does not permanently
+// strand the prefix: the next Decrement (or Increment) must retry and install a surviving peer.
+func TestAllowedIPs_SelfHealAfterSwapAddError(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p := mustPrefix(t, "10.44.8.0/24")
+
+	mustIncrement(t, c, p, "peerA")
+	mustIncrement(t, c, p, "peerB")
+	mustIncrement(t, c, p, "peerC")
+
+	// Removing the active peerA triggers a swap to a survivor; make the add fail once.
+	f.failAdd = true
+	if _, err := c.Decrement(p, "peerA"); err == nil {
+		t.Fatalf("expected error from failed swap add")
+	}
+	if _, ok := f.installed[p]; ok {
+		t.Fatalf("nothing should be installed after a failed swap add, got %q", f.installed[p])
+	}
+
+	// A later Decrement of a non-active survivor must retry the hand-off (self-heal), not stay stuck.
+	ref := mustDecrement(t, c, p, "peerC")
+	if got := f.installed[p]; got == "" {
+		t.Fatalf("self-heal failed: prefix left unrouted after add recovered")
+	}
+	if ref.Out == "" {
+		t.Fatalf("expected an active peer after self-heal, got empty")
+	}
+}
+
+// TestAllowedIPs_SelfHealAfterRemoveError ensures a failed remove during a swap is retried instead
+// of leaving e.active stuck on a peer that no longer holds references.
+func TestAllowedIPs_SelfHealAfterRemoveError(t *testing.T) {
+	f := newFakeWG()
+	c := f.counter()
+	p := mustPrefix(t, "10.44.8.0/24")
+
+	mustIncrement(t, c, p, "peerA")
+	mustIncrement(t, c, p, "peerB")
+
+	// Releasing active peerA must detach it (remove) then add peerB; fail the remove once.
+	f.failRemove = true
+	if _, err := c.Decrement(p, "peerA"); err == nil {
+		t.Fatalf("expected error from failed remove")
+	}
+
+	// Next Decrement of the non-active survivor retries: removes stale peerA, installs peerB.
+	mustDecrement(t, c, p, "peerB")
+	// peerB had only one ref, so after retry the prefix is fully released.
+	if _, ok := f.installed[p]; ok {
+		t.Fatalf("expected prefix released after self-heal, still on %q", f.installed[p])
+	}
+}

client/internal/routemanager/refcounter/types.go

@@ -5,5 +5,7 @@ import "net/netip"
 // RouteRefCounter is a Counter for Route, it doesn't take any input on Increment and doesn't use any output on Decrement
 type RouteRefCounter = Counter[netip.Prefix, struct{}, struct{}]
 
-// AllowedIPsRefCounter is a Counter for AllowedIPs, it takes a peer key on Increment and passes it back to Decrement
-type AllowedIPsRefCounter = Counter[netip.Prefix, string, string]
+// AllowedIPsRefCounter tracks WireGuard AllowedIPs per prefix. Unlike the generic Counter it is peer-aware:
+// a prefix can be claimed by several peers at once and WireGuard allows a given prefix on exactly one peer,
+// so the counter records the per-peer reference count and swaps the installed peer when the active one is released.
+// See allowedips.go.

client/internal/routemanager/static/route.go

@@ -15,6 +15,11 @@ type Route struct {
 	route                *route.Route
 	routeRefCounter      *refcounter.RouteRefCounter
 	allowedIPsRefcounter *refcounter.AllowedIPsRefCounter
+	// currentPeerKey is the routing peer this watcher currently has the prefix installed on
+	// (the HA winner elected by the watcher). It can differ from route.Peer and change on
+	// failover, so it is recorded on AddAllowedIPs and used on RemoveAllowedIPs to decrement
+	// the exact peer that was incremented.
+	currentPeerKey string
 }
 
 func NewRoute(params common.HandlerParams) *Route {
@@ -52,12 +57,15 @@ func (r *Route) AddAllowedIPs(peerKey string) error {
 			ref.Out,
 		)
 	}
+	r.currentPeerKey = peerKey
 	return nil
 }
 
 func (r *Route) RemoveAllowedIPs() error {
-	if _, err := r.allowedIPsRefcounter.Decrement(r.route.Network); err != nil {
-		return err
+	var err error
+	if _, decErr := r.allowedIPsRefcounter.Decrement(r.route.Network, r.currentPeerKey); decErr != nil {
+		err = fmt.Errorf("remove allowed IP %s: %w", r.route.Network, decErr)
 	}
-	return nil
+	r.currentPeerKey = ""
+	return err
 }

client/mdm/canonical_loaders.go

@@ -22,6 +22,7 @@ var allKeys = []string{
 	KeyDisableMetricsCollection,
 	KeyAllowServerSSH,
 	KeyDisableAutoConnect,
+	KeyDisableAutostart,
 	KeyPreSharedKey,
 	KeyRosenpassEnabled,
 	KeyRosenpassPermissive,

client/mdm/policy.go

@@ -20,10 +20,10 @@ import (
 // names (lowerCamelCase) so the daemon can map a Policy key directly to a
 // configuration field.
 const (
-	KeyManagementURL            = "managementURL"
-	KeyDisableUpdateSettings    = "disableUpdateSettings"
-	KeyDisableProfiles          = "disableProfiles"
-	KeyDisableNetworks          = "disableNetworks"
+	KeyManagementURL         = "managementURL"
+	KeyDisableUpdateSettings = "disableUpdateSettings"
+	KeyDisableProfiles       = "disableProfiles"
+	KeyDisableNetworks       = "disableNetworks"
 	// KeyDisableAdvancedView gates the advanced-view section in the
 	// upcoming UI revision. UI-only: NOT stored on Config, not
 	// applied by applyMDMPolicy, not rejectable via SetConfig. The
@@ -37,10 +37,16 @@ const (
 	KeyDisableMetricsCollection = "disableMetricsCollection"
 	KeyAllowServerSSH           = "allowServerSSH"
 	KeyDisableAutoConnect       = "disableAutoConnect"
-	KeyPreSharedKey             = "preSharedKey"
-	KeyRosenpassEnabled         = "rosenpassEnabled"
-	KeyRosenpassPermissive      = "rosenpassPermissive"
-	KeyWireguardPort            = "wireguardPort"
+	// KeyDisableAutostart suppresses the GUI's fresh-install
+	// launch-on-login default and marks the Settings toggle as
+	// MDM-managed. UI-only: NOT stored on Config and not applied by
+	// applyMDMPolicy; the GUI reads it directly and it appears in
+	// GetConfigResponse.mDMManagedFields when set.
+	KeyDisableAutostart    = "disableAutostart"
+	KeyPreSharedKey        = "preSharedKey"
+	KeyRosenpassEnabled    = "rosenpassEnabled"
+	KeyRosenpassPermissive = "rosenpassPermissive"
+	KeyWireguardPort       = "wireguardPort"
 
 	// Split tunnel is modeled as a single conceptual policy with two
 	// registry/plist values. KeySplitTunnelMode is the discriminator

client/proto/daemon_gateway_test.go

@@ -0,0 +1,80 @@
+package proto
+
+import (
+	"context"
+	"net"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	gatewayruntime "github.com/grpc-ecosystem/grpc-gateway/v2/runtime"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/credentials/insecure"
+	"google.golang.org/grpc/test/bufconn"
+)
+
+func TestGatewayServerRoutesCoverDaemonRPCs(t *testing.T) {
+	mux := gatewayruntime.NewServeMux()
+	if err := RegisterDaemonServiceHandlerServer(context.Background(), mux, UnimplementedDaemonServiceServer{}); err != nil {
+		t.Fatalf("register daemon gateway server handlers: %v", err)
+	}
+
+	assertAllDaemonGatewayRoutesRegistered(t, mux)
+}
+
+func TestGatewayClientRoutesCoverDaemonRPCs(t *testing.T) {
+	listener := bufconn.Listen(1024 * 1024)
+	server := grpc.NewServer()
+	RegisterDaemonServiceServer(server, UnimplementedDaemonServiceServer{})
+	go func() {
+		if err := server.Serve(listener); err != nil && err != grpc.ErrServerStopped {
+			t.Errorf("serve bufconn gRPC server: %v", err)
+		}
+	}()
+	t.Cleanup(func() {
+		server.Stop()
+		_ = listener.Close()
+	})
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	mux := gatewayruntime.NewServeMux()
+	opts := []grpc.DialOption{
+		grpc.WithContextDialer(func(context.Context, string) (net.Conn, error) {
+			return listener.Dial()
+		}),
+		grpc.WithTransportCredentials(insecure.NewCredentials()),
+	}
+	if err := RegisterDaemonServiceHandlerFromEndpoint(ctx, mux, "passthrough:///bufnet", opts); err != nil {
+		t.Fatalf("register daemon gateway client handlers: %v", err)
+	}
+
+	assertAllDaemonGatewayRoutesRegistered(t, mux)
+}
+
+func assertAllDaemonGatewayRoutesRegistered(t *testing.T, mux http.Handler) {
+	t.Helper()
+	for _, method := range DaemonService_ServiceDesc.Methods {
+		assertGatewayRouteRegistered(t, mux, method.MethodName)
+	}
+	for _, stream := range DaemonService_ServiceDesc.Streams {
+		assertGatewayRouteRegistered(t, mux, stream.StreamName)
+	}
+}
+
+func assertGatewayRouteRegistered(t *testing.T, mux http.Handler, methodName string) {
+	t.Helper()
+
+	path := "/daemon.DaemonService/" + methodName
+	req := httptest.NewRequest(http.MethodPost, path, strings.NewReader("{}"))
+	req.Header.Set("Content-Type", "application/json")
+	res := httptest.NewRecorder()
+
+	mux.ServeHTTP(res, req)
+
+	if res.Code == http.StatusNotFound {
+		t.Fatalf("gateway route for %s is not registered", methodName)
+	}
+}

client/proto/generate.sh

@@ -12,5 +12,11 @@ script_path=$(dirname "$(realpath "$0")")
 cd "$script_path"
 go install google.golang.org/protobuf/cmd/protoc-gen-go@v1.36.6
 go install google.golang.org/grpc/cmd/protoc-gen-go-grpc@v1.6.1
-protoc -I ./ ./daemon.proto --go_out=../ --go-grpc_out=../ --experimental_allow_proto3_optional
+go install github.com/grpc-ecosystem/grpc-gateway/v2/protoc-gen-grpc-gateway@v2.26.3
+protoc -I ./ ./daemon.proto \
+  --go_out=../ \
+  --go-grpc_out=../ \
+  --grpc-gateway_out=../ \
+  --grpc-gateway_opt=generate_unbound_methods=true \
+  --experimental_allow_proto3_optional
 cd "$old_pwd"

client/server/server.go

@@ -181,7 +181,7 @@ func (s *Server) Start() error {
 		log.Warnf("failed to redirect stderr: %v", err)
 	}
 
-	if err := restoreResidualState(s.rootCtx, s.profileManager.GetStatePath()); err != nil {
+	if err := RestoreResidualState(s.rootCtx, s.profileManager.GetStatePath()); err != nil {
 		log.Warnf(errRestoreResidualState, err)
 	}
 
@@ -551,7 +551,7 @@ func (s *Server) Login(callerCtx context.Context, msg *proto.LoginRequest) (*pro
 	s.actCancel = cancel
 	s.mutex.Unlock()
 
-	if err := restoreResidualState(s.rootCtx, s.profileManager.GetStatePath()); err != nil {
+	if err := RestoreResidualState(s.rootCtx, s.profileManager.GetStatePath()); err != nil {
 		log.Warnf(errRestoreResidualState, err)
 	}
 
@@ -828,6 +828,7 @@ func (s *Server) WaitSSOLogin(callerCtx context.Context, msg *proto.WaitSSOLogin
 		return nil, err
 	}
 
+	log.Infof("SSO login flow finished, returning success to caller")
 	return &proto.WaitSSOLoginResponse{
 		Email: tokenInfo.Email,
 	}, nil
@@ -835,6 +836,7 @@ func (s *Server) WaitSSOLogin(callerCtx context.Context, msg *proto.WaitSSOLogin
 
 // Up starts engine work in the daemon.
 func (s *Server) Up(callerCtx context.Context, msg *proto.UpRequest) (*proto.UpResponse, error) {
+	log.Infof("up request received")
 	s.mutex.Lock()
 	// clientRunning is the daemon-intent flag (set by previous Up/Start, cleared
 	// by Down). connectionGoroutineRunning() reports whether the previous retry-loop
@@ -856,7 +858,7 @@ func (s *Server) Up(callerCtx context.Context, msg *proto.UpRequest) (*proto.UpR
 
 		return s.waitForUp(callerCtx)
 	}
-	if err := restoreResidualState(callerCtx, s.profileManager.GetStatePath()); err != nil {
+	if err := RestoreResidualState(callerCtx, s.profileManager.GetStatePath()); err != nil {
 		log.Warnf(errRestoreResidualState, err)
 	}
 

client/server/state.go

@@ -46,7 +46,7 @@ func (s *Server) CleanState(ctx context.Context, req *proto.CleanStateRequest) (
 
 	if req.All {
 		// Reuse existing cleanup logic for all states
-		if err := restoreResidualState(ctx, statePath); err != nil {
+		if err := RestoreResidualState(ctx, statePath); err != nil {
 			return nil, status.Errorf(codes.Internal, "failed to clean all states: %v", err)
 		}
 
@@ -113,9 +113,9 @@ func (s *Server) DeleteState(ctx context.Context, req *proto.DeleteStateRequest)
 	}, nil
 }
 
-// restoreResidualState checks if the client was not shut down in a clean way and restores residual if required.
+// RestoreResidualState checks if the client was not shut down in a clean way and restores residual if required.
 // Otherwise, we might not be able to connect to the management server to retrieve new config.
-func restoreResidualState(ctx context.Context, statePath string) error {
+func RestoreResidualState(ctx context.Context, statePath string) error {
 	if statePath == "" {
 		return nil
 	}

client/status/status.go

@@ -746,6 +746,8 @@ func ToProtoFullStatus(fullStatus peer.FullStatus) *proto.FullStatus {
 		pbFullStatus.DnsServers = append(pbFullStatus.DnsServers, pbDnsState)
 	}
 
+	pbFullStatus.Events = fullStatus.Events
+
 	return &pbFullStatus
 }
 

client/system/process.go

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

client/system/process_test.go

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

client/test/json-socket-docker.sh

@@ -0,0 +1,223 @@
+#!/usr/bin/env bash
+set -eEuo pipefail
+
+usage() {
+  cat <<'EOF'
+Usage: client/test/json-socket-docker.sh [tcp|unix|both]
+
+Builds the NetBird client Docker image from the local source tree, starts
+`netbird service run` in a container with --enable-json-socket, and verifies
+that the HTTP/JSON daemon gateway responds to Status requests.
+
+Modes:
+  tcp   Validate tcp://0.0.0.0:8080 via a published localhost port (default)
+  unix  Validate unix:///sock/netbird-http.sock via a bind-mounted socket dir
+  both  Run both validations
+
+Environment:
+  CONTAINER_RUNTIME  docker or podman. Auto-detected if unset.
+  IMAGE              Image tag to build. Default: netbird-json-socket-test:local
+  TARGETARCH         Go/Docker target arch. Default: `go env GOARCH`
+  PLATFORM           Docker platform. Default: linux/$TARGETARCH
+  WAIT_TIMEOUT       Seconds to wait for the JSON socket. Default: 30
+EOF
+}
+
+if [[ "${1:-}" == "-h" || "${1:-}" == "--help" ]]; then
+  usage
+  exit 0
+fi
+
+MODE="${1:-tcp}"
+case "${MODE}" in
+  tcp|unix|both) ;;
+  *)
+    usage >&2
+    echo "invalid mode: ${MODE}" >&2
+    exit 2
+    ;;
+esac
+
+RUNTIME="${CONTAINER_RUNTIME:-}"
+if [[ -z "${RUNTIME}" ]]; then
+  if command -v docker >/dev/null 2>&1; then
+    RUNTIME=docker
+  elif command -v podman >/dev/null 2>&1; then
+    RUNTIME=podman
+  else
+    echo "docker or podman is required" >&2
+    exit 127
+  fi
+fi
+if ! command -v "${RUNTIME}" >/dev/null 2>&1; then
+  echo "container runtime not found: ${RUNTIME}" >&2
+  exit 127
+fi
+
+if ! command -v curl >/dev/null 2>&1; then
+  echo "curl is required" >&2
+  exit 127
+fi
+
+ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
+IMAGE="${IMAGE:-netbird-json-socket-test:local}"
+TARGETARCH="${TARGETARCH:-$(go env GOARCH)}"
+PLATFORM="${PLATFORM:-linux/${TARGETARCH}}"
+WAIT_TIMEOUT="${WAIT_TIMEOUT:-30}"
+TMP_DIR="$(mktemp -d)"
+CONTAINERS=()
+
+cleanup() {
+  local status=$?
+  for container in "${CONTAINERS[@]:-}"; do
+    "${RUNTIME}" rm -f "${container}" >/dev/null 2>&1 || true
+  done
+  rm -rf "${TMP_DIR}"
+  exit "${status}"
+}
+trap cleanup EXIT
+
+build_image() {
+  echo "==> Building Linux ${TARGETARCH} netbird binary"
+  mkdir -p "${TMP_DIR}/context/client"
+  cp "${ROOT_DIR}/client/Dockerfile" "${TMP_DIR}/context/Dockerfile"
+  cp "${ROOT_DIR}/client/netbird-entrypoint.sh" "${TMP_DIR}/context/client/netbird-entrypoint.sh"
+
+  (cd "${ROOT_DIR}" && CGO_ENABLED=0 GOOS=linux GOARCH="${TARGETARCH}" go build -o "${TMP_DIR}/context/netbird" ./client)
+
+  echo "==> Building ${IMAGE} for ${PLATFORM}"
+  "${RUNTIME}" build \
+    --platform "${PLATFORM}" \
+    --build-arg NETBIRD_BINARY=netbird \
+    -t "${IMAGE}" \
+    -f "${TMP_DIR}/context/Dockerfile" \
+    "${TMP_DIR}/context"
+}
+
+pick_port() {
+  python3 - <<'PY'
+import socket
+sock = socket.socket()
+sock.bind(("127.0.0.1", 0))
+print(sock.getsockname()[1])
+sock.close()
+PY
+}
+
+assert_status_json() {
+  local response_file="$1"
+  if command -v python3 >/dev/null 2>&1; then
+    python3 - "${response_file}" <<'PY'
+import json
+import sys
+with open(sys.argv[1], encoding="utf-8") as fh:
+    data = json.load(fh)
+if not data.get("status"):
+    raise SystemExit("missing non-empty status field")
+if "daemonVersion" not in data:
+    raise SystemExit("missing daemonVersion field")
+print(f"status={data['status']} daemonVersion={data['daemonVersion']}")
+PY
+  else
+    grep -q '"status"' "${response_file}"
+    grep -q '"daemonVersion"' "${response_file}"
+    cat "${response_file}"
+  fi
+}
+
+container_logs() {
+  local container="$1"
+  echo "---- ${container} logs ----" >&2
+  "${RUNTIME}" logs "${container}" >&2 || true
+  echo "--------------------------" >&2
+}
+
+wait_for_http_status() {
+  local container="$1"
+  local response="${TMP_DIR}/${container}.json"
+  local curl_err="${TMP_DIR}/${container}.curl.err"
+  shift
+  local deadline=$((SECONDS + WAIT_TIMEOUT))
+
+  while (( SECONDS < deadline )); do
+    if curl -fsS "$@" \
+      -X POST \
+      -H 'Content-Type: application/json' \
+      -d '{}' \
+      -o "${response}" \
+      2>"${curl_err}"; then
+      assert_status_json "${response}"
+      return 0
+    fi
+
+    if ! "${RUNTIME}" ps --format '{{.Names}}' | grep -Fxq "${container}"; then
+      echo "container exited before JSON socket became ready" >&2
+      container_logs "${container}"
+      return 1
+    fi
+    sleep 1
+  done
+
+  echo "timed out waiting for JSON socket after ${WAIT_TIMEOUT}s" >&2
+  cat "${curl_err}" >&2 || true
+  container_logs "${container}"
+  return 1
+}
+
+run_netbird_container() {
+  local container="$1"
+  local json_socket="$2"
+  shift 2
+
+  CONTAINERS+=("${container}")
+  "${RUNTIME}" run --rm -d \
+    --name "${container}" \
+    -e NB_STATE_DIR=/tmp/netbird-state \
+    --entrypoint /usr/local/bin/netbird \
+    "$@" \
+    "${IMAGE}" \
+    --log-file console \
+    --daemon-addr unix:///tmp/netbird.sock \
+    service run \
+    --enable-json-socket \
+    --json-socket "${json_socket}" >/dev/null
+}
+
+run_tcp_test() {
+  local port container
+  port="$(pick_port)"
+  container="nb-json-socket-tcp-$RANDOM-$RANDOM"
+
+  echo "==> Validating TCP JSON socket on 127.0.0.1:${port}"
+  run_netbird_container "${container}" "tcp://0.0.0.0:8080" -p "127.0.0.1:${port}:8080"
+  wait_for_http_status "${container}" "http://127.0.0.1:${port}/daemon.DaemonService/Status"
+}
+
+run_unix_test() {
+  local sock_dir sock_path container
+  sock_dir="${TMP_DIR}/sock"
+  sock_path="${sock_dir}/netbird-http.sock"
+  container="nb-json-socket-unix-$RANDOM-$RANDOM"
+  mkdir -p "${sock_dir}"
+
+  echo "==> Validating Unix JSON socket at ${sock_path}"
+  run_netbird_container "${container}" "unix:///sock/netbird-http.sock" -v "${sock_dir}:/sock"
+  wait_for_http_status "${container}" --unix-socket "${sock_path}" "http://unix/daemon.DaemonService/Status"
+}
+
+build_image
+
+case "${MODE}" in
+  tcp)
+    run_tcp_test
+    ;;
+  unix)
+    run_unix_test
+    ;;
+  both)
+    run_tcp_test
+    run_unix_test
+    ;;
+esac
+
+echo "==> Docker JSON socket validation passed (${MODE})"

client/ui/autostart_default.go

@@ -0,0 +1,107 @@
+//go:build !android && !ios && !freebsd && !js
+
+package main
+
+import (
+	"context"
+	"os"
+	"path/filepath"
+
+	log "github.com/sirupsen/logrus"
+
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mdm"
+	"github.com/netbirdio/netbird/client/ui/preferences"
+	"github.com/netbirdio/netbird/client/ui/services"
+)
+
+// autostartDefaultState carries the guard inputs of the one-time autostart
+// default decision so the decision itself stays a pure, testable function.
+type autostartDefaultState struct {
+	supported    bool
+	mdmDisabled  bool
+	priorInstall bool
+}
+
+// shouldEnableAutostartDefault applies the first-run guards in order and
+// returns whether autostart may be enabled, plus the reason when it may not.
+func shouldEnableAutostartDefault(s autostartDefaultState) (bool, string) {
+	switch {
+	case !s.supported:
+		return false, "autostart not supported on this platform"
+	case s.mdmDisabled:
+		return false, "autostart disabled by MDM policy"
+	case s.priorInstall:
+		return false, "existing NetBird installation"
+	}
+	return true, ""
+}
+
+// autostartDisabledByMDM reports whether the MDM policy manages the
+// disableAutostart key in a way that must suppress the default. An
+// unparseable managed value is treated as disabled to stay on the safe side.
+func autostartDisabledByMDM(policy *mdm.Policy) bool {
+	if !policy.HasKey(mdm.KeyDisableAutostart) {
+		return false
+	}
+	disabled, ok := policy.GetBool(mdm.KeyDisableAutostart)
+	return !ok || disabled
+}
+
+// netbirdFootprintExists reports whether the machine already carries NetBird
+// daemon config or state, meaning this is not a genuinely fresh install. It is
+// the update-safety gate for the autostart default: upgrading users always
+// have a footprint, so an update can never trigger a login-item write.
+func netbirdFootprintExists() bool {
+	candidates := []string{
+		profilemanager.DefaultConfigPath,
+		filepath.Join(profilemanager.DefaultConfigPathDir, "config.json"),
+		filepath.Join(profilemanager.DefaultConfigPathDir, "state.json"),
+	}
+	for _, path := range candidates {
+		if path != "" && fileExists(path) {
+			return true
+		}
+	}
+	return false
+}
+
+// applyAutostartDefault runs the one-time launch-on-login default for genuinely
+// fresh installs. The autostartInitialized marker is persisted before any
+// enable attempt so a crash mid-flow degrades to "never enabled" instead of
+// retrying login-item writes on every launch. A user's later disable in
+// Settings is never overridden: the marker guarantees at-most-once, ever.
+func applyAutostartDefault(ctx context.Context, autostart *services.Autostart, prefs *preferences.Store, prefsFileExisted bool) {
+	priorFootprint := netbirdFootprintExists() || prefsFileExisted
+
+	if prefs.Get().AutostartInitialized {
+		return
+	}
+	if err := prefs.SetAutostartInitialized(true); err != nil {
+		log.Warnf("persist autostart marker, skipping autostart default: %v", err)
+		return
+	}
+
+	state := autostartDefaultState{
+		supported:    autostart.Supported(ctx),
+		mdmDisabled:  autostartDisabledByMDM(mdm.LoadPolicy()),
+		priorInstall: priorFootprint,
+	}
+	enable, reason := shouldEnableAutostartDefault(state)
+	if !enable {
+		log.Debugf("skipping autostart default: %s", reason)
+		return
+	}
+
+	if err := autostart.SetEnabled(ctx, true); err != nil {
+		log.Warnf("enable autostart on fresh install: %v", err)
+		return
+	}
+	log.Info("autostart enabled by default on fresh install")
+}
+
+// fileExists reports whether path exists.
+func fileExists(path string) bool {
+	_, err := os.Stat(path)
+	return err == nil
+}

client/ui/autostart_default_test.go

@@ -0,0 +1,125 @@
+//go:build !android && !ios && !freebsd && !js
+
+package main
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	"github.com/netbirdio/netbird/client/mdm"
+)
+
+func TestShouldEnableAutostartDefault(t *testing.T) {
+	allPass := autostartDefaultState{
+		supported:    true,
+		mdmDisabled:  false,
+		priorInstall: false,
+	}
+
+	tests := []struct {
+		name       string
+		mutate     func(*autostartDefaultState)
+		wantEnable bool
+		wantReason string
+	}{
+		{
+			name:       "fresh install with all guards passing enables",
+			mutate:     func(*autostartDefaultState) {},
+			wantEnable: true,
+		},
+		{
+			name:       "unsupported platform skips",
+			mutate:     func(s *autostartDefaultState) { s.supported = false },
+			wantReason: "autostart not supported on this platform",
+		},
+		{
+			name:       "MDM disable skips",
+			mutate:     func(s *autostartDefaultState) { s.mdmDisabled = true },
+			wantReason: "autostart disabled by MDM policy",
+		},
+		{
+			name:       "existing installation (upgrade) skips",
+			mutate:     func(s *autostartDefaultState) { s.priorInstall = true },
+			wantReason: "existing NetBird installation",
+		},
+		{
+			name: "unsupported wins over every other guard",
+			mutate: func(s *autostartDefaultState) {
+				s.supported = false
+				s.mdmDisabled = true
+				s.priorInstall = true
+			},
+			wantReason: "autostart not supported on this platform",
+		},
+		{
+			name: "MDM disable wins over prior install",
+			mutate: func(s *autostartDefaultState) {
+				s.mdmDisabled = true
+				s.priorInstall = true
+			},
+			wantReason: "autostart disabled by MDM policy",
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			state := allPass
+			tc.mutate(&state)
+			enable, reason := shouldEnableAutostartDefault(state)
+			assert.Equal(t, tc.wantEnable, enable, "enable decision should match for state %+v", state)
+			assert.Equal(t, tc.wantReason, reason, "skip reason should identify the failing guard")
+		})
+	}
+}
+
+func TestAutostartDisabledByMDM(t *testing.T) {
+	tests := []struct {
+		name   string
+		values map[string]any
+		want   bool
+	}{
+		{
+			name:   "empty policy does not disable",
+			values: nil,
+			want:   false,
+		},
+		{
+			name:   "unrelated managed keys do not disable",
+			values: map[string]any{mdm.KeyDisableAutoConnect: true},
+			want:   false,
+		},
+		{
+			name:   "disableAutostart true disables",
+			values: map[string]any{mdm.KeyDisableAutostart: true},
+			want:   true,
+		},
+		{
+			name:   "disableAutostart registry DWORD 1 disables",
+			values: map[string]any{mdm.KeyDisableAutostart: int64(1)},
+			want:   true,
+		},
+		{
+			name:   "disableAutostart string true disables",
+			values: map[string]any{mdm.KeyDisableAutostart: "true"},
+			want:   true,
+		},
+		{
+			name:   "disableAutostart explicit false allows",
+			values: map[string]any{mdm.KeyDisableAutostart: false},
+			want:   false,
+		},
+		{
+			name:   "unparseable managed value is treated as disabled",
+			values: map[string]any{mdm.KeyDisableAutostart: "not-a-bool"},
+			want:   true,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			got := autostartDisabledByMDM(mdm.NewPolicy(tc.values))
+			assert.Equal(t, tc.want, got, "MDM disable decision should match for values %v", tc.values)
+		})
+	}
+}

client/ui/frontend/src/app.tsx

@@ -16,10 +16,13 @@ import LoginWaitingForBrowserDialog from "@/modules/login/LoginWaitingForBrowser
 import { initI18n } from "@/lib/i18n";
 import { initPlatform } from "@/lib/platform";
 import { initLogForwarding } from "@/lib/logs";
+import { initStallWatch } from "@/lib/stallwatch";
 
 // Must run first so even init-time logs reach the Go log pipeline.
 initLogForwarding();
 
+initStallWatch();
+
 welcome();
 
 Promise.all([

client/ui/frontend/src/lib/connection.ts

@@ -51,7 +51,14 @@ async function runSsoLogin(
     if (uri) await openBrowserLoginUri(uri);
 
     const cancelPromise = buildSsoCancelPromise(state, signal);
-    const waitPromise = Connection.WaitSSOLogin({ userCode: result.userCode, hostname: "" });
+    // Combine wait + up in Go so the connection comes up the moment SSO
+    // completes. During SSO the tray window is hidden and the webview is
+    // suspended, so a frontend-driven Up (a promise continuation) would not
+    // fire until the user woke the window (e.g. hovering the tray icon).
+    const waitPromise = Connection.WaitSSOLoginAndUp(
+        { userCode: result.userCode, hostname: "" },
+        { profileName: "", username: "" },
+    );
 
     try {
         await Promise.race([waitPromise, cancelPromise]);
@@ -89,13 +96,13 @@ export async function startConnection(onSettled?: () => void, signal?: AbortSign
         if (signal?.aborted) state.cancelled = true;
 
         if (!state.cancelled && result.needsSsoLogin) {
+            // runSsoLogin brings the connection up in Go once SSO completes.
             await runSsoLogin(result, state, signal);
-        }
-
-        if (!state.cancelled && signal?.aborted) state.cancelled = true;
-
-        if (!state.cancelled) {
-            await Connection.Up({ profileName: "", username: "" });
+        } else {
+            if (!state.cancelled && signal?.aborted) state.cancelled = true;
+            if (!state.cancelled) {
+                await Connection.Up({ profileName: "", username: "" });
+            }
         }
     } catch (e) {
         WindowManager.CloseBrowserLogin().catch(console.error);

client/ui/frontend/src/lib/stallwatch.ts

@@ -0,0 +1,31 @@
+// Detects webview suspension (macOS App Nap / hidden-window timer throttling).
+// While the webview is suspended no JS runs at all, so detection happens on
+// resume: a 1s interval measures wall-clock drift and reports how long timers
+// were frozen. Silent unless a stall actually occurred; a stalled webview is
+// what delays promise continuations such as the WaitSSOLogin → Up handoff.
+
+const INTERVAL_MS = 1000;
+const STALL_THRESHOLD_MS = 5000;
+const REPORT_COOLDOWN_MS = 60_000;
+
+let started = false;
+
+export function initStallWatch() {
+    if (started) return;
+    started = true;
+
+    let last = Date.now();
+    let lastReport = 0;
+    setInterval(() => {
+        const now = Date.now();
+        const stall = now - last - INTERVAL_MS;
+        last = now;
+        if (stall < STALL_THRESHOLD_MS) return;
+        if (now - lastReport < REPORT_COOLDOWN_MS) return;
+        lastReport = now;
+        console.warn(
+            `webview timers were suspended for ${(stall / 1000).toFixed(1)}s ` +
+                `(App Nap / hidden-window throttling); pending UI work ran late`,
+        );
+    }, INTERVAL_MS);
+}

client/ui/main.go

@@ -197,6 +197,9 @@ func main() {
 		// daemon may keep the main window from showing, so the OS toast is the
 		// only reliable signal the user gets.
 		go notifyIfDaemonOutdated(compat, notifier, localizer)
+		// One-time launch-on-login default for fresh installs; gated by the
+		// NetBird footprint check, MDM policy, and the persisted marker.
+		go applyAutostartDefault(context.Background(), services.NewAutostart(app.Autostart), prefStore, prefStore.ExistedAtLoad())
 	})
 
 	if err := app.Run(); err != nil {

client/ui/preferences/store.go

@@ -54,6 +54,10 @@ type UIPreferences struct {
 	Language            i18n.LanguageCode `json:"language"`
 	ViewMode            ViewMode          `json:"viewMode"`
 	OnboardingCompleted bool              `json:"onboardingCompleted"`
+	// AutostartInitialized records that the one-time autostart default
+	// decision has run for this OS user. It only ever transitions to true
+	// and is never reset, so the default-on flow runs at most once, ever.
+	AutostartInitialized bool `json:"autostartInitialized"`
 }
 
 // LanguageValidator rejects SetLanguage inputs with no shipped bundle.
@@ -72,8 +76,9 @@ type Emitter interface {
 type Store struct {
 	path string
 
-	mu      sync.RWMutex
-	current UIPreferences
+	mu            sync.RWMutex
+	current       UIPreferences
+	existedAtLoad bool
 
 	subsMu sync.Mutex
 	subs   []chan UIPreferences
@@ -157,6 +162,27 @@ func (s *Store) SetOnboardingCompleted(done bool) error {
 	return nil
 }
 
+// SetAutostartInitialized persists the one-time autostart decision marker.
+// No-op if unchanged.
+func (s *Store) SetAutostartInitialized(done bool) error {
+	s.mu.Lock()
+	if s.current.AutostartInitialized == done {
+		s.mu.Unlock()
+		return nil
+	}
+	next := s.current
+	next.AutostartInitialized = done
+	if err := s.persistLocked(next); err != nil {
+		s.mu.Unlock()
+		return fmt.Errorf("persist preferences: %w", err)
+	}
+	s.current = next
+	s.mu.Unlock()
+
+	s.broadcast(next)
+	return nil
+}
+
 // SetLanguage validates, persists, and broadcasts. No-op if unchanged.
 func (s *Store) SetLanguage(lang i18n.LanguageCode) error {
 	if lang == "" {
@@ -206,13 +232,29 @@ func (s *Store) Subscribe() (<-chan UIPreferences, func()) {
 	return ch, unsubscribe
 }
 
+// ExistedAtLoad reports whether the backing preferences file was present on
+// disk when the store loaded. It distinguishes a user who ran a prior GUI
+// version from a brand-new OS user with no preferences yet.
+func (s *Store) ExistedAtLoad() bool {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	return s.existedAtLoad
+}
+
 // load reads the file into current. A missing file is not an error (the
 // in-memory default stands); malformed contents return an error.
 func (s *Store) load() error {
-	if _, err := os.Stat(s.path); errors.Is(err, os.ErrNotExist) {
-		return nil
+	if _, err := os.Stat(s.path); err != nil {
+		if errors.Is(err, os.ErrNotExist) {
+			return nil
+		}
+		return fmt.Errorf("stat preferences: %w", err)
 	}
 
+	s.mu.Lock()
+	s.existedAtLoad = true
+	s.mu.Unlock()
+
 	var loaded UIPreferences
 	if _, err := util.ReadJson(s.path, &loaded); err != nil {
 		return err

client/ui/preferences/store_test.go

@@ -215,6 +215,46 @@ func TestStore_FileShapeIsJSON(t *testing.T) {
 	assert.Equal(t, i18n.LanguageCode("hu"), parsed.Language)
 }
 
+func TestStore_SetAutostartInitializedPersistsAcrossReload(t *testing.T) {
+	withTempConfigDir(t)
+	emitter := &recordingEmitter{}
+	s, err := NewStore(nil, emitter)
+	require.NoError(t, err)
+
+	assert.False(t, s.Get().AutostartInitialized, "marker must default to false when no file is on disk")
+
+	require.NoError(t, s.SetAutostartInitialized(true))
+	assert.True(t, s.Get().AutostartInitialized, "Get should reflect the persisted marker")
+	require.Len(t, emitter.calledWith(EventPreferencesChanged), 1, "first marker write should broadcast")
+
+	// Re-setting the same value must be a no-op: no disk write, no broadcast.
+	require.NoError(t, s.SetAutostartInitialized(true))
+	assert.Len(t, emitter.calledWith(EventPreferencesChanged), 1, "idempotent marker write should not broadcast again")
+
+	// A fresh Store (new GUI launch) must see the marker so the autostart
+	// default decision never runs twice.
+	reloaded, err := NewStore(nil, nil)
+	require.NoError(t, err)
+	assert.True(t, reloaded.Get().AutostartInitialized, "marker must survive a reload from disk")
+}
+
+func TestStore_ExistedAtLoad(t *testing.T) {
+	withTempConfigDir(t)
+
+	// Brand-new OS user: no preferences file on disk yet.
+	fresh, err := NewStore(nil, nil)
+	require.NoError(t, err)
+	assert.False(t, fresh.ExistedAtLoad(), "ExistedAtLoad must be false when no file is on disk")
+
+	// Persisting a value writes the file to disk.
+	require.NoError(t, fresh.SetLanguage("en"))
+
+	// A subsequent GUI launch reopens the now-present file.
+	reopened, err := NewStore(nil, nil)
+	require.NoError(t, err)
+	assert.True(t, reopened.ExistedAtLoad(), "ExistedAtLoad must be true after the store has persisted and is reopened")
+}
+
 func TestStore_ErrUnsupportedSentinel(t *testing.T) {
 	// Verifies callers can match on the sentinel error rather than parsing
 	// strings — protects against accidental %v -> %w changes that would

client/ui/services/connection.go

@@ -35,7 +35,7 @@ type LoginResult struct {
 	VerificationURIComplete string `json:"verificationUriComplete"`
 }
 
-// WaitSSOParams are the inputs to WaitSSOLogin.
+// WaitSSOParams are the inputs to waitSSOLogin.
 type WaitSSOParams struct {
 	UserCode string `json:"userCode"`
 	Hostname string `json:"hostname"`
@@ -116,6 +116,7 @@ func (s *Connection) Login(ctx context.Context, p LoginParams) (LoginResult, err
 	if err != nil {
 		return LoginResult{}, s.classifyDaemonError(err)
 	}
+	log.Infof("daemon login response received, needs SSO login: %v", resp.GetNeedsSSOLogin())
 	return LoginResult{
 		NeedsSSOLogin:           resp.GetNeedsSSOLogin(),
 		UserCode:                resp.GetUserCode(),
@@ -124,26 +125,12 @@ func (s *Connection) Login(ctx context.Context, p LoginParams) (LoginResult, err
 	}, nil
 }
 
-func (s *Connection) WaitSSOLogin(ctx context.Context, p WaitSSOParams) (string, error) {
-	cli, err := s.conn.Client()
-	if err != nil {
-		return "", err
-	}
-	resp, err := cli.WaitSSOLogin(ctx, &proto.WaitSSOLoginRequest{
-		UserCode: p.UserCode,
-		Hostname: p.Hostname,
-	})
-	if err != nil {
-		return "", s.classifyDaemonError(err)
-	}
-	return resp.GetEmail(), nil
-}
-
 func (s *Connection) Up(ctx context.Context, p UpParams) error {
 	cli, err := s.conn.Client()
 	if err != nil {
 		return err
 	}
+	log.Infof("sending up request to daemon")
 	// Always async: status updates flow via SubscribeStatus.
 	req := &proto.UpRequest{Async: true}
 	if p.ProfileName != "" {
@@ -158,6 +145,27 @@ func (s *Connection) Up(ctx context.Context, p UpParams) error {
 	return nil
 }
 
+// WaitSSOLoginAndUp blocks until the SSO login completes and then brings the
+// connection up, both from the Go side. Keeping the post-login Up here rather
+// than as a frontend continuation is deliberate: during SSO the tray window is
+// hidden and the webview is suspended (macOS App Nap / hidden-window timer
+// throttling), so a frontend-driven Up would not run until the user woke the
+// window (e.g. by hovering the tray icon). Doing it in Go connects the moment
+// the daemon reports SSO success. Returns the authenticated user's email.
+func (s *Connection) WaitSSOLoginAndUp(ctx context.Context, wait WaitSSOParams, up UpParams) (string, error) {
+	email, err := s.waitSSOLogin(ctx, wait)
+	if err != nil {
+		return "", err
+	}
+	if err := ctx.Err(); err != nil {
+		return "", err
+	}
+	if err := s.Up(ctx, up); err != nil {
+		return "", err
+	}
+	return email, nil
+}
+
 func (s *Connection) Down(ctx context.Context) error {
 	cli, err := s.conn.Client()
 	if err != nil {
@@ -217,6 +225,26 @@ func (s *Connection) Logout(ctx context.Context, p LogoutParams) error {
 	return nil
 }
 
+// waitSSOLogin blocks until the daemon reports the SSO login result and returns
+// the authenticated user's email. It is unexported because the frontend drives
+// SSO through the exported WaitSSOLoginAndUp.
+func (s *Connection) waitSSOLogin(ctx context.Context, p WaitSSOParams) (string, error) {
+	cli, err := s.conn.Client()
+	if err != nil {
+		return "", err
+	}
+	log.Infof("waiting for SSO login to complete")
+	resp, err := cli.WaitSSOLogin(ctx, &proto.WaitSSOLoginRequest{
+		UserCode: p.UserCode,
+		Hostname: p.Hostname,
+	})
+	if err != nil {
+		return "", s.classifyDaemonError(err)
+	}
+	log.Infof("SSO login completed, daemon reported success")
+	return resp.GetEmail(), nil
+}
+
 // classifyDaemonError maps a gRPC error to a localised ClientError.
 func (s *Connection) classifyDaemonError(err error) *ClientError {
 	return s.classifier.classify(err)

client/ui/services/settings.go

@@ -20,11 +20,12 @@ type MDMFields struct {
 	DisableServerRoutes      bool   `json:"disableServerRoutes"`
 	AllowServerSSH           *bool  `json:"allowServerSSH"`
 	DisableAutoConnect       bool   `json:"disableAutoConnect"`
+	DisableAutostart         bool   `json:"disableAutostart"`
 	BlockInbound             bool   `json:"blockInbound"`
 	DisableMetricsCollection bool   `json:"disableMetricsCollection"`
 	SplitTunnelMode          bool   `json:"splitTunnelMode"`
 	SplitTunnelApps          bool   `json:"splitTunnelApps"`
-	DisableAdvancedView bool `json:"disableAdvancedView"`
+	DisableAdvancedView      bool   `json:"disableAdvancedView"`
 }
 
 type Features struct {

client/wasm/internal/rdp/rdcleanpath.go

@@ -23,7 +23,7 @@ const (
 	RDCleanPathProxyHost   = "rdcleanpath.proxy.local"
 	RDCleanPathProxyScheme = "ws"
 
-	rdpDialTimeout = 15 * time.Second
+	rdpDialTimeout = 30 * time.Second
 
 	GeneralErrorCode = 1
 	WSAETimedOut     = 10060

e2e/agentnetwork/chat_test.go

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

e2e/agentnetwork/guardrail_test.go

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

e2e/harness/agentnetwork.go

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

e2e/harness/client.go

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

flow/client/client.go

@@ -109,7 +109,7 @@ func (c *GRPCClient) Close() error {
 func (c *GRPCClient) Send(event *proto.FlowEvent) error {
 	c.mu.Lock()
 	stream := c.stream
-	c.mu.Unlock()
+	defer c.mu.Unlock() // stream.Send() is not safe to call concurrently from multiple goroutines
 
 	if stream == nil {
 		return errors.New("stream not initialized")

flow/proto/flow.pb.go

@@ -134,9 +134,11 @@ type FlowEvent struct {
 	// When the event occurred
 	Timestamp *timestamppb.Timestamp `protobuf:"bytes,2,opt,name=timestamp,proto3" json:"timestamp,omitempty"`
 	// Public key of the sending peer
-	PublicKey   []byte      `protobuf:"bytes,3,opt,name=public_key,json=publicKey,proto3" json:"public_key,omitempty"`
-	FlowFields  *FlowFields `protobuf:"bytes,4,opt,name=flow_fields,json=flowFields,proto3" json:"flow_fields,omitempty"`
-	IsInitiator bool        `protobuf:"varint,5,opt,name=isInitiator,proto3" json:"isInitiator,omitempty"`
+	PublicKey   []byte                 `protobuf:"bytes,3,opt,name=public_key,json=publicKey,proto3" json:"public_key,omitempty"`
+	FlowFields  *FlowFields            `protobuf:"bytes,4,opt,name=flow_fields,json=flowFields,proto3" json:"flow_fields,omitempty"`
+	IsInitiator bool                   `protobuf:"varint,5,opt,name=isInitiator,proto3" json:"isInitiator,omitempty"`
+	WindowStart *timestamppb.Timestamp `protobuf:"bytes,6,opt,name=window_start,json=windowStart,proto3" json:"window_start,omitempty"`
+	WindowEnd   *timestamppb.Timestamp `protobuf:"bytes,7,opt,name=window_end,json=windowEnd,proto3" json:"window_end,omitempty"`
 }
 
 func (x *FlowEvent) Reset() {
@@ -206,6 +208,20 @@ func (x *FlowEvent) GetIsInitiator() bool {
 	return false
 }
 
+func (x *FlowEvent) GetWindowStart() *timestamppb.Timestamp {
+	if x != nil {
+		return x.WindowStart
+	}
+	return nil
+}
+
+func (x *FlowEvent) GetWindowEnd() *timestamppb.Timestamp {
+	if x != nil {
+		return x.WindowEnd
+	}
+	return nil
+}
+
 type FlowEventAck struct {
 	state         protoimpl.MessageState
 	sizeCache     protoimpl.SizeCache
@@ -284,7 +300,6 @@ type FlowFields struct {
 	// Layer 4 -specific information
 	//
 	// Types that are assignable to ConnectionInfo:
-	//
 	//	*FlowFields_PortInfo
 	//	*FlowFields_IcmpInfo
 	ConnectionInfo isFlowFields_ConnectionInfo `protobuf_oneof:"connection_info"`
@@ -297,6 +312,9 @@ type FlowFields struct {
 	// Resource ID
 	SourceResourceId []byte `protobuf:"bytes,14,opt,name=source_resource_id,json=sourceResourceId,proto3" json:"source_resource_id,omitempty"`
 	DestResourceId   []byte `protobuf:"bytes,15,opt,name=dest_resource_id,json=destResourceId,proto3" json:"dest_resource_id,omitempty"`
+	NumOfStarts      uint64 `protobuf:"varint,16,opt,name=num_of_starts,json=numOfStarts,proto3" json:"num_of_starts,omitempty"`
+	NumOfEnds        uint64 `protobuf:"varint,17,opt,name=num_of_ends,json=numOfEnds,proto3" json:"num_of_ends,omitempty"`
+	NumOfDrops       uint64 `protobuf:"varint,18,opt,name=num_of_drops,json=numOfDrops,proto3" json:"num_of_drops,omitempty"`
 }
 
 func (x *FlowFields) Reset() {
@@ -443,6 +461,27 @@ func (x *FlowFields) GetDestResourceId() []byte {
 	return nil
 }
 
+func (x *FlowFields) GetNumOfStarts() uint64 {
+	if x != nil {
+		return x.NumOfStarts
+	}
+	return 0
+}
+
+func (x *FlowFields) GetNumOfEnds() uint64 {
+	if x != nil {
+		return x.NumOfEnds
+	}
+	return 0
+}
+
+func (x *FlowFields) GetNumOfDrops() uint64 {
+	if x != nil {
+		return x.NumOfDrops
+	}
+	return 0
+}
+
 type isFlowFields_ConnectionInfo interface {
 	isFlowFields_ConnectionInfo()
 }
@@ -579,7 +618,7 @@ var file_flow_proto_rawDesc = []byte{
 	0x0a, 0x0a, 0x66, 0x6c, 0x6f, 0x77, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x12, 0x04, 0x66, 0x6c,
 	0x6f, 0x77, 0x1a, 0x1f, 0x67, 0x6f, 0x6f, 0x67, 0x6c, 0x65, 0x2f, 0x70, 0x72, 0x6f, 0x74, 0x6f,
 	0x62, 0x75, 0x66, 0x2f, 0x74, 0x69, 0x6d, 0x65, 0x73, 0x74, 0x61, 0x6d, 0x70, 0x2e, 0x70, 0x72,
-	0x6f, 0x74, 0x6f, 0x22, 0xd4, 0x01, 0x0a, 0x09, 0x46, 0x6c, 0x6f, 0x77, 0x45, 0x76, 0x65, 0x6e,
+	0x6f, 0x74, 0x6f, 0x22, 0xce, 0x02, 0x0a, 0x09, 0x46, 0x6c, 0x6f, 0x77, 0x45, 0x76, 0x65, 0x6e,
 	0x74, 0x12, 0x19, 0x0a, 0x08, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x5f, 0x69, 0x64, 0x18, 0x01, 0x20,
 	0x01, 0x28, 0x0c, 0x52, 0x07, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x49, 0x64, 0x12, 0x38, 0x0a, 0x09,
 	0x74, 0x69, 0x6d, 0x65, 0x73, 0x74, 0x61, 0x6d, 0x70, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0b, 0x32,
@@ -592,45 +631,59 @@ var file_flow_proto_rawDesc = []byte{
 	0x77, 0x2e, 0x46, 0x6c, 0x6f, 0x77, 0x46, 0x69, 0x65, 0x6c, 0x64, 0x73, 0x52, 0x0a, 0x66, 0x6c,
 	0x6f, 0x77, 0x46, 0x69, 0x65, 0x6c, 0x64, 0x73, 0x12, 0x20, 0x0a, 0x0b, 0x69, 0x73, 0x49, 0x6e,
 	0x69, 0x74, 0x69, 0x61, 0x74, 0x6f, 0x72, 0x18, 0x05, 0x20, 0x01, 0x28, 0x08, 0x52, 0x0b, 0x69,
-	0x73, 0x49, 0x6e, 0x69, 0x74, 0x69, 0x61, 0x74, 0x6f, 0x72, 0x22, 0x4b, 0x0a, 0x0c, 0x46, 0x6c,
-	0x6f, 0x77, 0x45, 0x76, 0x65, 0x6e, 0x74, 0x41, 0x63, 0x6b, 0x12, 0x19, 0x0a, 0x08, 0x65, 0x76,
-	0x65, 0x6e, 0x74, 0x5f, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x07, 0x65, 0x76,
-	0x65, 0x6e, 0x74, 0x49, 0x64, 0x12, 0x20, 0x0a, 0x0b, 0x69, 0x73, 0x49, 0x6e, 0x69, 0x74, 0x69,
-	0x61, 0x74, 0x6f, 0x72, 0x18, 0x02, 0x20, 0x01, 0x28, 0x08, 0x52, 0x0b, 0x69, 0x73, 0x49, 0x6e,
-	0x69, 0x74, 0x69, 0x61, 0x74, 0x6f, 0x72, 0x22, 0x9c, 0x04, 0x0a, 0x0a, 0x46, 0x6c, 0x6f, 0x77,
-	0x46, 0x69, 0x65, 0x6c, 0x64, 0x73, 0x12, 0x17, 0x0a, 0x07, 0x66, 0x6c, 0x6f, 0x77, 0x5f, 0x69,
-	0x64, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x06, 0x66, 0x6c, 0x6f, 0x77, 0x49, 0x64, 0x12,
-	0x1e, 0x0a, 0x04, 0x74, 0x79, 0x70, 0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0e, 0x32, 0x0a, 0x2e,
-	0x66, 0x6c, 0x6f, 0x77, 0x2e, 0x54, 0x79, 0x70, 0x65, 0x52, 0x04, 0x74, 0x79, 0x70, 0x65, 0x12,
-	0x17, 0x0a, 0x07, 0x72, 0x75, 0x6c, 0x65, 0x5f, 0x69, 0x64, 0x18, 0x03, 0x20, 0x01, 0x28, 0x0c,
-	0x52, 0x06, 0x72, 0x75, 0x6c, 0x65, 0x49, 0x64, 0x12, 0x2d, 0x0a, 0x09, 0x64, 0x69, 0x72, 0x65,
-	0x63, 0x74, 0x69, 0x6f, 0x6e, 0x18, 0x04, 0x20, 0x01, 0x28, 0x0e, 0x32, 0x0f, 0x2e, 0x66, 0x6c,
-	0x6f, 0x77, 0x2e, 0x44, 0x69, 0x72, 0x65, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x52, 0x09, 0x64, 0x69,
-	0x72, 0x65, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x12, 0x1a, 0x0a, 0x08, 0x70, 0x72, 0x6f, 0x74, 0x6f,
-	0x63, 0x6f, 0x6c, 0x18, 0x05, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x08, 0x70, 0x72, 0x6f, 0x74, 0x6f,
-	0x63, 0x6f, 0x6c, 0x12, 0x1b, 0x0a, 0x09, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x69, 0x70,
-	0x18, 0x06, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x08, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x49, 0x70,
-	0x12, 0x17, 0x0a, 0x07, 0x64, 0x65, 0x73, 0x74, 0x5f, 0x69, 0x70, 0x18, 0x07, 0x20, 0x01, 0x28,
-	0x0c, 0x52, 0x06, 0x64, 0x65, 0x73, 0x74, 0x49, 0x70, 0x12, 0x2d, 0x0a, 0x09, 0x70, 0x6f, 0x72,
-	0x74, 0x5f, 0x69, 0x6e, 0x66, 0x6f, 0x18, 0x08, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x0e, 0x2e, 0x66,
-	0x6c, 0x6f, 0x77, 0x2e, 0x50, 0x6f, 0x72, 0x74, 0x49, 0x6e, 0x66, 0x6f, 0x48, 0x00, 0x52, 0x08,
-	0x70, 0x6f, 0x72, 0x74, 0x49, 0x6e, 0x66, 0x6f, 0x12, 0x2d, 0x0a, 0x09, 0x69, 0x63, 0x6d, 0x70,
-	0x5f, 0x69, 0x6e, 0x66, 0x6f, 0x18, 0x09, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x0e, 0x2e, 0x66, 0x6c,
-	0x6f, 0x77, 0x2e, 0x49, 0x43, 0x4d, 0x50, 0x49, 0x6e, 0x66, 0x6f, 0x48, 0x00, 0x52, 0x08, 0x69,
-	0x63, 0x6d, 0x70, 0x49, 0x6e, 0x66, 0x6f, 0x12, 0x1d, 0x0a, 0x0a, 0x72, 0x78, 0x5f, 0x70, 0x61,
-	0x63, 0x6b, 0x65, 0x74, 0x73, 0x18, 0x0a, 0x20, 0x01, 0x28, 0x04, 0x52, 0x09, 0x72, 0x78, 0x50,
-	0x61, 0x63, 0x6b, 0x65, 0x74, 0x73, 0x12, 0x1d, 0x0a, 0x0a, 0x74, 0x78, 0x5f, 0x70, 0x61, 0x63,
-	0x6b, 0x65, 0x74, 0x73, 0x18, 0x0b, 0x20, 0x01, 0x28, 0x04, 0x52, 0x09, 0x74, 0x78, 0x50, 0x61,
-	0x63, 0x6b, 0x65, 0x74, 0x73, 0x12, 0x19, 0x0a, 0x08, 0x72, 0x78, 0x5f, 0x62, 0x79, 0x74, 0x65,
-	0x73, 0x18, 0x0c, 0x20, 0x01, 0x28, 0x04, 0x52, 0x07, 0x72, 0x78, 0x42, 0x79, 0x74, 0x65, 0x73,
-	0x12, 0x19, 0x0a, 0x08, 0x74, 0x78, 0x5f, 0x62, 0x79, 0x74, 0x65, 0x73, 0x18, 0x0d, 0x20, 0x01,
-	0x28, 0x04, 0x52, 0x07, 0x74, 0x78, 0x42, 0x79, 0x74, 0x65, 0x73, 0x12, 0x2c, 0x0a, 0x12, 0x73,
-	0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x72, 0x65, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x69,
-	0x64, 0x18, 0x0e, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x10, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x52,
-	0x65, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x49, 0x64, 0x12, 0x28, 0x0a, 0x10, 0x64, 0x65, 0x73,
-	0x74, 0x5f, 0x72, 0x65, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x69, 0x64, 0x18, 0x0f, 0x20,
-	0x01, 0x28, 0x0c, 0x52, 0x0e, 0x64, 0x65, 0x73, 0x74, 0x52, 0x65, 0x73, 0x6f, 0x75, 0x72, 0x63,
-	0x65, 0x49, 0x64, 0x42, 0x11, 0x0a, 0x0f, 0x63, 0x6f, 0x6e, 0x6e, 0x65, 0x63, 0x74, 0x69, 0x6f,
+	0x73, 0x49, 0x6e, 0x69, 0x74, 0x69, 0x61, 0x74, 0x6f, 0x72, 0x12, 0x3d, 0x0a, 0x0c, 0x77, 0x69,
+	0x6e, 0x64, 0x6f, 0x77, 0x5f, 0x73, 0x74, 0x61, 0x72, 0x74, 0x18, 0x06, 0x20, 0x01, 0x28, 0x0b,
+	0x32, 0x1a, 0x2e, 0x67, 0x6f, 0x6f, 0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62,
+	0x75, 0x66, 0x2e, 0x54, 0x69, 0x6d, 0x65, 0x73, 0x74, 0x61, 0x6d, 0x70, 0x52, 0x0b, 0x77, 0x69,
+	0x6e, 0x64, 0x6f, 0x77, 0x53, 0x74, 0x61, 0x72, 0x74, 0x12, 0x39, 0x0a, 0x0a, 0x77, 0x69, 0x6e,
+	0x64, 0x6f, 0x77, 0x5f, 0x65, 0x6e, 0x64, 0x18, 0x07, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x1a, 0x2e,
+	0x67, 0x6f, 0x6f, 0x67, 0x6c, 0x65, 0x2e, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62, 0x75, 0x66, 0x2e,
+	0x54, 0x69, 0x6d, 0x65, 0x73, 0x74, 0x61, 0x6d, 0x70, 0x52, 0x09, 0x77, 0x69, 0x6e, 0x64, 0x6f,
+	0x77, 0x45, 0x6e, 0x64, 0x22, 0x4b, 0x0a, 0x0c, 0x46, 0x6c, 0x6f, 0x77, 0x45, 0x76, 0x65, 0x6e,
+	0x74, 0x41, 0x63, 0x6b, 0x12, 0x19, 0x0a, 0x08, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x5f, 0x69, 0x64,
+	0x18, 0x01, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x07, 0x65, 0x76, 0x65, 0x6e, 0x74, 0x49, 0x64, 0x12,
+	0x20, 0x0a, 0x0b, 0x69, 0x73, 0x49, 0x6e, 0x69, 0x74, 0x69, 0x61, 0x74, 0x6f, 0x72, 0x18, 0x02,
+	0x20, 0x01, 0x28, 0x08, 0x52, 0x0b, 0x69, 0x73, 0x49, 0x6e, 0x69, 0x74, 0x69, 0x61, 0x74, 0x6f,
+	0x72, 0x22, 0x82, 0x05, 0x0a, 0x0a, 0x46, 0x6c, 0x6f, 0x77, 0x46, 0x69, 0x65, 0x6c, 0x64, 0x73,
+	0x12, 0x17, 0x0a, 0x07, 0x66, 0x6c, 0x6f, 0x77, 0x5f, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28,
+	0x0c, 0x52, 0x06, 0x66, 0x6c, 0x6f, 0x77, 0x49, 0x64, 0x12, 0x1e, 0x0a, 0x04, 0x74, 0x79, 0x70,
+	0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0e, 0x32, 0x0a, 0x2e, 0x66, 0x6c, 0x6f, 0x77, 0x2e, 0x54,
+	0x79, 0x70, 0x65, 0x52, 0x04, 0x74, 0x79, 0x70, 0x65, 0x12, 0x17, 0x0a, 0x07, 0x72, 0x75, 0x6c,
+	0x65, 0x5f, 0x69, 0x64, 0x18, 0x03, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x06, 0x72, 0x75, 0x6c, 0x65,
+	0x49, 0x64, 0x12, 0x2d, 0x0a, 0x09, 0x64, 0x69, 0x72, 0x65, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x18,
+	0x04, 0x20, 0x01, 0x28, 0x0e, 0x32, 0x0f, 0x2e, 0x66, 0x6c, 0x6f, 0x77, 0x2e, 0x44, 0x69, 0x72,
+	0x65, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x52, 0x09, 0x64, 0x69, 0x72, 0x65, 0x63, 0x74, 0x69, 0x6f,
+	0x6e, 0x12, 0x1a, 0x0a, 0x08, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x63, 0x6f, 0x6c, 0x18, 0x05, 0x20,
+	0x01, 0x28, 0x0d, 0x52, 0x08, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x63, 0x6f, 0x6c, 0x12, 0x1b, 0x0a,
+	0x09, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x69, 0x70, 0x18, 0x06, 0x20, 0x01, 0x28, 0x0c,
+	0x52, 0x08, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x49, 0x70, 0x12, 0x17, 0x0a, 0x07, 0x64, 0x65,
+	0x73, 0x74, 0x5f, 0x69, 0x70, 0x18, 0x07, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x06, 0x64, 0x65, 0x73,
+	0x74, 0x49, 0x70, 0x12, 0x2d, 0x0a, 0x09, 0x70, 0x6f, 0x72, 0x74, 0x5f, 0x69, 0x6e, 0x66, 0x6f,
+	0x18, 0x08, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x0e, 0x2e, 0x66, 0x6c, 0x6f, 0x77, 0x2e, 0x50, 0x6f,
+	0x72, 0x74, 0x49, 0x6e, 0x66, 0x6f, 0x48, 0x00, 0x52, 0x08, 0x70, 0x6f, 0x72, 0x74, 0x49, 0x6e,
+	0x66, 0x6f, 0x12, 0x2d, 0x0a, 0x09, 0x69, 0x63, 0x6d, 0x70, 0x5f, 0x69, 0x6e, 0x66, 0x6f, 0x18,
+	0x09, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x0e, 0x2e, 0x66, 0x6c, 0x6f, 0x77, 0x2e, 0x49, 0x43, 0x4d,
+	0x50, 0x49, 0x6e, 0x66, 0x6f, 0x48, 0x00, 0x52, 0x08, 0x69, 0x63, 0x6d, 0x70, 0x49, 0x6e, 0x66,
+	0x6f, 0x12, 0x1d, 0x0a, 0x0a, 0x72, 0x78, 0x5f, 0x70, 0x61, 0x63, 0x6b, 0x65, 0x74, 0x73, 0x18,
+	0x0a, 0x20, 0x01, 0x28, 0x04, 0x52, 0x09, 0x72, 0x78, 0x50, 0x61, 0x63, 0x6b, 0x65, 0x74, 0x73,
+	0x12, 0x1d, 0x0a, 0x0a, 0x74, 0x78, 0x5f, 0x70, 0x61, 0x63, 0x6b, 0x65, 0x74, 0x73, 0x18, 0x0b,
+	0x20, 0x01, 0x28, 0x04, 0x52, 0x09, 0x74, 0x78, 0x50, 0x61, 0x63, 0x6b, 0x65, 0x74, 0x73, 0x12,
+	0x19, 0x0a, 0x08, 0x72, 0x78, 0x5f, 0x62, 0x79, 0x74, 0x65, 0x73, 0x18, 0x0c, 0x20, 0x01, 0x28,
+	0x04, 0x52, 0x07, 0x72, 0x78, 0x42, 0x79, 0x74, 0x65, 0x73, 0x12, 0x19, 0x0a, 0x08, 0x74, 0x78,
+	0x5f, 0x62, 0x79, 0x74, 0x65, 0x73, 0x18, 0x0d, 0x20, 0x01, 0x28, 0x04, 0x52, 0x07, 0x74, 0x78,
+	0x42, 0x79, 0x74, 0x65, 0x73, 0x12, 0x2c, 0x0a, 0x12, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f,
+	0x72, 0x65, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x69, 0x64, 0x18, 0x0e, 0x20, 0x01, 0x28,
+	0x0c, 0x52, 0x10, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x52, 0x65, 0x73, 0x6f, 0x75, 0x72, 0x63,
+	0x65, 0x49, 0x64, 0x12, 0x28, 0x0a, 0x10, 0x64, 0x65, 0x73, 0x74, 0x5f, 0x72, 0x65, 0x73, 0x6f,
+	0x75, 0x72, 0x63, 0x65, 0x5f, 0x69, 0x64, 0x18, 0x0f, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x0e, 0x64,
+	0x65, 0x73, 0x74, 0x52, 0x65, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x49, 0x64, 0x12, 0x22, 0x0a,
+	0x0d, 0x6e, 0x75, 0x6d, 0x5f, 0x6f, 0x66, 0x5f, 0x73, 0x74, 0x61, 0x72, 0x74, 0x73, 0x18, 0x10,
+	0x20, 0x01, 0x28, 0x04, 0x52, 0x0b, 0x6e, 0x75, 0x6d, 0x4f, 0x66, 0x53, 0x74, 0x61, 0x72, 0x74,
+	0x73, 0x12, 0x1e, 0x0a, 0x0b, 0x6e, 0x75, 0x6d, 0x5f, 0x6f, 0x66, 0x5f, 0x65, 0x6e, 0x64, 0x73,
+	0x18, 0x11, 0x20, 0x01, 0x28, 0x04, 0x52, 0x09, 0x6e, 0x75, 0x6d, 0x4f, 0x66, 0x45, 0x6e, 0x64,
+	0x73, 0x12, 0x20, 0x0a, 0x0c, 0x6e, 0x75, 0x6d, 0x5f, 0x6f, 0x66, 0x5f, 0x64, 0x72, 0x6f, 0x70,
+	0x73, 0x18, 0x12, 0x20, 0x01, 0x28, 0x04, 0x52, 0x0a, 0x6e, 0x75, 0x6d, 0x4f, 0x66, 0x44, 0x72,
+	0x6f, 0x70, 0x73, 0x42, 0x11, 0x0a, 0x0f, 0x63, 0x6f, 0x6e, 0x6e, 0x65, 0x63, 0x74, 0x69, 0x6f,
 	0x6e, 0x5f, 0x69, 0x6e, 0x66, 0x6f, 0x22, 0x48, 0x0a, 0x08, 0x50, 0x6f, 0x72, 0x74, 0x49, 0x6e,
 	0x66, 0x6f, 0x12, 0x1f, 0x0a, 0x0b, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x5f, 0x70, 0x6f, 0x72,
 	0x74, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x0a, 0x73, 0x6f, 0x75, 0x72, 0x63, 0x65, 0x50,
@@ -683,17 +736,19 @@ var file_flow_proto_goTypes = []interface{}{
 var file_flow_proto_depIdxs = []int32{
 	7, // 0: flow.FlowEvent.timestamp:type_name -> google.protobuf.Timestamp
 	4, // 1: flow.FlowEvent.flow_fields:type_name -> flow.FlowFields
-	0, // 2: flow.FlowFields.type:type_name -> flow.Type
-	1, // 3: flow.FlowFields.direction:type_name -> flow.Direction
-	5, // 4: flow.FlowFields.port_info:type_name -> flow.PortInfo
-	6, // 5: flow.FlowFields.icmp_info:type_name -> flow.ICMPInfo
-	2, // 6: flow.FlowService.Events:input_type -> flow.FlowEvent
-	3, // 7: flow.FlowService.Events:output_type -> flow.FlowEventAck
-	7, // [7:8] is the sub-list for method output_type
-	6, // [6:7] is the sub-list for method input_type
-	6, // [6:6] is the sub-list for extension type_name
-	6, // [6:6] is the sub-list for extension extendee
-	0, // [0:6] is the sub-list for field type_name
+	7, // 2: flow.FlowEvent.window_start:type_name -> google.protobuf.Timestamp
+	7, // 3: flow.FlowEvent.window_end:type_name -> google.protobuf.Timestamp
+	0, // 4: flow.FlowFields.type:type_name -> flow.Type
+	1, // 5: flow.FlowFields.direction:type_name -> flow.Direction
+	5, // 6: flow.FlowFields.port_info:type_name -> flow.PortInfo
+	6, // 7: flow.FlowFields.icmp_info:type_name -> flow.ICMPInfo
+	2, // 8: flow.FlowService.Events:input_type -> flow.FlowEvent
+	3, // 9: flow.FlowService.Events:output_type -> flow.FlowEventAck
+	9, // [9:10] is the sub-list for method output_type
+	8, // [8:9] is the sub-list for method input_type
+	8, // [8:8] is the sub-list for extension type_name
+	8, // [8:8] is the sub-list for extension extendee
+	0, // [0:8] is the sub-list for field type_name
 }
 
 func init() { file_flow_proto_init() }

flow/proto/flow.proto

@@ -24,6 +24,9 @@ message FlowEvent {
   FlowFields flow_fields = 4;
 
   bool isInitiator = 5;
+
+  google.protobuf.Timestamp window_start = 6;
+  google.protobuf.Timestamp window_end = 7;
 }
 
 message FlowEventAck {
@@ -75,6 +78,9 @@ message FlowFields {
   bytes source_resource_id = 14;
   bytes dest_resource_id = 15;
 
+  uint64 num_of_starts = 16;
+  uint64 num_of_ends = 17;
+  uint64 num_of_drops = 18;
 }
 
 // Flow event types

flow/proto/flow_grpc.pb.go

@@ -1,4 +1,8 @@
 // Code generated by protoc-gen-go-grpc. DO NOT EDIT.
+// versions:
+// - protoc-gen-go-grpc v1.6.1
+// - protoc             v3.21.9
+// source: flow.proto
 
 package proto
 
@@ -11,15 +15,19 @@ import (
 
 // This is a compile-time assertion to ensure that this generated file
 // is compatible with the grpc package it is being compiled against.
-// Requires gRPC-Go v1.32.0 or later.
-const _ = grpc.SupportPackageIsVersion7
+// Requires gRPC-Go v1.64.0 or later.
+const _ = grpc.SupportPackageIsVersion9
+
+const (
+	FlowService_Events_FullMethodName = "/flow.FlowService/Events"
+)
 
 // FlowServiceClient is the client API for FlowService service.
 //
 // For semantics around ctx use and closing/ending streaming RPCs, please refer to https://pkg.go.dev/google.golang.org/grpc/?tab=doc#ClientConn.NewStream.
 type FlowServiceClient interface {
 	// Client to receiver streams of events and acknowledgements
-	Events(ctx context.Context, opts ...grpc.CallOption) (FlowService_EventsClient, error)
+	Events(ctx context.Context, opts ...grpc.CallOption) (grpc.BidiStreamingClient[FlowEvent, FlowEventAck], error)
 }
 
 type flowServiceClient struct {
@@ -30,54 +38,40 @@ func NewFlowServiceClient(cc grpc.ClientConnInterface) FlowServiceClient {
 	return &flowServiceClient{cc}
 }
 
-func (c *flowServiceClient) Events(ctx context.Context, opts ...grpc.CallOption) (FlowService_EventsClient, error) {
-	stream, err := c.cc.NewStream(ctx, &FlowService_ServiceDesc.Streams[0], "/flow.FlowService/Events", opts...)
+func (c *flowServiceClient) Events(ctx context.Context, opts ...grpc.CallOption) (grpc.BidiStreamingClient[FlowEvent, FlowEventAck], error) {
+	cOpts := append([]grpc.CallOption{grpc.StaticMethod()}, opts...)
+	stream, err := c.cc.NewStream(ctx, &FlowService_ServiceDesc.Streams[0], FlowService_Events_FullMethodName, cOpts...)
 	if err != nil {
 		return nil, err
 	}
-	x := &flowServiceEventsClient{stream}
+	x := &grpc.GenericClientStream[FlowEvent, FlowEventAck]{ClientStream: stream}
 	return x, nil
 }
 
-type FlowService_EventsClient interface {
-	Send(*FlowEvent) error
-	Recv() (*FlowEventAck, error)
-	grpc.ClientStream
-}
-
-type flowServiceEventsClient struct {
-	grpc.ClientStream
-}
-
-func (x *flowServiceEventsClient) Send(m *FlowEvent) error {
-	return x.ClientStream.SendMsg(m)
-}
-
-func (x *flowServiceEventsClient) Recv() (*FlowEventAck, error) {
-	m := new(FlowEventAck)
-	if err := x.ClientStream.RecvMsg(m); err != nil {
-		return nil, err
-	}
-	return m, nil
-}
+// This type alias is provided for backwards compatibility with existing code that references the prior non-generic stream type by name.
+type FlowService_EventsClient = grpc.BidiStreamingClient[FlowEvent, FlowEventAck]
 
 // FlowServiceServer is the server API for FlowService service.
 // All implementations must embed UnimplementedFlowServiceServer
-// for forward compatibility
+// for forward compatibility.
 type FlowServiceServer interface {
 	// Client to receiver streams of events and acknowledgements
-	Events(FlowService_EventsServer) error
+	Events(grpc.BidiStreamingServer[FlowEvent, FlowEventAck]) error
 	mustEmbedUnimplementedFlowServiceServer()
 }
 
-// UnimplementedFlowServiceServer must be embedded to have forward compatible implementations.
-type UnimplementedFlowServiceServer struct {
-}
+// UnimplementedFlowServiceServer must be embedded to have
+// forward compatible implementations.
+//
+// NOTE: this should be embedded by value instead of pointer to avoid a nil
+// pointer dereference when methods are called.
+type UnimplementedFlowServiceServer struct{}
 
-func (UnimplementedFlowServiceServer) Events(FlowService_EventsServer) error {
-	return status.Errorf(codes.Unimplemented, "method Events not implemented")
+func (UnimplementedFlowServiceServer) Events(grpc.BidiStreamingServer[FlowEvent, FlowEventAck]) error {
+	return status.Error(codes.Unimplemented, "method Events not implemented")
 }
 func (UnimplementedFlowServiceServer) mustEmbedUnimplementedFlowServiceServer() {}
+func (UnimplementedFlowServiceServer) testEmbeddedByValue()                     {}
 
 // UnsafeFlowServiceServer may be embedded to opt out of forward compatibility for this service.
 // Use of this interface is not recommended, as added methods to FlowServiceServer will
@@ -87,34 +81,22 @@ type UnsafeFlowServiceServer interface {
 }
 
 func RegisterFlowServiceServer(s grpc.ServiceRegistrar, srv FlowServiceServer) {
+	// If the following call panics, it indicates UnimplementedFlowServiceServer was
+	// embedded by pointer and is nil.  This will cause panics if an
+	// unimplemented method is ever invoked, so we test this at initialization
+	// time to prevent it from happening at runtime later due to I/O.
+	if t, ok := srv.(interface{ testEmbeddedByValue() }); ok {
+		t.testEmbeddedByValue()
+	}
 	s.RegisterService(&FlowService_ServiceDesc, srv)
 }
 
 func _FlowService_Events_Handler(srv interface{}, stream grpc.ServerStream) error {
-	return srv.(FlowServiceServer).Events(&flowServiceEventsServer{stream})
+	return srv.(FlowServiceServer).Events(&grpc.GenericServerStream[FlowEvent, FlowEventAck]{ServerStream: stream})
 }
 
-type FlowService_EventsServer interface {
-	Send(*FlowEventAck) error
-	Recv() (*FlowEvent, error)
-	grpc.ServerStream
-}
-
-type flowServiceEventsServer struct {
-	grpc.ServerStream
-}
-
-func (x *flowServiceEventsServer) Send(m *FlowEventAck) error {
-	return x.ServerStream.SendMsg(m)
-}
-
-func (x *flowServiceEventsServer) Recv() (*FlowEvent, error) {
-	m := new(FlowEvent)
-	if err := x.ServerStream.RecvMsg(m); err != nil {
-		return nil, err
-	}
-	return m, nil
-}
+// This type alias is provided for backwards compatibility with existing code that references the prior non-generic stream type by name.
+type FlowService_EventsServer = grpc.BidiStreamingServer[FlowEvent, FlowEventAck]
 
 // FlowService_ServiceDesc is the grpc.ServiceDesc for FlowService service.
 // It's only intended for direct use with grpc.RegisterService,

go.mod

@@ -2,7 +2,7 @@ module github.com/netbirdio/netbird
 
 go 1.25.5
 
-toolchain go1.25.11
+toolchain go1.25.12
 
 require (
 	cunicu.li/go-rosenpass v0.5.42
@@ -19,8 +19,8 @@ require (
 	github.com/spf13/cobra v1.10.2
 	github.com/spf13/pflag v1.0.10
 	github.com/vishvananda/netlink v1.3.1
-	golang.org/x/crypto v0.50.0
-	golang.org/x/sys v0.43.0
+	golang.org/x/crypto v0.54.0
+	golang.org/x/sys v0.47.0
 	golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173
 	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10
 	golang.zx2c4.com/wireguard/windows v0.5.3
@@ -66,6 +66,7 @@ require (
 	github.com/google/nftables v0.3.0
 	github.com/gopacket/gopacket v1.4.0
 	github.com/grpc-ecosystem/go-grpc-middleware/v2 v2.0.2-0.20240212192251-757544f21357
+	github.com/grpc-ecosystem/grpc-gateway/v2 v2.26.3
 	github.com/hashicorp/go-multierror v1.1.1
 	github.com/hashicorp/go-secure-stdlib/base62 v0.1.2
 	github.com/hashicorp/go-version v1.7.0
@@ -100,7 +101,7 @@ require (
 	github.com/quic-go/quic-go v0.55.0
 	github.com/redis/go-redis/v9 v9.7.3
 	github.com/rs/xid v1.3.0
-	github.com/shirou/gopsutil/v3 v3.24.4
+	github.com/shirou/gopsutil/v4 v4.25.8
 	github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966
 	github.com/songgao/water v0.0.0-20200317203138-2b4b6d7c09d8
 	github.com/stretchr/testify v1.11.1
@@ -125,11 +126,11 @@ require (
 	goauthentik.io/api/v3 v3.2023051.3
 	golang.org/x/exp v0.0.0-20260410095643-746e56fc9e2f
 	golang.org/x/mobile v0.0.0-20251113184115-a159579294ab
-	golang.org/x/mod v0.35.0
-	golang.org/x/net v0.53.0
+	golang.org/x/mod v0.37.0
+	golang.org/x/net v0.56.0
 	golang.org/x/oauth2 v0.36.0
-	golang.org/x/sync v0.20.0
-	golang.org/x/term v0.42.0
+	golang.org/x/sync v0.22.0
+	golang.org/x/term v0.45.0
 	golang.org/x/time v0.15.0
 	google.golang.org/api v0.276.0
 	gopkg.in/yaml.v3 v3.0.1
@@ -294,8 +295,6 @@ require (
 	github.com/prometheus/procfs v0.19.2 // indirect
 	github.com/russellhaering/goxmldsig v1.6.0 // indirect
 	github.com/ryanuber/go-glob v1.0.0 // indirect
-	github.com/shirou/gopsutil/v4 v4.25.8 // indirect
-	github.com/shoenig/go-m1cpu v0.2.1 // indirect
 	github.com/shopspring/decimal v1.4.0 // indirect
 	github.com/spf13/cast v1.10.0 // indirect
 	github.com/stretchr/objx v0.5.2 // indirect
@@ -315,9 +314,10 @@ require (
 	go.opentelemetry.io/otel/trace v1.43.0 // indirect
 	go.uber.org/multierr v1.11.0 // indirect
 	go.yaml.in/yaml/v2 v2.4.3 // indirect
-	golang.org/x/text v0.37.0 // indirect
-	golang.org/x/tools v0.44.0 // indirect
+	golang.org/x/text v0.40.0 // indirect
+	golang.org/x/tools v0.47.0 // indirect
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
+	google.golang.org/genproto/googleapis/api v0.0.0-20260319201613-d00831a3d3e7 // indirect
 	google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 // indirect
 	gopkg.in/square/go-jose.v2 v2.6.0 // indirect
 	gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7 // indirect

go.sum

@@ -271,9 +271,7 @@ github.com/google/go-cmp v0.3.1/go.mod h1:8QqcDgzrUqlUb/G2PQTWiueGozuR1884gddMyw
 github.com/google/go-cmp v0.4.0/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.2/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.5/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
-github.com/google/go-cmp v0.5.6/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.8/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
-github.com/google/go-cmp v0.5.9/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
 github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
 github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
@@ -417,7 +415,6 @@ github.com/libp2p/go-netroute v0.4.0 h1:sZZx9hyANYUx9PZyqcgE/E1GUG3iEtTZHUEvdtXT
 github.com/libp2p/go-netroute v0.4.0/go.mod h1:Nkd5ShYgSMS5MUKy/MU2T57xFoOKvvLR92Lic48LEyA=
 github.com/lrh3321/ipset-go v0.0.0-20250619021614-54a0a98ace81 h1:J56rFEfUTFT9j9CiRXhi1r8lUJ4W5idG3CiaBZGojNU=
 github.com/lrh3321/ipset-go v0.0.0-20250619021614-54a0a98ace81/go.mod h1:RD8ML/YdXctQ7qbcizZkw5mZ6l8Ogrl1dodBzVJduwI=
-github.com/lufia/plan9stats v0.0.0-20211012122336-39d0f177ccd0/go.mod h1:zJYVVT2jmtg6P3p1VtQj7WsuWi/y4VnjVBn7F8KPB3I=
 github.com/lufia/plan9stats v0.0.0-20240513124658-fba389f38bae h1:dIZY4ULFcto4tAFlj1FYZl8ztUZ13bdq+PLY+NOfbyI=
 github.com/lufia/plan9stats v0.0.0-20240513124658-fba389f38bae/go.mod h1:ilwx/Dta8jXAgpFYFvSWEMwxmbWXyiUHkd5FwyKhb5k=
 github.com/magiconair/properties v1.8.10 h1:s31yESBquKXCV9a/ScB3ESkOjUYYv+X0rg8SYxI99mE=
@@ -569,7 +566,6 @@ github.com/pkg/sftp v1.13.9/go.mod h1:OBN7bVXdstkFFN/gdnHPUb5TE8eb8G1Rp9wCItqjkk
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
-github.com/power-devops/perfstat v0.0.0-20210106213030-5aafc221ea8c/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
 github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55 h1:o4JXh1EVt9k/+g42oCprj/FisM4qX9L3sZB3upGN2ZU=
 github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
 github.com/pquerna/otp v1.5.0 h1:NMMR+WrmaqXU4EzdGJEE1aUUI0AMRzsp96fFFWNPwxs=
@@ -599,16 +595,8 @@ github.com/russellhaering/goxmldsig v1.6.0/go.mod h1:TrnaquDcYxWXfJrOjeMBTX4mLBe
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
 github.com/ryanuber/go-glob v1.0.0 h1:iQh3xXAumdQ+4Ufa5b25cRpC5TYKlno6hsv6Cb3pkBk=
 github.com/ryanuber/go-glob v1.0.0/go.mod h1:807d1WSdnB0XRJzKNil9Om6lcp/3a0v4qIHxIXzX/Yc=
-github.com/shirou/gopsutil/v3 v3.24.4 h1:dEHgzZXt4LMNm+oYELpzl9YCqV65Yr/6SfrvgRBtXeU=
-github.com/shirou/gopsutil/v3 v3.24.4/go.mod h1:lTd2mdiOspcqLgAnr9/nGi71NkeMpWKdmhuxm9GusH8=
 github.com/shirou/gopsutil/v4 v4.25.8 h1:NnAsw9lN7587WHxjJA9ryDnqhJpFH6A+wagYWTOH970=
 github.com/shirou/gopsutil/v4 v4.25.8/go.mod h1:q9QdMmfAOVIw7a+eF86P7ISEU6ka+NLgkUxlopV4RwI=
-github.com/shoenig/go-m1cpu v0.1.6/go.mod h1:1JJMcUBvfNwpq05QDQVAnx3gUHr9IYF7GNg9SUEw2VQ=
-github.com/shoenig/go-m1cpu v0.2.1 h1:yqRB4fvOge2+FyRXFkXqsyMoqPazv14Yyy+iyccT2E4=
-github.com/shoenig/go-m1cpu v0.2.1/go.mod h1:KkDOw6m3ZJQAPHbrzkZki4hnx+pDRR1Lo+ldA56wD5w=
-github.com/shoenig/test v0.6.4/go.mod h1:byHiCGXqrVaflBLAMq/srcZIHynQPQgeyvkvXnjqq0k=
-github.com/shoenig/test v1.7.0 h1:eWcHtTXa6QLnBvm0jgEabMRN/uJ4DMV3M8xUGgRkZmk=
-github.com/shoenig/test v1.7.0/go.mod h1:UxJ6u/x2v/TNs/LoLxBNJRV9DiwBBKYxXSyczsBHFoI=
 github.com/shopspring/decimal v1.4.0 h1:bxl37RwXBklmTi0C79JfXCEBD1cqqHt0bbgBAGFp81k=
 github.com/shopspring/decimal v1.4.0/go.mod h1:gawqmDU56v4yIKSwfBSFip1HdCCXN8/+DMd9qYNcwME=
 github.com/sirupsen/logrus v1.9.4 h1:TsZE7l11zFCLZnZ+teH4Umoq5BhEIfIzfRDZ1Uzql2w=
@@ -640,7 +628,6 @@ github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/
 github.com/stretchr/testify v1.8.0/go.mod h1:yNjHg4UonilssWZ8iaSj1OCr/vHnekPRkoO+kdMU+MU=
 github.com/stretchr/testify v1.8.3/go.mod h1:sz/lmYIOXD/1dqDmKjjqLyZ2RngseejIcXlSw2iwfAo=
 github.com/stretchr/testify v1.8.4/go.mod h1:sz/lmYIOXD/1dqDmKjjqLyZ2RngseejIcXlSw2iwfAo=
-github.com/stretchr/testify v1.9.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
 github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
 github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
 github.com/testcontainers/testcontainers-go v0.37.0 h1:L2Qc0vkTw2EHWQ08djon0D2uw7Z/PtHS/QzZZ5Ra/hg=
@@ -659,10 +646,8 @@ github.com/ti-mo/netfilter v0.5.2 h1:CTjOwFuNNeZ9QPdRXt1MZFLFUf84cKtiQutNauHWd40
 github.com/ti-mo/netfilter v0.5.2/go.mod h1:Btx3AtFiOVdHReTDmP9AE+hlkOcvIy403u7BXXbWZKo=
 github.com/tinylib/msgp v1.6.3 h1:bCSxiTz386UTgyT1i0MSCvdbWjVW+8sG3PjkGsZQt4s=
 github.com/tinylib/msgp v1.6.3/go.mod h1:RSp0LW9oSxFut3KzESt5Voq4GVWyS+PSulT77roAqEA=
-github.com/tklauser/go-sysconf v0.3.12/go.mod h1:Ho14jnntGE1fpdOqQEEaiKRpvIavV0hSfmBq8nJbHYI=
 github.com/tklauser/go-sysconf v0.3.15 h1:VE89k0criAymJ/Os65CSn1IXaol+1wrsFHEB8Ol49K4=
 github.com/tklauser/go-sysconf v0.3.15/go.mod h1:Dmjwr6tYFIseJw7a3dRLJfsHAMXZ3nEnL/aZY+0IuI4=
-github.com/tklauser/numcpus v0.6.1/go.mod h1:1XfjsgE2zo8GVw7POkMbHENHzVg3GzmoZ9fESEdAacY=
 github.com/tklauser/numcpus v0.10.0 h1:18njr6LDBk1zuna922MgdjQuJFjrdppsZG60sHGfjso=
 github.com/tklauser/numcpus v0.10.0/go.mod h1:BiTKazU708GQTYF4mB+cmlpT2Is1gLk7XVuEeem8LsQ=
 github.com/ugorji/go v1.1.7/go.mod h1:kZn38zHttfInRq0xu/PH0az30d+z6vm202qpg1oXVMw=
@@ -747,8 +732,8 @@ golang.org/x/crypto v0.18.0/go.mod h1:R0j02AL6hcrfOiy9T4ZYp/rcWeMxM3L6QYxlOuEG1m
 golang.org/x/crypto v0.19.0/go.mod h1:Iy9bg/ha4yyC70EfRS8jz+B6ybOBKMaSxLj6P6oBDfU=
 golang.org/x/crypto v0.23.0/go.mod h1:CKFgDieR+mRhux2Lsu27y0fO304Db0wZe70UKqHu0v8=
 golang.org/x/crypto v0.31.0/go.mod h1:kDsLvtWBEx7MV9tJOj9bnXsPbxwJQ6csT/x4KIN4Ssk=
-golang.org/x/crypto v0.50.0 h1:zO47/JPrL6vsNkINmLoo/PH1gcxpls50DNogFvB5ZGI=
-golang.org/x/crypto v0.50.0/go.mod h1:3muZ7vA7PBCE6xgPX7nkzzjiUq87kRItoJQM1Yo8S+Q=
+golang.org/x/crypto v0.54.0 h1:YLIA59K4fiNzHzjnZt2tUJQjQtUWfWbeHBqKtk3eScw=
+golang.org/x/crypto v0.54.0/go.mod h1:KWL8ny2AZdGR2cWmzeHrp2azQPGogOv+HeQaVEXC2dk=
 golang.org/x/exp v0.0.0-20260410095643-746e56fc9e2f h1:W3F4c+6OLc6H2lb//N1q4WpJkhzJCK5J6kUi1NTVXfM=
 golang.org/x/exp v0.0.0-20260410095643-746e56fc9e2f/go.mod h1:J1xhfL/vlindoeF/aINzNzt2Bket5bjo9sdOYzOsU80=
 golang.org/x/lint v0.0.0-20200302205851-738671d3881b/go.mod h1:3xt1FjdF8hUf6vQPIChWIBhFzV8gjjsPE/fR3IyQdNY=
@@ -763,8 +748,8 @@ golang.org/x/mod v0.8.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
 golang.org/x/mod v0.12.0/go.mod h1:iBbtSCu2XBx23ZKBPSOrRkjjQPZFPuis4dIYUhu/chs=
 golang.org/x/mod v0.15.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c=
 golang.org/x/mod v0.17.0/go.mod h1:hTbmBsO62+eylJbnUtE2MGJUyE7QWk4xUqPFrRgJ+7c=
-golang.org/x/mod v0.35.0 h1:Ww1D637e6Pg+Zb2KrWfHQUnH2dQRLBQyAtpr/haaJeM=
-golang.org/x/mod v0.35.0/go.mod h1:+GwiRhIInF8wPm+4AoT6L0FA1QWAad3OMdTRx4tFYlU=
+golang.org/x/mod v0.37.0 h1:vF1DjpVEshcIqoEaauuHebaLk1O1forxjxBaVn884JQ=
+golang.org/x/mod v0.37.0/go.mod h1:m8S8VeM9r4dzDwjrKO0a1sZP3YjeMamRRlD+fmR2Q/0=
 golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190603091049-60506f45cf65/go.mod h1:HSz+uSET+XFnRR8LxR5pz3Of3rY3CfYBVs4xY44aLks=
@@ -783,8 +768,8 @@ golang.org/x/net v0.15.0/go.mod h1:idbUs1IY1+zTqbi8yxTbhexhEEk5ur9LInksu6HrEpk=
 golang.org/x/net v0.20.0/go.mod h1:z8BVo6PvndSri0LbOE3hAn0apkU+1YvI6E70E9jsnvY=
 golang.org/x/net v0.21.0/go.mod h1:bIjVDfnllIU7BJ2DNgfnXvpSvtn8VRwhlsaeUTyUS44=
 golang.org/x/net v0.25.0/go.mod h1:JkAGAh7GEvH74S6FOH42FLoXpXbE/aqXSrIQjXgsiwM=
-golang.org/x/net v0.53.0 h1:d+qAbo5L0orcWAr0a9JweQpjXF19LMXJE8Ey7hwOdUA=
-golang.org/x/net v0.53.0/go.mod h1:JvMuJH7rrdiCfbeHoo3fCQU24Lf5JJwT9W3sJFulfgs=
+golang.org/x/net v0.56.0 h1:Rw8j/hFzGvJUZwNBXnAtf5sVDVt+65SK2C7IxCxZt5o=
+golang.org/x/net v0.56.0/go.mod h1:D3Ku6r+V6JROoZK144D2XfMHFcMq/0zSfLelVTCFKec=
 golang.org/x/oauth2 v0.8.0/go.mod h1:yr7u4HXZRm1R1kBWqr/xKNqewf0plRYoB7sla+BCIXE=
 golang.org/x/oauth2 v0.36.0 h1:peZ/1z27fi9hUOFCAZaHyrpWG5lwe0RJEEEeH0ThlIs=
 golang.org/x/oauth2 v0.36.0/go.mod h1:YDBUJMTkDnJS+A4BP4eZBjCqtokkg1hODuPjwiGPO7Q=
@@ -799,8 +784,8 @@ golang.org/x/sync v0.3.0/go.mod h1:FU7BRWz2tNW+3quACPkgCx/L+uEAv1htQ0V83Z9Rj+Y=
 golang.org/x/sync v0.6.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
 golang.org/x/sync v0.7.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
 golang.org/x/sync v0.10.0/go.mod h1:Czt+wKu1gCyEFDUtn0jG5QVvpJ6rzVqr5aXyt9drQfk=
-golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
-golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
+golang.org/x/sync v0.22.0 h1:SZjpbeLmrCk4xhRSZFNZW5gFUeCeFgjekvI/+gfScek=
+golang.org/x/sync v0.22.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20180909124046-d0be0721c37e/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
@@ -834,11 +819,10 @@ golang.org/x/sys v0.11.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.12.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.16.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/sys v0.17.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/sys v0.19.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/sys v0.20.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
 golang.org/x/sys v0.28.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/sys v0.43.0 h1:Rlag2XtaFTxp19wS8MXlJwTvoh8ArU6ezoyFsMyCTNI=
-golang.org/x/sys v0.43.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/sys v0.47.0 h1:o7XGOvZQCADBQQ4Y7VNq2dRWQR7JmOUW8Kxx4ZsNgWs=
+golang.org/x/sys v0.47.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/telemetry v0.0.0-20240228155512-f48c80bd79b2/go.mod h1:TeRTkGYfJXctD9OcfyVLyj2J3IxLnKwHJR8f4D8a3YE=
 golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
 golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
@@ -851,8 +835,8 @@ golang.org/x/term v0.16.0/go.mod h1:yn7UURbUtPyrVJPGPq404EukNFxcm/foM+bV/bfcDsY=
 golang.org/x/term v0.17.0/go.mod h1:lLRBjIVuehSbZlaOtGMbcMncT+aqLLLmKrsjNrUguwk=
 golang.org/x/term v0.20.0/go.mod h1:8UkIAJTvZgivsXaD6/pH6U9ecQzZ45awqEOzuCvwpFY=
 golang.org/x/term v0.27.0/go.mod h1:iMsnZpn0cago0GOrHO2+Y7u7JPn5AylBrcoWkElMTSM=
-golang.org/x/term v0.42.0 h1:UiKe+zDFmJobeJ5ggPwOshJIVt6/Ft0rcfrXZDLWAWY=
-golang.org/x/term v0.42.0/go.mod h1:Dq/D+snpsbazcBG5+F9Q1n2rXV8Ma+71xEjTRufARgY=
+golang.org/x/term v0.45.0 h1:NwWyBmoJCbfTHpxrWoZ9C6/VxOf7ic219I8xZZFdrf0=
+golang.org/x/term v0.45.0/go.mod h1:9aqxs0blBcrm/n0L9QW0aRVD+ktan8ssZromtqJC43w=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
@@ -864,8 +848,8 @@ golang.org/x/text v0.13.0/go.mod h1:TvPlkZtksWOMsz7fbANvkp4WM8x/WCo/om8BMLbz+aE=
 golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 golang.org/x/text v0.15.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
 golang.org/x/text v0.21.0/go.mod h1:4IBbMaMmOPCJ8SecivzSH54+73PCFmPWxNTLm+vZkEQ=
-golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
-golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
+golang.org/x/text v0.40.0 h1:Ub2Z6/xjgF1WrYQz2nuITOEegKFtiIy+rieRJ5lHZKs=
+golang.org/x/text v0.40.0/go.mod h1:hpnzDAfGV753zIKo+wk3u1bVKCGPbrnF7+7LBF/UHVY=
 golang.org/x/time v0.15.0 h1:bbrp8t3bGUeFOx08pvsMYRTCVSMk89u4tKbNOZbp88U=
 golang.org/x/time v0.15.0/go.mod h1:Y4YMaQmXwGQZoFaVFk4YpCt4FLQMYKZe9oeV/f4MSno=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
@@ -879,8 +863,8 @@ golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc
 golang.org/x/tools v0.6.0/go.mod h1:Xwgl3UAJ/d3gWutnCtw505GrjyAbvKui8lOU390QaIU=
 golang.org/x/tools v0.13.0/go.mod h1:HvlwmtVNQAhOuCjW7xxvovg8wbNq7LwfXh/k7wXUl58=
 golang.org/x/tools v0.21.1-0.20240508182429-e35e4ccd0d2d/go.mod h1:aiJjzUbINMkxbQROHiO6hDPo2LHcIPhhQsa9DLh0yGk=
-golang.org/x/tools v0.44.0 h1:UP4ajHPIcuMjT1GqzDWRlalUEoY+uzoZKnhOjbIPD2c=
-golang.org/x/tools v0.44.0/go.mod h1:KA0AfVErSdxRZIsOVipbv3rQhVXTnlU6UhKxHd1seDI=
+golang.org/x/tools v0.47.0 h1:7Kn5x/d1svx/PzryTsqeoZN4TZwqeH5pGWjefhLi/1Q=
+golang.org/x/tools v0.47.0/go.mod h1:dFHnyTvFWY212G+h7ZY4Vsp/K3U4/7W9TyVaAul8uCA=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=

management/internals/modules/peers/ephemeral/manager/ephemeral.go

@@ -215,7 +215,7 @@ func (e *EphemeralManager) cleanup(ctx context.Context) {
 	}
 
 	for accountID, peerIDs := range peerIDsPerAccount {
-		log.WithContext(ctx).Tracef("cleanup: deleting %d ephemeral peers for account %s", len(peerIDs), accountID)
+		log.WithContext(ctx).Debugf("cleanup: deleting %d ephemeral peers for account %s: %s", len(peerIDs), accountID, peerIDs)
 		err := e.peersManager.DeletePeers(ctx, accountID, peerIDs, activity.SystemInitiator, true)
 		if err != nil {
 			log.WithContext(ctx).Errorf("failed to delete ephemeral peers: %s", err)

management/internals/modules/peers/manager.go

@@ -184,6 +184,8 @@ func (m *managerImpl) DeletePeers(ctx context.Context, accountID string, peerIDs
 				return err
 			}
 
+			log.WithContext(ctx).Debugf("DeletePeers: deleted peer %s", peerID)
+
 			if !(peer.ProxyMeta.Embedded || peer.Meta.KernelVersion == "wasm") {
 				eventsToStore = append(eventsToStore, func() {
 					m.accountManager.StoreEvent(ctx, userID, peer.ID, accountID, activity.PeerRemovedByUser, peer.EventMeta(dnsDomain))
@@ -224,30 +226,6 @@ func (m *managerImpl) CreateProxyPeer(ctx context.Context, accountID string, pee
 		return nil
 	}
 
-	// Dedupe stale embedded peer records for the same (account, cluster).
-	// The proxy generates a fresh WireGuard keypair on every startup
-	// (proxy/internal/roundtrip/netbird.go), so without this sweep the
-	// prior embedded peer would linger forever — holding its CGNAT IP
-	// allocation, polluting other peers' rosters, and (most visibly)
-	// leaving the synth DNS pointing at the dead address. The
-	// (account, cluster) tuple identifies "the embedded peer for this
-	// proxy instance at this cluster"; any record matching that tuple
-	// with a different pubkey is by definition stale and must go.
-	staleIDs, err := m.findStaleEmbeddedProxyPeers(ctx, accountID, cluster, peerKey)
-	if err != nil {
-		return fmt.Errorf("scan for stale embedded proxy peers: %w", err)
-	}
-	if len(staleIDs) > 0 {
-		// userID="" + checkConnected=false: the deletion is initiated
-		// by management itself on behalf of the freshly-registering
-		// proxy, not by an end user; the stale peer may still be
-		// marked Connected from its prior session, but its session is
-		// dead by definition (its key no longer exists).
-		if err := m.DeletePeers(ctx, accountID, staleIDs, "", false); err != nil {
-			return fmt.Errorf("delete stale embedded proxy peers %v: %w", staleIDs, err)
-		}
-	}
-
 	name := fmt.Sprintf("proxy-%s", xid.New().String())
 	newPeer := &peer.Peer{
 		Ephemeral: true,
@@ -273,29 +251,3 @@ func (m *managerImpl) CreateProxyPeer(ctx context.Context, accountID string, pee
 
 	return nil
 }
-
-// findStaleEmbeddedProxyPeers returns the peer IDs of embedded proxy peer
-// records in accountID that target the same cluster but carry a different
-// WireGuard pubkey than the freshly-registering one. Used by CreateProxyPeer
-// to garbage-collect stale records left behind when the proxy restarts with a
-// regenerated keypair.
-func (m *managerImpl) findStaleEmbeddedProxyPeers(ctx context.Context, accountID, cluster, newKey string) ([]string, error) {
-	account, err := m.store.GetAccount(ctx, accountID)
-	if err != nil {
-		return nil, err
-	}
-	var stale []string
-	for _, p := range account.Peers {
-		if p == nil || !p.ProxyMeta.Embedded {
-			continue
-		}
-		if p.ProxyMeta.Cluster != cluster {
-			continue
-		}
-		if p.Key == newKey {
-			continue
-		}
-		stale = append(stale, p.ID)
-	}
-	return stale, nil
-}

management/internals/server/server.go

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

management/internals/shared/grpc/token_mgr.go

@@ -161,6 +161,8 @@ func (m *TimeBasedAuthSecretsManager) SetupRefresh(ctx context.Context, accountI
 		m.turnCancelMap[peerID] = turnCancel
 		go m.refreshTURNTokens(ctx, accountID, peerID, turnCancel)
 		log.WithContext(ctx).Debugf("starting TURN refresh for %s", peerID)
+	} else {
+		log.WithContext(ctx).Debugf("no TURN configuration, skipping TURN refresh for %s", peerID)
 	}
 
 	if m.relayCfg != nil {
@@ -168,6 +170,8 @@ func (m *TimeBasedAuthSecretsManager) SetupRefresh(ctx context.Context, accountI
 		m.relayCancelMap[peerID] = relayCancel
 		go m.refreshRelayTokens(ctx, accountID, peerID, relayCancel)
 		log.WithContext(ctx).Tracef("starting relay refresh for %s", peerID)
+	} else {
+		log.WithContext(ctx).Tracef("no relay configuration, skipping relay refresh for %s", peerID)
 	}
 }
 

management/main.go

@@ -1,19 +1,24 @@
 package main
 
 import (
-	"log"
 	"net/http"
 	// nolint:gosec
 	_ "net/http/pprof"
 	"os"
 
+	log "github.com/sirupsen/logrus"
+
 	"github.com/netbirdio/netbird/management/cmd"
 )
 
 func main() {
-	go func() {
-		log.Println(http.ListenAndServe("localhost:6060", nil))
-	}()
+	if pprofAddr := os.Getenv("NB_PPROF_ADDR"); pprofAddr != "" {
+		log.Infof("pprof enabled, listening on: %s", pprofAddr)
+		go func() {
+			log.Println(http.ListenAndServe(pprofAddr, nil))
+		}()
+	}
+
 	if err := cmd.Execute(); err != nil {
 		os.Exit(1)
 	}

management/server/agentnetwork_proxypeer_restart_test.go

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

management/server/http/handlers/accounts/accounts_handler.go

@@ -286,6 +286,21 @@ func (h *handler) updateAccountRequestSettings(req api.PutApiAccountsAccountIdJS
 	if req.Settings.MetricsPushEnabled != nil {
 		returnSettings.MetricsPushEnabled = *req.Settings.MetricsPushEnabled
 	}
+	if req.Settings.AgentNetworkOnly != nil {
+		returnSettings.AgentNetworkOnly = *req.Settings.AgentNetworkOnly
+	}
+	if req.Settings.DashboardFeatures != nil {
+		returnSettings.DashboardFeatures = &types.DashboardFeatures{
+			AgentNetwork: req.Settings.DashboardFeatures.AgentNetwork,
+		}
+	}
+
+	if returnSettings.AgentNetworkOnly &&
+		(returnSettings.DashboardFeatures == nil ||
+			returnSettings.DashboardFeatures.AgentNetwork == nil ||
+			!*returnSettings.DashboardFeatures.AgentNetwork) {
+		return nil, status.Errorf(status.InvalidArgument, "agent network only mode requires dashboard_features.agent_network to be enabled")
+	}
 
 	return returnSettings, nil
 }
@@ -417,6 +432,7 @@ func toAccountResponse(accountID string, settings *types.Settings, meta *types.A
 		AutoUpdateAlways:                &settings.AutoUpdateAlways,
 		Ipv6EnabledGroups:               &settings.IPv6EnabledGroups,
 		MetricsPushEnabled:              &settings.MetricsPushEnabled,
+		AgentNetworkOnly:                &settings.AgentNetworkOnly,
 		EmbeddedIdpEnabled:              &settings.EmbeddedIdpEnabled,
 		LocalAuthDisabled:               &settings.LocalAuthDisabled,
 		LocalMfaEnabled:                 &settings.LocalMfaEnabled,
@@ -430,6 +446,11 @@ func toAccountResponse(accountID string, settings *types.Settings, meta *types.A
 		networkRangeV6Str := settings.NetworkRangeV6.String()
 		apiSettings.NetworkRangeV6 = &networkRangeV6Str
 	}
+	if settings.DashboardFeatures != nil {
+		apiSettings.DashboardFeatures = &api.AccountDashboardFeatures{
+			AgentNetwork: settings.DashboardFeatures.AgentNetwork,
+		}
+	}
 
 	apiOnboarding := api.AccountOnboarding{
 		OnboardingFlowPending: onboarding.OnboardingFlowPending,

management/server/http/handlers/accounts/accounts_handler_test.go

@@ -130,6 +130,7 @@ func TestAccounts_AccountsHandler(t *testing.T) {
 				AutoUpdateAlways:                br(false),
 				AutoUpdateVersion:               sr(""),
 				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
 				EmbeddedIdpEnabled:              br(false),
 				LocalAuthDisabled:               br(false),
 				LocalMfaEnabled:                 br(false),
@@ -158,6 +159,7 @@ func TestAccounts_AccountsHandler(t *testing.T) {
 				AutoUpdateAlways:                br(false),
 				AutoUpdateVersion:               sr(""),
 				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
 				EmbeddedIdpEnabled:              br(false),
 				LocalAuthDisabled:               br(false),
 				LocalMfaEnabled:                 br(false),
@@ -186,6 +188,7 @@ func TestAccounts_AccountsHandler(t *testing.T) {
 				AutoUpdateAlways:                br(false),
 				AutoUpdateVersion:               sr("latest"),
 				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
 				EmbeddedIdpEnabled:              br(false),
 				LocalAuthDisabled:               br(false),
 				LocalMfaEnabled:                 br(false),
@@ -214,6 +217,7 @@ func TestAccounts_AccountsHandler(t *testing.T) {
 				AutoUpdateAlways:                br(false),
 				AutoUpdateVersion:               sr(""),
 				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
 				EmbeddedIdpEnabled:              br(false),
 				LocalAuthDisabled:               br(false),
 				LocalMfaEnabled:                 br(false),
@@ -242,6 +246,7 @@ func TestAccounts_AccountsHandler(t *testing.T) {
 				AutoUpdateAlways:                br(false),
 				AutoUpdateVersion:               sr(""),
 				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
 				EmbeddedIdpEnabled:              br(false),
 				LocalAuthDisabled:               br(false),
 				LocalMfaEnabled:                 br(false),
@@ -270,6 +275,109 @@ func TestAccounts_AccountsHandler(t *testing.T) {
 				AutoUpdateAlways:                br(false),
 				AutoUpdateVersion:               sr(""),
 				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
+				EmbeddedIdpEnabled:              br(false),
+				LocalAuthDisabled:               br(false),
+				LocalMfaEnabled:                 br(false),
+			},
+			expectedArray: false,
+			expectedID:    accountID,
+		},
+		{
+			name:           "PutAccount OK enabling agent_network_only",
+			expectedBody:   true,
+			requestType:    http.MethodPut,
+			requestPath:    "/api/accounts/" + accountID,
+			requestBody:    bytes.NewBufferString("{\"settings\": {\"peer_login_expiration\": 15552000,\"peer_login_expiration_enabled\": true,\"agent_network_only\": true,\"dashboard_features\": {\"agent_network\": true}},\"onboarding\": {\"onboarding_flow_pending\": true,\"signup_form_pending\": true}}"),
+			expectedStatus: http.StatusOK,
+			expectedSettings: api.AccountSettings{
+				PeerLoginExpiration:             15552000,
+				PeerLoginExpirationEnabled:      true,
+				GroupsPropagationEnabled:        br(false),
+				JwtGroupsClaimName:              sr(""),
+				JwtGroupsEnabled:                br(false),
+				JwtAllowGroups:                  &[]string{},
+				RegularUsersViewBlocked:         false,
+				RoutingPeerDnsResolutionEnabled: br(false),
+				LazyConnectionEnabled:           br(false),
+				DnsDomain:                       sr(""),
+				AutoUpdateAlways:                br(false),
+				AutoUpdateVersion:               sr(""),
+				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(true),
+				DashboardFeatures: &api.AccountDashboardFeatures{
+					AgentNetwork: br(true),
+				},
+				EmbeddedIdpEnabled: br(false),
+				LocalAuthDisabled:  br(false),
+				LocalMfaEnabled:    br(false),
+			},
+			expectedArray: false,
+			expectedID:    accountID,
+		},
+		{
+			name:           "PutAccount fails enabling agent_network_only without dashboard_features",
+			expectedBody:   true,
+			requestType:    http.MethodPut,
+			requestPath:    "/api/accounts/" + accountID,
+			requestBody:    bytes.NewBufferString("{\"settings\": {\"peer_login_expiration\": 15552000,\"peer_login_expiration_enabled\": true,\"agent_network_only\": true},\"onboarding\": {\"onboarding_flow_pending\": true,\"signup_form_pending\": true}}"),
+			expectedStatus: http.StatusUnprocessableEntity,
+			expectedArray:  false,
+		},
+		{
+			name:           "PutAccount OK setting dashboard_features agent_network",
+			expectedBody:   true,
+			requestType:    http.MethodPut,
+			requestPath:    "/api/accounts/" + accountID,
+			requestBody:    bytes.NewBufferString("{\"settings\": {\"peer_login_expiration\": 15552000,\"peer_login_expiration_enabled\": true,\"dashboard_features\": {\"agent_network\": true}},\"onboarding\": {\"onboarding_flow_pending\": true,\"signup_form_pending\": true}}"),
+			expectedStatus: http.StatusOK,
+			expectedSettings: api.AccountSettings{
+				PeerLoginExpiration:             15552000,
+				PeerLoginExpirationEnabled:      true,
+				GroupsPropagationEnabled:        br(false),
+				JwtGroupsClaimName:              sr(""),
+				JwtGroupsEnabled:                br(false),
+				JwtAllowGroups:                  &[]string{},
+				RegularUsersViewBlocked:         false,
+				RoutingPeerDnsResolutionEnabled: br(false),
+				LazyConnectionEnabled:           br(false),
+				DnsDomain:                       sr(""),
+				AutoUpdateAlways:                br(false),
+				AutoUpdateVersion:               sr(""),
+				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
+				DashboardFeatures: &api.AccountDashboardFeatures{
+					AgentNetwork: br(true),
+				},
+				EmbeddedIdpEnabled: br(false),
+				LocalAuthDisabled:  br(false),
+				LocalMfaEnabled:    br(false),
+			},
+			expectedArray: false,
+			expectedID:    accountID,
+		},
+		{
+			name:           "PutAccount OK disabling agent_network_only again",
+			expectedBody:   true,
+			requestType:    http.MethodPut,
+			requestPath:    "/api/accounts/" + accountID,
+			requestBody:    bytes.NewBufferString("{\"settings\": {\"peer_login_expiration\": 15552000,\"peer_login_expiration_enabled\": true,\"agent_network_only\": false},\"onboarding\": {\"onboarding_flow_pending\": true,\"signup_form_pending\": true}}"),
+			expectedStatus: http.StatusOK,
+			expectedSettings: api.AccountSettings{
+				PeerLoginExpiration:             15552000,
+				PeerLoginExpirationEnabled:      true,
+				GroupsPropagationEnabled:        br(false),
+				JwtGroupsClaimName:              sr(""),
+				JwtGroupsEnabled:                br(false),
+				JwtAllowGroups:                  &[]string{},
+				RegularUsersViewBlocked:         false,
+				RoutingPeerDnsResolutionEnabled: br(false),
+				LazyConnectionEnabled:           br(false),
+				DnsDomain:                       sr(""),
+				AutoUpdateAlways:                br(false),
+				AutoUpdateVersion:               sr(""),
+				MetricsPushEnabled:              br(false),
+				AgentNetworkOnly:                br(false),
 				EmbeddedIdpEnabled:              br(false),
 				LocalAuthDisabled:               br(false),
 				LocalMfaEnabled:                 br(false),

management/server/peer.go

@@ -106,11 +106,13 @@ func (am *DefaultAccountManager) MarkPeerConnected(ctx context.Context, peerPubK
 	}
 	if !updated {
 		am.metrics.AccountManagerMetrics().CountPeerStatusUpdate(telemetry.PeerStatusConnect, telemetry.PeerStatusStale)
-		log.WithContext(ctx).Tracef("peer %s already has a newer session in store, skipping connect", peer.ID)
+		log.WithContext(ctx).Debugf("peer %s already has a newer session in store, skipping connect", peer.ID)
 		return nil
 	}
 	am.metrics.AccountManagerMetrics().CountPeerStatusUpdate(telemetry.PeerStatusConnect, telemetry.PeerStatusApplied)
 
+	log.WithContext(ctx).Debugf("mark peer %s connected", peer.ID)
+
 	if err = am.schedulePeerExpirations(ctx, accountID, peer); err != nil {
 		return err
 	}
@@ -180,12 +182,14 @@ func (am *DefaultAccountManager) MarkPeerDisconnected(ctx context.Context, peerP
 	}
 	if !updated {
 		am.metrics.AccountManagerMetrics().CountPeerStatusUpdate(telemetry.PeerStatusDisconnect, telemetry.PeerStatusStale)
-		log.WithContext(ctx).Tracef("peer %s session token mismatch on disconnect (token=%d), skipping",
+		log.WithContext(ctx).Debugf("peer %s session token mismatch on disconnect (token=%d), skipping",
 			peer.ID, sessionStartedAt)
 		return nil
 	}
 	am.metrics.AccountManagerMetrics().CountPeerStatusUpdate(telemetry.PeerStatusDisconnect, telemetry.PeerStatusApplied)
 
+	log.WithContext(ctx).Debugf("mark peer %s disconnected", peer.ID)
+
 	// Symmetric with MarkPeerConnected: when an embedded proxy peer goes
 	// offline, refresh the peers that had synthesized records pointing at
 	// it so they pull the stale entries instead of waiting out TTL.

management/server/store/sql_store.go

@@ -1605,7 +1605,8 @@ func (s *SqlStore) getAccount(ctx context.Context, accountID string) (*types.Acc
 			settings_jwt_groups_enabled, settings_jwt_groups_claim_name, settings_jwt_allow_groups,
 			settings_routing_peer_dns_resolution_enabled, settings_dns_domain, settings_network_range,
 			settings_network_range_v6, settings_ipv6_enabled_groups, settings_lazy_connection_enabled,
-			settings_local_mfa_enabled, settings_metrics_push_enabled,
+			settings_local_mfa_enabled, settings_metrics_push_enabled, settings_agent_network_only,
+			settings_dashboard_features,
 			-- Embedded ExtraSettings
 			settings_extra_peer_approval_enabled, settings_extra_user_approval_required,
 			settings_extra_integrated_validator, settings_extra_integrated_validator_groups
@@ -1629,6 +1630,8 @@ func (s *SqlStore) getAccount(ctx context.Context, accountID string) (*types.Acc
 		sLazyConnectionEnabled           sql.NullBool
 		sLocalMFAEnabled                 sql.NullBool
 		sMetricsPushEnabled              sql.NullBool
+		sAgentNetworkOnly                sql.NullBool
+		sDashboardFeatures               sql.NullString
 		sExtraPeerApprovalEnabled        sql.NullBool
 		sExtraUserApprovalRequired       sql.NullBool
 		sExtraIntegratedValidator        sql.NullString
@@ -1651,7 +1654,8 @@ func (s *SqlStore) getAccount(ctx context.Context, accountID string) (*types.Acc
 		&sJWTGroupsEnabled, &sJWTGroupsClaimName, &sJWTAllowGroups,
 		&sRoutingPeerDNSResolutionEnabled, &sDNSDomain, &sNetworkRange,
 		&sNetworkRangeV6, &sIPv6EnabledGroups, &sLazyConnectionEnabled,
-		&sLocalMFAEnabled, &sMetricsPushEnabled,
+		&sLocalMFAEnabled, &sMetricsPushEnabled, &sAgentNetworkOnly,
+		&sDashboardFeatures,
 		&sExtraPeerApprovalEnabled, &sExtraUserApprovalRequired,
 		&sExtraIntegratedValidator, &sExtraIntegratedValidatorGroups,
 	)
@@ -1720,6 +1724,14 @@ func (s *SqlStore) getAccount(ctx context.Context, accountID string) (*types.Acc
 	if sMetricsPushEnabled.Valid {
 		account.Settings.MetricsPushEnabled = sMetricsPushEnabled.Bool
 	}
+	if sAgentNetworkOnly.Valid {
+		account.Settings.AgentNetworkOnly = sAgentNetworkOnly.Bool
+	}
+	if sDashboardFeatures.Valid && sDashboardFeatures.String != "" {
+		if err := json.Unmarshal([]byte(sDashboardFeatures.String), &account.Settings.DashboardFeatures); err != nil {
+			log.WithContext(ctx).Warnf("failed to unmarshal dashboard features for account %s: %v", accountID, err)
+		}
+	}
 	if sJWTAllowGroups.Valid {
 		_ = json.Unmarshal([]byte(sJWTAllowGroups.String), &account.Settings.JWTAllowGroups)
 	}

management/server/store/sql_store_test.go

@@ -1245,6 +1245,61 @@ func TestSqlite_CreateAndGetObjectInTransaction(t *testing.T) {
 	assert.NoError(t, err)
 }
 
+func TestSqlStore_SaveAccountPersistsAgentNetworkOnly(t *testing.T) {
+	store, cleanup, err := NewTestStoreFromSQL(context.Background(), "../testdata/store.sql", t.TempDir())
+	t.Cleanup(cleanup)
+	require.NoError(t, err)
+
+	accountID := "bf1c8084-ba50-4ce7-9439-34653001fc3b"
+	account, err := store.GetAccount(context.Background(), accountID)
+	require.NoError(t, err)
+	require.False(t, account.Settings.AgentNetworkOnly, "setting should default to false")
+
+	account.Settings.AgentNetworkOnly = true
+	require.NoError(t, store.SaveAccount(context.Background(), account))
+
+	reloaded, err := store.GetAccount(context.Background(), accountID)
+	require.NoError(t, err)
+	require.True(t, reloaded.Settings.AgentNetworkOnly, "setting should survive a save/load round-trip")
+
+	reloaded.Settings.AgentNetworkOnly = false
+	require.NoError(t, store.SaveAccount(context.Background(), reloaded))
+
+	disabled, err := store.GetAccount(context.Background(), accountID)
+	require.NoError(t, err)
+	require.False(t, disabled.Settings.AgentNetworkOnly, "disabling should persist")
+}
+
+func TestSqlStore_SaveAccountPersistsDashboardFeatures(t *testing.T) {
+	store, cleanup, err := NewTestStoreFromSQL(context.Background(), "../testdata/store.sql", t.TempDir())
+	t.Cleanup(cleanup)
+	require.NoError(t, err)
+
+	accountID := "bf1c8084-ba50-4ce7-9439-34653001fc3b"
+	account, err := store.GetAccount(context.Background(), accountID)
+	require.NoError(t, err)
+	require.Nil(t, account.Settings.DashboardFeatures, "dashboard features should default to unset")
+
+	agentNetwork := true
+	account.Settings.DashboardFeatures = &types.DashboardFeatures{AgentNetwork: &agentNetwork}
+	require.NoError(t, store.SaveAccount(context.Background(), account))
+
+	reloaded, err := store.GetAccount(context.Background(), accountID)
+	require.NoError(t, err)
+	require.NotNil(t, reloaded.Settings.DashboardFeatures, "dashboard features should survive a save/load round-trip")
+	require.NotNil(t, reloaded.Settings.DashboardFeatures.AgentNetwork, "agent network flag should be set")
+	require.True(t, *reloaded.Settings.DashboardFeatures.AgentNetwork, "agent network flag should persist as true")
+
+	disabled := false
+	reloaded.Settings.DashboardFeatures = &types.DashboardFeatures{AgentNetwork: &disabled}
+	require.NoError(t, store.SaveAccount(context.Background(), reloaded))
+
+	reloadedDisabled, err := store.GetAccount(context.Background(), accountID)
+	require.NoError(t, err)
+	require.NotNil(t, reloadedDisabled.Settings.DashboardFeatures.AgentNetwork, "agent network flag should remain set")
+	require.False(t, *reloadedDisabled.Settings.DashboardFeatures.AgentNetwork, "explicit false should persist")
+}
+
 func TestSqlStore_GetAccountUsers(t *testing.T) {
 	store, cleanup, err := NewTestStoreFromSQL(context.Background(), "../testdata/extended-store.sql", t.TempDir())
 	t.Cleanup(cleanup)

management/server/types/account.go

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

management/server/types/networkmap_components.go

@@ -7,6 +7,7 @@ import (
 	"slices"
 	"strconv"
 	"strings"
+	"sync"
 	"time"
 
 	"github.com/netbirdio/netbird/client/ssh/auth"
@@ -42,6 +43,14 @@ type NetworkMapComponents struct {
 	PostureFailedPeers map[string]map[string]struct{}
 
 	RouterPeers map[string]*nbpeer.Peer
+
+	routesByPeerOnce sync.Once
+	routesByPeerIdx  map[string][]routeIndexEntry
+}
+
+type routeIndexEntry struct {
+	route    *route.Route
+	viaGroup bool
 }
 
 type AccountSettingsInfo struct {
@@ -530,33 +539,43 @@ func (c *NetworkMapComponents) getRoutingPeerRoutes(peerID string) (enabledRoute
 		disabledRoutes = append(disabledRoutes, r)
 	}
 
-	for _, r := range c.Routes {
-		for _, groupID := range r.PeerGroups {
-			group := c.GetGroupInfo(groupID)
-			if group == nil {
-				continue
-			}
-			for _, id := range group.Peers {
-				if id != peerID {
-					continue
-				}
-
-				newPeerRoute := r.Copy()
-				newPeerRoute.Peer = id
-				newPeerRoute.PeerGroups = nil
-				newPeerRoute.ID = route.ID(string(r.ID) + ":" + id)
-				takeRoute(newPeerRoute)
-				break
-			}
-		}
-		if r.Peer == peerID {
-			takeRoute(r.Copy())
+	for _, entry := range c.routesByPeer()[peerID] {
+		if entry.viaGroup {
+			newPeerRoute := entry.route.Copy()
+			newPeerRoute.PeerGroups = nil
+			newPeerRoute.ID = route.ID(string(entry.route.ID) + ":" + peerID)
+			takeRoute(newPeerRoute)
+			continue
 		}
+		takeRoute(entry.route.Copy())
 	}
 
 	return enabledRoutes, disabledRoutes
 }
 
+func (c *NetworkMapComponents) routesByPeer() map[string][]routeIndexEntry {
+	c.routesByPeerOnce.Do(func() {
+		idx := make(map[string][]routeIndexEntry)
+		for _, r := range c.Routes {
+			for _, groupID := range r.PeerGroups {
+				group := c.GetGroupInfo(groupID)
+				if group == nil {
+					continue
+				}
+				for _, id := range group.Peers {
+					idx[id] = append(idx[id], routeIndexEntry{route: r, viaGroup: true})
+				}
+			}
+			if r.Peer != "" {
+				idx[r.Peer] = append(idx[r.Peer], routeIndexEntry{route: r})
+			}
+		}
+		c.routesByPeerIdx = idx
+	})
+
+	return c.routesByPeerIdx
+}
+
 func (c *NetworkMapComponents) filterRoutesByGroups(routes []*route.Route, groupListMap LookupMap) []*route.Route {
 	var filteredRoutes []*route.Route
 	for _, r := range routes {

management/server/types/settings.go

@@ -76,6 +76,15 @@ type Settings struct {
 	// MetricsPushEnabled globally enables or disables client metrics push for the account
 	MetricsPushEnabled bool `gorm:"default:false"`
 
+	// AgentNetworkOnly limits the dashboard to the Agent Network surface for this account.
+	// Set for accounts created via netbird.ai signups; users can disable it later.
+	AgentNetworkOnly bool `gorm:"default:false"`
+
+	// DashboardFeatures holds per-account dashboard section visibility overrides.
+	// It serializes to a single JSON column so new sections can be added without
+	// a schema change.
+	DashboardFeatures *DashboardFeatures `gorm:"serializer:json"`
+
 	// EmbeddedIdpEnabled indicates if the embedded identity provider is enabled.
 	// This is a runtime-only field, not stored in the database.
 	EmbeddedIdpEnabled bool `gorm:"-"`
@@ -114,6 +123,7 @@ func (s *Settings) Copy() *Settings {
 		AutoUpdateAlways:                s.AutoUpdateAlways,
 		IPv6EnabledGroups:               slices.Clone(s.IPv6EnabledGroups),
 		MetricsPushEnabled:              s.MetricsPushEnabled,
+		AgentNetworkOnly:                s.AgentNetworkOnly,
 		EmbeddedIdpEnabled:              s.EmbeddedIdpEnabled,
 		LocalAuthDisabled:               s.LocalAuthDisabled,
 		LocalMfaEnabled:                 s.LocalMfaEnabled,
@@ -121,9 +131,31 @@ func (s *Settings) Copy() *Settings {
 	if s.Extra != nil {
 		settings.Extra = s.Extra.Copy()
 	}
+	if s.DashboardFeatures != nil {
+		settings.DashboardFeatures = s.DashboardFeatures.Copy()
+	}
 	return settings
 }
 
+// DashboardFeatures holds per-account dashboard section visibility overrides.
+// Nil fields are unset and follow the default dashboard behavior; an explicit
+// value forces that section shown or hidden for the account.
+type DashboardFeatures struct {
+	// AgentNetwork, when set, forces the Agent Network menu shown (true) or
+	// hidden (false) regardless of the deployment feature flag.
+	AgentNetwork *bool `json:"agent_network,omitempty"`
+}
+
+// Copy returns a deep copy of the DashboardFeatures struct.
+func (d *DashboardFeatures) Copy() *DashboardFeatures {
+	c := &DashboardFeatures{}
+	if d.AgentNetwork != nil {
+		v := *d.AgentNetwork
+		c.AgentNetwork = &v
+	}
+	return c
+}
+
 type ExtraSettings struct {
 	// PeerApprovalEnabled enables or disables the need for peers bo be approved by an administrator
 	PeerApprovalEnabled bool

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

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

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

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

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

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

shared/management/http/api/openapi.yml

@@ -375,6 +375,12 @@ components:
           description: Enables or disables client metrics push for all peers in the account
           type: boolean
           example: false
+        agent_network_only:
+          description: Limits the dashboard to the Agent Network surface for this account. Set for accounts created via netbird.ai signups and can be disabled later. Enabling this requires dashboard_features.agent_network to be true in the same request.
+          type: boolean
+          example: false
+        dashboard_features:
+          $ref: '#/components/schemas/AccountDashboardFeatures'
         embedded_idp_enabled:
           description: Indicates whether the embedded identity provider (Dex) is enabled for this account. This is a read-only field.
           type: boolean
@@ -403,6 +409,14 @@ components:
         - regular_users_view_blocked
         - peer_expose_enabled
         - peer_expose_groups
+    AccountDashboardFeatures:
+      description: Per-account dashboard section visibility overrides. Omitted keys follow the default dashboard behavior.
+      type: object
+      properties:
+        agent_network:
+          description: Controls the Agent Network menu for the account regardless of the deployment feature flag. When true the menu is shown, when false it is hidden, and when omitted the default behavior applies. Must be true when agent_network_only is enabled.
+          type: boolean
+          example: true
     AccountExtraSettings:
       type: object
       properties:
@@ -2769,6 +2783,28 @@ components:
           type: integer
           description: "Number of packets transmitted."
           example: 5
+        num_of_starts:
+          type: integer
+          description: "Number of start events."
+          example: 3
+        num_of_ends:
+          type: integer
+          description: "Number of end events."
+          example: 4
+        num_of_drops:
+          type: integer
+          description: "Number of drop events."
+          example: 5
+        window_start:
+          type: string
+          format: date-time
+          description: Timestamp of the start of the aggregation window.
+          example: 2025-03-20T16:23:58.125397Z
+        window_end:
+          type: string
+          format: date-time
+          description: Timestamp of the end of the aggregation window.
+          example: 2025-03-20T16:23:58.125397Z
         events:
           type: array
           description: "List of events that are correlated to this flow (e.g., start, end)."
@@ -2790,6 +2826,11 @@ components:
         - rx_packets
         - tx_bytes
         - tx_packets
+        - num_of_starts
+        - num_of_ends
+        - num_of_drops
+        - window_start
+        - window_end
         - events
     NetworkTrafficEventsResponse:
       type: object
@@ -9300,6 +9341,18 @@ paths:
           required: false
           schema:
             type: string
+        - name: source_id
+          in: query
+          description: Filter by source endpoint ID
+          required: false
+          schema:
+            type: string
+        - name: destination_id
+          in: query
+          description: Filter by destination endpoint ID
+          required: false
+          schema:
+            type: string
         - name: protocol
           in: query
           description: Filter by protocol
@@ -10451,7 +10504,7 @@ paths:
         - EDR Intune Integrations
       summary: Delete EDR Intune Integration
       description: Deletes an EDR Intune Integration by its ID.
-      operationId: deleteIntegration
+      operationId: deleteEDRIntuneIntegration
       responses:
         '200':
           description: Integration deleted successfully. Returns an empty object.
@@ -12535,7 +12588,7 @@ paths:
         - Event Streaming Integrations
       summary: Delete Event Streaming Integration
       description: Deletes an event streaming integration by its ID.
-      operationId: deleteIntegration
+      operationId: deleteEventStreamingIntegration
       responses:
         '200':
           description: Integration deleted successfully. Returns an empty object.

shared/management/http/api/types.gen.go

@@ -1612,6 +1612,12 @@ type Account struct {
 	Settings   AccountSettings   `json:"settings"`
 }
 
+// AccountDashboardFeatures Per-account dashboard section visibility overrides. Omitted keys follow the default dashboard behavior.
+type AccountDashboardFeatures struct {
+	// AgentNetwork Controls the Agent Network menu for the account regardless of the deployment feature flag. When true the menu is shown, when false it is hidden, and when omitted the default behavior applies. Must be true when agent_network_only is enabled.
+	AgentNetwork *bool `json:"agent_network,omitempty"`
+}
+
 // AccountExtraSettings defines model for AccountExtraSettings.
 type AccountExtraSettings struct {
 	// NetworkTrafficLogsEnabled Enables or disables network traffic logging. If enabled, all network traffic events from peers will be stored.
@@ -1647,12 +1653,18 @@ type AccountRequest struct {
 
 // AccountSettings defines model for AccountSettings.
 type AccountSettings struct {
+	// AgentNetworkOnly Limits the dashboard to the Agent Network surface for this account. Set for accounts created via netbird.ai signups and can be disabled later. Enabling this requires dashboard_features.agent_network to be true in the same request.
+	AgentNetworkOnly *bool `json:"agent_network_only,omitempty"`
+
 	// AutoUpdateAlways When true, updates are installed automatically in the background. When false, updates require user interaction from the UI.
 	AutoUpdateAlways *bool `json:"auto_update_always,omitempty"`
 
 	// AutoUpdateVersion Set Clients auto-update version. "latest", "disabled", or a specific version (e.g "0.50.1")
 	AutoUpdateVersion *string `json:"auto_update_version,omitempty"`
 
+	// DashboardFeatures Per-account dashboard section visibility overrides. Omitted keys follow the default dashboard behavior.
+	DashboardFeatures *AccountDashboardFeatures `json:"dashboard_features,omitempty"`
+
 	// DnsDomain Allows to define a custom dns domain for the account
 	DnsDomain *string `json:"dns_domain,omitempty"`
 
@@ -3709,9 +3721,18 @@ type NetworkTrafficEvent struct {
 	Events []NetworkTrafficSubEvent `json:"events"`
 
 	// FlowId FlowID is the ID of the connection flow. Not unique because it can be the same for multiple events (e.g., start and end of the connection).
-	FlowId string               `json:"flow_id"`
-	Icmp   NetworkTrafficICMP   `json:"icmp"`
-	Policy NetworkTrafficPolicy `json:"policy"`
+	FlowId string             `json:"flow_id"`
+	Icmp   NetworkTrafficICMP `json:"icmp"`
+
+	// NumOfDrops Number of drop events.
+	NumOfDrops int `json:"num_of_drops"`
+
+	// NumOfEnds Number of end events.
+	NumOfEnds int `json:"num_of_ends"`
+
+	// NumOfStarts Number of start events.
+	NumOfStarts int                  `json:"num_of_starts"`
+	Policy      NetworkTrafficPolicy `json:"policy"`
 
 	// Protocol Protocol is the protocol of the traffic (e.g. 1 = ICMP, 6 = TCP, 17 = UDP, etc.).
 	Protocol int `json:"protocol"`
@@ -3732,6 +3753,12 @@ type NetworkTrafficEvent struct {
 	// TxPackets Number of packets transmitted.
 	TxPackets int                `json:"tx_packets"`
 	User      NetworkTrafficUser `json:"user"`
+
+	// WindowEnd Timestamp of the end of the aggregation window.
+	WindowEnd time.Time `json:"window_end"`
+
+	// WindowStart Timestamp of the start of the aggregation window.
+	WindowStart time.Time `json:"window_start"`
 }
 
 // NetworkTrafficEventsResponse defines model for NetworkTrafficEventsResponse.
@@ -5842,6 +5869,12 @@ type GetApiEventsNetworkTrafficParams struct {
 	// ReporterId Filter by reporter ID
 	ReporterId *string `form:"reporter_id,omitempty" json:"reporter_id,omitempty"`
 
+	// SourceId Filter by source endpoint ID
+	SourceId *string `form:"source_id,omitempty" json:"source_id,omitempty"`
+
+	// DestinationId Filter by destination endpoint ID
+	DestinationId *string `form:"destination_id,omitempty" json:"destination_id,omitempty"`
+
 	// Protocol Filter by protocol
 	Protocol *int `form:"protocol,omitempty" json:"protocol,omitempty"`
 

shared/relay/client/dialer/ws/close_generic.go

@@ -0,0 +1,9 @@
+//go:build !js
+
+package ws
+
+// closeConn closes the underlying WebSocket immediately, skipping the close
+// handshake.
+func (c *Conn) closeConn() error {
+	return c.Conn.CloseNow()
+}

shared/relay/client/dialer/ws/close_js.go

@@ -0,0 +1,25 @@
+//go:build js
+
+package ws
+
+import (
+	"github.com/coder/websocket"
+	log "github.com/sirupsen/logrus"
+)
+
+// closeConn closes the browser WebSocket without blocking the caller.
+//
+// The browser close API only accepts codes 1000 and 3000-4999, so CloseNow's
+// 1001 (going away) throws an InvalidAccessError. Close with a valid code
+// waits for the browser close event before returning, which can park the
+// calling goroutine (the relay teardown path holds its mutexes while closing)
+// until the close handshake finishes. Run the close in the background and
+// report success; a teardown close error is not actionable.
+func (c *Conn) closeConn() error {
+	go func() {
+		if err := c.Conn.Close(websocket.StatusNormalClosure, ""); err != nil {
+			log.Debugf("failed to close relay websocket: %v", err)
+		}
+	}()
+	return nil
+}

shared/relay/client/dialer/ws/conn.go

@@ -77,5 +77,5 @@ func (c *Conn) SetDeadline(t time.Time) error {
 }
 
 func (c *Conn) Close() error {
-	return c.Conn.CloseNow()
+	return c.closeConn()
 }

shared/relay/client/early_msg_buffer.go

@@ -10,7 +10,7 @@ import (
 
 const (
 	earlyMsgTTL      = 5 * time.Second
-	earlyMsgCapacity = 1000
+	earlyMsgCapacity = 10000
 )
 
 // earlyMsgBuffer buffers transport messages that arrive before the corresponding

shared/relay/client/manager.go

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

shared/relay/client/manager_cleanup_test.go

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

shared/relay/client/manager_relaystates_test.go

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

signal/cmd/run.go

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