Merge branch 'main' into file-share

# Conflicts:
#	client/ios/NetBirdSDK/client.go

.devcontainer/Dockerfile

@@ -1,4 +1,4 @@
-FROM golang:1.25-bookworm
+FROM golang:1.26.7-bookworm
 
 RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
     && apt-get -y install --no-install-recommends\

.github/workflows/golang-test-linux.yml

@@ -233,7 +233,7 @@ jobs:
             -e GOCACHE=${CONTAINER_GOCACHE} \
             -e GOMODCACHE=${CONTAINER_GOMODCACHE} \
             -e CONTAINER=${CONTAINER} \
-            golang:1.25-alpine \
+            golang:1.26.7-alpine \
             sh -c ' \
               apk update; apk add --no-cache \
                 ca-certificates iptables ip6tables dbus dbus-dev libpcap-dev build-base; \
@@ -730,6 +730,11 @@ jobs:
       - name: Install modules
         run: go mod tidy
 
+      - name: Run Mage
+        uses: magefile/mage-action@a662bd8c29d8106879588cfff83b2faf6e6f59db # v4.0.0
+        with:
+          install-only: true
+
       - name: check git status
         run: git --no-pager diff --exit-code
 
@@ -738,9 +743,7 @@ jobs:
           CGO_ENABLED=1 GOARCH=${{ matrix.arch }} \
           NETBIRD_STORE_ENGINE=${{ matrix.store }} \
           CI=true \
-          go test -tags=integration -coverprofile=coverage.txt \
-          -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE' \
-          -timeout 20m ./management/server/http/...
+          mage integrationtest:all -gotestflags="-coverprofile=coverage.txt"
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'

.github/workflows/release.yml

@@ -215,7 +215,7 @@ jobs:
           echo "GPG_RPM_KEY_FILE=/tmp/gpg-rpm-signing-key.asc" >> $GITHUB_ENV
 
       - name: Install goversioninfo
-        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
+        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@b66839b # v1.7.0
       - name: Generate windows syso amd64
         run: goversioninfo  -icon client/ui/build/windows/icon.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_amd64.syso
       - name: Generate windows syso arm64
@@ -435,7 +435,7 @@ jobs:
           tar -xf llvm-mingw-20250709-ucrt-ubuntu-22.04-x86_64.tar.xz
           echo "/tmp/llvm-mingw-20250709-ucrt-ubuntu-22.04-x86_64/bin" >> $GITHUB_PATH
       - name: Install goversioninfo
-        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
+        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@b66839b # v1.7.0
       - name: Install wails3 CLI
         # Version derived from go.mod so the binding generator always matches
         # the wails runtime the binary links against.

CONTRIBUTING.md

@@ -192,7 +192,7 @@ dependencies are installed. Here is a short guide on how that can be done.
 
 ### Requirements
 
-#### Go 1.25
+#### Go 1.26
 
 Follow the installation guide from https://go.dev/
 
@@ -200,7 +200,7 @@ Follow the installation guide from https://go.dev/
 
 The desktop UI client (`client/ui`) is built with [Wails v3](https://v3.wails.io/) and a React frontend rendered in a WebView. To build it you need:
 
-- Go ≥ 1.25
+- Go ≥ 1.26
 - Node ≥ 20 and **pnpm** (`corepack enable && corepack prepare pnpm@latest --activate`)
 - The `wails3` CLI: `go install github.com/wailsapp/wails/v3/cmd/wails3@latest`
 - The `task` runner: `go install github.com/go-task/task/v3/cmd/task@latest`

README.md

@@ -26,7 +26,7 @@
   <strong>
     Start using NetBird at <a href="https://netbird.io/pricing">netbird.io</a>
     <br/>
-    See <a href="https://netbird.io/docs/">Documentation</a>
+    See <a href="https://docs.netbird.io/">Documentation</a>
     <br/>
     Join our <a href="https://docs.netbird.io/slack-url">Slack channel</a> or our <a href="https://forum.netbird.io">Community forum</a>
   </strong>

agent-network/README.md

@@ -96,6 +96,42 @@ components:
   — the management-side control plane: providers, policies, guardrails, limits, routing,
   and usage/access logs.
 
+## Access roles
+
+Agent Network permissions build on the account permission matrix
+([`management/server/permissions/`](../management/server/permissions)). The
+`agent_network` area is split into dotted submodules (`agent_network.providers`,
+`.policies`, `.guardrails`, `.budgets`, `.usage`, `.logs`, `.settings`); a role may
+grant a single submodule or the parent, which cascades to all of them.
+
+Two roles delegate Agent Network access without account-admin rights:
+
+- **`agent_network_admin`** — full control over the whole `agent_network` area plus
+  read-only users, groups, peers, and account info (needed to build policies).
+  Nothing else in the account.
+- **`usage_viewer`** — the regular User baseline plus read on
+  `agent_network.usage` (the aggregated usage and cost overview) and read-only
+  access to the resources the usage filters resolve against: users, groups,
+  peers, and the provider list (connection config redacted — no upstream URLs
+  or operator-supplied header values). No policies, and no account-wide
+  request-level access logs; like any caller, it still reads its own requests
+  through the self-scoped endpoints below.
+
+Every authenticated user, regardless of role, can read the caller-scoped
+self-service endpoint `GET /api/agent-network/agent-config` (the endpoint, providers,
+and models the caller's own policies allow — what a local AI tool needs and nothing
+more). The regular usage and access-log endpoints self-scope instead of denying:
+a caller without the account-wide grant gets their own rows back, so "my usage"
+and "my requests" are the same endpoints the admin dashboard uses. The provider
+list self-scopes the same way — a caller without the providers grant gets the
+providers their own policies authorize, reduced to the display surface, with
+each provider's model list cut to what the caller's policy guardrails and the
+provider's declared models effectively permit (the same computation the setup
+answer and the proxy use). This feeds the dashboard's provider and model
+filters. Role
+definitions live in
+[`management/server/permissions/roles/`](../management/server/permissions/roles).
+
 ## Documentation
 
 Full documentation, architecture, and quickstart:

client/android/split_tunnel.go

@@ -0,0 +1,106 @@
+package android
+
+// Split tunnelling modes, stored as strings so an unknown value written by a
+// newer build degrades to "off" rather than to some other mode's behaviour.
+const (
+	SplitTunnelModeOff     = "off"
+	SplitTunnelModeExclude = "exclude"
+	SplitTunnelModeInclude = "include"
+)
+
+type splitTunnelSection struct {
+	Mode     string   `json:"mode"`
+	Excluded []string `json:"excluded"`
+	Included []string `json:"included"`
+}
+
+// PackageList wraps []string for gomobile compatibility.
+type PackageList struct {
+	items []string
+}
+
+// NewPackageList creates an empty list to fill via Add.
+func NewPackageList() *PackageList {
+	return &PackageList{}
+}
+
+// Add appends a package name, ignoring empty ones.
+func (l *PackageList) Add(s string) {
+	if s == "" {
+		return
+	}
+	l.items = append(l.items, s)
+}
+
+// Size returns the number of entries.
+func (l *PackageList) Size() int {
+	return len(l.items)
+}
+
+// Get returns the entry at index i, or an empty string when out of range.
+func (l *PackageList) Get(i int) string {
+	if i < 0 || i >= len(l.items) {
+		return ""
+	}
+	return l.items[i]
+}
+
+// SplitTunnelSettings is one profile's choice of which applications the tunnel
+// carries. The two selections are kept apart because the platform applies one
+// or the other and never both, and so that switching mode does not throw away
+// the picks made in the other one.
+type SplitTunnelSettings struct {
+	Mode     string
+	Excluded *PackageList
+	Included *PackageList
+}
+
+// NewSplitTunnelSettings creates settings that carry every application.
+func NewSplitTunnelSettings() *SplitTunnelSettings {
+	return &SplitTunnelSettings{
+		Mode:     SplitTunnelModeOff,
+		Excluded: NewPackageList(),
+		Included: NewPackageList(),
+	}
+}
+
+func packagesOf(list *PackageList) []string {
+	if list == nil {
+		return nil
+	}
+	out := make([]string, 0, len(list.items))
+	out = append(out, list.items...)
+	return out
+}
+
+func normalizeSplitTunnelMode(mode string) string {
+	switch mode {
+	case SplitTunnelModeExclude, SplitTunnelModeInclude:
+		return mode
+	default:
+		return SplitTunnelModeOff
+	}
+}
+
+func settingsFromSection(section splitTunnelSection) *SplitTunnelSettings {
+	out := NewSplitTunnelSettings()
+	out.Mode = normalizeSplitTunnelMode(section.Mode)
+	for _, pkg := range section.Excluded {
+		out.Excluded.Add(pkg)
+	}
+	for _, pkg := range section.Included {
+		out.Included.Add(pkg)
+	}
+	return out
+}
+
+func sectionFromSettings(settings *SplitTunnelSettings) splitTunnelSection {
+	if settings == nil {
+		settings = NewSplitTunnelSettings()
+	}
+	return splitTunnelSection{
+		Mode:     normalizeSplitTunnelMode(settings.Mode),
+		Excluded: packagesOf(settings.Excluded),
+		Included: packagesOf(settings.Included),
+	}
+}

client/android/split_tunnel_store.go

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

client/android/split_tunnel_test.go

@@ -0,0 +1,109 @@
+package android
+
+import (
+	"reflect"
+	"testing"
+)
+
+func TestNormalizeSplitTunnelMode(t *testing.T) {
+	tests := []struct {
+		name string
+		mode string
+		want string
+	}{
+		{name: "exclude is kept", mode: SplitTunnelModeExclude, want: SplitTunnelModeExclude},
+		{name: "include is kept", mode: SplitTunnelModeInclude, want: SplitTunnelModeInclude},
+		{name: "off is kept", mode: SplitTunnelModeOff, want: SplitTunnelModeOff},
+		{name: "empty falls back to off", mode: "", want: SplitTunnelModeOff},
+		{name: "a mode from a newer build falls back to off", mode: "only-work-apps", want: SplitTunnelModeOff},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := normalizeSplitTunnelMode(tt.mode); got != tt.want {
+				t.Errorf("normalizeSplitTunnelMode(%q) = %q, want %q", tt.mode, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestSettingsFromSection(t *testing.T) {
+	got := settingsFromSection(splitTunnelSection{
+		Mode:     SplitTunnelModeExclude,
+		Excluded: []string{"com.example.a", "com.example.b"},
+		Included: []string{"com.example.c"},
+	})
+
+	if got.Mode != SplitTunnelModeExclude {
+		t.Errorf("mode = %q, want %q", got.Mode, SplitTunnelModeExclude)
+	}
+	if got.Excluded.Size() != 2 || got.Excluded.Get(0) != "com.example.a" {
+		t.Errorf("excluded = %v, want the two stored packages", packagesOf(got.Excluded))
+	}
+	if got.Included.Size() != 1 || got.Included.Get(0) != "com.example.c" {
+		t.Errorf("included = %v, want the stored package", packagesOf(got.Included))
+	}
+}
+
+// A profile that has never stored anything decodes into an empty section, and
+// must come back as settings that carry every application rather than as nil
+// lists the caller would have to guard against.
+func TestSettingsFromEmptySectionCarriesEverything(t *testing.T) {
+	got := settingsFromSection(splitTunnelSection{})
+
+	if got.Mode != SplitTunnelModeOff {
+		t.Errorf("mode = %q, want %q", got.Mode, SplitTunnelModeOff)
+	}
+	if got.Excluded == nil || got.Included == nil {
+		t.Fatal("both selections must be usable lists, not nil")
+	}
+	if got.Excluded.Size() != 0 || got.Included.Size() != 0 {
+		t.Errorf("selections = %v/%v, want both empty", packagesOf(got.Excluded), packagesOf(got.Included))
+	}
+}
+
+func TestSectionFromSettingsRoundTrip(t *testing.T) {
+	settings := NewSplitTunnelSettings()
+	settings.Mode = SplitTunnelModeInclude
+	settings.Included.Add("com.example.a")
+	settings.Excluded.Add("com.example.b")
+
+	section := sectionFromSettings(settings)
+	back := settingsFromSection(section)
+
+	if back.Mode != SplitTunnelModeInclude {
+		t.Errorf("mode = %q, want %q", back.Mode, SplitTunnelModeInclude)
+	}
+	if !reflect.DeepEqual(packagesOf(back.Included), []string{"com.example.a"}) {
+		t.Errorf("included = %v, want [com.example.a]", packagesOf(back.Included))
+	}
+	// The inactive selection survives, so switching mode back does not make the
+	// user pick their applications again.
+	if !reflect.DeepEqual(packagesOf(back.Excluded), []string{"com.example.b"}) {
+		t.Errorf("excluded = %v, want [com.example.b]", packagesOf(back.Excluded))
+	}
+}
+
+func TestSectionFromNilSettings(t *testing.T) {
+	section := sectionFromSettings(nil)
+
+	if section.Mode != SplitTunnelModeOff {
+		t.Errorf("mode = %q, want %q", section.Mode, SplitTunnelModeOff)
+	}
+	if len(section.Excluded) != 0 || len(section.Included) != 0 {
+		t.Errorf("selections = %v/%v, want both empty", section.Excluded, section.Included)
+	}
+}
+
+func TestPackageListIgnoresEmptyAndBounds(t *testing.T) {
+	list := NewPackageList()
+	list.Add("com.example.a")
+	list.Add("")
+
+	if list.Size() != 1 {
+		t.Errorf("size = %d, want 1", list.Size())
+	}
+	if list.Get(-1) != "" || list.Get(5) != "" {
+		t.Error("out of range access must return an empty string")
+	}
+}

client/cmd/debug.go

@@ -3,7 +3,6 @@ package cmd
 import (
 	"context"
 	"fmt"
-	"os/user"
 	"strings"
 	"time"
 
@@ -114,7 +113,7 @@ func debugConfigDump(cmd *cobra.Command, _ []string) error {
 	if err != nil {
 		return fmt.Errorf("get active profile: %v", err)
 	}
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}

client/cmd/jobs.go

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

client/cmd/login.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"fmt"
 	"os"
-	"os/user"
 	"strings"
 
 	log "github.com/sirupsen/logrus"
@@ -53,7 +52,7 @@ var loginCmd = &cobra.Command{
 			// nolint
 			ctx = context.WithValue(ctx, system.DeviceNameCtxKey, hostName)
 		}
-		username, err := user.Current()
+		username, err := profilemanager.InvokingUser()
 		if err != nil {
 			return fmt.Errorf("get current user: %v", err)
 		}
@@ -74,7 +73,7 @@ var loginCmd = &cobra.Command{
 			if providedSetupKey != "" {
 				return fmt.Errorf("--extend cannot be combined with a setup key; setup keys can only enrol new peers")
 			}
-			if err := doExtendSession(ctx, cmd); err != nil {
+			if err := doExtendSession(ctx, cmd, activeProf); err != nil {
 				return fmt.Errorf("extend session failed: %v", err)
 			}
 			return nil
@@ -176,7 +175,7 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
 // (browser + verification URL) and the resulting JWT is forwarded to the
 // management server's ExtendAuthSession RPC. The tunnel stays up
 // throughout — no Down/Up, no network-map resync.
-func doExtendSession(ctx context.Context, cmd *cobra.Command) error {
+func doExtendSession(ctx context.Context, cmd *cobra.Command, activeProf *profilemanager.Profile) error {
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
 		//nolint
@@ -190,14 +189,12 @@ func doExtendSession(ctx context.Context, cmd *cobra.Command) error {
 
 	// the CLI runs in the user's session, the daemon does not: tell it what we can see
 	req := &proto.RequestExtendAuthSessionRequest{HasGraphicalSession: util.HasGraphicalSession()}
-	// Pre-fill the IdP login hint from the active profile so the user
+	// Pre-fill the IdP login hint from the resolved profile so the user
 	// doesn't have to retype their email. Best-effort: we still proceed
 	// without a hint if the lookup fails.
 	pm := profilemanager.NewProfileManager()
-	if active, perr := pm.GetActiveProfile(); perr == nil {
-		if profState, sperr := pm.GetProfileState(active.ID); sperr == nil && profState.Email != "" {
-			req.Hint = &profState.Email
-		}
+	if profState, perr := pm.GetProfileState(activeProf.ID); perr == nil && profState.Email != "" {
+		req.Hint = &profState.Email
 	}
 
 	startResp, err := client.RequestExtendAuthSession(ctx, req)
@@ -235,9 +232,11 @@ func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, pr
 	// switch profile if provided
 
 	if profileName != "" {
-		if err := switchProfileOnDaemon(ctx, pm, profileName, username); err != nil {
+		prof, err := switchProfileOnDaemon(ctx, pm, profileName, username)
+		if err != nil {
 			return nil, fmt.Errorf("switch profile: %v", err)
 		}
+		return prof, nil
 	}
 
 	activeProf, err := pm.GetActiveProfile()
@@ -251,20 +250,19 @@ func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, pr
 	return activeProf, nil
 }
 
-func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManager, handle string, username string) error {
+func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManager, handle string, username string) (*profilemanager.Profile, error) {
 	resolvedID, err := switchProfile(ctx, handle, username)
 	if err != nil {
-		return fmt.Errorf("switch profile on daemon: %v", err)
+		return nil, fmt.Errorf("switch profile on daemon: %v", err)
 	}
 
 	if err := pm.SwitchProfile(resolvedID); err != nil {
-		return fmt.Errorf("switch profile: %v", err)
+		return nil, fmt.Errorf("switch profile: %v", err)
 	}
 
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
-		log.Errorf("failed to connect to service CLI interface %v", err)
-		return err
+		return nil, fmt.Errorf("connect to service CLI interface: %w", err)
 	}
 	defer conn.Close()
 
@@ -272,17 +270,17 @@ func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManage
 
 	status, err := client.Status(ctx, &proto.StatusRequest{})
 	if err != nil {
-		return fmt.Errorf("unable to get daemon status: %v", err)
+		return nil, fmt.Errorf("unable to get daemon status: %v", err)
 	}
 
 	if status.Status == string(internal.StatusConnected) {
 		if _, err := client.Down(ctx, &proto.DownRequest{}); err != nil {
 			log.Errorf("call service down method: %v", err)
-			return err
+			return nil, err
 		}
 	}
 
-	return nil
+	return &profilemanager.Profile{ID: resolvedID}, nil
 }
 
 // switchProfile asks the daemon to switch to the profile identified by

client/cmd/logout.go

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

client/cmd/profile.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"errors"
 	"fmt"
-	"os/user"
 	"strings"
 	"text/tabwriter"
 	"time"
@@ -97,7 +96,7 @@ func listProfilesFunc(cmd *cobra.Command, _ []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -138,7 +137,7 @@ func addProfileFunc(cmd *cobra.Command, args []string) error {
 		return err
 	}
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -179,7 +178,7 @@ func renameProfileFunc(cmd *cobra.Command, args []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -233,7 +232,7 @@ func removeProfileFunc(cmd *cobra.Command, args []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -261,7 +260,7 @@ func selectProfileFunc(cmd *cobra.Command, args []string) error {
 	profileManager := profilemanager.NewProfileManager()
 	handle := args[0]
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}

client/cmd/root.go

@@ -23,6 +23,7 @@ import (
 
 	"github.com/netbirdio/netbird/client/anonymize"
 	daddr "github.com/netbirdio/netbird/client/internal/daemonaddr"
+	"github.com/netbirdio/netbird/client/internal/localmetrics"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 )
 
@@ -31,6 +32,8 @@ const (
 	dnsResolverAddress       = "dns-resolver-address"
 	enableRosenpassFlag      = "enable-rosenpass"
 	rosenpassPermissiveFlag  = "rosenpass-permissive"
+	enableLocalMetricsFlag   = "enable-local-metrics"
+	localMetricsAddressFlag  = "local-metrics-address"
 	preSharedKeyFlag         = "preshared-key"
 	interfaceNameFlag        = "interface-name"
 	wireguardPortFlag        = "wireguard-port"
@@ -80,6 +83,8 @@ var (
 	updateSettingsDisabled bool
 	captureEnabled         bool
 	networksDisabled       bool
+	localMetricsEnabled    bool
+	localMetricsAddr       string
 
 	rootCmd = &cobra.Command{
 		Use:          "netbird",
@@ -215,6 +220,8 @@ func init() {
 	upCmd.PersistentFlags().BoolVar(&rosenpassEnabled, enableRosenpassFlag, false, "[Experimental] Enable Rosenpass feature. If enabled, the connection will be post-quantum secured via Rosenpass.")
 	upCmd.PersistentFlags().BoolVar(&rosenpassPermissive, rosenpassPermissiveFlag, false, "[Experimental] Enable Rosenpass in permissive mode to allow this peer to accept WireGuard connections without requiring Rosenpass functionality from peers that do not have Rosenpass enabled.")
 	upCmd.PersistentFlags().BoolVar(&autoConnectDisabled, disableAutoConnectFlag, false, "Disables auto-connect feature. If enabled, then the client won't connect automatically when the service starts.")
+	upCmd.PersistentFlags().BoolVar(&localMetricsEnabled, enableLocalMetricsFlag, false, "Enables a local Prometheus /metrics endpoint exposing connection state (peers, latency, P2P vs relay).")
+	upCmd.PersistentFlags().StringVar(&localMetricsAddr, localMetricsAddressFlag, localmetrics.DefaultListenAddress, "Listen address of the local Prometheus /metrics endpoint.")
 	upCmd.PersistentFlags().BoolVar(&lazyConnEnabled, enableLazyConnectionFlag, false, "Deprecated: no longer used. Lazy connections are controlled by the server and the NB_LAZY_CONN environment variable.")
 	_ = upCmd.PersistentFlags().MarkDeprecated(enableLazyConnectionFlag, "no longer used; lazy connections are controlled by the server and the NB_LAZY_CONN environment variable")
 

client/cmd/testutil_test.go

@@ -124,7 +124,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 
 	updateManager := update_channel.NewPeersUpdateManager(metrics)
 	requestBuffer := mgmt.NewAccountRequestBuffer(ctx, store)
-	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config)
+	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config, nil)
 
 	accountManager, err := mgmt.BuildManager(ctx, config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false, cacheStore)
 	if err != nil {

client/cmd/up.go

@@ -2,10 +2,10 @@ package cmd
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"net"
 	"net/netip"
-	"os/user"
 	"runtime"
 	"strings"
 	"time"
@@ -48,6 +48,8 @@ const (
 	profileNameDesc = "profile name to use for the login. If not specified, the last used profile will be used."
 )
 
+var errDaemonActiveProfileUnsupported = errors.New("daemon does not support active profile lookup")
+
 var (
 	foregroundMode     bool
 	dnsLabels          []string
@@ -122,23 +124,25 @@ func upFunc(cmd *cobra.Command, args []string) error {
 
 	pm := profilemanager.NewProfileManager()
 
-	username, err := user.Current()
+	username, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
 
+	var activeProf *profilemanager.Profile
 	var profileSwitched bool
 	// switch profile if provided
 	if profileName != "" {
-		if err := switchOrCreateProfile(cmd.Context(), pm, profileName, username.Username); err != nil {
+		activeProf, err = switchOrCreateProfile(cmd.Context(), pm, profileName, username.Username)
+		if err != nil {
 			return fmt.Errorf("switch profile: %v", err)
 		}
 		profileSwitched = true
-	}
-
-	activeProf, err := pm.GetActiveProfile()
-	if err != nil {
-		return fmt.Errorf("get active profile: %v", err)
+	} else {
+		activeProf, err = pm.GetActiveProfile()
+		if err != nil {
+			return fmt.Errorf("get active profile: %v", err)
+		}
 	}
 
 	if foregroundMode {
@@ -150,13 +154,15 @@ func upFunc(cmd *cobra.Command, args []string) error {
 // switchOrCreateProfile switches the active profile to the one identified by
 // handle, creating it first when it does not exist yet. This restores the
 // pre-0.73 behaviour where `netbird up --profile <name>` auto-creates a
-// missing profile instead of failing.
-func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManager, handle, username string) error {
+// missing profile instead of failing. Returns the daemon-resolved profile so
+// callers act on it directly instead of re-reading the local state, which is
+// not updated under sudo.
+func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManager, handle, username string) (*profilemanager.Profile, error) {
 	resolvedID, err := switchProfile(ctx, handle, username)
 	if err != nil {
 		st, ok := gstatus.FromError(err)
 		if !ok || st.Code() != codes.NotFound {
-			return err
+			return nil, err
 		}
 		// Don't fail immediately on a create error: a concurrent run may
 		// have created the profile between the NotFound above and this
@@ -165,16 +171,16 @@ func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManage
 		_, createErr := createProfile(ctx, handle, username)
 		if resolvedID, err = switchProfile(ctx, handle, username); err != nil {
 			if createErr != nil {
-				return fmt.Errorf("create profile: %w", createErr)
+				return nil, fmt.Errorf("create profile: %w", createErr)
 			}
-			return err
+			return nil, err
 		}
 	}
 
 	if err := pm.SwitchProfile(resolvedID); err != nil {
-		return err
+		return nil, err
 	}
-	return nil
+	return &profilemanager.Profile{ID: resolvedID}, nil
 }
 
 // createProfile dials the daemon and creates a new profile with the given
@@ -302,6 +308,30 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 		return fmt.Errorf("unable to get daemon status: %v", err)
 	}
 
+	// Under sudo the invoking user's local active-profile mirror is never
+	// written (the SwitchProfile write is a no-op), and plain root has no
+	// invoking user at all — so the mirror read into activeProf above is stale
+	// or defaulted and must not drive the daemon. With no --profile to make the
+	// choice explicit, take the profile the daemon already holds for this user
+	// instead: it stays on the user's current profile rather than silently
+	// switching to the mirror's default, and refuses when the daemon is on
+	// another user's profile.
+	if profileName == "" && !profilemanager.MirrorIsAuthoritative() {
+		u, err := profilemanager.InvokingUser()
+		if err != nil {
+			return fmt.Errorf("get current user: %v", err)
+		}
+		resolved, err := daemonActiveProfileForUser(ctx, client, u.Username)
+		switch {
+		case errors.Is(err, errDaemonActiveProfileUnsupported):
+			log.Warnf("keeping the locally resolved profile: %v", err)
+		case err != nil:
+			return err
+		default:
+			activeProf = resolved
+		}
+	}
+
 	if status.Status == string(internal.StatusConnected) {
 		if !profileSwitched {
 			cmd.Println("Already connected")
@@ -314,7 +344,7 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 		}
 	}
 
-	username, err := user.Current()
+	username, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
@@ -398,26 +428,21 @@ func doDaemonUp(ctx context.Context, cmd *cobra.Command, client proto.DaemonServ
 	return nil
 }
 
-func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, profileName, username string) *proto.SetConfigRequest {
-	var req proto.SetConfigRequest
-	req.ProfileName = profileName
-	req.Username = username
-
-	req.ManagementUrl = managementURL
-	req.AdminURL = adminURL
-	req.NatExternalIPs = natExternalIPs
-	req.CustomDNSAddress = customDNSAddressConverted
-	req.ExtraIFaceBlacklist = extraIFaceBlackList
-	req.DnsLabels = dnsLabelsValidated.ToPunycodeList()
-	req.CleanDNSLabels = dnsLabels != nil && len(dnsLabels) == 0
-	req.CleanNATExternalIPs = natExternalIPs != nil && len(natExternalIPs) == 0
-
-	if cmd.Flag(enableRosenpassFlag).Changed {
-		req.RosenpassEnabled = &rosenpassEnabled
-	}
-	if cmd.Flag(rosenpassPermissiveFlag).Changed {
-		req.RosenpassPermissive = &rosenpassPermissive
+// setBoolPtrIfChanged points dst at a copy of val when the named bool flag was
+// explicitly set on cmd. It collapses the repeated
+// "if cmd.Flag(x).Changed { field = &val }" pattern in the request builders into
+// a single call, keeping their cognitive complexity within bounds.
+func setBoolPtrIfChanged(cmd *cobra.Command, name string, dst **bool, val bool) {
+	if cmd.Flag(name).Changed {
+		dst2 := val
+		*dst = &dst2
 	}
+}
+
+// setSSHSetConfigFields copies the SSH server flags the user actually
+// passed into req, leaving the rest unset so the daemon keeps the
+// persisted values.
+func setSSHSetConfigFields(req *proto.SetConfigRequest, cmd *cobra.Command) {
 	if cmd.Flag(serverSSHAllowedFlag).Changed {
 		req.ServerSSHAllowed = &serverSSHAllowed
 	}
@@ -440,6 +465,31 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
 		req.SshJWTCacheTTL = &sshJWTCacheTTL32
 	}
+}
+
+func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, profileName, username string) *proto.SetConfigRequest {
+	var req proto.SetConfigRequest
+	req.ProfileName = profileName
+	req.Username = username
+
+	req.ManagementUrl = managementURL
+	req.AdminURL = adminURL
+	req.NatExternalIPs = natExternalIPs
+	req.CustomDNSAddress = customDNSAddressConverted
+	req.ExtraIFaceBlacklist = extraIFaceBlackList
+	req.DnsLabels = dnsLabelsValidated.ToPunycodeList()
+	req.CleanDNSLabels = dnsLabels != nil && len(dnsLabels) == 0
+	req.CleanNATExternalIPs = natExternalIPs != nil && len(natExternalIPs) == 0
+
+	if cmd.Flag(enableRosenpassFlag).Changed {
+		req.RosenpassEnabled = &rosenpassEnabled
+	}
+	if cmd.Flag(rosenpassPermissiveFlag).Changed {
+		req.RosenpassPermissive = &rosenpassPermissive
+	}
+	setSSHSetConfigFields(&req, cmd)
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &req.RemoteJobsAllowed, remoteJobsAllowed)
+
 	if cmd.Flag(interfaceNameFlag).Changed {
 		if err := parseInterfaceName(interfaceName); err != nil {
 			log.Errorf("parse interface name: %v", err)
@@ -499,6 +549,13 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		req.DisableIpv6 = &disableIPv6
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		req.EnableLocalMetrics = &localMetricsEnabled
+	}
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		req.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	return &req
 }
 
@@ -523,6 +580,7 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 	if cmd.Flag(serverSSHAllowedFlag).Changed {
 		ic.ServerSSHAllowed = &serverSSHAllowed
 	}
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &ic.RemoteJobsAllowed, remoteJobsAllowed)
 
 	if cmd.Flag(enableSSHRootFlag).Changed {
 		ic.EnableSSHRoot = &enableSSHRoot
@@ -616,9 +674,45 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 		ic.DisableIPv6 = &disableIPv6
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		ic.LocalMetricsEnabled = &localMetricsEnabled
+	}
+
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		ic.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	return &ic, nil
 }
 
+// setSSHLoginFields copies the SSH server flags the user actually passed
+// into req, leaving the rest unset so the daemon keeps the persisted
+// values.
+func setSSHLoginFields(req *proto.LoginRequest, cmd *cobra.Command) {
+	if cmd.Flag(serverSSHAllowedFlag).Changed {
+		req.ServerSSHAllowed = &serverSSHAllowed
+	}
+	if cmd.Flag(enableSSHRootFlag).Changed {
+		req.EnableSSHRoot = &enableSSHRoot
+	}
+	if cmd.Flag(enableSSHSFTPFlag).Changed {
+		req.EnableSSHSFTP = &enableSSHSFTP
+	}
+	if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
+		req.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
+	}
+	if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
+		req.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
+	}
+	if cmd.Flag(disableSSHAuthFlag).Changed {
+		req.DisableSSHAuth = &disableSSHAuth
+	}
+	if cmd.Flag(sshJWTCacheTTLFlag).Changed {
+		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
+		req.SshJWTCacheTTL = &sshJWTCacheTTL32
+	}
+}
+
 func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte, cmd *cobra.Command) (*proto.LoginRequest, error) {
 	loginRequest := proto.LoginRequest{
 		SetupKey:            providedSetupKey,
@@ -645,39 +739,21 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		loginRequest.RosenpassPermissive = &rosenpassPermissive
 	}
 
-	if cmd.Flag(serverSSHAllowedFlag).Changed {
-		loginRequest.ServerSSHAllowed = &serverSSHAllowed
-	}
-
-	if cmd.Flag(enableSSHRootFlag).Changed {
-		loginRequest.EnableSSHRoot = &enableSSHRoot
-	}
-
-	if cmd.Flag(enableSSHSFTPFlag).Changed {
-		loginRequest.EnableSSHSFTP = &enableSSHSFTP
-	}
-
-	if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
-		loginRequest.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
-	}
-
-	if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
-		loginRequest.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
-	}
-
-	if cmd.Flag(disableSSHAuthFlag).Changed {
-		loginRequest.DisableSSHAuth = &disableSSHAuth
-	}
-
-	if cmd.Flag(sshJWTCacheTTLFlag).Changed {
-		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
-		loginRequest.SshJWTCacheTTL = &sshJWTCacheTTL32
-	}
+	setSSHLoginFields(&loginRequest, cmd)
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &loginRequest.RemoteJobsAllowed, remoteJobsAllowed)
 
 	if cmd.Flag(disableAutoConnectFlag).Changed {
 		loginRequest.DisableAutoConnect = &autoConnectDisabled
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		loginRequest.EnableLocalMetrics = &localMetricsEnabled
+	}
+
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		loginRequest.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	if cmd.Flag(interfaceNameFlag).Changed {
 		if err := parseInterfaceName(interfaceName); err != nil {
 			return nil, err
@@ -849,3 +925,31 @@ func isValidAddrPort(input string) bool {
 	_, err := netip.ParseAddrPort(input)
 	return err == nil
 }
+
+// daemonActiveProfileForUser returns the profile the daemon currently holds for
+// username, for the no --profile case where the local mirror is not
+// authoritative (sudo or plain root). It returns that profile when the daemon
+// owns it for this user or when the profile is unowned (empty username, as on a
+// fresh install), so the caller acts on the daemon's real state instead of the
+// stale mirror. It denies with a --profile hint when the daemon is on another
+// user's profile, when the lookup fails, or when the daemon reports no active
+// profile. Returns errDaemonActiveProfileUnsupported when the daemon predates
+// the RPC; the caller keeps the mirror-derived profile in that case.
+func daemonActiveProfileForUser(ctx context.Context, client proto.DaemonServiceClient, username string) (*profilemanager.Profile, error) {
+	active, err := client.GetActiveProfile(ctx, &proto.GetActiveProfileRequest{})
+	if err != nil {
+		if st, ok := gstatus.FromError(err); ok && st.Code() == codes.Unimplemented {
+			return nil, fmt.Errorf("%w: %v", errDaemonActiveProfileUnsupported, err)
+		}
+		return nil, fmt.Errorf("pass --profile to choose the profile explicitly: the daemon's active profile could not be verified: %v", err)
+	}
+	if active.GetId() == "" {
+		return nil, fmt.Errorf("pass --profile to choose the profile explicitly: the daemon reported no active profile")
+	}
+	if active.GetUsername() != "" && active.GetUsername() != username {
+		return nil, fmt.Errorf(
+			"pass --profile to choose the profile explicitly: the daemon's active profile is %q (user %q) but this invocation runs for %q",
+			active.GetProfileName(), active.GetUsername(), username)
+	}
+	return &profilemanager.Profile{ID: profilemanager.ID(active.GetId())}, nil
+}

client/cmd/up_test.go

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

client/embed/embed_test.go

@@ -146,7 +146,7 @@ func startManagement(t *testing.T, signalAddr string) string {
 
 	updateManager := update_channel.NewPeersUpdateManager(metrics)
 	requestBuffer := mgmt.NewAccountRequestBuffer(context.Background(), testStore)
-	networkMapController := controller.NewController(context.Background(), testStore, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(testStore, peersManager), cfg)
+	networkMapController := controller.NewController(context.Background(), testStore, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(testStore, peersManager), cfg, nil)
 	accountManager, err := mgmt.BuildManager(context.Background(), cfg, testStore, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManager, false, cacheStore)
 	require.NoError(t, err)
 

client/internal/auth/auth.go

@@ -368,6 +368,7 @@ func (a *Auth) setSystemInfoFlags(info *system.Info) {
 		a.config.EnableSSHLocalPortForwarding,
 		a.config.EnableSSHRemotePortForwarding,
 		a.config.DisableSSHAuth,
+		a.config.RemoteJobsAllowed,
 	)
 }
 

client/internal/auth/device_flow.go

@@ -304,6 +304,16 @@ func (d *DeviceAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowIn
 				return TokenInfo{}, fmt.Errorf("validate access token failed with error: %v", err)
 			}
 
+			// Same as the PKCE flow: the account the token belongs to is what
+			// callers store to send back as the login_hint. Without it a client
+			// driven through the device flow — Android TV and tvOS — never binds
+			// an account to its profile and every later login goes out blind.
+			if email, err := parseEmailFromIDToken(tokenInfo.IDToken); err != nil {
+				log.Warnf("failed to parse email from ID token: %v", err)
+			} else {
+				tokenInfo.Email = email
+			}
+
 			log.Infof("device flow: user authorization confirmed after %d polls in %s", polls, time.Since(start).Round(time.Second))
 			return tokenInfo, err
 		}

client/internal/connect.go

@@ -250,7 +250,7 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 	wrapErr := state.Wrap
 	myPrivateKey, err := wgtypes.ParseKey(c.config.PrivateKey)
 	if err != nil {
-		log.Errorf("failed parsing Wireguard key %s: [%s]", c.config.PrivateKey, err.Error())
+		log.Errorf("failed parsing Wireguard key: %s", err)
 		return wrapErr(err)
 	}
 
@@ -661,6 +661,7 @@ func createEngineConfig(key wgtypes.Key, config *profilemanager.Config, peerConf
 		RosenpassEnabled:              config.RosenpassEnabled,
 		RosenpassPermissive:           config.RosenpassPermissive,
 		ServerSSHAllowed:              util.ReturnBoolWithDefaultTrue(config.ServerSSHAllowed),
+		RemoteJobsAllowed:             util.ReturnBoolWithDefaultFalse(config.RemoteJobsAllowed),
 		EnableSSHRoot:                 config.EnableSSHRoot,
 		EnableSSHSFTP:                 config.EnableSSHSFTP,
 		EnableSSHLocalPortForwarding:  config.EnableSSHLocalPortForwarding,
@@ -758,6 +759,7 @@ func loginToManagement(ctx context.Context, client mgm.Client, pubSSHKey []byte,
 		config.EnableSSHLocalPortForwarding,
 		config.EnableSSHRemotePortForwarding,
 		config.DisableSSHAuth,
+		config.RemoteJobsAllowed,
 	)
 	return client.Login(sysInfo, pubSSHKey, config.DNSLabels)
 }

client/internal/debug/debug.go

@@ -711,6 +711,9 @@ func (g *BundleGenerator) addCommonConfigFields(configContent *strings.Builder)
 	if g.internalConfig.ServerSSHAllowed != nil {
 		configContent.WriteString(fmt.Sprintf("ServerSSHAllowed: %v\n", *g.internalConfig.ServerSSHAllowed))
 	}
+	if g.internalConfig.RemoteJobsAllowed != nil {
+		configContent.WriteString(fmt.Sprintf("RemoteJobsAllowed: %v\n", *g.internalConfig.RemoteJobsAllowed))
+	}
 	if g.internalConfig.EnableSSHRoot != nil {
 		configContent.WriteString(fmt.Sprintf("EnableSSHRoot: %v\n", *g.internalConfig.EnableSSHRoot))
 	}
@@ -737,6 +740,8 @@ func (g *BundleGenerator) addCommonConfigFields(configContent *strings.Builder)
 	configContent.WriteString(fmt.Sprintf("BlockLANAccess: %v\n", g.internalConfig.BlockLANAccess))
 	configContent.WriteString(fmt.Sprintf("BlockInbound: %v\n", g.internalConfig.BlockInbound))
 	configContent.WriteString(fmt.Sprintf("DisableIPv6: %v\n", g.internalConfig.DisableIPv6))
+	configContent.WriteString(fmt.Sprintf("LocalMetricsEnabled: %v\n", g.internalConfig.LocalMetricsEnabled))
+	configContent.WriteString(fmt.Sprintf("LocalMetricsAddress: %s\n", g.internalConfig.LocalMetricsAddress))
 	configContent.WriteString(fmt.Sprintf("SyncMessageVersion: %v\n", g.internalConfig.SyncMessageVersion))
 
 	if g.internalConfig.DisableNotifications != nil {

client/internal/debug/debug_test.go

@@ -839,12 +839,13 @@ COMMIT`
 // the excluded set with a justification.
 func TestAddConfig_AllFieldsCovered(t *testing.T) {
 	excluded := map[string]string{
-		"PrivateKey":        "sensitive: WireGuard private key",
-		"PreSharedKey":      "sensitive: WireGuard pre-shared key",
-		"SSHKey":            "sensitive: SSH private key",
-		"ClientCertKeyPair": "non-config: parsed cert pair, not serialized",
-		"Name":              "non-config: profile name is not needed for debug purposes",
-		"policy":            "non-config: in-memory MDM policy snapshot, surfaced via Config.Policy() / GetConfigResponse.MDMManagedFields",
+		"PrivateKey":           "sensitive: WireGuard private key",
+		"PreSharedKey":         "sensitive: WireGuard pre-shared key",
+		"SSHKey":               "sensitive: SSH private key",
+		"ClientCertKeyPair":    "non-config: parsed cert pair, not serialized",
+		"Name":                 "non-config: profile name is not needed for debug purposes",
+		"policy":               "non-config: in-memory MDM policy snapshot, surfaced via Config.Policy() / GetConfigResponse.MDMManagedFields",
+		"DebugBundleUploadURL": "sensitive: MDM-provided upload URL may carry credentials or query tokens; kept out of the shared bundle",
 	}
 
 	mURL, _ := url.Parse("https://api.example.com:443")
@@ -864,6 +865,7 @@ func TestAddConfig_AllFieldsCovered(t *testing.T) {
 		RosenpassEnabled:              true,
 		RosenpassPermissive:           true,
 		ServerSSHAllowed:              &bTrue,
+		RemoteJobsAllowed:             &bTrue,
 		EnableSSHRoot:                 &bTrue,
 		EnableSSHSFTP:                 &bTrue,
 		EnableSSHLocalPortForwarding:  &bTrue,
@@ -886,6 +888,7 @@ func TestAddConfig_AllFieldsCovered(t *testing.T) {
 		ClientCertPath:                "/tmp/cert",
 		ClientCertKeyPath:             "/tmp/key",
 		LazyConnection:                "on",
+		DebugBundleUploadURL:          "https://upload.example.test/bundle?token=secret",
 		MTU:                           1280,
 		DisableIPv6:                   true,
 		SyncMessageVersion:            func(v int) *int { return &v }(1),
@@ -903,6 +906,13 @@ func TestAddConfig_AllFieldsCovered(t *testing.T) {
 			g.addCommonConfigFields(&sb)
 			rendered := sb.String() + renderAddConfigSpecific(g)
 
+			// DebugBundleUploadURL is an MDM-provided value that can carry
+			// credentials or signed query tokens. It is deliberately excluded
+			// above; assert it never reaches the rendered bundle — neither the
+			// field name nor the token — in either anonymize mode.
+			assert.NotContains(t, rendered, "DebugBundleUploadURL:", "MDM upload URL field must not be serialized into the debug bundle")
+			assert.NotContains(t, rendered, "token=secret", "MDM upload URL value must not leak into the debug bundle")
+
 			val := reflect.ValueOf(cfg).Elem()
 			typ := val.Type()
 			var missing []string

client/internal/dns/host_windows.go

@@ -6,8 +6,10 @@ import (
 	"fmt"
 	"io"
 	"net/netip"
+	"os"
 	"os/exec"
 	"slices"
+	"strconv"
 	"strings"
 	"syscall"
 	"time"
@@ -34,10 +36,16 @@ var (
 // Registry locations of the host DNS configuration this package programs,
 // exported so a diagnostic reader reports the same locations that are written.
 const (
-	// NRPTKeyPrefix starts the name of every NRPT rule key this client creates.
-	// Older versions used different layouts under the same prefix: a single
-	// unsuffixed key, then one key per domain, now one key per batch of domains.
-	NRPTKeyPrefix = "NetBird-Match"
+	// NRPTKeyPrefix starts the name of every NRPT rule key this client creates:
+	// the match rules, the catch-all, and the .local exemption. Cleanup
+	// enumerates by this prefix, so a new kind of rule is removed by existing
+	// code as long as its key starts here.
+	NRPTKeyPrefix = "NetBird-"
+
+	// nrptMatchKeyName names the match-domain rules. Older versions used
+	// different layouts under the same name: a single unsuffixed key, then one
+	// key per domain, now one key per batch of domains.
+	nrptMatchKeyName = NRPTKeyPrefix + "Match"
 
 	// DNSPolicyConfigRoot holds the NRPT rules of the local policy store.
 	DNSPolicyConfigRoot = `SYSTEM\CurrentControlSet\Services\Dnscache\Parameters\DnsPolicyConfig`
@@ -53,8 +61,24 @@ const (
 )
 
 const (
-	dnsPolicyConfigMatchPath    = DNSPolicyConfigRoot + `\` + NRPTKeyPrefix
-	gpoDnsPolicyConfigMatchPath = GPODNSPolicyConfigRoot + `\` + NRPTKeyPrefix
+	dnsPolicyConfigMatchPath    = DNSPolicyConfigRoot + `\` + nrptMatchKeyName
+	gpoDnsPolicyConfigMatchPath = GPODNSPolicyConfigRoot + `\` + nrptMatchKeyName
+
+	dnsPolicyConfigExemptLocalPath    = DNSPolicyConfigRoot + `\` + NRPTKeyPrefix + `ExemptLocal`
+	gpoDnsPolicyConfigExemptLocalPath = GPODNSPolicyConfigRoot + `\` + NRPTKeyPrefix + `ExemptLocal`
+
+	nrptCatchAllNamespace = "."
+	// nrptLocalNamespace is reserved for multicast DNS by RFC 6762: a unicast
+	// resolver must not answer for it. The catch-all rule would hand it to us
+	// anyway, so it gets an exemption rule of its own.
+	nrptLocalNamespace = ".local"
+
+	// envLegacyDNSResolution restores the pre-catch-all behaviour: the adapter's
+	// NameServer alone, leaving the OS free to query other adapters' resolvers in
+	// parallel. An escape hatch for setups that depend on a resolver of theirs
+	// still being reachable while connected, at the cost of the leak and of the
+	// race the catch-all rule exists to close.
+	envLegacyDNSResolution = "NB_USE_LEGACY_DNS_RESOLUTION"
 
 	dnsPolicyConfigVersionKey           = "Version"
 	dnsPolicyConfigVersionValue         = 2
@@ -293,6 +317,13 @@ func (r *registryConfigurator) disableWINSForInterface() error {
 }
 
 func (r *registryConfigurator) applyDNSConfig(config HostDNSConfig, stateManager *statemanager.Manager) error {
+	// Clear every rule the previous apply installed before installing any new
+	// one, including a leftover catch-all: removal is unconditional so a rule
+	// from an earlier run cannot survive into a config that no longer wants it.
+	if err := r.removeDNSMatchPolicies(); err != nil {
+		log.Errorf("cleanup old dns match policies: %s", err)
+	}
+
 	if config.RouteAll {
 		if err := r.addDNSSetupForAll(config.ServerIP); err != nil {
 			return fmt.Errorf("add dns setup: %w", err)
@@ -318,8 +349,22 @@ func (r *registryConfigurator) applyDNSConfig(config HostDNSConfig, stateManager
 		matchDomains = append(matchDomains, "."+strings.TrimSuffix(dConf.Domain, "."))
 	}
 
-	if err := r.removeDNSMatchPolicies(); err != nil {
-		log.Errorf("cleanup old dns match policies: %s", err)
+	// The root namespace is a match domain like any other: it just happens to
+	// match every name. Without it the adapter's NameServer only adds one more
+	// resolver to the set Windows queries in parallel, keeping whichever answer
+	// comes back first — which leaks every query to the local network and lets a
+	// resolver other than ours answer for a name we are authoritative for.
+	if config.RouteAll {
+		if parseBoolEnv(envLegacyDNSResolution) {
+			log.Infof("%s is set, leaving DNS resolution shared with the other adapters' resolvers instead of forcing it through %s", envLegacyDNSResolution, config.ServerIP)
+		} else {
+			matchDomains = append(matchDomains, nrptCatchAllNamespace)
+			log.Infof("routing every namespace through %s: DNS resolution is now exclusive to NetBird", config.ServerIP)
+
+			if err := r.addDNSExemptLocalPolicy(); err != nil {
+				return fmt.Errorf("add dns exempt policy: %w", err)
+			}
+		}
 	}
 
 	if len(matchDomains) != 0 {
@@ -397,6 +442,42 @@ func (r *registryConfigurator) addDNSMatchPolicy(domains []string, ip netip.Addr
 	return nil
 }
 
+// addDNSExemptLocalPolicy carves .local back out of the catch-all. RFC 6762
+// reserves it for multicast DNS, so forwarding those names to a unicast
+// upstream answers NXDOMAIN for hosts that do exist - printers, NAS boxes, and
+// anything else announcing itself on the link - and the answer is authoritative
+// enough that Windows stops looking. A rule naming the namespace with no
+// servers hands it back to the DNS client untouched. A more specific rule still
+// wins, so a match domain under .local keeps going through us.
+func (r *registryConfigurator) addDNSExemptLocalPolicy() error {
+	var noServers netip.Addr
+
+	if err := r.configureDNSPolicy(dnsPolicyConfigExemptLocalPath, []string{nrptLocalNamespace}, noServers); err != nil {
+		return fmt.Errorf("configure exempt policy for %s: %w", nrptLocalNamespace, err)
+	}
+
+	if r.gpo {
+		if err := r.configureDNSPolicy(gpoDnsPolicyConfigExemptLocalPath, []string{nrptLocalNamespace}, noServers); err != nil {
+			return fmt.Errorf("configure gpo exempt policy for %s: %w", nrptLocalNamespace, err)
+		}
+		if err := refreshGroupPolicy(); err != nil {
+			log.Warnf("failed to refresh group policy: %v", err)
+		}
+	}
+
+	log.Infof("added NRPT exemption for %s, leaving it to the OS resolver", nrptLocalNamespace)
+	return nil
+}
+
+// configureDNSPolicy writes one NRPT rule. An invalid ip writes an exemption
+// rule: the namespace with an empty server list, which tells the DNS client to
+// resolve those names the way it would without any rule at all.
+//
+// The empty string is the whole difference, and it has to be written: dropping
+// the value and clearing ConfigOptions instead produces a rule Windows treats
+// as a no-op, keeps out of Get-DnsClientNrptPolicy -Effective, and ignores in
+// favour of the catch-all. 0x8 says the server list is the meaningful part of
+// the rule, and an empty list then means "no server, resolve normally".
 func (r *registryConfigurator) configureDNSPolicy(policyPath string, domains []string, ip netip.Addr) error {
 	if err := removeRegistryKeyFromDNSPolicyConfig(policyPath); err != nil {
 		return fmt.Errorf("remove existing dns policy: %w", err)
@@ -416,7 +497,11 @@ func (r *registryConfigurator) configureDNSPolicy(policyPath string, domains []s
 		return fmt.Errorf("set %s: %w", dnsPolicyConfigNameKey, err)
 	}
 
-	if err := regKey.SetStringValue(dnsPolicyConfigGenericDNSServersKey, ip.String()); err != nil {
+	var servers string
+	if ip.IsValid() {
+		servers = ip.String()
+	}
+	if err := regKey.SetStringValue(dnsPolicyConfigGenericDNSServersKey, servers); err != nil {
 		return fmt.Errorf("set %s: %w", dnsPolicyConfigGenericDNSServersKey, err)
 	}
 
@@ -514,8 +599,11 @@ func (r *registryConfigurator) getInterfaceRegistryKey() (registry.Key, error) {
 }
 
 func (r *registryConfigurator) restoreHostDNS() error {
+	// Propagated, unlike in applyDNSConfig: there we are about to write fresh
+	// rules over whatever survived, here we are leaving, and a rule left behind
+	// keeps sending every query to an address that is about to disappear.
 	if err := r.removeDNSMatchPolicies(); err != nil {
-		log.Errorf("remove dns match policies: %s", err)
+		return fmt.Errorf("remove dns match policies: %w", err)
 	}
 
 	if err := r.deleteInterfaceRegistryKeyProperty(interfaceConfigSearchListKey); err != nil {
@@ -598,9 +686,17 @@ func listNRPTRuleKeys(root string) ([]string, error) {
 
 func removeRegistryKeyFromDNSPolicyConfig(regKeyPath string) error {
 	k, err := registry.OpenKey(registry.LOCAL_MACHINE, regKeyPath, registry.QUERY_VALUE)
-	if err != nil {
-		log.Debugf("failed to open HKEY_LOCAL_MACHINE\\%s: %v", regKeyPath, err)
+	switch {
+	case errors.Is(err, registry.ErrNotExist), errors.Is(err, syscall.ERROR_PATH_NOT_FOUND):
+		// nothing to remove, which is the normal case for a rule this config
+		// never installed
+		log.Debugf("HKEY_LOCAL_MACHINE\\%s does not exist", regKeyPath)
 		return nil
+	case err != nil:
+		// anything else has to reach the caller: reporting success here would
+		// leave the rule in force while claiming it was removed, which is how a
+		// stale rule outlives the interface it points at
+		return fmt.Errorf("open HKEY_LOCAL_MACHINE\\%s: %w", regKeyPath, err)
 	}
 
 	closer(k)
@@ -636,6 +732,20 @@ func refreshGroupPolicy() error {
 	return nil
 }
 
+func parseBoolEnv(key string) bool {
+	val := os.Getenv(key)
+	if val == "" {
+		return false
+	}
+
+	parsed, err := strconv.ParseBool(val)
+	if err != nil {
+		log.Warnf("failed to parse %s=%q: %v", key, val, err)
+		return false
+	}
+	return parsed
+}
+
 func closer(closer io.Closer) {
 	if err := closer.Close(); err != nil {
 		log.Errorf("failed to close: %s", err)

client/internal/dns/host_windows_test.go

@@ -94,6 +94,145 @@ func TestNRPTEntriesCleanupOnConfigChange(t *testing.T) {
 	assert.False(t, exists, "NRPT rule 2 should NOT exist after reducing to 75 domains")
 }
 
+// TestNRPTCatchAllRule verifies that RouteAll adds the root namespace to the
+// match rule instead of a rule of its own, that .local is carved back out with
+// an empty server list, and that both go away when RouteAll is cleared or the
+// host DNS is restored.
+func TestNRPTCatchAllRule(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping registry integration test in short mode")
+	}
+
+	defer cleanupRegistryKeys(t)
+	cleanupRegistryKeys(t)
+
+	testIP := netip.MustParseAddr("100.64.0.1")
+	testGUID := "{12345678-1234-1234-1234-123456789ABC}"
+	interfacePath := InterfaceConfigPath + `\` + testGUID
+	testKey, _, err := registry.CreateKey(registry.LOCAL_MACHINE, interfacePath, registry.SET_VALUE)
+	require.NoError(t, err, "Should create test interface registry key")
+	require.NoError(t, testKey.Close(), "close test interface registry key")
+	defer func() {
+		assert.NoError(t, registry.DeleteKey(registry.LOCAL_MACHINE, interfacePath), "delete test interface registry key")
+	}()
+
+	cfg := &registryConfigurator{guid: testGUID}
+
+	matchOnly := HostDNSConfig{
+		ServerIP: testIP,
+		Domains:  []DomainConfig{{Domain: "example.com", MatchOnly: true}},
+	}
+	primary := HostDNSConfig{
+		ServerIP: testIP,
+		RouteAll: true,
+		Domains:  []DomainConfig{{Domain: "example.com", MatchOnly: true}},
+	}
+	firstRule := fmt.Sprintf("%s-0", dnsPolicyConfigMatchPath)
+
+	// The root namespace is not a rule of its own: it rides in the match rule,
+	// which is the point of it not being a special case.
+	require.NoError(t, cfg.applyDNSConfig(matchOnly, nil))
+	names := ruleNamespaces(t, firstRule)
+	assert.Contains(t, names, ".example.com")
+	assert.NotContains(t, names, nrptCatchAllNamespace, "a match-only config must not claim every namespace")
+
+	require.NoError(t, cfg.applyDNSConfig(primary, nil))
+	names = ruleNamespaces(t, firstRule)
+	assert.Contains(t, names, ".example.com")
+	assert.Contains(t, names, nrptCatchAllNamespace, "RouteAll should add the root namespace to the match rule")
+
+	k, err := registry.OpenKey(registry.LOCAL_MACHINE, firstRule, registry.QUERY_VALUE)
+	require.NoError(t, err)
+	servers, _, err := k.GetStringValue(dnsPolicyConfigGenericDNSServersKey)
+	require.NoError(t, err)
+	assert.Equal(t, testIP.String(), servers, "every namespace in the rule resolves through our resolver")
+	require.NoError(t, k.Close(), "close match rule key")
+
+	// .local is carved back out: RFC 6762 reserves it for mDNS, so it needs a
+	// rule of its own — it is the one rule with a different server list.
+	ek, err := registry.OpenKey(registry.LOCAL_MACHINE, dnsPolicyConfigExemptLocalPath, registry.QUERY_VALUE)
+	require.NoError(t, err, "exemption rule should exist once the root namespace is claimed")
+
+	exemptNames, _, err := ek.GetStringsValue(dnsPolicyConfigNameKey)
+	require.NoError(t, err)
+	assert.Equal(t, []string{nrptLocalNamespace}, exemptNames, "the exemption should name only the mDNS namespace")
+
+	exemptServers, _, err := ek.GetStringValue(dnsPolicyConfigGenericDNSServersKey)
+	require.NoError(t, err, "the value has to be present, empty: without it Windows drops the rule")
+	assert.Empty(t, exemptServers, "an exemption rule lists no servers")
+
+	exemptOpts, _, err := ek.GetIntegerValue(dnsPolicyConfigConfigOptionsKey)
+	require.NoError(t, err)
+	assert.EqualValues(t, dnsPolicyConfigConfigOptionsValue, exemptOpts, "same options as a normal rule; the empty server list is what makes it an exemption")
+	require.NoError(t, ek.Close(), "close exemption rule key")
+
+	require.NoError(t, cfg.applyDNSConfig(matchOnly, nil))
+	names = ruleNamespaces(t, firstRule)
+	assert.NotContains(t, names, nrptCatchAllNamespace, "clearing RouteAll should drop the root namespace")
+
+	exists, err := registryKeyExists(dnsPolicyConfigExemptLocalPath)
+	require.NoError(t, err)
+	assert.False(t, exists, "exemption rule should go with the namespace it carves out of")
+
+	require.NoError(t, cfg.applyDNSConfig(primary, nil))
+	require.NoError(t, cfg.restoreHostDNS())
+	exists, err = registryKeyExists(firstRule)
+	require.NoError(t, err)
+	assert.False(t, exists, "restore should leave no rule behind")
+}
+
+// ruleNamespaces returns the namespaces an NRPT rule key claims.
+func ruleNamespaces(t *testing.T, path string) []string {
+	t.Helper()
+	k, err := registry.OpenKey(registry.LOCAL_MACHINE, path, registry.QUERY_VALUE)
+	require.NoError(t, err, "rule key %s should exist", path)
+	defer k.Close()
+
+	names, _, err := k.GetStringsValue(dnsPolicyConfigNameKey)
+	require.NoError(t, err)
+	return names
+}
+
+// TestNRPTCatchAllRuleLegacyEnv verifies that NB_USE_LEGACY_DNS_RESOLUTION
+// leaves the root namespace unclaimed, so no rule is written for a RouteAll
+// config that carries no match domains.
+func TestNRPTCatchAllRuleLegacyEnv(t *testing.T) {
+	if testing.Short() {
+		t.Skip("skipping registry integration test in short mode")
+	}
+
+	defer cleanupRegistryKeys(t)
+	cleanupRegistryKeys(t)
+
+	t.Setenv(envLegacyDNSResolution, "true")
+
+	testGUID := "{12345678-1234-1234-1234-123456789ABC}"
+	interfacePath := InterfaceConfigPath + `\` + testGUID
+	testKey, _, err := registry.CreateKey(registry.LOCAL_MACHINE, interfacePath, registry.SET_VALUE)
+	require.NoError(t, err, "Should create test interface registry key")
+	require.NoError(t, testKey.Close(), "close test interface registry key")
+	defer func() {
+		assert.NoError(t, registry.DeleteKey(registry.LOCAL_MACHINE, interfacePath), "delete test interface registry key")
+	}()
+
+	cfg := &registryConfigurator{guid: testGUID}
+	config := HostDNSConfig{
+		ServerIP: netip.MustParseAddr("100.64.0.1"),
+		RouteAll: true,
+	}
+
+	require.NoError(t, cfg.applyDNSConfig(config, nil))
+
+	// RouteAll with no match domains and the switch set leaves nothing to write.
+	exists, err := registryKeyExists(fmt.Sprintf("%s-0", dnsPolicyConfigMatchPath))
+	require.NoError(t, err)
+	assert.False(t, exists, "no rule should be written when the legacy env var is set")
+
+	exists, err = registryKeyExists(dnsPolicyConfigExemptLocalPath)
+	require.NoError(t, err)
+	assert.False(t, exists, "no exemption without a claimed root namespace")
+}
+
 func registryKeyExists(path string) (bool, error) {
 	k, err := registry.OpenKey(registry.LOCAL_MACHINE, path, registry.QUERY_VALUE)
 	if err != nil {

client/internal/dns_test.go

@@ -8,7 +8,9 @@ import (
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
 	nbdns "github.com/netbirdio/netbird/dns"
+	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
 )
 
 func TestCreatePTRRecord_IPv4(t *testing.T) {
@@ -136,3 +138,88 @@ func TestAddReverseZone_IPv6(t *testing.T) {
 	assert.Len(t, reverseZone.Records, 1)
 	assert.Equal(t, int(dns.TypePTR), reverseZone.Records[0].Type)
 }
+
+// TestToDNSConfig_ZoneFlagsPreserved pins the per-zone NonAuthoritative flag
+// through the legacy DNSConfig path. A non-authoritative zone is match-only:
+// the local resolver falls through to the upstream for an in-zone name it does
+// not define. The built-in peer zone is the authoritative one and must stay
+// that way, so the flag has to travel per zone rather than be derived.
+func TestToDNSConfig_ZoneFlagsPreserved(t *testing.T) {
+	config := toDNSConfig(&mgmProto.DNSConfig{
+		ServiceEnable: true,
+		CustomZones: []*mgmProto.CustomZone{
+			{
+				Domain: "netbird.cloud.",
+				Records: []*mgmProto.SimpleRecord{
+					{Name: "peer1.netbird.cloud.", Type: int64(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.1"},
+				},
+			},
+			{
+				Domain:               "corp.internal.",
+				NonAuthoritative:     true,
+				SearchDomainDisabled: true,
+				Records: []*mgmProto.SimpleRecord{
+					{Name: "db.corp.internal.", Type: int64(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.10.0.5"},
+				},
+			},
+		},
+	}, wgaddr.Address{
+		IP:      netip.MustParseAddr("100.64.0.1"),
+		Network: netip.MustParsePrefix("100.64.0.0/16"),
+	})
+
+	zones := make(map[string]nbdns.CustomZone, len(config.CustomZones))
+	for _, zone := range config.CustomZones {
+		zones[zone.Domain] = zone
+	}
+
+	peerZone, ok := zones["netbird.cloud."]
+	require.True(t, ok, "peer zone must survive")
+	assert.False(t, peerZone.NonAuthoritative, "the built-in peer zone owns the account domain and stays authoritative")
+
+	accountZone, ok := zones["corp.internal."]
+	require.True(t, ok, "account zone must survive")
+	assert.True(t, accountZone.NonAuthoritative, "an account zone stays match-only, else undefined in-zone names get black-holed")
+	assert.True(t, accountZone.SearchDomainDisabled)
+}
+
+// TestToDNSConfig_SingleZoneForcedAuthoritative pins the compatibility clause
+// in toDNSConfig: a config carrying exactly one zone is treated as
+// authoritative no matter what the server said, because servers that predate
+// the NonAuthoritative field send only the peer FQDN zone.
+//
+// The clause can only ever downgrade an explicit true to false, so a server
+// that legitimately sends a single non-authoritative zone — an account whose
+// only zone is a custom one, with no peer records to build the built-in zone
+// from — gets that zone's whole apex black-holed on the client. Real accounts
+// always carry the peer zone alongside, which is why this is latent. Narrowing
+// it needs a way to tell "unset" from "false" on the wire, or the account
+// domain passed down here; until then this test states the contract so a
+// change to it is deliberate.
+func TestToDNSConfig_SingleZoneForcedAuthoritative(t *testing.T) {
+	config := toDNSConfig(&mgmProto.DNSConfig{
+		ServiceEnable: true,
+		CustomZones: []*mgmProto.CustomZone{
+			{
+				Domain:           "corp.internal.",
+				NonAuthoritative: true,
+				Records: []*mgmProto.SimpleRecord{
+					{Name: "db.corp.internal.", Type: int64(dns.TypeA), Class: nbdns.DefaultClass, TTL: 300, RData: "10.10.0.5"},
+				},
+			},
+		},
+	}, wgaddr.Address{
+		IP:      netip.MustParseAddr("100.64.0.1"),
+		Network: netip.MustParsePrefix("100.64.0.0/16"),
+	})
+
+	require.NotEmpty(t, config.CustomZones)
+	assert.Equal(t, "corp.internal.", config.CustomZones[0].Domain)
+	assert.False(t, config.CustomZones[0].NonAuthoritative,
+		"a lone zone is forced authoritative for pre-NonAuthoritative servers")
+
+	// The reverse zone the config gains afterwards must not feed back into the
+	// decision: the compat gate counts the zones the server sent.
+	require.Len(t, config.CustomZones, 2, "a reverse zone is appended for the overlay prefix")
+	assert.Equal(t, "64.100.in-addr.arpa.", config.CustomZones[1].Domain)
+}

client/internal/engine.go

@@ -138,6 +138,7 @@ type EngineConfig struct {
 	RosenpassPermissive bool
 
 	ServerSSHAllowed              bool
+	RemoteJobsAllowed             bool
 	EnableSSHRoot                 *bool
 	EnableSSHSFTP                 *bool
 	EnableSSHLocalPortForwarding  *bool
@@ -1268,6 +1269,7 @@ func (e *Engine) applyInfoFlags(info *system.Info) {
 		e.config.EnableSSHLocalPortForwarding,
 		e.config.EnableSSHRemotePortForwarding,
 		e.config.DisableSSHAuth,
+		&e.config.RemoteJobsAllowed,
 	)
 }
 
@@ -1355,6 +1357,13 @@ func (e *Engine) receiveJobEvents() {
 				ID:     msg.ID,
 				Status: mgmProto.JobStatus_failed,
 			}
+			// Remote jobs are an explicit opt-in. When not enabled on this
+			// peer, every job is refused before any work is done.
+			if !e.config.RemoteJobsAllowed {
+				log.Warnf("refusing remote job: remote jobs are not enabled on this peer (enable with --allow-remote-jobs)")
+				resp.Reason = []byte("remote jobs are not enabled on this peer")
+				return &resp
+			}
 			switch params := msg.WorkloadParameters.(type) {
 			case *mgmProto.JobRequest_Bundle:
 				bundleResult, err := e.handleBundle(params.Bundle)
@@ -1384,7 +1393,25 @@ func (e *Engine) receiveJobEvents() {
 }
 
 func (e *Engine) handleBundle(params *mgmProto.BundleParameters) (*mgmProto.JobResponse_Bundle, error) {
-	log.Infof("handle remote debug bundle request: %s", params.String())
+	// The upload URL can carry a host, credentials, or query tokens, so it is
+	// kept out of the info-level line; the full parameters stay available at
+	// debug level for troubleshooting.
+	log.Infof("handle remote debug bundle request: anonymize=%v anonymize_level=%q log_file_count=%d bundle_for=%v bundle_for_time=%d",
+		params.GetAnonymize(), params.GetAnonymizeLevel(), params.GetLogFileCount(), params.GetBundleFor(), params.GetBundleForTime())
+	log.Debugf("remote debug bundle request parameters: %s", params.String())
+
+	// Resolve the upload destination: an MDM override, when set, takes
+	// precedence over the management-supplied URL. Both are validated the same
+	// way; an empty result falls back to the default upload server downstream.
+	uploadURL := params.GetUploadUrl()
+	if override := e.config.ProfileConfig.DebugBundleUploadURL; override != "" {
+		log.Infof("using MDM debug bundle upload URL override instead of the management-supplied value")
+		uploadURL = override
+	}
+	if err := validateBundleUploadURL(uploadURL); err != nil {
+		return nil, err
+	}
+
 	syncResponse, err := e.GetLatestSyncResponse()
 	if err != nil {
 		log.Warnf("get latest sync response: %v", err)
@@ -1412,7 +1439,7 @@ func (e *Engine) handleBundle(params *mgmProto.BundleParameters) (*mgmProto.JobR
 
 	waitFor := time.Duration(params.BundleForTime) * time.Minute
 
-	uploadKey, err := e.jobExecutor.BundleJob(e.ctx, bundleDeps, bundleJobParams, waitFor, e.config.ProfileConfig.ManagementURL.String())
+	uploadKey, err := e.jobExecutor.BundleJob(e.ctx, bundleDeps, bundleJobParams, waitFor, e.config.ProfileConfig.ManagementURL.String(), uploadURL)
 	if err != nil {
 		return nil, err
 	}
@@ -1425,6 +1452,16 @@ func (e *Engine) handleBundle(params *mgmProto.BundleParameters) (*mgmProto.JobR
 	return response, nil
 }
 
+// validateBundleUploadURL sanity-checks a management-supplied upload URL for a
+// remote debug bundle job. It delegates to profilemanager.ValidateBundleUploadURL
+// so the executor and the MDM policy override share one definition of the rule
+// (empty accepted; otherwise a well-formed https URL with a host) and cannot
+// drift. The host is deliberately left unconstrained pending a decision on
+// management-directed uploads.
+func validateBundleUploadURL(raw string) error {
+	return profilemanager.ValidateBundleUploadURL(raw)
+}
+
 // receiveManagementEvents connects to the Management Service event stream to receive updates from the management service
 // E.g. when a new peer has been registered and we are allowed to connect to it.
 func (e *Engine) receiveManagementEvents() {

client/internal/engine_bundle_test.go

@@ -0,0 +1,36 @@
+package internal
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// TestValidateBundleUploadURL covers the sanity check applied to a
+// management-supplied upload URL before a remote debug bundle is generated.
+func TestValidateBundleUploadURL(t *testing.T) {
+	for _, tc := range []struct {
+		name    string
+		raw     string
+		wantErr bool
+	}{
+		{name: "empty falls back to default", raw: ""},
+		{name: "https with host", raw: "https://upload.debug.netbird.io/upload"},
+		{name: "https self-hosted host", raw: "https://upload.example.com"},
+		{name: "plaintext rejected", raw: "http://upload.example.com", wantErr: true},
+		{name: "missing host rejected", raw: "https:///upload", wantErr: true},
+		{name: "port-only authority rejected", raw: "https://:443", wantErr: true},
+		{name: "non-url scheme rejected", raw: "ftp://upload.example.com", wantErr: true},
+		{name: "garbage rejected", raw: "://not a url", wantErr: true},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			err := validateBundleUploadURL(tc.raw)
+			if tc.wantErr {
+				require.Error(t, err, "an invalid upload URL must be rejected")
+				return
+			}
+			assert.NoError(t, err, "a valid or empty upload URL must be accepted")
+		})
+	}
+}

client/internal/engine_privileged_test.go

@@ -519,7 +519,7 @@ func startManagement(t *testing.T, dataDir, testFile string) (*grpc.Server, stri
 
 	updateManager := update_channel.NewPeersUpdateManager(metrics)
 	requestBuffer := server.NewAccountRequestBuffer(context.Background(), store)
-	networkMapController := controller.NewController(context.Background(), store, metrics, updateManager, requestBuffer, server.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersManager), config)
+	networkMapController := controller.NewController(context.Background(), store, metrics, updateManager, requestBuffer, server.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersManager), config, nil)
 	accountManager, err := server.BuildManager(context.Background(), config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, ia, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManager, false, cacheStore)
 	if err != nil {
 		return nil, "", err

client/internal/localmetrics/localmetrics.go

@@ -0,0 +1,274 @@
+// Package localmetrics exposes client connection state as a local
+// Prometheus /metrics endpoint.
+package localmetrics
+
+import (
+	"context"
+	"errors"
+	"net"
+	"net/http"
+	"net/netip"
+	"sync"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/promhttp"
+	dto "github.com/prometheus/client_model/go"
+	log "github.com/sirupsen/logrus"
+
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+// DefaultListenAddress is used when local metrics are enabled without an explicit address.
+const DefaultListenAddress = "127.0.0.1:9191"
+
+const (
+	shutdownTimeout   = 3 * time.Second
+	readHeaderTimeout = 5 * time.Second
+	readTimeout       = 10 * time.Second
+	writeTimeout      = 30 * time.Second
+	idleTimeout       = time.Minute
+)
+
+// statusSource provides the connection state snapshots the collector reads on scrape.
+type statusSource interface {
+	GetPeerStates() []peer.State
+	GetManagementState() peer.ManagementState
+	GetSignalState() peer.SignalState
+}
+
+// GathererProvider returns the current client metrics gatherer, or nil when
+// no engine is running. It is called on every scrape.
+type GathererProvider func() prometheus.Gatherer
+
+// Manager runs the local /metrics HTTP endpoint according to the active
+// client configuration. Reconcile is safe to call on every config change.
+type Manager struct {
+	status        statusSource
+	clientMetrics GathererProvider
+
+	mu   sync.Mutex
+	srv  *http.Server
+	addr string
+}
+
+// NewManager creates a manager that serves metrics from status and
+// clientMetrics and shuts down when ctx is canceled.
+func NewManager(ctx context.Context, status statusSource, clientMetrics GathererProvider) *Manager {
+	m := &Manager{status: status, clientMetrics: clientMetrics}
+	go func() {
+		<-ctx.Done()
+		m.Stop()
+	}()
+	return m
+}
+
+// Reconcile starts, stops, or restarts the metrics endpoint to match the
+// desired state. An empty addr falls back to DefaultListenAddress.
+func (m *Manager) Reconcile(enabled bool, addr string) {
+	if addr == "" {
+		addr = DefaultListenAddress
+	}
+	warnIfNotLoopback(addr)
+
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if !enabled {
+		m.stop()
+		return
+	}
+	if m.srv != nil && m.addr == addr {
+		return
+	}
+	m.stop()
+
+	registry := prometheus.NewRegistry()
+	registry.MustRegister(newCollector(m.status))
+
+	gatherers := prometheus.Gatherers{registry, prometheus.GathererFunc(func() ([]*dto.MetricFamily, error) {
+		if m.clientMetrics == nil {
+			return nil, nil
+		}
+		g := m.clientMetrics()
+		if g == nil {
+			return nil, nil
+		}
+		return g.Gather()
+	})}
+
+	mux := http.NewServeMux()
+	mux.Handle("/metrics", promhttp.HandlerFor(gatherers, promhttp.HandlerOpts{}))
+
+	srv := &http.Server{
+		Addr:              addr,
+		Handler:           mux,
+		ReadHeaderTimeout: readHeaderTimeout,
+		ReadTimeout:       readTimeout,
+		WriteTimeout:      writeTimeout,
+		IdleTimeout:       idleTimeout,
+	}
+	m.srv = srv
+	m.addr = addr
+
+	log.Infof("serving local metrics on http://%s/metrics", addr)
+	go func() {
+		if err := srv.ListenAndServe(); err != nil && !errors.Is(err, http.ErrServerClosed) {
+			log.Errorf("failed to serve local metrics on %s: %v", addr, err)
+			m.clear(srv)
+		}
+	}()
+}
+
+// clear drops the reference to srv so a later Reconcile with the same
+// address restarts it. A newer server may already have replaced it, in
+// which case the reference must stay.
+func (m *Manager) clear(srv *http.Server) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if m.srv != srv {
+		return
+	}
+	m.srv = nil
+	m.addr = ""
+}
+
+// Stop shuts down the metrics endpoint if it is running.
+func (m *Manager) Stop() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.stop()
+}
+
+// stop shuts down the running server. Callers must hold m.mu.
+func (m *Manager) stop() {
+	if m.srv == nil {
+		return
+	}
+	ctx, cancel := context.WithTimeout(context.Background(), shutdownTimeout)
+	defer cancel()
+	if err := m.srv.Shutdown(ctx); err != nil {
+		log.Debugf("failed to shut down local metrics server: %v", err)
+	}
+	m.srv = nil
+	m.addr = ""
+}
+
+// collector converts status recorder snapshots into Prometheus metrics at scrape time.
+type collector struct {
+	status statusSource
+
+	managementConnected *prometheus.Desc
+	signalConnected     *prometheus.Desc
+	peersTotal          *prometheus.Desc
+	peersConnected      *prometheus.Desc
+	peerLatency         *prometheus.Desc
+}
+
+func newCollector(status statusSource) *collector {
+	return &collector{
+		status: status,
+		managementConnected: prometheus.NewDesc(
+			"netbird_management_connected",
+			"Whether the client is connected to the management service (1 connected, 0 disconnected).",
+			nil, nil,
+		),
+		signalConnected: prometheus.NewDesc(
+			"netbird_signal_connected",
+			"Whether the client is connected to the signal service (1 connected, 0 disconnected).",
+			nil, nil,
+		),
+		peersTotal: prometheus.NewDesc(
+			"netbird_peers",
+			"Number of peers known to this client.",
+			nil, nil,
+		),
+		peersConnected: prometheus.NewDesc(
+			"netbird_peers_connected",
+			"Number of connected peers by connection type.",
+			[]string{"connection_type"}, nil,
+		),
+		peerLatency: prometheus.NewDesc(
+			"netbird_peer_latency_seconds",
+			"Round-trip latency per directly connected peer; relayed connections have no latency measurement.",
+			[]string{"peer"}, nil,
+		),
+	}
+}
+
+// Describe implements prometheus.Collector.
+func (c *collector) Describe(ch chan<- *prometheus.Desc) {
+	ch <- c.managementConnected
+	ch <- c.signalConnected
+	ch <- c.peersTotal
+	ch <- c.peersConnected
+	ch <- c.peerLatency
+}
+
+// Collect implements prometheus.Collector.
+func (c *collector) Collect(ch chan<- prometheus.Metric) {
+	ch <- prometheus.MustNewConstMetric(c.managementConnected, prometheus.GaugeValue, boolToFloat(c.status.GetManagementState().Connected))
+	ch <- prometheus.MustNewConstMetric(c.signalConnected, prometheus.GaugeValue, boolToFloat(c.status.GetSignalState().Connected))
+
+	peers := c.status.GetPeerStates()
+	ch <- prometheus.MustNewConstMetric(c.peersTotal, prometheus.GaugeValue, float64(len(peers)))
+
+	var p2p, relayed float64
+	for _, p := range peers {
+		if p.ConnStatus != peer.StatusConnected {
+			continue
+		}
+		if p.Relayed {
+			relayed++
+			continue
+		}
+		p2p++
+
+		if latency := p.Latency.Seconds(); latency > 0 {
+			ch <- prometheus.MustNewConstMetric(c.peerLatency, prometheus.GaugeValue, latency, p.FQDN)
+		}
+	}
+	ch <- prometheus.MustNewConstMetric(c.peersConnected, prometheus.GaugeValue, p2p, "p2p")
+	ch <- prometheus.MustNewConstMetric(c.peersConnected, prometheus.GaugeValue, relayed, "relay")
+}
+
+func boolToFloat(b bool) float64 {
+	if b {
+		return 1
+	}
+	return 0
+}
+
+// IsLoopback reports whether addr binds the endpoint to the local host only.
+// An empty address means DefaultListenAddress. It fails closed: an address
+// that cannot be confirmed loopback, including an unparseable one, is not.
+func IsLoopback(addr string) bool {
+	if addr == "" {
+		addr = DefaultListenAddress
+	}
+
+	host, _, err := net.SplitHostPort(addr)
+	if err != nil {
+		return false
+	}
+	if host == "localhost" {
+		return true
+	}
+
+	ip, err := netip.ParseAddr(host)
+	if err != nil {
+		return false
+	}
+	return ip.Unmap().IsLoopback()
+}
+
+// warnIfNotLoopback logs a warning when the listen address cannot be
+// confirmed to be local-only, since the endpoint exposes peer and
+// connectivity details without authentication.
+func warnIfNotLoopback(addr string) {
+	if IsLoopback(addr) {
+		return
+	}
+	log.Warnf("local metrics endpoint listens on non-loopback address %s and is reachable from the network without authentication", addr)
+}

client/internal/localmetrics/localmetrics_test.go

@@ -0,0 +1,151 @@
+package localmetrics
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"net"
+	"net/http"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus/testutil"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+type stubStatus struct {
+	peers      []peer.State
+	management peer.ManagementState
+	signal     peer.SignalState
+}
+
+func (s *stubStatus) GetPeerStates() []peer.State              { return s.peers }
+func (s *stubStatus) GetManagementState() peer.ManagementState { return s.management }
+func (s *stubStatus) GetSignalState() peer.SignalState         { return s.signal }
+
+func testStatus() *stubStatus {
+	return &stubStatus{
+		management: peer.ManagementState{Connected: true},
+		signal:     peer.SignalState{Connected: true},
+		peers: []peer.State{
+			{FQDN: "peer-a.netbird.cloud", IP: "100.90.0.1", ConnStatus: peer.StatusConnected, Relayed: false, Latency: 12 * time.Millisecond},
+			{FQDN: "peer-b.netbird.cloud", IP: "100.90.0.2", ConnStatus: peer.StatusConnected, Relayed: false, Latency: 36 * time.Millisecond},
+			{FQDN: "peer-c.netbird.cloud", IP: "100.90.0.3", ConnStatus: peer.StatusConnected, Relayed: true},
+			{FQDN: "peer-d.netbird.cloud", IP: "100.90.0.4", ConnStatus: peer.StatusIdle},
+		},
+	}
+}
+
+func TestCollector(t *testing.T) {
+	c := newCollector(testStatus())
+
+	expected := `
+# HELP netbird_management_connected Whether the client is connected to the management service (1 connected, 0 disconnected).
+# TYPE netbird_management_connected gauge
+netbird_management_connected 1
+# HELP netbird_peer_latency_seconds Round-trip latency per directly connected peer; relayed connections have no latency measurement.
+# TYPE netbird_peer_latency_seconds gauge
+netbird_peer_latency_seconds{peer="peer-a.netbird.cloud"} 0.012
+netbird_peer_latency_seconds{peer="peer-b.netbird.cloud"} 0.036
+# HELP netbird_peers Number of peers known to this client.
+# TYPE netbird_peers gauge
+netbird_peers 4
+# HELP netbird_peers_connected Number of connected peers by connection type.
+# TYPE netbird_peers_connected gauge
+netbird_peers_connected{connection_type="p2p"} 2
+netbird_peers_connected{connection_type="relay"} 1
+# HELP netbird_signal_connected Whether the client is connected to the signal service (1 connected, 0 disconnected).
+# TYPE netbird_signal_connected gauge
+netbird_signal_connected 1
+`
+	require.NoError(t, testutil.CollectAndCompare(c, strings.NewReader(expected)))
+}
+
+func TestServe(t *testing.T) {
+	ln, err := net.Listen("tcp", "127.0.0.1:0")
+	require.NoError(t, err, "must find a free port")
+	addr := ln.Addr().String()
+	require.NoError(t, ln.Close())
+
+	ctx, cancel := context.WithCancel(context.Background())
+	t.Cleanup(cancel)
+	m := NewManager(ctx, testStatus(), nil)
+	m.Reconcile(true, addr)
+
+	var body string
+	require.Eventually(t, func() bool {
+		resp, err := http.Get(fmt.Sprintf("http://%s/metrics", addr))
+		if err != nil {
+			return false
+		}
+		defer resp.Body.Close()
+		data, err := io.ReadAll(resp.Body)
+		if err != nil || resp.StatusCode != http.StatusOK {
+			return false
+		}
+		body = string(data)
+		return true
+	}, 2*time.Second, 50*time.Millisecond, "metrics endpoint should come up")
+
+	assert.Contains(t, body, "netbird_peers 4")
+	assert.Contains(t, body, `netbird_peers_connected{connection_type="relay"} 1`)
+	assert.Contains(t, body, `netbird_peer_latency_seconds{peer="peer-a.netbird.cloud"} 0.012`)
+}
+
+// A server that never came up must not be remembered, otherwise reconciling the
+// same address again is a no-op and the endpoint never recovers.
+func TestReconcileForgetsAFailedServer(t *testing.T) {
+	blocker, err := net.Listen("tcp", "127.0.0.1:0")
+	require.NoError(t, err, "must find a free port")
+	t.Cleanup(func() { _ = blocker.Close() })
+	addr := blocker.Addr().String()
+
+	ctx, cancel := context.WithCancel(context.Background())
+	t.Cleanup(cancel)
+	m := NewManager(ctx, testStatus(), nil)
+	m.Reconcile(true, addr)
+
+	require.Eventually(t, func() bool {
+		m.mu.Lock()
+		defer m.mu.Unlock()
+		return m.srv == nil && m.addr == ""
+	}, 2*time.Second, 20*time.Millisecond, "the failed server should be dropped")
+
+	require.NoError(t, blocker.Close())
+	m.Reconcile(true, addr)
+
+	require.Eventually(t, func() bool {
+		resp, err := http.Get(fmt.Sprintf("http://%s/metrics", addr))
+		if err != nil {
+			return false
+		}
+		defer resp.Body.Close()
+		return resp.StatusCode == http.StatusOK
+	}, 2*time.Second, 50*time.Millisecond, "reconciling the same address should retry the bind")
+}
+
+func TestIsLoopback(t *testing.T) {
+	tests := map[string]bool{
+		"":                        true,
+		"127.0.0.1:9191":          true,
+		"127.9.9.9:9191":          true,
+		"[::1]:9191":              true,
+		"[::ffff:127.0.0.1]:9191": true,
+		"localhost:9191":          true,
+		"0.0.0.0:9191":            false,
+		"[::]:9191":               false,
+		"192.168.1.10:9191":       false,
+		"not-an-address":          false,
+		"example.com:9191":        false,
+	}
+
+	for addr, want := range tests {
+		t.Run(addr, func(t *testing.T) {
+			assert.Equal(t, want, IsLoopback(addr), "loopback verdict for %q", addr)
+		})
+	}
+}

client/internal/metrics/influxdb.go

@@ -45,30 +45,13 @@ func (m *influxDBMetrics) RecordConnectionStages(
 	isReconnection bool,
 	timestamps ConnectionStageTimestamps,
 ) {
-	var signalingReceivedToConnection, connectionToWgHandshake, totalDuration float64
-
-	if !timestamps.SignalingReceived.IsZero() && !timestamps.ConnectionReady.IsZero() {
-		signalingReceivedToConnection = timestamps.ConnectionReady.Sub(timestamps.SignalingReceived).Seconds()
-	}
-
-	if !timestamps.ConnectionReady.IsZero() && !timestamps.WgHandshakeSuccess.IsZero() {
-		connectionToWgHandshake = timestamps.WgHandshakeSuccess.Sub(timestamps.ConnectionReady).Seconds()
-	}
-
-	if !timestamps.SignalingReceived.IsZero() && !timestamps.WgHandshakeSuccess.IsZero() {
-		totalDuration = timestamps.WgHandshakeSuccess.Sub(timestamps.SignalingReceived).Seconds()
-	}
-
-	attemptType := "initial"
-	if isReconnection {
-		attemptType = "reconnection"
-	}
+	signalingReceivedToConnection, connectionToWgHandshake, totalDuration := timestamps.Durations()
 
 	connTypeStr := connectionType.String()
 	tags := fmt.Sprintf("deployment_type=%s,connection_type=%s,attempt_type=%s,version=%s,os=%s,arch=%s,peer_id=%s,connection_pair_id=%s",
 		agentInfo.DeploymentType.String(),
 		connTypeStr,
-		attemptType,
+		attemptType(isReconnection),
 		agentInfo.Version,
 		agentInfo.OS,
 		agentInfo.Arch,
@@ -94,7 +77,7 @@ func (m *influxDBMetrics) RecordConnectionStages(
 	m.trimLocked()
 
 	log.Tracef("peer connection metrics [%s, %s, %s]: signalingReceived→connection: %.3fs, connection→wg_handshake: %.3fs, total: %.3fs",
-		agentInfo.DeploymentType.String(), connTypeStr, attemptType, signalingReceivedToConnection, connectionToWgHandshake, totalDuration)
+		agentInfo.DeploymentType.String(), connTypeStr, attemptType(isReconnection), signalingReceivedToConnection, connectionToWgHandshake, totalDuration)
 }
 
 func (m *influxDBMetrics) RecordSyncDuration(_ context.Context, agentInfo AgentInfo, duration time.Duration) {

client/internal/metrics/metrics.go

@@ -89,6 +89,21 @@ type ConnectionStageTimestamps struct {
 	WgHandshakeSuccess time.Time
 }
 
+// Durations returns the stage durations in seconds. A duration is zero when
+// either of its timestamps is missing.
+func (c ConnectionStageTimestamps) Durations() (signalingToConnection, connectionToWgHandshake, total float64) {
+	if !c.SignalingReceived.IsZero() && !c.ConnectionReady.IsZero() {
+		signalingToConnection = c.ConnectionReady.Sub(c.SignalingReceived).Seconds()
+	}
+	if !c.ConnectionReady.IsZero() && !c.WgHandshakeSuccess.IsZero() {
+		connectionToWgHandshake = c.WgHandshakeSuccess.Sub(c.ConnectionReady).Seconds()
+	}
+	if !c.SignalingReceived.IsZero() && !c.WgHandshakeSuccess.IsZero() {
+		total = c.WgHandshakeSuccess.Sub(c.SignalingReceived).Seconds()
+	}
+	return signalingToConnection, connectionToWgHandshake, total
+}
+
 // String returns a human-readable representation of the connection stage timestamps
 func (c ConnectionStageTimestamps) String() string {
 	return fmt.Sprintf("ConnectionStageTimestamps{SignalingReceived=%v, ConnectionReady=%v, WgHandshakeSuccess=%v}",
@@ -279,3 +294,11 @@ func (c *ClientMetrics) stopPushLocked() {
 	c.wg.Wait()
 	c.push.Store(nil)
 }
+
+// attemptType returns the metric label for an initial vs reconnection attempt.
+func attemptType(isReconnection bool) string {
+	if isReconnection {
+		return "reconnection"
+	}
+	return "initial"
+}

client/internal/metrics/metrics_default.go

@@ -2,10 +2,24 @@
 
 package metrics
 
+import "github.com/prometheus/client_golang/prometheus"
+
 // NewClientMetrics creates a new ClientMetrics instance
 func NewClientMetrics(agentInfo AgentInfo) *ClientMetrics {
 	return &ClientMetrics{
-		impl:      newInfluxDBMetrics(),
+		impl:      newPrometheusMetrics(newInfluxDBMetrics()),
 		agentInfo: agentInfo,
 	}
 }
+
+// PrometheusGatherer returns the registry with the mirrored Prometheus
+// metrics, or nil when unavailable.
+func (c *ClientMetrics) PrometheusGatherer() prometheus.Gatherer {
+	if c == nil {
+		return nil
+	}
+	if pm, ok := c.impl.(*prometheusMetrics); ok {
+		return pm.Gatherer()
+	}
+	return nil
+}

client/internal/metrics/prometheus.go

@@ -0,0 +1,119 @@
+//go:build !js
+
+package metrics
+
+import (
+	"context"
+	"io"
+	"strconv"
+	"time"
+
+	"github.com/prometheus/client_golang/prometheus"
+)
+
+// prometheusMetrics mirrors recorded client metrics into a Prometheus
+// registry for the local /metrics endpoint, then delegates to the wrapped
+// implementation. Export and Reset pass through untouched: Prometheus
+// metrics are cumulative and pull-based.
+type prometheusMetrics struct {
+	next     metricsImplementation
+	registry *prometheus.Registry
+
+	connectionStages  *prometheus.HistogramVec
+	syncDuration      prometheus.Histogram
+	syncPhaseDuration *prometheus.HistogramVec
+	loginDuration     *prometheus.HistogramVec
+}
+
+func newPrometheusMetrics(next metricsImplementation) *prometheusMetrics {
+	connectionBuckets := []float64{.05, .1, .25, .5, 1, 2.5, 5, 10, 30, 60}
+
+	m := &prometheusMetrics{
+		next:     next,
+		registry: prometheus.NewRegistry(),
+		connectionStages: prometheus.NewHistogramVec(prometheus.HistogramOpts{
+			Name:    "netbird_peer_connection_stage_duration_seconds",
+			Help:    "Duration of peer connection establishment stages.",
+			Buckets: connectionBuckets,
+		}, []string{"stage", "connection_type", "attempt_type"}),
+		syncDuration: prometheus.NewHistogram(prometheus.HistogramOpts{
+			Name:    "netbird_sync_duration_seconds",
+			Help:    "Duration of management sync message processing.",
+			Buckets: prometheus.DefBuckets,
+		}),
+		syncPhaseDuration: prometheus.NewHistogramVec(prometheus.HistogramOpts{
+			Name:    "netbird_sync_phase_duration_seconds",
+			Help:    "Duration of individual sync processing phases.",
+			Buckets: prometheus.DefBuckets,
+		}, []string{"phase"}),
+		loginDuration: prometheus.NewHistogramVec(prometheus.HistogramOpts{
+			Name:    "netbird_login_duration_seconds",
+			Help:    "Duration of logins to the management service.",
+			Buckets: prometheus.DefBuckets,
+		}, []string{"success"}),
+	}
+
+	m.registry.MustRegister(m.connectionStages, m.syncDuration, m.syncPhaseDuration, m.loginDuration)
+	return m
+}
+
+// Gatherer returns the registry holding the mirrored metrics.
+func (m *prometheusMetrics) Gatherer() prometheus.Gatherer {
+	return m.registry
+}
+
+// RecordConnectionStages implements metricsImplementation.
+func (m *prometheusMetrics) RecordConnectionStages(
+	ctx context.Context,
+	agentInfo AgentInfo,
+	connectionPairID string,
+	connectionType ConnectionType,
+	isReconnection bool,
+	timestamps ConnectionStageTimestamps,
+) {
+	attempt := attemptType(isReconnection)
+	connType := connectionType.String()
+
+	signalingToConnection, connectionToWgHandshake, total := timestamps.Durations()
+	if signalingToConnection > 0 {
+		m.connectionStages.WithLabelValues("signaling_to_connection", connType, attempt).Observe(signalingToConnection)
+	}
+	if connectionToWgHandshake > 0 {
+		m.connectionStages.WithLabelValues("connection_to_wg_handshake", connType, attempt).Observe(connectionToWgHandshake)
+	}
+	if total > 0 {
+		m.connectionStages.WithLabelValues("total", connType, attempt).Observe(total)
+	}
+
+	m.next.RecordConnectionStages(ctx, agentInfo, connectionPairID, connectionType, isReconnection, timestamps)
+}
+
+// RecordSyncDuration implements metricsImplementation.
+func (m *prometheusMetrics) RecordSyncDuration(ctx context.Context, agentInfo AgentInfo, duration time.Duration) {
+	m.syncDuration.Observe(duration.Seconds())
+	m.next.RecordSyncDuration(ctx, agentInfo, duration)
+}
+
+// RecordSyncPhase implements metricsImplementation.
+func (m *prometheusMetrics) RecordSyncPhase(ctx context.Context, agentInfo AgentInfo, phase string, duration time.Duration) {
+	m.syncPhaseDuration.WithLabelValues(phase).Observe(duration.Seconds())
+	m.next.RecordSyncPhase(ctx, agentInfo, phase, duration)
+}
+
+// RecordLoginDuration implements metricsImplementation.
+func (m *prometheusMetrics) RecordLoginDuration(ctx context.Context, agentInfo AgentInfo, duration time.Duration, success bool) {
+	m.loginDuration.WithLabelValues(strconv.FormatBool(success)).Observe(duration.Seconds())
+	m.next.RecordLoginDuration(ctx, agentInfo, duration, success)
+}
+
+// Export implements metricsImplementation by delegating to the wrapped
+// implementation; Prometheus metrics are pulled via the registry instead.
+func (m *prometheusMetrics) Export(w io.Writer) error {
+	return m.next.Export(w)
+}
+
+// Reset implements metricsImplementation by delegating to the wrapped
+// implementation; Prometheus metrics must not be cleared on push.
+func (m *prometheusMetrics) Reset() {
+	m.next.Reset()
+}

client/internal/peer/status.go

@@ -1167,6 +1167,18 @@ func (d *Status) GetResolvedDomainsStates() map[domain.Domain]ResolvedDomainInfo
 	return maps.Clone(d.resolvedDomainsStates)
 }
 
+// GetPeerStates returns a snapshot of all known peer states, including offline peers.
+func (d *Status) GetPeerStates() []State {
+	d.mux.RLock()
+	defer d.mux.RUnlock()
+
+	states := make([]State, 0, d.numOfPeers())
+	for _, state := range d.peers {
+		states = append(states, state)
+	}
+	return append(states, d.offlinePeers...)
+}
+
 // GetFullStatus gets full status
 func (d *Status) GetFullStatus() FullStatus {
 	fullStatus := FullStatus{

client/internal/peer/status_test.go

@@ -129,6 +129,28 @@ func TestStatus_PeerStateByIP_RemovedPeer(t *testing.T) {
 	req.False(ok, "removed peer must not resolve by IPv6 tunnel address")
 }
 
+// TestStatus_GetPeerStates_IncludesOfflinePeers keeps the snapshot in line with
+// GetFullStatus: offline peers are known peers, so a consumer counting peers
+// must see the same total the status command reports.
+func TestStatus_GetPeerStates_IncludesOfflinePeers(t *testing.T) {
+	status := NewRecorder("https://mgm")
+	req := require.New(t)
+
+	req.NoError(status.AddPeer("pk-online", "online.netbird", "100.64.0.10", "fd00::1"))
+	status.ReplaceOfflinePeers([]State{
+		{PubKey: "pk-offline", FQDN: "offline.netbird", IP: "100.64.0.20", ConnStatus: StatusIdle},
+	})
+
+	states := status.GetPeerStates()
+	req.Len(states, 2, "snapshot must carry both the online and the offline peer")
+
+	keys := make([]string, 0, len(states))
+	for _, s := range states {
+		keys = append(keys, s.PubKey)
+	}
+	req.ElementsMatch([]string{"pk-online", "pk-offline"}, keys, "snapshot must carry both peers")
+}
+
 func TestStatus_UpdatePeerFQDN(t *testing.T) {
 	key := "abc"
 	fqdn := "peer-a.netbird.local"

client/internal/peer/worker_ice.go

@@ -64,6 +64,9 @@ type WorkerICE struct {
 
 	// portForwardAttempted tracks if we've already tried port forwarding this session
 	portForwardAttempted bool
+
+	// dialFunc, when non-nil, replaces agentDial in connect(). Only for tests.
+	dialFunc func(ctx context.Context, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *OfferAnswer) (net.Conn, error)
 }
 
 func NewWorkerICE(ctx context.Context, log *log.Entry, config ConnConfig, conn *Conn, signaler *Signaler, ifaceDiscover stdnet.ExternalIFaceDiscover, statusRecorder *Status, hasRelayOnLocally bool) (*WorkerICE, error) {
@@ -123,7 +126,7 @@ func (w *WorkerICE) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
 			w.log.Errorf("failed to create new session ID: %s", err)
 		}
 		w.sessionID = sessionID
-		w.agent = nil
+		w.abandonNegotiation()
 	}
 
 	var preferredCandidateTypes []ice.CandidateType
@@ -151,7 +154,9 @@ func (w *WorkerICE) OnNewOffer(remoteOfferAnswer *OfferAnswer) {
 		w.remoteSessionID = ""
 	}
 
-	go w.connect(dialerCtx, agent, remoteOfferAnswer)
+	// Capture the cancel func at spawn time: connect reads it from the argument
+	// instead of the field, which a newer OnNewOffer may already have replaced.
+	go w.connect(dialerCtx, dialerCancel, agent, remoteOfferAnswer)
 }
 
 // OnRemoteCandidate Handles ICE connection Candidate provided by the remote peer.
@@ -200,16 +205,16 @@ func (w *WorkerICE) Close() {
 	w.muxAgent.Lock()
 	defer w.muxAgent.Unlock()
 
-	if w.agent == nil {
-		return
+	if w.agent != nil {
+		w.agentDialerCancel()
+		if err := w.agent.Close(); err != nil {
+			w.log.Warnf("failed to close ICE agent: %s", err)
+		}
 	}
-
-	w.agentDialerCancel()
-	if err := w.agent.Close(); err != nil {
-		w.log.Warnf("failed to close ICE agent: %s", err)
-	}
-
-	w.agent = nil
+	// Unconditional: a dial goroutine racing this Close skips its own cleanup
+	// (closeAgent finds a nil agent), so the flags must be dropped here too or
+	// the reconnection guard reads the stale state as Connected forever.
+	w.abandonNegotiation()
 }
 
 func (w *WorkerICE) reCreateAgent(dialerCancel context.CancelFunc, candidates []ice.CandidateType) (*icemaker.ThreadSafeAgent, error) {
@@ -247,31 +252,52 @@ func (w *WorkerICE) SessionID() ICESessionID {
 // will block until connection succeeded
 // but it won't release if ICE Agent went into Disconnected or Failed state,
 // so we have to cancel it with the provided context once agent detected a broken connection
-func (w *WorkerICE) connect(ctx context.Context, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *OfferAnswer) {
+func (w *WorkerICE) connect(ctx context.Context, dialerCancel context.CancelFunc, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *OfferAnswer) {
 	w.log.Debugf("gather candidates")
 	if err := agent.GatherCandidates(); err != nil {
 		w.log.Warnf("failed to gather candidates: %s", err)
-		w.closeAgent(agent, w.agentDialerCancel)
+		w.closeAgent(agent, dialerCancel)
 		return
 	}
 
 	w.log.Debugf("agent dial")
-	remoteConn, err := w.agentDial(ctx, agent, remoteOfferAnswer)
+	dial := func(ctx context.Context, agent *icemaker.ThreadSafeAgent, remoteOfferAnswer *OfferAnswer) (net.Conn, error) {
+		return w.agentDial(ctx, agent, remoteOfferAnswer)
+	}
+	if w.dialFunc != nil {
+		dial = w.dialFunc
+	}
+	remoteConn, err := dial(ctx, agent, remoteOfferAnswer)
 	if err != nil {
 		w.log.Debugf("failed to dial the remote peer: %s", err)
-		w.closeAgent(agent, w.agentDialerCancel)
+		w.closeAgent(agent, dialerCancel)
 		return
 	}
 	w.log.Debugf("agent dial succeeded")
 
+	// A newer negotiation may have replaced our agent while agentDial was
+	// blocked. Drop the dead connection before running pair retrieval, port
+	// punching or candidate work against a closed agent. The commit-point
+	// check below still guards a replacement arriving after this point.
+	w.muxAgent.Lock()
+	stale := w.agent != agent
+	w.muxAgent.Unlock()
+	if stale {
+		if err := remoteConn.Close(); err != nil {
+			w.log.Warnf("failed to close stale ICE connection: %s", err)
+		}
+		w.log.Warnf("discarding connection from a stale ICE negotiation")
+		return
+	}
+
 	pair, err := agent.GetSelectedCandidatePair()
 	if err != nil {
-		w.closeAgent(agent, w.agentDialerCancel)
+		w.closeAgent(agent, dialerCancel)
 		return
 	}
 	if pair == nil {
 		w.log.Warnf("selected candidate pair is nil, cannot proceed")
-		w.closeAgent(agent, w.agentDialerCancel)
+		w.closeAgent(agent, dialerCancel)
 		return
 	}
 
@@ -301,11 +327,27 @@ func (w *WorkerICE) connect(ctx context.Context, agent *icemaker.ThreadSafeAgent
 
 	w.log.Infof("connection succeeded with offer session: %s", remoteOfferAnswer.SessionIDString())
 	w.muxAgent.Lock()
+	// Authoritative ownership guard: a negotiation that lost w.agent to a newer
+	// one between the post-dial check and the commit must not clear agentConnecting,
+	// record lastSuccess or report the connection, so the state commit has to be
+	// atomic with the check.
+	if w.agent != agent {
+		w.muxAgent.Unlock()
+		if err := remoteConn.Close(); err != nil {
+			w.log.Warnf("failed to close stale ICE connection: %s", err)
+		}
+		w.log.Warnf("discarding connection from a stale ICE negotiation")
+		return
+	}
 	w.agentConnecting = false
 	w.lastSuccess = time.Now()
 	w.muxAgent.Unlock()
 
 	// todo: the potential problem is a race between the onConnectionStateChange
+	// and the delivery below: after this unlock, a newer offer can replace
+	// w.agent before onICEConnectionIsReady runs, delivering this (now stale)
+	// connection. The newer negotiation overwrites it with its own delivery,
+	// so the window only ever downgrades an endpoint transiently.
 	w.conn.onICEConnectionIsReady(selectedPriority(pair), ci)
 }
 
@@ -321,20 +363,32 @@ func (w *WorkerICE) closeAgent(agent *icemaker.ThreadSafeAgent, cancel context.C
 	sessionChanged := w.remoteSessionChanged
 	w.remoteSessionChanged = false
 
+	// Only the owner of the current session may reset its state: a stale dial
+	// goroutine waking after a newer attempt must not clobber it.
 	if w.agent == agent {
-		// consider to remove from here and move to the OnNewOffer
 		sessionID, err := NewICESessionID()
 		if err != nil {
 			w.log.Errorf("failed to create new session ID: %s", err)
 		}
 		w.sessionID = sessionID
-		w.agent = nil
-		w.agentConnecting = false
-		w.remoteSessionID = ""
+		w.abandonNegotiation()
 	}
 	return sessionChanged
 }
 
+// abandonNegotiation drops all recorded ICE session state so the worker treats the
+// next offer as a fresh start instead of a duplicate of a dead negotiation. The
+// agent and agentConnecting flags must change together: leaving one stale wedges
+// the reconnection guard into reporting Connected forever. It neither cancels an
+// in-flight dial nor closes an agent — callers dispose of those themselves first,
+// so a stale goroutine can never cancel another session's dial through this path.
+// Caller must hold muxAgent.
+func (w *WorkerICE) abandonNegotiation() {
+	w.agent = nil
+	w.agentConnecting = false
+	w.remoteSessionID = ""
+}
+
 func (w *WorkerICE) punchRemoteWGPort(pair *ice.CandidatePair, remoteWgPort int) {
 	// wait local endpoint configuration
 	time.Sleep(time.Second)

client/internal/peer/worker_ice_close_test.go

@@ -0,0 +1,257 @@
+package peer
+
+import (
+	"context"
+	"net"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	log "github.com/sirupsen/logrus"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+
+	icemaker "github.com/netbirdio/netbird/client/internal/peer/ice"
+	signal "github.com/netbirdio/netbird/shared/signal/client"
+	sProto "github.com/netbirdio/netbird/shared/signal/proto"
+)
+
+// stubSignalClient satisfies signal.Client as a no-op so the candidate
+// goroutine spawned by a real GatherCandidates never dereferences a nil
+// signaler in tests.
+type stubSignalClient struct{}
+
+func (stubSignalClient) Close() error                                               { return nil }
+func (stubSignalClient) StreamConnected() bool                                      { return false }
+func (stubSignalClient) GetStatus() signal.Status                                   { return signal.StreamDisconnected }
+func (stubSignalClient) Receive(context.Context, func(*sProto.Message) error) error { return nil }
+func (stubSignalClient) Ready() bool                                                { return false }
+func (stubSignalClient) IsHealthy() bool                                            { return false }
+func (stubSignalClient) WaitStreamConnected(context.Context)                        {}
+func (stubSignalClient) SendToStream(*sProto.EncryptedMessage) error                { return nil }
+func (stubSignalClient) Send(*sProto.Message) error                                 { return nil }
+func (stubSignalClient) SetOnReconnectedListener(func())                            {}
+
+// newTestWorkerICE builds a worker with real pion plumbing and no-op signaling.
+func newTestWorkerICE(t *testing.T) *WorkerICE {
+	t.Helper()
+
+	config := connConf
+	stunTurn := &icemaker.StunTurn{}
+	stunTurn.Store(nil)
+	config.ICEConfig.StunTurn = stunTurn
+
+	w, err := NewWorkerICE(context.Background(), log.WithField("test", t.Name()), config, nil,
+		NewSignaler(stubSignalClient{}, wgtypes.Key{}), nil, nil, false)
+	require.NoError(t, err, "worker setup must succeed")
+	return w
+}
+
+// TestWorkerICE_CloseDuringDial_ClearsConnectingFlag drives the teardown race
+// through the real dial goroutine instead of simulating its cleanup.
+//
+// The real-world sequence this models:
+//  1. OnNewOffer starts a negotiation: agent set, agentConnecting = true,
+//     go connect()
+//  2. The network dies and connect() stays blocked inside GatherCandidates/Dial
+//  3. A WG handshake timeout calls Close(): the agent is released and the dial
+//     context cancelled, but agentConnecting is not reset
+//  4. The real goroutine wakes with an error and runs its own cleanup
+//     (closeAgent), where `w.agent == agent` is now false, so the flag reset
+//     is skipped
+//
+// There is no remote responder, so Dial can never succeed: whatever point the
+// goroutine is at, closing first forces it down the error path. Before the fix
+// the flag stays true forever and the deadline below expires.
+func TestWorkerICE_CloseDuringDial_ClearsConnectingFlag(t *testing.T) {
+	w := newTestWorkerICE(t)
+
+	sid := ICESessionID("test-session-id")
+	w.OnNewOffer(&OfferAnswer{
+		IceCredentials: IceCredentials{
+			UFrag: "testufrag",
+			Pwd:   "testpwdtestpwdtestpwd12",
+		},
+		SessionID: &sid,
+	})
+	require.True(t, w.InProgress(), "OnNewOffer must mark the negotiation as in progress")
+
+	// Teardown wins the race while connect() is still running.
+	w.Close()
+
+	// Close drops the flags synchronously, so the assertion below does not
+	// converge on the goroutine: the deadline only absorbs the dial goroutine
+	// waking up in the background, proving nothing re-wedges it afterwards.
+	require.Eventually(t, func() bool {
+		return !w.InProgress()
+	}, 10*time.Second, 50*time.Millisecond,
+		"Close must leave the negotiation idle even while the dial goroutine is still winding down")
+
+	// abandonNegotiation owns these three fields together; the worker is idle
+	// only when all of them are dropped.
+	w.muxAgent.Lock()
+	defer w.muxAgent.Unlock()
+	assert.Nil(t, w.agent, "no agent may survive the teardown")
+	assert.False(t, w.agentConnecting, "the connecting flag must match the nil agent")
+	assert.Empty(t, w.remoteSessionID, "a dead session's remote ID must not linger")
+}
+
+// TestWorkerICE_CloseClearsResidualConnectingState covers Close on a worker whose
+// agent is already gone but whose flag is stuck on true, e.g. after an aborted
+// recreate in OnNewOffer or after a first Close raced a dial goroutine.
+func TestWorkerICE_CloseClearsResidualConnectingState(t *testing.T) {
+	w := newTestWorkerICE(t)
+
+	w.muxAgent.Lock()
+	w.agentConnecting = true
+	w.muxAgent.Unlock()
+
+	w.Close()
+
+	assert.False(t, w.InProgress(), "Close must drop residual connecting state even without a live agent")
+
+	w.muxAgent.Lock()
+	defer w.muxAgent.Unlock()
+	assert.Nil(t, w.agent)
+	assert.False(t, w.agentConnecting)
+	assert.Empty(t, w.remoteSessionID)
+}
+
+// TestWorkerICE_StaleCloseAgentKeepsCurrentSession pins the ownership guard in
+// closeAgent: a late-waking dial goroutine from an older session must not reset
+// the state of a newer negotiation that reused the worker. The newer session
+// must survive wholesale - agent, flag and remote session identity alike.
+func TestWorkerICE_StaleCloseAgentKeepsCurrentSession(t *testing.T) {
+	w := newTestWorkerICE(t)
+	t.Cleanup(w.Close)
+
+	sidA := ICESessionID("session-a")
+	w.OnNewOffer(&OfferAnswer{
+		IceCredentials: IceCredentials{UFrag: "ufragaaaa", Pwd: "pwdpwdpwdpwdpwdpwdpwdp1"},
+		SessionID:      &sidA,
+	})
+	w.muxAgent.Lock()
+	oldAgent := w.agent
+	oldCancel := w.agentDialerCancel
+	w.muxAgent.Unlock()
+	require.NotNil(t, oldAgent, "OnNewOffer must have created an ICE agent")
+
+	w.Close()
+
+	sidB := ICESessionID("session-b")
+	w.OnNewOffer(&OfferAnswer{
+		IceCredentials: IceCredentials{UFrag: "ufragbbbb", Pwd: "pwdpwdpwdpwdpwdpwdpwdp2"},
+		SessionID:      &sidB,
+	})
+	require.True(t, w.InProgress(), "the second negotiation must be in flight")
+
+	w.muxAgent.Lock()
+	newAgent := w.agent
+	w.muxAgent.Unlock()
+
+	// The old dial goroutine finally wakes and cleans up its captured agent.
+	w.closeAgent(oldAgent, oldCancel)
+
+	w.muxAgent.Lock()
+	defer w.muxAgent.Unlock()
+	assert.Same(t, newAgent, w.agent, "the current agent must be untouched by the stale cleanup")
+	assert.True(t, w.agentConnecting, "the current negotiation must stay in flight")
+	// Read live under the lock: a snapshot captured before the stale cleanup
+	// would pass even if the cleanup wiped current state.
+	assert.Equal(t, sidB, w.remoteSessionID, "the remote session identity must be preserved")
+}
+
+// closeTrackConn records Close calls so a test can assert that a discarded
+// connection was actually released.
+type closeTrackConn struct {
+	net.Conn
+	closed atomic.Bool
+}
+
+func (c *closeTrackConn) Close() error {
+	c.closed.Store(true)
+	return c.Conn.Close()
+}
+
+// TestWorkerICE_StaleDialSuccessKeepsNewerNegotiation pins the ownership guard
+// in connect()'s success path: a dial that came back after a newer negotiation
+// replaced the agent must discard its connection and leave the newer session's
+// state - agent, agentConnecting, remoteSessionID, lastSuccess - intact.
+//
+// The dial hook holds session A's goroutine open until session B is installed,
+// then returns a live connection, mimicking the vendored pion dial which hands
+// out a live *ice.Conn when a pair is selected without checking afterwards
+// whether the agent was replaced meanwhile. Releasing A's dial therefore
+// exercises the stale-success commit path deterministically instead of racing
+// real ICE.
+func TestWorkerICE_StaleDialSuccessKeepsNewerNegotiation(t *testing.T) {
+	w := newTestWorkerICE(t)
+	t.Cleanup(w.Close)
+
+	dialStarted := make(chan struct{})
+	releaseDial := make(chan struct{})
+	staleConn := &closeTrackConn{}
+
+	var calls atomic.Int32
+	w.dialFunc = func(ctx context.Context, _ *icemaker.ThreadSafeAgent, _ *OfferAnswer) (net.Conn, error) {
+		if calls.Add(1) == 1 {
+			// Session A: hold the goroutine open until session B is installed,
+			// then return a live connection, mimicking the vendored pion dial
+			// which hands out a live *ice.Conn once a pair is selected without
+			// re-checking whether the agent was replaced meanwhile. Releasing
+			// the dial therefore exercises the stale-success commit path
+			// deterministically instead of racing real ICE.
+			close(dialStarted)
+			<-releaseDial
+			client, _ := net.Pipe()
+			staleConn.Conn = client
+			return staleConn, nil
+		}
+		// A newer negotiation parks on its dialer context, cancelled by the
+		// t.Cleanup Close at test end.
+		<-ctx.Done()
+		return nil, ctx.Err()
+	}
+
+	sidA := ICESessionID("session-a")
+	w.OnNewOffer(&OfferAnswer{
+		IceCredentials: IceCredentials{UFrag: "ufragaaaa", Pwd: "pwdpwdpwdpwdpwdpwdpwdp1"},
+		SessionID:      &sidA,
+	})
+	require.True(t, w.InProgress(), "session A must be in flight")
+
+	// Session A's goroutine is now parked in the dial hook.
+	<-dialStarted
+
+	sidB := ICESessionID("session-b")
+	w.OnNewOffer(&OfferAnswer{
+		IceCredentials: IceCredentials{UFrag: "ufragbbbb", Pwd: "pwdpwdpwdpwdpwdpwdpwdp2"},
+		SessionID:      &sidB,
+	})
+
+	w.muxAgent.Lock()
+	agentB := w.agent
+	w.lastSuccess = time.Time{}
+	w.muxAgent.Unlock()
+	require.NotNil(t, agentB, "session B must have created an ICE agent")
+	require.True(t, w.InProgress(), "session B must be in flight")
+
+	// Release session A's dial: it must be recognized as stale and discarded.
+	close(releaseDial)
+	require.Eventually(t, func() bool {
+		return staleConn.closed.Load()
+	}, 10*time.Second, 10*time.Millisecond,
+		"the stale connection must be closed by the ownership guard")
+
+	w.muxAgent.Lock()
+	defer w.muxAgent.Unlock()
+	assert.Same(t, agentB, w.agent, "session A must not uninstall session B's agent")
+	assert.True(t, w.agentConnecting, "session A must not clear session B's connecting flag")
+	assert.Equal(t, sidB, w.remoteSessionID, "session A must not clear session B's remote session identity")
+	assert.True(t, w.lastSuccess.IsZero(), "session A must not record a success for session B")
+	// The commit block guards agentConnecting, lastSuccess and
+	// onICEConnectionIsReady together, so the state assertions above imply the
+	// callback never ran for session A; the nil conn would have panicked the
+	// stale goroutine on any invocation.
+}

client/internal/profilemanager/config.go

@@ -70,6 +70,7 @@ type ConfigInput struct {
 	StateFilePath                 string
 	PreSharedKey                  *string
 	ServerSSHAllowed              *bool
+	RemoteJobsAllowed             *bool
 	EnableSSHRoot                 *bool
 	EnableSSHSFTP                 *bool
 	EnableSSHLocalPortForwarding  *bool
@@ -103,6 +104,9 @@ type ConfigInput struct {
 	DNSLabels domain.List
 
 	MTU *uint16
+
+	LocalMetricsEnabled *bool
+	LocalMetricsAddress *string
 }
 
 // Config Configuration type
@@ -124,6 +128,7 @@ type Config struct {
 	RosenpassEnabled              bool
 	RosenpassPermissive           bool
 	ServerSSHAllowed              *bool
+	RemoteJobsAllowed             *bool
 	EnableSSHRoot                 *bool
 	EnableSSHSFTP                 *bool
 	EnableSSHLocalPortForwarding  *bool
@@ -144,6 +149,11 @@ type Config struct {
 
 	DNSLabels domain.List
 
+	// LocalMetricsEnabled enables the local Prometheus /metrics endpoint.
+	LocalMetricsEnabled bool
+	// LocalMetricsAddress is the listen address of the local /metrics endpoint.
+	LocalMetricsAddress string
+
 	// SSHKey is a private SSH key in a PEM format
 	SSHKey string
 
@@ -184,6 +194,12 @@ type Config struct {
 	// Runtime-only: re-derived from MDM policy on each load, never persisted.
 	LazyConnection string `json:"-"`
 
+	// DebugBundleUploadURL is the MDM-managed debug-bundle upload URL override.
+	// When set, it takes precedence over the management-supplied upload URL for
+	// remote debug bundle jobs. Runtime-only: re-derived from MDM policy on each
+	// load, never persisted.
+	DebugBundleUploadURL string `json:"-"`
+
 	MTU uint16
 
 	// policy is the MDM policy that produced the currently-set values for
@@ -217,6 +233,12 @@ func getConfigDir() (string, error) {
 	}
 
 	configDir := filepath.Join(base, "netbird")
+	// Under sudo this is the invoking user's directory and strictly read-only:
+	// anything root creates in it would be root-owned and break the user's own
+	// runs. Reads of a missing directory fall through to defaults.
+	if sudoActive() {
+		return configDir, nil
+	}
 	if err := os.MkdirAll(configDir, 0o755); err != nil {
 		return "", err
 	}
@@ -224,6 +246,16 @@ func getConfigDir() (string, error) {
 }
 
 func baseConfigDir() (string, error) {
+	if u, ok := sudoInvokingUser(); ok {
+		return userBaseConfigDir(u)
+	}
+	// Fail closed instead of falling through to root's own config directory:
+	// reading root's active-profile and email state for what is actually the
+	// invoking user's invocation is the very confusion this resolution exists
+	// to prevent.
+	if sudoActive() {
+		return "", fmt.Errorf("resolve sudo invoking user %q: refusing to fall back to root's config directory", os.Getenv(envSudoUser))
+	}
 	if runtime.GOOS == "darwin" {
 		if u, err := user.Current(); err == nil && u.HomeDir != "" {
 			return filepath.Join(u.HomeDir, "Library", "Application Support"), nil
@@ -265,7 +297,10 @@ func createNewConfig(input ConfigInput) (*Config, error) {
 	config := &Config{
 		// defaults to false only for new (post 0.26) configurations
 		ServerSSHAllowed: util.False(),
-		WgPort:           iface.DefaultWgPort,
+		// Remote jobs are an explicit opt-in and default off, including for
+		// legacy configs (a nil value materializes to false at connect time).
+		RemoteJobsAllowed: util.False(),
+		WgPort:            iface.DefaultWgPort,
 	}
 
 	if _, err := config.apply(input); err != nil {
@@ -388,6 +423,18 @@ func (config *Config) apply(input ConfigInput) (updated bool, err error) {
 		updated = true
 	}
 
+	if input.LocalMetricsEnabled != nil && *input.LocalMetricsEnabled != config.LocalMetricsEnabled {
+		log.Infof("switching local metrics to %t", *input.LocalMetricsEnabled)
+		config.LocalMetricsEnabled = *input.LocalMetricsEnabled
+		updated = true
+	}
+
+	if input.LocalMetricsAddress != nil && *input.LocalMetricsAddress != config.LocalMetricsAddress {
+		log.Infof("switching local metrics address to %s", *input.LocalMetricsAddress)
+		config.LocalMetricsAddress = *input.LocalMetricsAddress
+		updated = true
+	}
+
 	if input.NetworkMonitor != nil && (config.NetworkMonitor == nil || *input.NetworkMonitor != *config.NetworkMonitor) {
 		log.Infof("switching Network Monitor to %t", *input.NetworkMonitor)
 		config.NetworkMonitor = input.NetworkMonitor
@@ -456,6 +503,21 @@ func (config *Config) apply(input ConfigInput) (updated bool, err error) {
 		updated = true
 	}
 
+	if input.RemoteJobsAllowed != nil && (config.RemoteJobsAllowed == nil || *input.RemoteJobsAllowed != *config.RemoteJobsAllowed) {
+		if *input.RemoteJobsAllowed {
+			log.Infof("enabling remote jobs")
+		} else {
+			log.Infof("disabling remote jobs")
+		}
+		config.RemoteJobsAllowed = input.RemoteJobsAllowed
+		updated = true
+	} else if config.RemoteJobsAllowed == nil {
+		// Remote jobs are an explicit opt-in: unlike SSH, a pre-existing config
+		// with no value defaults to disabled rather than being turned on.
+		config.RemoteJobsAllowed = util.False()
+		updated = true
+	}
+
 	if input.EnableSSHRoot != nil && (config.EnableSSHRoot == nil || *input.EnableSSHRoot != *config.EnableSSHRoot) {
 		if *input.EnableSSHRoot {
 			log.Infof("enabling SSH root login")
@@ -665,6 +727,14 @@ func (config *Config) apply(input ConfigInput) (updated bool, err error) {
 // for the key, so per-field rejection of user writes still applies).
 func (config *Config) applyMDMPolicy(policy *mdm.Policy) {
 	config.policy = policy
+
+	// DebugBundleUploadURL is a runtime-only override re-derived from MDM on
+	// every apply. Resolve it unconditionally (before the IsEmpty early return)
+	// so a policy that drops the key, becomes empty, or carries an invalid
+	// value can never leave a previously-enforced upload target active on a
+	// reused Config instance.
+	config.DebugBundleUploadURL = mdmDebugBundleUploadURL(policy)
+
 	if policy.IsEmpty() {
 		return
 	}
@@ -712,12 +782,19 @@ func (config *Config) applyMDMPolicy(policy *mdm.Policy) {
 	}
 
 	applyBool(mdm.KeyAllowServerSSH, func(v bool) { bv := v; config.ServerSSHAllowed = &bv })
+	applyBool(mdm.KeyRemoteJobsAllowed, func(v bool) { bv := v; config.RemoteJobsAllowed = &bv })
 	applyBool(mdm.KeyDisableClientRoutes, func(v bool) { config.DisableClientRoutes = v })
 	applyBool(mdm.KeyDisableServerRoutes, func(v bool) { config.DisableServerRoutes = v })
 	applyBool(mdm.KeyBlockInbound, func(v bool) { config.BlockInbound = v })
 	applyBool(mdm.KeyDisableAutoConnect, func(v bool) { config.DisableAutoConnect = v })
 	applyBool(mdm.KeyRosenpassEnabled, func(v bool) { config.RosenpassEnabled = v })
 	applyBool(mdm.KeyRosenpassPermissive, func(v bool) { config.RosenpassPermissive = v })
+	applyBool(mdm.KeyEnableLocalMetrics, func(v bool) { config.LocalMetricsEnabled = v })
+
+	if v, ok := policy.GetString(mdm.KeyLocalMetricsAddress); ok {
+		config.LocalMetricsAddress = v
+		logApplied(mdm.KeyLocalMetricsAddress, v)
+	}
 
 	if v, ok := policy.GetInt(mdm.KeyWireguardPort); ok {
 		// REG_DWORD is 32-bit; UDP port range is 1-65535. Clamp at the
@@ -739,6 +816,52 @@ func (config *Config) applyMDMPolicy(policy *mdm.Policy) {
 		config.LazyConnection = state
 		logApplied(mdm.KeyLazyConnection, state)
 	}
+
+}
+
+// ValidateBundleUploadURL sanity-checks a debug-bundle upload URL. An empty
+// value is accepted — the executor falls back to the default upload service. A
+// non-empty value must be a well-formed https URL with a host; a malformed
+// value or a plaintext scheme is rejected. It deliberately does not constrain
+// which host may receive the bundle. This is the single source of truth for the
+// rule, shared by the remote-job executor (client/internal) and the MDM policy
+// override below so the two validation paths cannot drift.
+func ValidateBundleUploadURL(raw string) error {
+	if raw == "" {
+		return nil
+	}
+	parsed, err := url.Parse(raw)
+	if err != nil {
+		return fmt.Errorf("parse upload URL: %w", err)
+	}
+	// Hostname(), not Host: an authority like ":443" is non-empty but has no
+	// host, and would fail the actual upload.
+	if parsed.Scheme != "https" || parsed.Hostname() == "" {
+		return fmt.Errorf("upload URL must be an https URL with a host")
+	}
+	return nil
+}
+
+// mdmDebugBundleUploadURL resolves the MDM-enforced debug-bundle upload URL
+// override from the policy, returning the empty string when the policy does
+// not carry a valid KeyBundleUploadURL. An absent or invalid value fails
+// closed to "" so it falls back to the management-supplied or default upload
+// target rather than a previously-enforced one. The URL is never logged: it
+// can embed credentials or signed query tokens (KeyBundleUploadURL is in
+// mdm.SecretKeys).
+func mdmDebugBundleUploadURL(policy *mdm.Policy) string {
+	v, ok := policy.GetString(mdm.KeyBundleUploadURL)
+	if !ok || v == "" {
+		return ""
+	}
+	// Must be a well-formed https URL with a host, matching the client's
+	// remote-job upload-URL validation (shared validator, single source of truth).
+	if err := ValidateBundleUploadURL(v); err != nil {
+		log.Warnf("MDM debug bundle upload URL is invalid (must be an https URL with a host); ignoring the override")
+		return ""
+	}
+	log.Infof("MDM override %s = ********** (secret)", mdm.KeyBundleUploadURL)
+	return v
 }
 
 // parseURL parses and validates the URL for the named service. The URL

client/internal/profilemanager/config_mdm_test.go

@@ -130,6 +130,32 @@ func TestApply_MDMBoolKeysOverrideOnDiskValue(t *testing.T) {
 	assert.True(t, cfg.Policy().HasKey(mdm.KeyRosenpassEnabled))
 }
 
+func TestApply_MDMLocalMetrics(t *testing.T) {
+	tmp := filepath.Join(t.TempDir(), "config.json")
+
+	// Seed without MDM.
+	withMDMPolicy(t, mdm.NewPolicy(nil))
+	_, err := UpdateOrCreateConfig(ConfigInput{
+		ConfigPath:          tmp,
+		LocalMetricsEnabled: boolPtr(false),
+	})
+	require.NoError(t, err)
+
+	withMDMPolicy(t, mdm.NewPolicy(map[string]any{
+		mdm.KeyEnableLocalMetrics:  true,
+		mdm.KeyLocalMetricsAddress: "127.0.0.1:9292",
+	}))
+
+	cfg, err := UpdateOrCreateConfig(ConfigInput{ConfigPath: tmp})
+	require.NoError(t, err)
+	require.NotNil(t, cfg)
+
+	assert.True(t, cfg.LocalMetricsEnabled, "MDM override should flip on-disk false to true")
+	assert.Equal(t, "127.0.0.1:9292", cfg.LocalMetricsAddress)
+	assert.True(t, cfg.Policy().HasKey(mdm.KeyEnableLocalMetrics))
+	assert.True(t, cfg.Policy().HasKey(mdm.KeyLocalMetricsAddress))
+}
+
 func TestApply_MDMLazyConnection(t *testing.T) {
 	cases := []struct {
 		name string

client/internal/profilemanager/config_test.go

@@ -14,6 +14,7 @@ import (
 
 	"github.com/netbirdio/netbird/client/iface"
 	"github.com/netbirdio/netbird/client/internal/routemanager/dynamic"
+	"github.com/netbirdio/netbird/client/mdm"
 	"github.com/netbirdio/netbird/util"
 )
 
@@ -271,6 +272,83 @@ func TestUpdateConfigServerSSHAllowedNotSet(t *testing.T) {
 	}
 }
 
+func TestUpdateConfigRemoteJobsAllowed(t *testing.T) {
+	// Unlike SSH (which defaults on for legacy configs), remote jobs are an
+	// explicit opt-in: a pre-existing config with no value materializes to off.
+	t.Run("legacy config defaults off", func(t *testing.T) {
+		configPath := filepath.Join(t.TempDir(), "config.json")
+		require.NoError(t, os.WriteFile(configPath, []byte("{}"), 0600))
+
+		config, err := UpdateConfig(ConfigInput{ConfigPath: configPath})
+		require.NoError(t, err)
+		require.NotNil(t, config.RemoteJobsAllowed, "RemoteJobsAllowed should be materialized")
+		assert.False(t, *config.RemoteJobsAllowed, "remote jobs must default off")
+	})
+
+	for _, tt := range []struct {
+		name  string
+		input *bool
+		want  bool
+	}{
+		{"enable", util.True(), true},
+		{"disable", util.False(), false},
+	} {
+		t.Run(tt.name, func(t *testing.T) {
+			configPath := filepath.Join(t.TempDir(), "config.json")
+			require.NoError(t, os.WriteFile(configPath, []byte("{}"), 0600))
+
+			config, err := UpdateConfig(ConfigInput{ConfigPath: configPath, RemoteJobsAllowed: tt.input})
+			require.NoError(t, err)
+			require.NotNil(t, config.RemoteJobsAllowed)
+			assert.Equal(t, tt.want, *config.RemoteJobsAllowed)
+		})
+	}
+}
+
+func TestApplyMDMPolicyRemoteJobs(t *testing.T) {
+	t.Run("enables remote jobs and sets the upload URL override", func(t *testing.T) {
+		cfg := &Config{}
+		cfg.applyMDMPolicy(mdm.NewPolicy(map[string]any{
+			mdm.KeyRemoteJobsAllowed: true,
+			mdm.KeyBundleUploadURL:   "https://upload.example.com",
+		}))
+		require.NotNil(t, cfg.RemoteJobsAllowed)
+		assert.True(t, *cfg.RemoteJobsAllowed, "MDM allowRemoteJobs must enable the flag")
+		assert.Equal(t, "https://upload.example.com", cfg.DebugBundleUploadURL, "MDM upload URL override must be applied")
+	})
+
+	t.Run("a non-https upload URL is rejected", func(t *testing.T) {
+		cfg := &Config{}
+		cfg.applyMDMPolicy(mdm.NewPolicy(map[string]any{
+			mdm.KeyBundleUploadURL: "http://insecure.example.com",
+		}))
+		assert.Empty(t, cfg.DebugBundleUploadURL, "a non-https upload URL must be skipped")
+	})
+
+	t.Run("dropping the key clears a previously-applied override", func(t *testing.T) {
+		cfg := &Config{DebugBundleUploadURL: "https://old.example.com"}
+		// A replacement policy that no longer carries the key must not leave
+		// the old upload target directing bundles.
+		cfg.applyMDMPolicy(mdm.NewPolicy(map[string]any{mdm.KeyRemoteJobsAllowed: true}))
+		assert.Empty(t, cfg.DebugBundleUploadURL, "the stale upload URL override must be cleared")
+	})
+
+	t.Run("an empty replacement policy clears a previously-applied override", func(t *testing.T) {
+		cfg := &Config{DebugBundleUploadURL: "https://old.example.com"}
+		// A policy that becomes empty entirely hits the IsEmpty early return;
+		// the override must still be cleared rather than surviving on the
+		// reused Config instance.
+		cfg.applyMDMPolicy(mdm.NewPolicy(map[string]any{}))
+		assert.Empty(t, cfg.DebugBundleUploadURL, "the stale upload URL override must be cleared when the policy empties")
+	})
+
+	t.Run("an invalid upload URL clears a previously-applied override (fail closed)", func(t *testing.T) {
+		cfg := &Config{DebugBundleUploadURL: "https://old.example.com"}
+		cfg.applyMDMPolicy(mdm.NewPolicy(map[string]any{mdm.KeyBundleUploadURL: "not-a-url"}))
+		assert.Empty(t, cfg.DebugBundleUploadURL, "an invalid override must fail closed, not keep the stale target")
+	})
+}
+
 func TestUpdateOldManagementURL(t *testing.T) {
 	origProber := newMgmProber
 	newMgmProber = func(_ context.Context, _ string, _ wgtypes.Key, _ bool) (mgmProber, error) {

client/internal/profilemanager/invoking_user.go

@@ -0,0 +1,100 @@
+package profilemanager
+
+import (
+	"fmt"
+	"os"
+	"os/user"
+	"path/filepath"
+	"runtime"
+
+	log "github.com/sirupsen/logrus"
+)
+
+const envSudoUser = "SUDO_USER"
+
+var (
+	geteuid    = os.Geteuid
+	lookupUser = user.Lookup
+)
+
+// InvokingUser returns the user a CLI invocation acts for. Under sudo that is
+// the user who ran sudo, not root: privileged flags force commands through
+// sudo, and resolving profiles as root would silently switch the daemon to
+// root's (default) profile instead of the invoking user's. Privilege decisions
+// are not made here — those stay on the kernel credentials of the daemon
+// connection, which SUDO_USER (a plain environment variable) can never
+// influence; a forged value only selects a profile root could select anyway.
+func InvokingUser() (*user.User, error) {
+	if u, ok := sudoInvokingUser(); ok {
+		return u, nil
+	}
+	// Fail closed instead of falling through to root: every caller feeds this
+	// username into profile-path resolution, so a lookup failure would resolve
+	// (and create) a root-owned profile namespace and switch the daemon onto it
+	// behind the invoking user's back.
+	if sudoActive() {
+		return nil, fmt.Errorf("resolve sudo invoking user %q: refusing to fall back to root", os.Getenv(envSudoUser))
+	}
+	return user.Current()
+}
+
+// IsPlainRoot reports that the process runs as root with no usable sudo
+// context: there is no invoking user to act for, so per-user resolution falls
+// back to root's own (empty) state. Callers use it to refuse ambiguous
+// operations instead of silently acting on the wrong profile.
+func IsPlainRoot() bool {
+	if geteuid() != 0 {
+		return false
+	}
+	_, ok := sudoInvokingUser()
+	return !ok
+}
+
+// MirrorIsAuthoritative reports whether the invoking user's local
+// active-profile mirror can be trusted as the profile selector. It cannot under
+// sudo (writes to it are skipped, so it goes stale) or as plain root (there is
+// no invoking user, so it falls back to root's own default). Callers use it to
+// decide whether to read the profile from the mirror or from the daemon.
+func MirrorIsAuthoritative() bool {
+	return !sudoActive() && !IsPlainRoot()
+}
+
+// sudoInvokingUser resolves SUDO_USER when the process runs as root under
+// sudo. Returns false whenever the sudo context is absent or unusable, in
+// which case callers fall back to the process user.
+func sudoInvokingUser() (*user.User, bool) {
+	if !sudoActive() {
+		return nil, false
+	}
+	name := os.Getenv(envSudoUser)
+	u, err := lookupUser(name)
+	if err != nil {
+		log.Warnf("sudo invoking user %q lookup: %v", name, err)
+		return nil, false
+	}
+	return u, true
+}
+
+// sudoActive reports a sudo context from the environment alone: write-skip
+// decisions key off it so a transient user lookup failure can never flip a
+// run from read-only to writing root-owned files into the user's directory.
+func sudoActive() bool {
+	if geteuid() != 0 {
+		return false
+	}
+	name := os.Getenv(envSudoUser)
+	return name != "" && name != "root"
+}
+
+// userBaseConfigDir mirrors os.UserConfigDir for a user other than the process
+// owner. Environment overrides (XDG_CONFIG_HOME) cannot be honoured here: under
+// sudo the environment is root's, not the invoking user's.
+func userBaseConfigDir(u *user.User) (string, error) {
+	if u.HomeDir == "" {
+		return "", fmt.Errorf("user %s has no home directory", u.Username)
+	}
+	if runtime.GOOS == "darwin" {
+		return filepath.Join(u.HomeDir, "Library", "Application Support"), nil
+	}
+	return filepath.Join(u.HomeDir, ".config"), nil
+}

client/internal/profilemanager/invoking_user_test.go

@@ -0,0 +1,230 @@
+package profilemanager
+
+import (
+	"errors"
+	"io/fs"
+	"os"
+	"os/user"
+	"path/filepath"
+	"runtime"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestInvokingUserFallsBackToProcessUser(t *testing.T) {
+	t.Setenv(envSudoUser, "")
+
+	got, err := InvokingUser()
+	require.NoError(t, err)
+
+	current, err := user.Current()
+	require.NoError(t, err)
+	assert.Equal(t, current.Username, got.Username)
+}
+
+func TestSudoInvokingUserInactiveWithoutSudoContext(t *testing.T) {
+	t.Setenv(envSudoUser, "")
+	_, ok := sudoInvokingUser()
+	assert.False(t, ok)
+}
+
+func TestSudoInvokingUserIgnoresRoot(t *testing.T) {
+	t.Setenv(envSudoUser, "root")
+	origEuid := geteuid
+	geteuid = func() int { return 0 }
+	t.Cleanup(func() { geteuid = origEuid })
+
+	_, ok := sudoInvokingUser()
+	assert.False(t, ok, "sudo from a root shell must not redirect anything")
+	assert.False(t, sudoActive())
+	assert.True(t, IsPlainRoot())
+}
+
+func TestSudoInvokingUserResolvesInvokingUser(t *testing.T) {
+	fakeSudo(t, filepath.Join("/home", "misha"))
+
+	u, ok := sudoInvokingUser()
+	require.True(t, ok)
+	assert.Equal(t, "misha", u.Username)
+
+	got, err := InvokingUser()
+	require.NoError(t, err)
+	assert.Equal(t, "misha", got.Username)
+
+	assert.False(t, IsPlainRoot())
+}
+
+func TestInvokingUserFailsClosedWhenSudoLookupFails(t *testing.T) {
+	fakeSudo(t, filepath.Join("/home", "misha"))
+	lookupUser = func(string) (*user.User, error) { return nil, errors.New("nss unavailable") }
+
+	got, err := InvokingUser()
+	require.Error(t, err)
+	assert.Nil(t, got, "must not resolve to the root process user")
+}
+
+func TestProfileFilePathFailsClosedWhenSudoLookupFails(t *testing.T) {
+	profilesRoot := t.TempDir()
+	fakeSudo(t, filepath.Join("/home", "misha"))
+	lookupUser = func(string) (*user.User, error) { return nil, errors.New("nss unavailable") }
+
+	origDir := DefaultConfigPathDir
+	DefaultConfigPathDir = profilesRoot
+	t.Cleanup(func() { DefaultConfigPathDir = origDir })
+
+	p := &Profile{ID: "0123456789abcdef0123456789abcdef"}
+	_, err := p.FilePath()
+	require.Error(t, err)
+	assertNoEntries(t, profilesRoot)
+}
+
+func TestSudoActiveSurvivesLookupFailure(t *testing.T) {
+	fakeSudo(t, filepath.Join("/home", "misha"))
+	lookupUser = func(string) (*user.User, error) { return nil, errors.New("nss unavailable") }
+
+	_, ok := sudoInvokingUser()
+	assert.False(t, ok)
+	assert.True(t, sudoActive())
+	assert.True(t, IsPlainRoot())
+}
+
+func TestGetConfigDirUnderSudoIsReadOnly(t *testing.T) {
+	home := t.TempDir()
+	fakeSudo(t, home)
+
+	base, err := baseConfigDir()
+	require.NoError(t, err)
+	if runtime.GOOS == "darwin" {
+		assert.Equal(t, filepath.Join(home, "Library", "Application Support"), base)
+	} else {
+		assert.Equal(t, filepath.Join(home, ".config"), base)
+	}
+
+	dir, err := getConfigDir()
+	require.NoError(t, err)
+	assert.Equal(t, filepath.Join(base, "netbird"), dir)
+	assert.NoDirExists(t, dir)
+}
+
+func TestBaseConfigDirFailsClosedWhenSudoLookupFails(t *testing.T) {
+	fakeSudo(t, filepath.Join("/home", "misha"))
+	lookupUser = func(string) (*user.User, error) { return nil, errors.New("nss unavailable") }
+
+	_, err := baseConfigDir()
+	require.Error(t, err)
+
+	_, err = getConfigDir()
+	require.Error(t, err)
+}
+
+func TestSwitchProfileSkipsStateWriteUnderSudo(t *testing.T) {
+	home := t.TempDir()
+	fakeSudo(t, home)
+
+	pm := NewProfileManager()
+	require.NoError(t, pm.SwitchProfile(defaultProfileName))
+	assertNoEntries(t, home)
+}
+
+func TestSetProfileStateSkipsWriteUnderSudo(t *testing.T) {
+	home := t.TempDir()
+	fakeSudo(t, home)
+
+	pm := NewProfileManager()
+	require.NoError(t, pm.SetProfileState(defaultProfileName, &ProfileState{Email: "misha@example.com"}))
+	assertNoEntries(t, home)
+}
+
+func TestRemoveProfileStateSkipsRemoveUnderSudo(t *testing.T) {
+	home := t.TempDir()
+	stateDir := filepath.Join(home, ".config", "netbird")
+	if runtime.GOOS == "darwin" {
+		stateDir = filepath.Join(home, "Library", "Application Support", "netbird")
+	}
+	require.NoError(t, os.MkdirAll(stateDir, 0o700))
+	stateFile := filepath.Join(stateDir, "default.state.json")
+	require.NoError(t, os.WriteFile(stateFile, []byte(`{"email":"misha@example.com"}`), 0o600))
+
+	fakeSudo(t, home)
+	pm := NewProfileManager()
+	require.NoError(t, pm.RemoveProfileState("default"))
+	assert.FileExists(t, stateFile)
+}
+
+func TestUserBaseConfigDir(t *testing.T) {
+	u := &user.User{Username: "misha", HomeDir: filepath.Join("/home", "misha")}
+	dir, err := userBaseConfigDir(u)
+	require.NoError(t, err)
+	if runtime.GOOS == "darwin" {
+		assert.Equal(t, filepath.Join(u.HomeDir, "Library", "Application Support"), dir)
+	} else {
+		assert.Equal(t, filepath.Join(u.HomeDir, ".config"), dir)
+	}
+
+	_, err = userBaseConfigDir(&user.User{Username: "nohome"})
+	require.Error(t, err)
+}
+
+func TestIsPlainRoot(t *testing.T) {
+	t.Setenv(envSudoUser, "")
+	origEuid := geteuid
+	t.Cleanup(func() { geteuid = origEuid })
+
+	geteuid = func() int { return 1000 }
+	assert.False(t, IsPlainRoot())
+
+	geteuid = func() int { return 0 }
+	assert.True(t, IsPlainRoot())
+}
+
+func TestMirrorIsAuthoritative(t *testing.T) {
+	t.Setenv(envSudoUser, "")
+	origEuid := geteuid
+	t.Cleanup(func() { geteuid = origEuid })
+
+	geteuid = func() int { return 1000 }
+	assert.True(t, MirrorIsAuthoritative(), "a normal user's own mirror is authoritative")
+
+	geteuid = func() int { return 0 }
+	assert.False(t, MirrorIsAuthoritative(), "plain root has no authoritative mirror")
+}
+
+func TestMirrorIsAuthoritativeFalseUnderSudo(t *testing.T) {
+	fakeSudo(t, filepath.Join("/home", "misha"))
+	assert.False(t, MirrorIsAuthoritative(), "the sudo mirror is frozen, so it is not authoritative")
+}
+
+func fakeSudo(t *testing.T, home string) {
+	t.Helper()
+	t.Setenv(envSudoUser, "misha")
+
+	origEuid := geteuid
+	origLookup := lookupUser
+	origOverride := ConfigDirOverride
+	geteuid = func() int { return 0 }
+	lookupUser = func(name string) (*user.User, error) {
+		return &user.User{Username: name, Uid: "1234", Gid: "1234", HomeDir: home}, nil
+	}
+	ConfigDirOverride = ""
+	t.Cleanup(func() {
+		geteuid = origEuid
+		lookupUser = origLookup
+		ConfigDirOverride = origOverride
+	})
+}
+
+func assertNoEntries(t *testing.T, root string) {
+	t.Helper()
+	err := filepath.WalkDir(root, func(path string, _ fs.DirEntry, err error) error {
+		if err != nil {
+			return err
+		}
+		if path != root {
+			t.Errorf("unexpected entry created under %s: %s", root, path)
+		}
+		return nil
+	})
+	require.NoError(t, err)
+}

client/internal/profilemanager/profilemanager.go

@@ -3,7 +3,6 @@ package profilemanager
 import (
 	"fmt"
 	"os"
-	"os/user"
 	"path/filepath"
 	"strings"
 	"sync"
@@ -54,7 +53,7 @@ func (p *Profile) FilePath() (string, error) {
 		return "", fmt.Errorf("invalid profile ID: %q", id)
 	}
 
-	username, err := user.Current()
+	username, err := InvokingUser()
 	if err != nil {
 		return "", fmt.Errorf("failed to get current user: %w", err)
 	}
@@ -130,7 +129,7 @@ func (pm *ProfileManager) getActiveProfileState() ID {
 	if err != nil {
 		if !os.IsNotExist(err) {
 			log.Warnf("failed to read active profile state: %v", err)
-		} else {
+		} else if !sudoActive() {
 			if err := pm.setActiveProfileState(defaultProfileName); err != nil {
 				log.Warnf("failed to set default profile state: %v", err)
 			}
@@ -148,6 +147,13 @@ func (pm *ProfileManager) getActiveProfileState() ID {
 }
 
 func (pm *ProfileManager) setActiveProfileState(id ID) error {
+	// The invoking user's state is read-only under sudo — a root-owned file in
+	// the user's directory would break their own runs. The daemon still records
+	// the switch on its side; only the user-local bookkeeping is skipped.
+	if sudoActive() {
+		log.Infof("running under sudo: not persisting active profile %q for user %s", id, os.Getenv(envSudoUser))
+		return nil
+	}
 
 	configDir, err := getConfigDir()
 	if err != nil {

client/internal/profilemanager/state.go

@@ -7,6 +7,8 @@ import (
 	"os"
 	"path/filepath"
 
+	log "github.com/sirupsen/logrus"
+
 	"github.com/netbirdio/netbird/util"
 )
 
@@ -63,6 +65,15 @@ func (pm *ProfileManager) SetProfileState(id ID, state *ProfileState) error {
 		return fmt.Errorf("invalid profile ID: %q", id)
 	}
 
+	// The invoking user's state is read-only under sudo. The file only carries
+	// the account email for the login hint and display, so skipping the write
+	// costs at most one extra account prompt later — a root-owned file in the
+	// user's directory would cost every later update instead.
+	if sudoActive() {
+		log.Debugf("running under sudo: not persisting profile state for user %s", os.Getenv(envSudoUser))
+		return nil
+	}
+
 	stateFile := filepath.Join(configDir, id.String()+".state.json")
 	if err := util.WriteJsonWithRestrictedPermission(context.Background(), stateFile, state); err != nil {
 		return fmt.Errorf("write profile state: %w", err)
@@ -92,6 +103,11 @@ func (pm *ProfileManager) SetActiveProfileState(state *ProfileState) error {
 // equivalent to clearing it; the next SSO login recreates it. A missing file
 // is not an error.
 func (pm *ProfileManager) RemoveProfileState(profileName string) error {
+	if sudoActive() {
+		log.Debugf("running under sudo: not removing profile state for user %s", os.Getenv(envSudoUser))
+		return nil
+	}
+
 	configDir, err := getConfigDir()
 	if err != nil {
 		return fmt.Errorf("get config directory: %w", err)

client/internal/routemanager/selection.go

@@ -17,23 +17,30 @@ import (
 // are mutually exclusive: if the selection activates an exit node, every other
 // available exit node is deselected so two can't be active at once. With
 // appendRoute=false the previous selection is replaced instead of extended.
+// A partial failure (e.g. an unknown ID mixed with valid ones) still applies
+// the valid IDs to the routing table; the unknown ones are reported in the
+// returned error.
 func (m *DefaultManager) SelectRoutes(ids []route.NetID, appendRoute bool) error {
-	if err := m.selectRoutes(ids, appendRoute); err != nil {
-		return err
-	}
+	err := m.selectRoutes(ids, appendRoute)
+	// Apply regardless of err: selectRoutes already selects the valid part of a
+	// partial request, and skipping this on error would leave those routes
+	// selected in the selector but never installed in the routing table.
 	m.TriggerSelection(m.GetClientRoutes())
-	return nil
+	return err
 }
 
 // DeselectRoutes removes the routes with the given network IDs from the
 // selection and applies the change. V4/v6 exit-node pairs are expanded
-// automatically.
+// automatically. A partial failure (e.g. an unknown ID mixed with valid ones)
+// still applies the valid IDs to the routing table; the unknown ones are
+// reported in the returned error.
 func (m *DefaultManager) DeselectRoutes(ids []route.NetID) error {
-	if err := m.deselectRoutes(ids); err != nil {
-		return err
-	}
+	err := m.deselectRoutes(ids)
+	// Apply regardless of err: deselectRoutes already deselects the valid part
+	// of a partial request, and skipping this on error would leave those routes
+	// installed in the routing table despite being marked deselected.
 	m.TriggerSelection(m.GetClientRoutes())
-	return nil
+	return err
 }
 
 func (m *DefaultManager) deselectRoutes(ids []route.NetID) error {

client/internal/routemanager/selection_test.go

@@ -1,12 +1,17 @@
 package routemanager
 
 import (
+	"context"
 	"net/netip"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
+	"golang.org/x/exp/maps"
 
+	"github.com/netbirdio/netbird/client/internal/peer"
+	"github.com/netbirdio/netbird/client/internal/routemanager/client"
+	"github.com/netbirdio/netbird/client/internal/routemanager/notifier"
 	"github.com/netbirdio/netbird/client/internal/routeselector"
 	"github.com/netbirdio/netbird/route"
 )
@@ -112,6 +117,75 @@ func TestSelectRoutes_UnknownRoute(t *testing.T) {
 	assert.Error(t, m.deselectRoutes([]route.NetID{"missing"}), "deselecting an unavailable route must fail")
 }
 
+// newPartialFailureTestManager exercises the real install/remove path without
+// touching the system: the noop refcounter absorbs the route changes, and every
+// route already has a watcher, so none is started.
+func newPartialFailureTestManager() *DefaultManager {
+	ctx := context.Background()
+
+	m := &DefaultManager{
+		ctx: ctx,
+		clientRoutes: route.HAMap{
+			"lan|192.168.1.0/24": {{NetID: "lan", Network: netip.MustParsePrefix("192.168.1.0/24"), Peer: "p1"}},
+			"other|10.1.2.0/24":  {{NetID: "other", Network: netip.MustParsePrefix("10.1.2.0/24"), Peer: "p2"}},
+		},
+		routeSelector:  routeselector.NewRouteSelector(),
+		notifier:       notifier.NewNotifier(),
+		statusRecorder: peer.NewRecorder("https://mgm"),
+		activeRoutes:   make(map[route.HAUniqueID]client.RouteHandler),
+		clientNetworks: map[route.HAUniqueID]*client.Watcher{
+			"lan|192.168.1.0/24": client.NewWatcher(client.WatcherConfig{Context: ctx}),
+			"other|10.1.2.0/24":  client.NewWatcher(client.WatcherConfig{Context: ctx}),
+		},
+	}
+	m.setupRefCounters(true)
+	return m
+}
+
+// Regression for the reported symptom: a partial failure returned before
+// TriggerSelection ran, so the valid route was marked selected while never
+// reaching the routing table (activeRoutes/ip route).
+func TestSelectRoutes_PartialFailureStillInstallsValidRoute(t *testing.T) {
+	m := newPartialFailureTestManager()
+
+	err := m.SelectRoutes([]route.NetID{"missing", "lan"}, false)
+
+	assert.Error(t, err, "the unknown id must still be reported")
+	assert.Contains(t, m.activeRoutes, route.HAUniqueID("lan|192.168.1.0/24"), "the valid route must be installed despite the error")
+	assert.NotContains(t, m.activeRoutes, route.HAUniqueID("other|10.1.2.0/24"), "the deselected route must not be installed")
+}
+
+// Mirror of the case above: a partial failure must remove the valid route from
+// the routing table, not just mark it deselected in the selector.
+func TestDeselectRoutes_PartialFailureStillRemovesValidRoute(t *testing.T) {
+	m := newPartialFailureTestManager()
+
+	require.NoError(t, m.SelectRoutes([]route.NetID{"lan", "other"}, false))
+	require.Contains(t, m.activeRoutes, route.HAUniqueID("lan|192.168.1.0/24"))
+	require.Contains(t, m.activeRoutes, route.HAUniqueID("other|10.1.2.0/24"))
+
+	err := m.DeselectRoutes([]route.NetID{"missing", "other"})
+
+	assert.Error(t, err, "the unknown id must still be reported")
+	assert.NotContains(t, m.activeRoutes, route.HAUniqueID("other|10.1.2.0/24"), "the deselected route must be removed")
+	assert.Contains(t, m.activeRoutes, route.HAUniqueID("lan|192.168.1.0/24"), "the untouched route stays installed")
+}
+
+// The selection now runs on every request, including one where no ID is known
+// and the selector stays untouched. Nothing may be torn down or reinstalled on
+// that path.
+func TestSelectRoutes_TotalFailureLeavesInstalledRoutesAlone(t *testing.T) {
+	m := newPartialFailureTestManager()
+
+	require.NoError(t, m.SelectRoutes([]route.NetID{"lan", "other"}, false))
+	installed := maps.Keys(m.activeRoutes)
+
+	err := m.SelectRoutes([]route.NetID{"missing"}, false)
+
+	assert.Error(t, err, "the unknown id must still be reported")
+	assert.ElementsMatch(t, installed, maps.Keys(m.activeRoutes), "a fully invalid request must not disturb the routing table")
+}
+
 func TestExitNodeSelectionHelpers(t *testing.T) {
 	routesMap := map[route.NetID][]*route.Route{
 		"exitA": {{Network: netip.MustParsePrefix("0.0.0.0/0")}},

client/internal/routeselector/routeselector.go

@@ -32,6 +32,22 @@ func (rs *RouteSelector) SelectRoutes(routes []route.NetID, appendRoute bool, al
 	rs.mu.Lock()
 	defer rs.mu.Unlock()
 
+	// Validate before mutating: a non-append selection wipes the current selection
+	// first, so a request of only unavailable routes would deselect everything and
+	// put nothing back. An empty request means deselect all, so it still goes through.
+	var err *multierror.Error
+	available := make([]route.NetID, 0, len(routes))
+	for _, r := range routes {
+		if !slices.Contains(allRoutes, r) {
+			err = multierror.Append(err, fmt.Errorf("route '%s' is not available", r))
+			continue
+		}
+		available = append(available, r)
+	}
+	if len(available) == 0 && err != nil {
+		return errors.FormatErrorOrNil(err)
+	}
+
 	if !appendRoute || rs.deselectAll {
 		if rs.deselectedRoutes == nil {
 			rs.deselectedRoutes = map[route.NetID]struct{}{}
@@ -46,14 +62,9 @@ func (rs *RouteSelector) SelectRoutes(routes []route.NetID, appendRoute bool, al
 		}
 	}
 
-	var err *multierror.Error
-	for _, route := range routes {
-		if !slices.Contains(allRoutes, route) {
-			err = multierror.Append(err, fmt.Errorf("route '%s' is not available", route))
-			continue
-		}
-		delete(rs.deselectedRoutes, route)
-		rs.selectedRoutes[route] = struct{}{}
+	for _, r := range available {
+		delete(rs.deselectedRoutes, r)
+		rs.selectedRoutes[r] = struct{}{}
 	}
 
 	rs.deselectAll = false

client/internal/routeselector/routeselector_test.go

@@ -887,3 +887,70 @@ func TestRouteSelector_EnableExitNodeKeepsOtherRoutes(t *testing.T) {
 	assert.True(t, rs.IsSelected("lan1"), "non-exit route must stay selected")
 	assert.True(t, rs.IsSelected("lan2"), "non-exit route must stay selected")
 }
+
+// A non-append selection clears the current selection before applying the requested
+// one, so an all-unavailable request used to leave nothing selected while returning
+// an error. Requests with at least one available route are unaffected.
+func TestRouteSelector_SelectRoutes_AllUnavailableKeepsSelection(t *testing.T) {
+	allRoutes := []route.NetID{"route1", "route2", "route3"}
+
+	rs := routeselector.NewRouteSelector()
+	require.NoError(t, rs.SelectRoutes([]route.NetID{"route1"}, false, allRoutes))
+
+	err := rs.SelectRoutes([]route.NetID{"Route1", "route4"}, false, allRoutes)
+
+	assert.Error(t, err, "an unavailable route ID must still be reported")
+	assert.True(t, rs.IsSelected("route1"), "the previous selection must survive a fully invalid request")
+	for _, id := range []route.NetID{"route2", "route3"} {
+		assert.False(t, rs.IsSelected(id), "no other route may become selected")
+	}
+}
+
+// Boundary of the check above: an empty request is the caller deselecting everything,
+// not a failed lookup, so it must keep working.
+func TestRouteSelector_SelectRoutes_EmptyRequestStillDeselectsAll(t *testing.T) {
+	allRoutes := []route.NetID{"route1", "route2", "route3"}
+
+	rs := routeselector.NewRouteSelector()
+	require.NoError(t, rs.SelectRoutes([]route.NetID{"route1"}, false, allRoutes))
+
+	require.NoError(t, rs.SelectRoutes(nil, false, allRoutes))
+
+	for _, id := range allRoutes {
+		assert.False(t, rs.IsSelected(id), "an empty selection request must deselect everything")
+	}
+}
+
+// Mobile clients always call SelectRoutes with append=true. On that path an
+// all-unavailable request was never destructive to begin with (append skips the
+// wipe regardless of the guard above), but the behavior has no coverage yet.
+func TestRouteSelector_SelectRoutes_AppendAllUnavailableKeepsSelection(t *testing.T) {
+	allRoutes := []route.NetID{"route1", "route2", "route3"}
+
+	rs := routeselector.NewRouteSelector()
+	require.NoError(t, rs.SelectRoutes([]route.NetID{"route1"}, false, allRoutes))
+
+	err := rs.SelectRoutes([]route.NetID{"missing"}, true, allRoutes)
+
+	assert.Error(t, err, "an unavailable route ID must still be reported")
+	assert.True(t, rs.IsSelected("route1"), "the previous selection must survive a fully invalid request")
+	for _, id := range []route.NetID{"route2", "route3"} {
+		assert.False(t, rs.IsSelected(id), "no other route may become selected")
+	}
+}
+
+// The early return for an all-unavailable request must not clear deselectAll,
+// or a typo'd network ID would silently drop the "nothing selected, including
+// future networks" policy.
+func TestRouteSelector_SelectRoutes_AllUnavailableAfterDeselectAllKeepsPolicy(t *testing.T) {
+	allRoutes := []route.NetID{"route1", "route2"}
+
+	rs := routeselector.NewRouteSelector()
+	rs.DeselectAllRoutes()
+
+	err := rs.SelectRoutes([]route.NetID{"missing"}, false, allRoutes)
+
+	assert.Error(t, err, "an unavailable route ID must still be reported")
+	assert.True(t, rs.IsDeselectAll(), "deselect-all policy must survive a fully invalid request")
+	assert.False(t, rs.IsSelected("route3"), "deselect-all must still cover networks not present in allRoutes yet")
+}

client/ios/NetBirdSDK/client.go

@@ -4,12 +4,14 @@ package NetBirdSDK
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"net/netip"
 	"os"
 	"sort"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	log "github.com/sirupsen/logrus"
@@ -37,6 +39,8 @@ const (
 	AnonymizeLevelStrict  = nbAnonymize.LevelStrictString
 )
 
+var errClientAlreadyRunning = errors.New("client is already running")
+
 // RouteListener export internal RouteListener for mobile
 type NetworkChangeListener interface {
 	listener.NetworkChangeListener
@@ -74,15 +78,13 @@ type Client struct {
 	cacheDir              string
 	logFilePath           string
 	recorder              *peer.Status
-	ctxCancel             context.CancelFunc
-	ctxCancelLock         *sync.Mutex
 	deviceName            string
 	osName                string
 	osVersion             string
 	networkChangeListener listener.NetworkChangeListener
 	onHostDnsFn           func([]string)
 	dnsManager            dns.IosDnsManager
-	loginComplete         bool
+	loginComplete         atomic.Bool
 	// netMgr outlives engine restarts: it mirrors the OS connectivity, not
 	// the engine lifecycle. Run injects its state and sweeper into each new
 	// ConnectClient.
@@ -90,9 +92,16 @@ type Client struct {
 	// preloadedConfig holds config loaded from JSON (used on tvOS where file writes are blocked)
 	preloadedConfig *profilemanager.Config
 
+	// stateMu guards the run lifecycle as one unit: the cancel installed by
+	// the current run, the channel it closes on exit, and the state it
+	// published. One run at a time: startRun refuses a second Run while the
+	// previous one has not exited, and the platform serializes Stop before
+	// Start, so no generation tracking is needed.
 	stateMu       sync.RWMutex
 	connectClient *internal.ConnectClient
 	config        *profilemanager.Config
+	runDone       chan struct{}
+	ctxCancel     context.CancelFunc
 
 	fileDropMu sync.Mutex
 	fileDrop   *FileDrop
@@ -110,7 +119,6 @@ func NewClient(cfgFile, stateFile, cacheDir, logFilePath, deviceName string, osV
 		osName:                osName,
 		osVersion:             osVersion,
 		recorder:              recorder,
-		ctxCancelLock:         &sync.Mutex{},
 		networkChangeListener: networkChangeListener,
 		dnsManager:            dnsManager,
 		netMgr:                netevents.NewManager(recorder),
@@ -159,17 +167,21 @@ func (c *Client) Run(fd int32, interfaceName string, envList *EnvList) error {
 	c.recorder.UpdateManagementAddress(cfg.ManagementURL.String())
 	c.recorder.UpdateRosenpass(cfg.RosenpassEnabled, cfg.RosenpassPermissive)
 
-	var ctx context.Context
 	//nolint
 	ctxWithValues := context.WithValue(context.Background(), system.DeviceNameCtxKey, c.deviceName)
 	//nolint
 	ctxWithValues = context.WithValue(ctxWithValues, system.OsNameCtxKey, c.osName)
 	//nolint
 	ctxWithValues = context.WithValue(ctxWithValues, system.OsVersionCtxKey, c.osVersion)
-	c.ctxCancelLock.Lock()
-	ctx, c.ctxCancel = context.WithCancel(ctxWithValues)
-	defer c.ctxCancel()
-	c.ctxCancelLock.Unlock()
+	runCtx, runCancel := context.WithCancel(ctxWithValues)
+	defer runCancel()
+
+	done, err := c.startRun(runCancel)
+	if err != nil {
+		return err
+	}
+	defer c.finishRun(done)
+	ctx := runCtx
 
 	// No login pre-flight here. The engine's own loginToManagement (connect.go) performs
 	// the authoritative Login immediately before the first Sync, so a LoginSync() call at
@@ -219,16 +231,40 @@ func (c *Client) NotifyNetworkChange() {
 	c.netMgr.NotifyNetworkChange()
 }
 
-// Stop the internal client and free the resources
+// Stop cancels the running client and waits for the run loop to exit, so a
+// caller that restarts immediately cannot race the outgoing teardown.
 func (c *Client) Stop() {
-	c.ctxCancelLock.Lock()
-	defer c.ctxCancelLock.Unlock()
-	if c.ctxCancel == nil {
+	done := c.cancelRun()
+	if done == nil {
 		return
 	}
 
-	c.ctxCancel()
-	c.setState(nil, nil)
+	select {
+	case <-done:
+	case <-time.After(stopRunWaitTimeout):
+		log.Warnf("Stop: timed out waiting for the run loop to exit")
+	}
+}
+
+// StopWithoutWait cancels the running client without waiting for the run loop.
+// Use it where the caller is on a deadline the wait could overrun, such as
+// NEPacketTunnelProvider.stopTunnel, which iOS gives only a few seconds
+// before it kills the extension.
+func (c *Client) StopWithoutWait() {
+	c.cancelRun()
+}
+
+func (c *Client) cancelRun() chan struct{} {
+	c.stateMu.RLock()
+	done := c.runDone
+	cancel := c.ctxCancel
+	c.stateMu.RUnlock()
+
+	if cancel != nil {
+		cancel()
+	}
+
+	return done
 }
 
 // DebugBundle generates a debug bundle, uploads it and returns the upload key.
@@ -380,16 +416,14 @@ func (c *Client) IsLoginRequiredCached() bool {
 }
 
 func (c *Client) IsLoginRequired() bool {
-	var ctx context.Context
 	//nolint
 	ctxWithValues := context.WithValue(context.Background(), system.DeviceNameCtxKey, c.deviceName)
 	//nolint
 	ctxWithValues = context.WithValue(ctxWithValues, system.OsNameCtxKey, c.osName)
 	//nolint
 	ctxWithValues = context.WithValue(ctxWithValues, system.OsVersionCtxKey, c.osVersion)
-	c.ctxCancelLock.Lock()
-	defer c.ctxCancelLock.Unlock()
-	ctx, c.ctxCancel = context.WithCancel(ctxWithValues)
+	ctx, cancel := context.WithCancel(ctxWithValues)
+	defer cancel()
 
 	var cfg *profilemanager.Config
 	var err error
@@ -437,17 +471,22 @@ func (c *Client) IsLoginRequired() bool {
 // loginForMobileAuthTimeout is the timeout for requesting auth info from the server
 const loginForMobileAuthTimeout = 30 * time.Second
 
+const stopRunWaitTimeout = 20 * time.Second
+
 func (c *Client) LoginForMobile() string {
-	var ctx context.Context
 	//nolint
 	ctxWithValues := context.WithValue(context.Background(), system.DeviceNameCtxKey, c.deviceName)
 	//nolint
 	ctxWithValues = context.WithValue(ctxWithValues, system.OsNameCtxKey, c.osName)
 	//nolint
 	ctxWithValues = context.WithValue(ctxWithValues, system.OsVersionCtxKey, c.osVersion)
-	c.ctxCancelLock.Lock()
-	defer c.ctxCancelLock.Unlock()
-	ctx, c.ctxCancel = context.WithCancel(ctxWithValues)
+	ctx, cancel := context.WithCancel(ctxWithValues)
+	loginDone := false
+	defer func() {
+		if !loginDone {
+			cancel()
+		}
+	}()
 
 	// Use DirectUpdateOrCreateConfig to avoid atomic file operations (temp file + rename)
 	// which are blocked by the tvOS sandbox in App Group containers
@@ -474,7 +513,9 @@ func (c *Client) LoginForMobile() string {
 	}
 
 	// This could cause a potential race condition with loading the extension which need to be handled on swift side
+	loginDone = true
 	go func() {
+		defer cancel()
 		tokenInfo, err := oAuthFlow.WaitToken(ctx, flowInfo)
 		if err != nil {
 			log.Errorf("LoginForMobile: WaitToken failed: %v", err)
@@ -491,18 +532,18 @@ func (c *Client) LoginForMobile() string {
 			log.Errorf("LoginForMobile: Login failed: %v", err)
 			return
 		}
-		c.loginComplete = true
+		c.loginComplete.Store(true)
 	}()
 
 	return flowInfo.VerificationURIComplete
 }
 
 func (c *Client) IsLoginComplete() bool {
-	return c.loginComplete
+	return c.loginComplete.Load()
 }
 
 func (c *Client) ClearLoginComplete() {
-	c.loginComplete = false
+	c.loginComplete.Store(false)
 }
 
 func (c *Client) GetRoutesSelectionDetails() (*RoutesSelectionDetails, error) {
@@ -722,13 +763,36 @@ func (c *Client) DeselectRoute(id string) error {
 	return nil
 }
 
-// setState stores the running engine state so DebugBundle can reuse the live
-// config and ConnectClient. It is cleared on Stop.
-func (c *Client) setState(cfg *profilemanager.Config, cc *internal.ConnectClient) {
+func (c *Client) startRun(cancel context.CancelFunc) (chan struct{}, error) {
 	c.stateMu.Lock()
 	defer c.stateMu.Unlock()
+
+	if c.runDone != nil {
+		return nil, errClientAlreadyRunning
+	}
+
+	done := make(chan struct{})
+	c.runDone = done
+	c.ctxCancel = cancel
+	return done, nil
+}
+
+func (c *Client) finishRun(done chan struct{}) {
+	c.stateMu.Lock()
+	c.connectClient = nil
+	c.config = nil
+	c.runDone = nil
+	c.ctxCancel = nil
+	c.stateMu.Unlock()
+
+	close(done)
+}
+
+func (c *Client) setState(cfg *profilemanager.Config, cc *internal.ConnectClient) {
+	c.stateMu.Lock()
 	c.config = cfg
 	c.connectClient = cc
+	c.stateMu.Unlock()
 }
 
 // stateSnapshot returns the current config and ConnectClient under the lock.

client/ios/NetBirdSDK/login.go

@@ -11,6 +11,7 @@ import (
 
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mobile"
 	"github.com/netbirdio/netbird/client/system"
 )
 
@@ -284,12 +285,14 @@ func (a *Auth) login(urlOpener URLOpener, forceDeviceAuth bool, deviceName strin
 	}
 
 	jwtToken := ""
+	email := ""
 	if needsLogin {
 		tokenInfo, err := a.foregroundGetTokenInfo(authClient, urlOpener, forceDeviceAuth)
 		if err != nil {
 			return fmt.Errorf("interactive sso login failed: %v", err)
 		}
 		jwtToken = tokenInfo.GetTokenToUse()
+		email = tokenInfo.Email
 	}
 
 	err, isAuthError := authClient.Login(ctx, "", jwtToken)
@@ -301,6 +304,14 @@ func (a *Auth) login(urlOpener URLOpener, forceDeviceAuth bool, deviceName strin
 		return fmt.Errorf("login failed: %v", err)
 	}
 
+	// Stored after Login, not before: a rejected token must not leave a hint
+	// pointing at an account that cannot be used.
+	if email != "" && a.cfgPath != "" {
+		if err := mobile.WriteProfileEmail(a.cfgPath, email); err != nil {
+			log.Warnf("failed to store profile account email: %v", err)
+		}
+	}
+
 	// Save the config before notifying success to ensure persistence completes
 	// before the callback potentially triggers teardown on the Swift side.
 	// Note: This differs from Android which doesn't save config after login.
@@ -320,10 +331,24 @@ func (a *Auth) login(urlOpener URLOpener, forceDeviceAuth bool, deviceName strin
 	return nil
 }
 
+// profileLoginHint returns the stored account email for the profile at cfgPath,
+// so a re-login targets the account the profile already belongs to instead of
+// whatever session the shared browser cookie jar happens to hold.
+//
+// An empty hint is deliberate, not a fallback: a fresh profile leaves the
+// choice to the IdP. Switching accounts is done by switching or removing
+// profiles, not by logging out — logout keeps the email.
+func profileLoginHint(cfgPath string) string {
+	if cfgPath == "" {
+		return ""
+	}
+	return mobile.ReadProfileEmail(cfgPath)
+}
+
 const authInfoRequestTimeout = 30 * time.Second
 
 func (a *Auth) foregroundGetTokenInfo(authClient *auth.Auth, urlOpener URLOpener, forceDeviceAuth bool) (*auth.TokenInfo, error) {
-	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, forceDeviceAuth, "")
+	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, forceDeviceAuth, profileLoginHint(a.cfgPath))
 	if err != nil {
 		return nil, fmt.Errorf("failed to get OAuth flow: %v", err)
 	}

client/jobexec/executor.go

@@ -28,7 +28,11 @@ func NewExecutor() *Executor {
 	return &Executor{}
 }
 
-func (e *Executor) BundleJob(ctx context.Context, debugBundleDependencies debug.GeneratorDependencies, params debug.BundleConfig, waitForDuration time.Duration, mgmURL string) (string, error) {
+func (e *Executor) BundleJob(ctx context.Context, debugBundleDependencies debug.GeneratorDependencies, params debug.BundleConfig, waitForDuration time.Duration, mgmURL, uploadURL string) (string, error) {
+	if uploadURL == "" {
+		uploadURL = types.DefaultBundleURL
+	}
+
 	if waitForDuration > MaxBundleWaitTime {
 		log.Warnf("bundle wait time %v exceeds maximum %v, capping to maximum", waitForDuration, MaxBundleWaitTime)
 		waitForDuration = MaxBundleWaitTime
@@ -54,7 +58,7 @@ func (e *Executor) BundleJob(ctx context.Context, debugBundleDependencies debug.
 		}
 	}()
 
-	key, err := debug.UploadDebugBundle(ctx, types.DefaultBundleURL, mgmURL, path, false)
+	key, err := debug.UploadDebugBundle(ctx, uploadURL, mgmURL, path, false)
 	if err != nil {
 		log.Errorf("failed to upload debug bundle: %v", err)
 		return "", fmt.Errorf("upload debug bundle: %w", err)

client/mdm/canonical_loaders.go

@@ -27,9 +27,13 @@ var allKeys = []string{
 	KeyRosenpassEnabled,
 	KeyRosenpassPermissive,
 	KeyWireguardPort,
+	KeyEnableLocalMetrics,
+	KeyLocalMetricsAddress,
 	KeySplitTunnelMode,
 	KeySplitTunnelApps,
 	KeyLazyConnection,
+	KeyRemoteJobsAllowed,
+	KeyBundleUploadURL,
 }
 
 // canonicalKey maps the lowercase form of a managed-config value name to

client/mdm/canonical_loaders_test.go

@@ -0,0 +1,52 @@
+//go:build windows || darwin
+
+package mdm
+
+import (
+	"go/ast"
+	"go/parser"
+	"go/token"
+	"slices"
+	"strconv"
+	"testing"
+)
+
+// TestAllKeysCoversEveryPolicyKey guards against the drift that adding a Key*
+// constant without listing it in allKeys causes: the desktop loaders resolve
+// value names through canonicalKey, so an unlisted key is silently discarded as
+// unknown. policy.go is parsed rather than hand-mirrored so the test cannot go
+// stale in the same way.
+func TestAllKeysCoversEveryPolicyKey(t *testing.T) {
+	file, err := parser.ParseFile(token.NewFileSet(), "policy.go", nil, 0)
+	if err != nil {
+		t.Fatalf("parse policy.go: %v", err)
+	}
+
+	for _, decl := range file.Decls {
+		gen, ok := decl.(*ast.GenDecl)
+		if !ok || gen.Tok != token.CONST {
+			continue
+		}
+		for _, spec := range gen.Specs {
+			value, ok := spec.(*ast.ValueSpec)
+			if !ok || len(value.Names) != 1 || len(value.Values) != 1 {
+				continue
+			}
+			name := value.Names[0].Name
+			if len(name) < 4 || name[:3] != "Key" {
+				continue
+			}
+			lit, ok := value.Values[0].(*ast.BasicLit)
+			if !ok || lit.Kind != token.STRING {
+				continue
+			}
+			key, err := strconv.Unquote(lit.Value)
+			if err != nil {
+				t.Fatalf("unquote %s: %v", name, err)
+			}
+			if !slices.Contains(allKeys, key) {
+				t.Errorf("%s (%q) is missing from allKeys, so the desktop loaders discard it as unknown", name, key)
+			}
+		}
+	}
+}

client/mdm/policy.go

@@ -47,6 +47,8 @@ const (
 	KeyRosenpassEnabled    = "rosenpassEnabled"
 	KeyRosenpassPermissive = "rosenpassPermissive"
 	KeyWireguardPort       = "wireguardPort"
+	KeyEnableLocalMetrics  = "enableLocalMetrics"
+	KeyLocalMetricsAddress = "localMetricsAddress"
 
 	// Split tunnel is modeled as a single conceptual policy with two
 	// registry/plist values. KeySplitTunnelMode is the discriminator
@@ -60,6 +62,17 @@ const (
 	// the management feature flag. Read as a bool (native bool, or on/off,
 	// true/false, 1/0, yes/no); absent = defer to management.
 	KeyLazyConnection = "lazyConnection"
+
+	// KeyRemoteJobsAllowed opts the peer into management-requested remote jobs
+	// (e.g. debug bundles). Read as a bool; absent = defer to the local config
+	// (which defaults to disabled). Stored on Config as RemoteJobsAllowed.
+	KeyRemoteJobsAllowed = "allowRemoteJobs"
+
+	// KeyBundleUploadURL overrides the debug-bundle upload service URL for
+	// remote jobs, taking precedence over the management-supplied value. Read
+	// as a string; must be an https URL with a host. Absent = defer to the
+	// management-supplied URL (or the default upload server).
+	KeyBundleUploadURL = "debugBundleUploadURL"
 )
 
 // Split-tunnel mode literals (KeySplitTunnelMode values).
@@ -71,6 +84,8 @@ const (
 // SecretKeys lists keys whose values must be redacted in logs.
 var SecretKeys = map[string]struct{}{
 	KeyPreSharedKey: {},
+	// The upload URL can embed credentials or signed query tokens.
+	KeyBundleUploadURL: {},
 }
 
 // boolStringLiterals enumerates the textual boolean encodings the

client/mobile/profile_state.go

@@ -105,8 +105,14 @@ func WriteProfileEmail(configPath string, email string) error {
 		return fmt.Errorf("resolve profile account path: %w", err)
 	}
 
+	// DirectWriteJson, not the atomic writers: those create a temp file and
+	// rename it over the target, which the tvOS App Group sandbox blocks. It is
+	// the same reason the config next to this file goes through
+	// DirectWriteOutConfig. The file is rewritten whole from one key, so losing
+	// atomicity costs nothing beyond a torn write on a crash mid-write, which
+	// reads back as "no email" and is recovered by the next login.
 	state := profilemanager.ProfileState{Email: email}
-	if err := util.WriteJsonWithRestrictedPermission(context.Background(), accountPath, state); err != nil {
+	if err := util.DirectWriteJson(context.Background(), accountPath, state); err != nil {
 		return fmt.Errorf("write profile account: %w", err)
 	}
 

client/proto/daemon.pb.go

@@ -552,8 +552,13 @@ type LoginRequest struct {
 	DisableSSHAuth                *bool   `protobuf:"varint,38,opt,name=disableSSHAuth,proto3,oneof" json:"disableSSHAuth,omitempty"`
 	SshJWTCacheTTL                *int32  `protobuf:"varint,39,opt,name=sshJWTCacheTTL,proto3,oneof" json:"sshJWTCacheTTL,omitempty"`
 	DisableIpv6                   *bool   `protobuf:"varint,40,opt,name=disable_ipv6,json=disableIpv6,proto3,oneof" json:"disable_ipv6,omitempty"`
-	unknownFields                 protoimpl.UnknownFields
-	sizeCache                     protoimpl.SizeCache
+	EnableLocalMetrics            *bool   `protobuf:"varint,41,opt,name=enable_local_metrics,json=enableLocalMetrics,proto3,oneof" json:"enable_local_metrics,omitempty"`
+	LocalMetricsAddress           *string `protobuf:"bytes,42,opt,name=local_metrics_address,json=localMetricsAddress,proto3,oneof" json:"local_metrics_address,omitempty"`
+	// remoteJobsAllowed opts the peer into management-requested remote jobs
+	// (e.g. debug bundles). Absent leaves the stored value unchanged.
+	RemoteJobsAllowed *bool `protobuf:"varint,43,opt,name=remoteJobsAllowed,proto3,oneof" json:"remoteJobsAllowed,omitempty"`
+	unknownFields     protoimpl.UnknownFields
+	sizeCache         protoimpl.SizeCache
 }
 
 func (x *LoginRequest) Reset() {
@@ -867,6 +872,27 @@ func (x *LoginRequest) GetDisableIpv6() bool {
 	return false
 }
 
+func (x *LoginRequest) GetEnableLocalMetrics() bool {
+	if x != nil && x.EnableLocalMetrics != nil {
+		return *x.EnableLocalMetrics
+	}
+	return false
+}
+
+func (x *LoginRequest) GetLocalMetricsAddress() string {
+	if x != nil && x.LocalMetricsAddress != nil {
+		return *x.LocalMetricsAddress
+	}
+	return ""
+}
+
+func (x *LoginRequest) GetRemoteJobsAllowed() bool {
+	if x != nil && x.RemoteJobsAllowed != nil {
+		return *x.RemoteJobsAllowed
+	}
+	return false
+}
+
 type LoginResponse struct {
 	state                   protoimpl.MessageState `protogen:"open.v1"`
 	NeedsSSOLogin           bool                   `protobuf:"varint,1,opt,name=needsSSOLogin,proto3" json:"needsSSOLogin,omitempty"`
@@ -1424,6 +1450,7 @@ type GetConfigResponse struct {
 	DisableSSHAuth                bool   `protobuf:"varint,25,opt,name=disableSSHAuth,proto3" json:"disableSSHAuth,omitempty"`
 	SshJWTCacheTTL                int32  `protobuf:"varint,26,opt,name=sshJWTCacheTTL,proto3" json:"sshJWTCacheTTL,omitempty"`
 	DisableIpv6                   bool   `protobuf:"varint,27,opt,name=disable_ipv6,json=disableIpv6,proto3" json:"disable_ipv6,omitempty"`
+	RemoteJobsAllowed             bool   `protobuf:"varint,29,opt,name=remoteJobsAllowed,proto3" json:"remoteJobsAllowed,omitempty"`
 	// mDMManagedFields lists the names of configuration keys whose value is
 	// currently enforced by an MDM policy. Names match mdm.Key* constants
 	// (e.g. "managementURL", "disableClientRoutes"). UI/CLI clients should
@@ -1653,6 +1680,13 @@ func (x *GetConfigResponse) GetDisableIpv6() bool {
 	return false
 }
 
+func (x *GetConfigResponse) GetRemoteJobsAllowed() bool {
+	if x != nil {
+		return x.RemoteJobsAllowed
+	}
+	return false
+}
+
 func (x *GetConfigResponse) GetMDMManagedFields() []string {
 	if x != nil {
 		return x.MDMManagedFields
@@ -4442,8 +4476,13 @@ type SetConfigRequest struct {
 	DisableSSHAuth                *bool                `protobuf:"varint,33,opt,name=disableSSHAuth,proto3,oneof" json:"disableSSHAuth,omitempty"`
 	SshJWTCacheTTL                *int32               `protobuf:"varint,34,opt,name=sshJWTCacheTTL,proto3,oneof" json:"sshJWTCacheTTL,omitempty"`
 	DisableIpv6                   *bool                `protobuf:"varint,35,opt,name=disable_ipv6,json=disableIpv6,proto3,oneof" json:"disable_ipv6,omitempty"`
-	unknownFields                 protoimpl.UnknownFields
-	sizeCache                     protoimpl.SizeCache
+	EnableLocalMetrics            *bool                `protobuf:"varint,36,opt,name=enable_local_metrics,json=enableLocalMetrics,proto3,oneof" json:"enable_local_metrics,omitempty"`
+	LocalMetricsAddress           *string              `protobuf:"bytes,37,opt,name=local_metrics_address,json=localMetricsAddress,proto3,oneof" json:"local_metrics_address,omitempty"`
+	// remoteJobsAllowed opts the peer into management-requested remote jobs
+	// (e.g. debug bundles). Absent leaves the stored value unchanged.
+	RemoteJobsAllowed *bool `protobuf:"varint,38,opt,name=remoteJobsAllowed,proto3,oneof" json:"remoteJobsAllowed,omitempty"`
+	unknownFields     protoimpl.UnknownFields
+	sizeCache         protoimpl.SizeCache
 }
 
 func (x *SetConfigRequest) Reset() {
@@ -4721,6 +4760,27 @@ func (x *SetConfigRequest) GetDisableIpv6() bool {
 	return false
 }
 
+func (x *SetConfigRequest) GetEnableLocalMetrics() bool {
+	if x != nil && x.EnableLocalMetrics != nil {
+		return *x.EnableLocalMetrics
+	}
+	return false
+}
+
+func (x *SetConfigRequest) GetLocalMetricsAddress() string {
+	if x != nil && x.LocalMetricsAddress != nil {
+		return *x.LocalMetricsAddress
+	}
+	return ""
+}
+
+func (x *SetConfigRequest) GetRemoteJobsAllowed() bool {
+	if x != nil && x.RemoteJobsAllowed != nil {
+		return *x.RemoteJobsAllowed
+	}
+	return false
+}
+
 type SetConfigResponse struct {
 	state         protoimpl.MessageState `protogen:"open.v1"`
 	unknownFields protoimpl.UnknownFields
@@ -8165,7 +8225,7 @@ var File_daemon_proto protoreflect.FileDescriptor
 const file_daemon_proto_rawDesc = "" +
 	"\n" +
 	"\fdaemon.proto\x12\x06daemon\x1a google/protobuf/descriptor.proto\x1a\x1fgoogle/protobuf/timestamp.proto\x1a\x1egoogle/protobuf/duration.proto\"\x0e\n" +
-	"\fEmptyRequest\"\xef\x12\n" +
+	"\fEmptyRequest\"\xdb\x14\n" +
 	"\fLoginRequest\x12\x1a\n" +
 	"\bsetupKey\x18\x01 \x01(\tR\bsetupKey\x12&\n" +
 	"\fpreSharedKey\x18\x02 \x01(\tB\x02\x18\x01R\fpreSharedKey\x12$\n" +
@@ -8210,7 +8270,10 @@ const file_daemon_proto_rawDesc = "" +
 	"\x1denableSSHRemotePortForwarding\x18% \x01(\bH\x18R\x1denableSSHRemotePortForwarding\x88\x01\x01\x12+\n" +
 	"\x0edisableSSHAuth\x18& \x01(\bH\x19R\x0edisableSSHAuth\x88\x01\x01\x12+\n" +
 	"\x0esshJWTCacheTTL\x18' \x01(\x05H\x1aR\x0esshJWTCacheTTL\x88\x01\x01\x12&\n" +
-	"\fdisable_ipv6\x18( \x01(\bH\x1bR\vdisableIpv6\x88\x01\x01B\x13\n" +
+	"\fdisable_ipv6\x18( \x01(\bH\x1bR\vdisableIpv6\x88\x01\x01\x125\n" +
+	"\x14enable_local_metrics\x18) \x01(\bH\x1cR\x12enableLocalMetrics\x88\x01\x01\x127\n" +
+	"\x15local_metrics_address\x18* \x01(\tH\x1dR\x13localMetricsAddress\x88\x01\x01\x121\n" +
+	"\x11remoteJobsAllowed\x18+ \x01(\bH\x1eR\x11remoteJobsAllowed\x88\x01\x01B\x13\n" +
 	"\x11_rosenpassEnabledB\x10\n" +
 	"\x0e_interfaceNameB\x10\n" +
 	"\x0e_wireguardPortB\x17\n" +
@@ -8238,7 +8301,10 @@ const file_daemon_proto_rawDesc = "" +
 	"\x1e_enableSSHRemotePortForwardingB\x11\n" +
 	"\x0f_disableSSHAuthB\x11\n" +
 	"\x0f_sshJWTCacheTTLB\x0f\n" +
-	"\r_disable_ipv6\"\xb5\x01\n" +
+	"\r_disable_ipv6B\x17\n" +
+	"\x15_enable_local_metricsB\x18\n" +
+	"\x16_local_metrics_addressB\x14\n" +
+	"\x12_remoteJobsAllowed\"\xb5\x01\n" +
 	"\rLoginResponse\x12$\n" +
 	"\rneedsSSOLogin\x18\x01 \x01(\bR\rneedsSSOLogin\x12\x1a\n" +
 	"\buserCode\x18\x02 \x01(\tR\buserCode\x12(\n" +
@@ -8273,7 +8339,7 @@ const file_daemon_proto_rawDesc = "" +
 	"\fDownResponse\"P\n" +
 	"\x10GetConfigRequest\x12 \n" +
 	"\vprofileName\x18\x01 \x01(\tR\vprofileName\x12\x1a\n" +
-	"\busername\x18\x02 \x01(\tR\busername\"\xaa\t\n" +
+	"\busername\x18\x02 \x01(\tR\busername\"\xd8\t\n" +
 	"\x11GetConfigResponse\x12$\n" +
 	"\rmanagementUrl\x18\x01 \x01(\tR\rmanagementUrl\x12\x1e\n" +
 	"\n" +
@@ -8305,7 +8371,8 @@ const file_daemon_proto_rawDesc = "" +
 	"\x1denableSSHRemotePortForwarding\x18\x17 \x01(\bR\x1denableSSHRemotePortForwarding\x12&\n" +
 	"\x0edisableSSHAuth\x18\x19 \x01(\bR\x0edisableSSHAuth\x12&\n" +
 	"\x0esshJWTCacheTTL\x18\x1a \x01(\x05R\x0esshJWTCacheTTL\x12!\n" +
-	"\fdisable_ipv6\x18\x1b \x01(\bR\vdisableIpv6\x12*\n" +
+	"\fdisable_ipv6\x18\x1b \x01(\bR\vdisableIpv6\x12,\n" +
+	"\x11remoteJobsAllowed\x18\x1d \x01(\bR\x11remoteJobsAllowed\x12*\n" +
 	"\x10mDMManagedFields\x18\x1c \x03(\tR\x10mDMManagedFields\"\x92\x06\n" +
 	"\tPeerState\x12\x0e\n" +
 	"\x02IP\x18\x01 \x01(\tR\x02IP\x12\x16\n" +
@@ -8533,7 +8600,7 @@ const file_daemon_proto_rawDesc = "" +
 	"\f_profileNameB\v\n" +
 	"\t_username\"'\n" +
 	"\x15SwitchProfileResponse\x12\x0e\n" +
-	"\x02id\x18\x01 \x01(\tR\x02id\"\x98\x11\n" +
+	"\x02id\x18\x01 \x01(\tR\x02id\"\x84\x13\n" +
 	"\x10SetConfigRequest\x12\x1a\n" +
 	"\busername\x18\x01 \x01(\tR\busername\x12 \n" +
 	"\vprofileName\x18\x02 \x01(\tR\vprofileName\x12$\n" +
@@ -8573,7 +8640,10 @@ const file_daemon_proto_rawDesc = "" +
 	"\x1denableSSHRemotePortForwarding\x18  \x01(\bH\x15R\x1denableSSHRemotePortForwarding\x88\x01\x01\x12+\n" +
 	"\x0edisableSSHAuth\x18! \x01(\bH\x16R\x0edisableSSHAuth\x88\x01\x01\x12+\n" +
 	"\x0esshJWTCacheTTL\x18\" \x01(\x05H\x17R\x0esshJWTCacheTTL\x88\x01\x01\x12&\n" +
-	"\fdisable_ipv6\x18# \x01(\bH\x18R\vdisableIpv6\x88\x01\x01B\x13\n" +
+	"\fdisable_ipv6\x18# \x01(\bH\x18R\vdisableIpv6\x88\x01\x01\x125\n" +
+	"\x14enable_local_metrics\x18$ \x01(\bH\x19R\x12enableLocalMetrics\x88\x01\x01\x127\n" +
+	"\x15local_metrics_address\x18% \x01(\tH\x1aR\x13localMetricsAddress\x88\x01\x01\x121\n" +
+	"\x11remoteJobsAllowed\x18& \x01(\bH\x1bR\x11remoteJobsAllowed\x88\x01\x01B\x13\n" +
 	"\x11_rosenpassEnabledB\x10\n" +
 	"\x0e_interfaceNameB\x10\n" +
 	"\x0e_wireguardPortB\x17\n" +
@@ -8598,7 +8668,10 @@ const file_daemon_proto_rawDesc = "" +
 	"\x1e_enableSSHRemotePortForwardingB\x11\n" +
 	"\x0f_disableSSHAuthB\x11\n" +
 	"\x0f_sshJWTCacheTTLB\x0f\n" +
-	"\r_disable_ipv6\"\x13\n" +
+	"\r_disable_ipv6B\x17\n" +
+	"\x15_enable_local_metricsB\x18\n" +
+	"\x16_local_metrics_addressB\x14\n" +
+	"\x12_remoteJobsAllowed\"\x13\n" +
 	"\x11SetConfigResponse\"Q\n" +
 	"\x11AddProfileRequest\x12\x1a\n" +
 	"\busername\x18\x01 \x01(\tR\busername\x12 \n" +

client/proto/daemon.proto

@@ -268,6 +268,12 @@ message LoginRequest {
   optional bool disableSSHAuth = 38;
   optional int32 sshJWTCacheTTL = 39;
   optional bool disable_ipv6 = 40;
+
+  optional bool enable_local_metrics = 41;
+  optional string local_metrics_address = 42;
+  // remoteJobsAllowed opts the peer into management-requested remote jobs
+  // (e.g. debug bundles). Absent leaves the stored value unchanged.
+  optional bool remoteJobsAllowed = 43;
 }
 
 message LoginResponse {
@@ -388,6 +394,8 @@ message GetConfigResponse {
 
   bool disable_ipv6 = 27;
 
+  bool remoteJobsAllowed = 29;
+
   // mDMManagedFields lists the names of configuration keys whose value is
   // currently enforced by an MDM policy. Names match mdm.Key* constants
   // (e.g. "managementURL", "disableClientRoutes"). UI/CLI clients should
@@ -792,6 +800,12 @@ message SetConfigRequest {
   optional bool disableSSHAuth = 33;
   optional int32 sshJWTCacheTTL = 34;
   optional bool disable_ipv6 = 35;
+
+  optional bool enable_local_metrics = 36;
+  optional string local_metrics_address = 37;
+  // remoteJobsAllowed opts the peer into management-requested remote jobs
+  // (e.g. debug bundles). Absent leaves the stored value unchanged.
+  optional bool remoteJobsAllowed = 38;
 }
 
 message SetConfigResponse{}

client/server/logout_gate_test.go

@@ -0,0 +1,200 @@
+package server
+
+import (
+	"context"
+	"path/filepath"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+	"google.golang.org/grpc/codes"
+	gstatus "google.golang.org/grpc/status"
+
+	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/proto"
+)
+
+// unreachableManagementURL keeps a test that is expected to stop at a gate from
+// reaching the network if the gate ever regresses: the profiles a logout must
+// not touch point here, so a leak fails fast instead of contacting a real
+// management server.
+const unreachableManagementURL = "https://127.0.0.1:9"
+
+// enableSSHOnProfile rewrites the profile config at cfgPath with the SSH server
+// enabled. Deregistering an SSH-enabled profile is a privileged change, so an
+// unprivileged caller is refused by requirePrivilegeForDeregistration before any
+// management connection is attempted, which is what keeps these tests offline.
+func enableSSHOnProfile(t *testing.T, cfgPath string) {
+	t.Helper()
+	_, err := profilemanager.UpdateOrCreateConfig(profilemanager.ConfigInput{
+		ConfigPath:       cfgPath,
+		ManagementURL:    "https://api.netbird.io:443",
+		ServerSSHAllowed: boolPtr(true),
+	})
+	require.NoError(t, err)
+}
+
+// Logging out of the profile the daemon is already running is a deregistration,
+// not profile management, so the profiles-disabled kill switch must not block
+// it. The desktop UI always addresses logout by profile (both the profile menu
+// and the session-expiration dialog), so gating it left users with
+// disableProfiles enforced unable to log out at all.
+func TestLogout_ActiveProfileAllowedWhenProfilesDisabled(t *testing.T) {
+	s, _, activeProfile, username, cfgPath := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+	enableSSHOnProfile(t, cfgPath)
+
+	s.profilesDisabled = true
+
+	_, err := s.Logout(userCtx(), &proto.LogoutRequest{
+		ProfileName: &activeProfile,
+		Username:    &username,
+	})
+
+	require.Error(t, err, "the SSH privilege gate is expected to refuse this unprivileged caller")
+	require.Equal(t, codes.PermissionDenied, gstatus.Code(err),
+		"logout of the active profile must reach the deregistration path, not be refused as profile management: %v", err)
+	require.NotContains(t, gstatus.Convert(err).Message(), errProfilesDisabled)
+}
+
+// A profile-addressed logout that targets some *other* profile does manage
+// profiles, so it stays gated: with profiles disabled the daemon must not
+// deregister a peer the user is not currently running.
+func TestLogout_OtherProfileStaysGatedWhenProfilesDisabled(t *testing.T) {
+	s, _, _, username, _ := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+
+	other := "other-profile"
+	_, err := profilemanager.UpdateOrCreateConfig(profilemanager.ConfigInput{
+		ConfigPath:    filepath.Join(profilemanager.DefaultConfigPathDir, other+".json"),
+		ManagementURL: unreachableManagementURL,
+	})
+	require.NoError(t, err)
+
+	s.profilesDisabled = true
+
+	_, err = s.Logout(userCtx(), &proto.LogoutRequest{
+		ProfileName: &other,
+		Username:    &username,
+	})
+
+	require.Error(t, err)
+	require.Equal(t, codes.Unavailable, gstatus.Code(err), "want the profiles-disabled refusal, got %v", err)
+	require.Contains(t, gstatus.Convert(err).Message(), errProfilesDisabled)
+}
+
+// A legacy profile ID is a display name, so two users can hold the same ID in
+// their own profile directories. Matching on the ID alone would let one user's
+// logout pass the gate against the other user's active profile, so the username
+// is part of the comparison.
+func TestLogout_ForeignUserProfileStaysGatedWhenProfilesDisabled(t *testing.T) {
+	s, _, _, username, _ := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+
+	// A legacy-style profile whose ID is its filename stem, and an active state
+	// claiming that same ID for a different user.
+	shared := "shared-legacy-name"
+	_, err := profilemanager.UpdateOrCreateConfig(profilemanager.ConfigInput{
+		ConfigPath:    filepath.Join(profilemanager.DefaultConfigPathDir, shared+".json"),
+		ManagementURL: unreachableManagementURL,
+	})
+	require.NoError(t, err)
+	require.NoError(t, s.profileManager.SetActiveProfileState(&profilemanager.ActiveProfileState{
+		ID:       profilemanager.ID(shared),
+		Username: "someone-else",
+	}))
+
+	s.profilesDisabled = true
+
+	_, err = s.Logout(userCtx(), &proto.LogoutRequest{
+		ProfileName: &shared,
+		Username:    &username,
+	})
+
+	require.Error(t, err)
+	require.Equal(t, codes.Unavailable, gstatus.Code(err),
+		"another user's profile must not pass the gate on an ID match alone: %v", err)
+}
+
+// Deregistering a namesake profile must not go out with the running config.
+// logoutFromProfile reuses the connected client's config when the target is the
+// active profile, and on an ID-only match a shared legacy ID made it reuse it
+// for another user's profile, deregistering the active peer instead.
+func TestLogout_ForeignUserProfileDoesNotUseTheRunningConfig(t *testing.T) {
+	s, _, _, username, cfgPath := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+
+	// The running config has the SSH server enabled, so reusing it would be
+	// refused with PermissionDenied. The namesake profile does not, so the
+	// correct path gets as far as dialing its own unreachable management URL.
+	enableSSHOnProfile(t, cfgPath)
+	running, err := profilemanager.GetConfig(cfgPath)
+	require.NoError(t, err)
+	s.config = running
+	s.connectClient = newDummyConnectClient(context.Background())
+
+	shared := "shared-legacy-name"
+	_, err = profilemanager.UpdateOrCreateConfig(profilemanager.ConfigInput{
+		ConfigPath:    filepath.Join(profilemanager.DefaultConfigPathDir, shared+".json"),
+		ManagementURL: unreachableManagementURL,
+	})
+	require.NoError(t, err)
+	require.NoError(t, s.profileManager.SetActiveProfileState(&profilemanager.ActiveProfileState{
+		ID:       profilemanager.ID(shared),
+		Username: "someone-else",
+	}))
+
+	// Bounded so the deregistration the fixed path attempts fails on the dial
+	// rather than sitting in gRPC backoff for the whole test timeout.
+	ctx, cancel := context.WithTimeout(userCtx(), 2*time.Second)
+	t.Cleanup(cancel)
+
+	_, err = s.Logout(ctx, &proto.LogoutRequest{
+		ProfileName: &shared,
+		Username:    &username,
+	})
+
+	require.Error(t, err)
+	require.NotEqual(t, codes.PermissionDenied, gstatus.Code(err),
+		"the namesake profile was deregistered with the running config: %v", err)
+}
+
+// The connection teardown follows the profile that is active when the logout
+// completes, not the one seen before it started: Login switches profiles under
+// guardedConfigMu, which the logout path does not hold, so a login that landed
+// meanwhile must keep its connection.
+func TestCleanupAfterProfileLogout_FollowsTheCurrentActiveProfile(t *testing.T) {
+	s, _, activeProfile, username, _ := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+
+	state := internal.CtxGetState(s.rootCtx)
+
+	s.cleanupAfterProfileLogout("some-other-profile", username)
+	status, err := state.Status()
+	require.NoError(t, err)
+	require.NotEqual(t, internal.StatusNeedsLogin, status,
+		"logging out of a profile that is not active must not ask for a new login")
+
+	s.cleanupAfterProfileLogout(profilemanager.ID(activeProfile), username)
+	status, err = state.Status()
+	require.NoError(t, err)
+	require.Equal(t, internal.StatusNeedsLogin, status,
+		"logging out of the active profile must ask for a new login")
+}
+
+// With profiles enabled the gate is out of the way on both surfaces; the active
+// profile still reaches the deregistration path.
+func TestLogout_ActiveProfileAllowedWhenProfilesEnabled(t *testing.T) {
+	s, _, activeProfile, username, cfgPath := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+	enableSSHOnProfile(t, cfgPath)
+
+	_, err := s.Logout(userCtx(), &proto.LogoutRequest{
+		ProfileName: &activeProfile,
+		Username:    &username,
+	})
+
+	require.Error(t, err, "the SSH privilege gate is expected to refuse this unprivileged caller")
+	require.Equal(t, codes.PermissionDenied, gstatus.Code(err), "want the privilege refusal, got %v", err)
+}

client/server/mdm.go

@@ -233,6 +233,24 @@ func conflictString(key, got string) conflictCheck {
 	}
 }
 
+// conflictStringPtr is conflictString for optional proto fields, where an
+// explicit empty value is still a request to change the setting. If p is
+// nil the field is treated as matching (no override requested); otherwise
+// the check returns true only when the policy contains the key and its
+// value equals *p.
+func conflictStringPtr(key string, p *string) conflictCheck {
+	return conflictCheck{
+		key: key,
+		check: func(pol *mdm.Policy) bool {
+			if p == nil {
+				return true
+			}
+			want, ok := pol.GetString(key)
+			return ok && want == *p
+		},
+	}
+}
+
 // conflictInt64 builds a conflictCheck for an integer MDM key. If p is
 // nil the field is treated as matching; otherwise the check returns
 // true only when the policy contains the key and its int value equals *p.
@@ -297,10 +315,13 @@ func mdmManagedFieldConflicts(msg *proto.SetConfigRequest, policy *mdm.Policy) [
 		conflictBool(mdm.KeyRosenpassPermissive, msg.RosenpassPermissive),
 		conflictBool(mdm.KeyDisableAutoConnect, msg.DisableAutoConnect),
 		conflictBool(mdm.KeyAllowServerSSH, msg.ServerSSHAllowed),
+		conflictBool(mdm.KeyRemoteJobsAllowed, msg.RemoteJobsAllowed),
 		conflictBool(mdm.KeyDisableClientRoutes, msg.DisableClientRoutes),
 		conflictBool(mdm.KeyDisableServerRoutes, msg.DisableServerRoutes),
 		conflictBool(mdm.KeyBlockInbound, msg.BlockInbound),
 		conflictInt64(mdm.KeyWireguardPort, msg.WireguardPort),
+		conflictBool(mdm.KeyEnableLocalMetrics, msg.EnableLocalMetrics),
+		conflictStringPtr(mdm.KeyLocalMetricsAddress, msg.LocalMetricsAddress),
 	})
 }
 
@@ -332,6 +353,7 @@ func setConfigRequestHasConfigOverrides(msg *proto.SetConfigRequest) bool {
 		msg.Mtu != nil ||
 		msg.DisableAutoConnect != nil ||
 		msg.ServerSSHAllowed != nil ||
+		msg.RemoteJobsAllowed != nil ||
 		msg.NetworkMonitor != nil ||
 		msg.DisableClientRoutes != nil ||
 		msg.DisableServerRoutes != nil ||
@@ -346,7 +368,9 @@ func setConfigRequestHasConfigOverrides(msg *proto.SetConfigRequest) bool {
 		msg.EnableSSHLocalPortForwarding != nil ||
 		msg.EnableSSHRemotePortForwarding != nil ||
 		msg.DisableSSHAuth != nil ||
-		msg.SshJWTCacheTTL != nil
+		msg.SshJWTCacheTTL != nil ||
+		msg.EnableLocalMetrics != nil ||
+		msg.LocalMetricsAddress != nil
 }
 
 // loginRequestHasConfigOverrides reports whether the LoginRequest
@@ -370,6 +394,7 @@ func loginRequestHasConfigOverrides(msg *proto.LoginRequest) bool {
 		msg.WireguardPort != nil ||
 		msg.DisableAutoConnect != nil ||
 		msg.ServerSSHAllowed != nil ||
+		msg.RemoteJobsAllowed != nil ||
 		msg.RosenpassPermissive != nil ||
 		len(msg.ExtraIFaceBlacklist) > 0 ||
 		msg.NetworkMonitor != nil ||
@@ -381,7 +406,9 @@ func loginRequestHasConfigOverrides(msg *proto.LoginRequest) bool {
 		msg.BlockLanAccess != nil ||
 		msg.DisableNotifications != nil ||
 		len(msg.DnsLabels) > 0 || msg.CleanDNSLabels ||
-		msg.BlockInbound != nil
+		msg.BlockInbound != nil ||
+		msg.EnableLocalMetrics != nil ||
+		msg.LocalMetricsAddress != nil
 }
 
 // loginRequestMDMConflicts mirrors mdmManagedFieldConflicts but for the
@@ -418,10 +445,13 @@ func loginRequestMDMConflicts(msg *proto.LoginRequest, policy *mdm.Policy) []str
 		conflictBool(mdm.KeyRosenpassPermissive, msg.RosenpassPermissive),
 		conflictBool(mdm.KeyDisableAutoConnect, msg.DisableAutoConnect),
 		conflictBool(mdm.KeyAllowServerSSH, msg.ServerSSHAllowed),
+		conflictBool(mdm.KeyRemoteJobsAllowed, msg.RemoteJobsAllowed),
 		conflictBool(mdm.KeyDisableClientRoutes, msg.DisableClientRoutes),
 		conflictBool(mdm.KeyDisableServerRoutes, msg.DisableServerRoutes),
 		conflictBool(mdm.KeyBlockInbound, msg.BlockInbound),
 		conflictInt64(mdm.KeyWireguardPort, msg.WireguardPort),
+		conflictBool(mdm.KeyEnableLocalMetrics, msg.EnableLocalMetrics),
+		conflictStringPtr(mdm.KeyLocalMetricsAddress, msg.LocalMetricsAddress),
 	})
 }
 

client/server/network.go

@@ -232,4 +232,3 @@ func toNetIDs(routes []string) []route.NetID {
 	}
 	return netIDs
 }
-

client/server/server.go

@@ -23,6 +23,9 @@ import (
 
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/expose"
+	"github.com/prometheus/client_golang/prometheus"
+
+	"github.com/netbirdio/netbird/client/internal/localmetrics"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	sleephandler "github.com/netbirdio/netbird/client/internal/sleep/handler"
 	"github.com/netbirdio/netbird/client/mdm"
@@ -36,6 +39,7 @@ import (
 	"github.com/netbirdio/netbird/client/internal/statemanager"
 	"github.com/netbirdio/netbird/client/internal/updater"
 	"github.com/netbirdio/netbird/client/proto"
+	"github.com/netbirdio/netbird/util"
 	"github.com/netbirdio/netbird/util/capture"
 	"github.com/netbirdio/netbird/version"
 )
@@ -109,6 +113,7 @@ type Server struct {
 
 	statusRecorder *peer.Status
 	sessionWatcher *internal.SessionWatcher
+	localMetrics   *localmetrics.Manager
 
 	fileDrop *filedrop.Manager
 
@@ -174,9 +179,28 @@ func New(ctx context.Context, logFile string, configFile string, profilesDisable
 	s.sleepHandler = sleephandler.New(agent)
 	s.startSleepDetector()
 
+	s.localMetrics = localmetrics.NewManager(ctx, s.statusRecorder, s.clientMetricsGatherer)
+
 	return s
 }
 
+// clientMetricsGatherer returns the Prometheus gatherer of the running
+// engine's client metrics, or nil when no engine is running.
+func (s *Server) clientMetricsGatherer() prometheus.Gatherer {
+	s.mutex.Lock()
+	connectClient := s.connectClient
+	s.mutex.Unlock()
+
+	if connectClient == nil {
+		return nil
+	}
+	engine := connectClient.Engine()
+	if engine == nil {
+		return nil
+	}
+	return engine.GetClientMetrics().PrometheusGatherer()
+}
+
 func (s *Server) Start() error {
 	s.mutex.Lock()
 	defer s.mutex.Unlock()
@@ -257,6 +281,7 @@ func (s *Server) Start() error {
 
 	s.statusRecorder.UpdateManagementAddress(config.ManagementURL.String())
 	s.statusRecorder.UpdateRosenpass(config.RosenpassEnabled, config.RosenpassPermissive)
+	s.localMetrics.Reconcile(config.LocalMetricsEnabled, config.LocalMetricsAddress)
 
 	if s.sessionWatcher == nil {
 		s.sessionWatcher = internal.NewSessionWatcher(s.rootCtx, s.statusRecorder)
@@ -480,11 +505,18 @@ func (s *Server) SetConfig(callerCtx context.Context, msg *proto.SetConfigReques
 		return nil, err
 	}
 
-	if _, err := profilemanager.UpdateConfig(config); err != nil {
+	updatedConf, err := profilemanager.UpdateConfig(config)
+	if err != nil {
 		log.Errorf("failed to update profile config: %v", err)
 		return nil, fmt.Errorf("failed to update profile config: %w", err)
 	}
 
+	if activeProf, err := s.profileManager.GetActiveProfileState(); err == nil {
+		if activePath, err := activeProf.FilePath(); err == nil && activePath == config.ConfigPath {
+			s.localMetrics.Reconcile(updatedConf.LocalMetricsEnabled, updatedConf.LocalMetricsAddress)
+		}
+	}
+
 	return &proto.SetConfigResponse{}, nil
 }
 
@@ -554,8 +586,11 @@ func (s *Server) setConfigInputFromRequest(msg *proto.SetConfigRequest) (profile
 
 	config.RosenpassEnabled = msg.RosenpassEnabled
 	config.RosenpassPermissive = msg.RosenpassPermissive
+	config.LocalMetricsEnabled = msg.EnableLocalMetrics
+	config.LocalMetricsAddress = msg.LocalMetricsAddress
 	config.DisableAutoConnect = msg.DisableAutoConnect
 	config.ServerSSHAllowed = msg.ServerSSHAllowed
+	config.RemoteJobsAllowed = msg.RemoteJobsAllowed
 	config.NetworkMonitor = msg.NetworkMonitor
 	config.DisableClientRoutes = msg.DisableClientRoutes
 	config.DisableServerRoutes = msg.DisableServerRoutes
@@ -660,6 +695,8 @@ func (s *Server) Login(callerCtx context.Context, msg *proto.LoginRequest) (*pro
 	s.config = config
 	s.mutex.Unlock()
 
+	s.localMetrics.Reconcile(config.LocalMetricsEnabled, config.LocalMetricsAddress)
+
 	// A probe that errors leaves the login undecided: Management unreachable, a
 	// restart mid-request, an internal error. Those are returned for the caller
 	// to retry, because turning them into an SSO prompt asks the user to solve
@@ -678,54 +715,7 @@ func (s *Server) Login(callerCtx context.Context, msg *proto.LoginRequest) (*pro
 	}
 
 	if msg.SetupKey == "" {
-		hint := ""
-		if msg.Hint != nil {
-			hint = *msg.Hint
-		}
-		oAuthFlow, err := auth.NewOAuthFlow(ctx, config, msg.IsUnixDesktopClient, false, hint)
-		if err != nil {
-			state.Set(internal.StatusLoginFailed)
-			return nil, err
-		}
-
-		if s.oauthAuthFlow.flow != nil && s.oauthAuthFlow.flow.GetClientID(ctx) == oAuthFlow.GetClientID(ctx) {
-			if s.oauthAuthFlow.expiresAt.After(time.Now().Add(90 * time.Second)) {
-				log.Debugf("using previous oauth flow info")
-				state.Set(internal.StatusNeedsLogin)
-				return &proto.LoginResponse{
-					NeedsSSOLogin:           true,
-					VerificationURI:         s.oauthAuthFlow.info.VerificationURI,
-					VerificationURIComplete: s.oauthAuthFlow.info.VerificationURIComplete,
-					UserCode:                s.oauthAuthFlow.info.UserCode,
-				}, nil
-			} else {
-				log.Warnf("canceling previous waiting execution")
-				if s.oauthAuthFlow.waitCancel != nil {
-					s.oauthAuthFlow.waitCancel()
-				}
-			}
-		}
-
-		authInfo, err := oAuthFlow.RequestAuthInfo(ctx)
-		if err != nil {
-			log.Errorf("getting a request OAuth flow failed: %v", err)
-			return nil, err
-		}
-
-		s.mutex.Lock()
-		s.oauthAuthFlow.flow = oAuthFlow
-		s.oauthAuthFlow.info = authInfo
-		s.oauthAuthFlow.expiresAt = time.Now().Add(time.Duration(authInfo.ExpiresIn) * time.Second)
-		s.mutex.Unlock()
-
-		state.Set(internal.StatusNeedsLogin)
-
-		return &proto.LoginResponse{
-			NeedsSSOLogin:           true,
-			VerificationURI:         authInfo.VerificationURI,
-			VerificationURIComplete: authInfo.VerificationURIComplete,
-			UserCode:                authInfo.UserCode,
-		}, nil
+		return s.beginSSOLogin(ctx, config, msg)
 	}
 
 	// Setup-key path: we are about to dial Management with the key, so the
@@ -741,6 +731,76 @@ func (s *Server) Login(callerCtx context.Context, msg *proto.LoginRequest) (*pro
 	return &proto.LoginResponse{}, nil
 }
 
+// beginSSOLogin starts the browser leg of a login that carries no setup key and
+// returns the response that parks the caller on it.
+func (s *Server) beginSSOLogin(ctx context.Context, config *profilemanager.Config, msg *proto.LoginRequest) (*proto.LoginResponse, error) {
+	state := internal.CtxGetState(s.rootCtx)
+
+	hint := ""
+	if msg.Hint != nil {
+		hint = *msg.Hint
+	}
+	oAuthFlow, err := auth.NewOAuthFlow(ctx, config, msg.IsUnixDesktopClient, false, hint)
+	if err != nil {
+		state.Set(internal.StatusLoginFailed)
+		return nil, err
+	}
+
+	if resp := s.pendingOAuthFlowResponse(ctx, oAuthFlow); resp != nil {
+		state.Set(internal.StatusNeedsLogin)
+		return resp, nil
+	}
+
+	authInfo, err := oAuthFlow.RequestAuthInfo(ctx)
+	if err != nil {
+		log.Errorf("getting a request OAuth flow failed: %v", err)
+		return nil, err
+	}
+
+	s.mutex.Lock()
+	s.oauthAuthFlow.flow = oAuthFlow
+	s.oauthAuthFlow.info = authInfo
+	s.oauthAuthFlow.expiresAt = time.Now().Add(time.Duration(authInfo.ExpiresIn) * time.Second)
+	s.mutex.Unlock()
+
+	state.Set(internal.StatusNeedsLogin)
+
+	return &proto.LoginResponse{
+		NeedsSSOLogin:           true,
+		VerificationURI:         authInfo.VerificationURI,
+		VerificationURIComplete: authInfo.VerificationURIComplete,
+		UserCode:                authInfo.UserCode,
+	}, nil
+}
+
+// pendingOAuthFlowResponse returns the in-flight flow's response when it
+// targets the same IdP client and has enough time left for the user to finish
+// the browser leg, so a second login joins the pending flow instead of opening
+// a competing one. A flow too close to expiry has its waiter cancelled and nil
+// returned, leaving the caller to start a fresh flow.
+func (s *Server) pendingOAuthFlowResponse(ctx context.Context, oAuthFlow auth.OAuthFlow) *proto.LoginResponse {
+	if s.oauthAuthFlow.flow == nil || s.oauthAuthFlow.flow.GetClientID(ctx) != oAuthFlow.GetClientID(ctx) {
+		return nil
+	}
+
+	if s.oauthAuthFlow.expiresAt.After(time.Now().Add(90 * time.Second)) {
+		log.Debugf("using previous oauth flow info")
+		return &proto.LoginResponse{
+			NeedsSSOLogin:           true,
+			VerificationURI:         s.oauthAuthFlow.info.VerificationURI,
+			VerificationURIComplete: s.oauthAuthFlow.info.VerificationURIComplete,
+			UserCode:                s.oauthAuthFlow.info.UserCode,
+		}
+	}
+
+	log.Warnf("canceling previous waiting execution")
+	if s.oauthAuthFlow.waitCancel != nil {
+		s.oauthAuthFlow.waitCancel()
+	}
+
+	return nil
+}
+
 // WaitSSOLogin validates the supplied userCode against the in-flight OAuth
 // device/PKCE flow and blocks until the user finishes the browser leg.
 //
@@ -1010,6 +1070,7 @@ func (s *Server) Up(callerCtx context.Context, msg *proto.UpRequest) (*proto.UpR
 
 	s.statusRecorder.UpdateManagementAddress(s.config.ManagementURL.String())
 	s.statusRecorder.UpdateRosenpass(s.config.RosenpassEnabled, s.config.RosenpassPermissive)
+	s.localMetrics.Reconcile(s.config.LocalMetricsEnabled, s.config.LocalMetricsAddress)
 
 	s.clientRunning = true
 	s.clientRunningChan = make(chan struct{})
@@ -1187,6 +1248,7 @@ func (s *Server) SwitchProfile(callerCtx context.Context, msg *proto.SwitchProfi
 	}
 
 	s.config = config
+	s.localMetrics.Reconcile(config.LocalMetricsEnabled, config.LocalMetricsAddress)
 
 	if msg != nil && msg.ProfileName != nil {
 		s.publishProfileListChanged(*msg.ProfileName)
@@ -1313,11 +1375,16 @@ func (s *Server) handleProfileLogout(ctx context.Context, msg *proto.LogoutReque
 		return nil, err
 	}
 
-	if err := s.validateProfileOperation(resolved.ID, true); err != nil {
+	activeProf, err := s.profileManager.GetActiveProfileState()
+	if err != nil {
+		return nil, gstatus.Errorf(codes.FailedPrecondition, "failed to get active profile state: %v", err)
+	}
+
+	if err := s.validateProfileLogout(resolved.ID, isActiveProfile(activeProf, resolved.ID, username)); err != nil {
 		return nil, err
 	}
 
-	if err := s.logoutFromProfile(ctx, resolved); err != nil {
+	if err := s.logoutFromProfile(ctx, resolved, username); err != nil {
 		log.Errorf("failed to logout from profile %s: %v", resolved.ID, err)
 		// A refused deregistration is already a status error carrying the reason
 		// and the command to run; rewrapping it as Internal would flatten both
@@ -1328,18 +1395,35 @@ func (s *Server) handleProfileLogout(ctx context.Context, msg *proto.LogoutReque
 		return nil, gstatus.Errorf(codes.Internal, "logout: %v", err)
 	}
 
-	activeProf, _ := s.profileManager.GetActiveProfileState()
-	if activeProf != nil && activeProf.ID == resolved.ID {
-		if err := s.cleanupConnection(); err != nil && !errors.Is(err, ErrServiceNotUp) {
-			log.Errorf("failed to cleanup connection: %v", err)
-		}
-		state := internal.CtxGetState(s.rootCtx)
-		state.Set(internal.StatusNeedsLogin)
-	}
+	s.cleanupAfterProfileLogout(resolved.ID, username)
 
 	return &proto.LogoutResponse{}, nil
 }
 
+// cleanupAfterProfileLogout tears the connection down and asks for a new login
+// when the profile that was just deregistered is the one the daemon is running.
+// The active profile is read again here rather than reused from the pre-flight
+// check: Login switches profiles under guardedConfigMu, which this path does not
+// hold, so a login that landed meanwhile must not have its fresh connection
+// dropped by a logout that targeted the profile it replaced.
+func (s *Server) cleanupAfterProfileLogout(id profilemanager.ID, username string) {
+	activeProf, err := s.profileManager.GetActiveProfileState()
+	if err != nil {
+		log.Errorf("failed to get active profile state after logout from profile %s: %v", id, err)
+		return
+	}
+
+	if !isActiveProfile(activeProf, id, username) {
+		return
+	}
+
+	if err := s.cleanupConnection(); err != nil && !errors.Is(err, ErrServiceNotUp) {
+		log.Errorf("failed to cleanup connection: %v", err)
+	}
+	state := internal.CtxGetState(s.rootCtx)
+	state.Set(internal.StatusNeedsLogin)
+}
+
 func (s *Server) handleActiveProfileLogout(ctx context.Context) (*proto.LogoutResponse, error) {
 	if s.config == nil {
 		activeProf, err := s.profileManager.GetActiveProfileState()
@@ -1391,40 +1475,47 @@ func (s *Server) getConfig(activeProf *profilemanager.ActiveProfileState) (*prof
 	return config, configExisted, nil
 }
 
-func (s *Server) canRemoveProfile(id profilemanager.ID) error {
-	if id == profilemanager.DefaultProfileName {
-		return fmt.Errorf("remove profile with reserved name: %s", profilemanager.DefaultProfileName)
-	}
-
-	activeProf, err := s.profileManager.GetActiveProfileState()
-	if err == nil && activeProf.ID == id {
-		return fmt.Errorf("remove active profile: %s", id)
-	}
-
-	return nil
-}
-
-func (s *Server) validateProfileOperation(id profilemanager.ID, allowActiveProfile bool) error {
-	if s.checkProfilesDisabled() {
-		return gstatus.Errorf(codes.Unavailable, errProfilesDisabled)
-	}
-
+// validateProfileLogout gates a profile-addressed logout. Deregistering the
+// profile the daemon already runs is what a plain `netbird logout` does, so the
+// profiles-disabled kill switch must not block it. Logging out of any other
+// profile is profile management and stays gated.
+func (s *Server) validateProfileLogout(id profilemanager.ID, isActive bool) error {
 	if id == "" {
 		return gstatus.Errorf(codes.InvalidArgument, "profile name must be provided")
 	}
 
-	if !allowActiveProfile {
-		if err := s.canRemoveProfile(id); err != nil {
-			return gstatus.Errorf(codes.InvalidArgument, "%v", err)
-		}
+	if isActive {
+		return nil
+	}
+
+	if s.checkProfilesDisabled() {
+		return gstatus.Errorf(codes.Unavailable, errProfilesDisabled)
 	}
 
 	return nil
 }
 
-func (s *Server) logoutFromProfile(ctx context.Context, profile *profilemanager.Profile) error {
+// isActiveProfile reports whether id is the profile the daemon runs for
+// username. The username is part of the comparison because legacy profile IDs
+// are display names, which two users can both hold; the default profile is
+// shared by every user and carries no username.
+func isActiveProfile(activeProf *profilemanager.ActiveProfileState, id profilemanager.ID, username string) bool {
+	if activeProf == nil || activeProf.ID != id {
+		return false
+	}
+
+	return id == profilemanager.DefaultProfileName || activeProf.Username == username
+}
+
+// logoutFromProfile deregisters profile, reusing the running config when
+// profile is the one the daemon is connected with. The username takes part in
+// that decision for the same reason it does in the logout gate: a legacy
+// profile ID is a display name two users can share, and sending the running
+// config for a namesake would deregister the active peer instead of the
+// requested one.
+func (s *Server) logoutFromProfile(ctx context.Context, profile *profilemanager.Profile, username string) error {
 	activeProf, err := s.profileManager.GetActiveProfileState()
-	if err == nil && activeProf.ID == profile.ID && s.connectClient != nil {
+	if err == nil && isActiveProfile(activeProf, profile.ID, username) && s.connectClient != nil {
 		return s.sendLogoutRequest(ctx)
 	}
 
@@ -2103,6 +2194,7 @@ func (s *Server) GetConfig(ctx context.Context, req *proto.GetConfigRequest) (*p
 		Mtu:                           int64(cfg.MTU),
 		DisableAutoConnect:            cfg.DisableAutoConnect,
 		ServerSSHAllowed:              *cfg.ServerSSHAllowed,
+		RemoteJobsAllowed:             util.ReturnBoolWithDefaultFalse(cfg.RemoteJobsAllowed),
 		RosenpassEnabled:              cfg.RosenpassEnabled,
 		RosenpassPermissive:           cfg.RosenpassPermissive,
 		BlockInbound:                  cfg.BlockInbound,
@@ -2193,7 +2285,7 @@ func (s *Server) RemoveProfile(ctx context.Context, msg *proto.RemoveProfileRequ
 		return nil, err
 	}
 
-	if err := s.logoutFromProfile(ctx, resolved); err != nil {
+	if err := s.logoutFromProfile(ctx, resolved, msg.Username); err != nil {
 		// Deregistration is best-effort here: the local profile is removed
 		// either way, so an unprivileged caller leaves the peer registered on
 		// the management server rather than being blocked from removing it.

client/server/server_privileged_test.go

@@ -200,7 +200,7 @@ func startManagement(t *testing.T, signalAddr string, counter *int) (*grpc.Serve
 
 	requestBuffer := server.NewAccountRequestBuffer(context.Background(), store)
 	peersUpdateManager := update_channel.NewPeersUpdateManager(metrics)
-	networkMapController := controller.NewController(context.Background(), store, metrics, peersUpdateManager, requestBuffer, server.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersManager), config)
+	networkMapController := controller.NewController(context.Background(), store, metrics, peersUpdateManager, requestBuffer, server.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersManager), config, nil)
 	accountManager, err := server.BuildManager(context.Background(), config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, ia, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false, cacheStore)
 	if err != nil {
 		return nil, "", err

client/server/setconfig_mdm_test.go

@@ -136,6 +136,51 @@ func TestSetConfig_MDMReject_MultipleFields(t *testing.T) {
 	}, v.GetFields())
 }
 
+func TestSetConfig_MDMReject_LocalMetrics(t *testing.T) {
+	withMDMPolicy(t, mdm.NewPolicy(map[string]any{
+		mdm.KeyEnableLocalMetrics:  true,
+		mdm.KeyLocalMetricsAddress: "127.0.0.1:9191",
+	}))
+
+	s, ctx, profName, username, _ := setupServerWithProfile(t)
+
+	enabled := false
+	addr := "0.0.0.0:9999"
+	_, err := s.SetConfig(ctx, &proto.SetConfigRequest{
+		ProfileName:         profName,
+		Username:            username,
+		EnableLocalMetrics:  &enabled,
+		LocalMetricsAddress: &addr,
+	})
+
+	v := extractViolation(t, err)
+	assert.ElementsMatch(t, []string{
+		mdm.KeyEnableLocalMetrics,
+		mdm.KeyLocalMetricsAddress,
+	}, v.GetFields())
+}
+
+// An explicitly empty address still changes the effective listen address
+// (the manager falls back to the default), so presence must be honored
+// rather than collapsed to "field not set".
+func TestSetConfig_MDMReject_LocalMetricsEmptyAddress(t *testing.T) {
+	withMDMPolicy(t, mdm.NewPolicy(map[string]any{
+		mdm.KeyLocalMetricsAddress: "127.0.0.1:9999",
+	}))
+
+	s, ctx, profName, username, _ := setupServerWithProfile(t)
+
+	addr := ""
+	_, err := s.SetConfig(ctx, &proto.SetConfigRequest{
+		ProfileName:         profName,
+		Username:            username,
+		LocalMetricsAddress: &addr,
+	})
+
+	v := extractViolation(t, err)
+	assert.ElementsMatch(t, []string{mdm.KeyLocalMetricsAddress}, v.GetFields())
+}
+
 func TestSetConfig_MDMReject_AllOrNothing(t *testing.T) {
 	// MDM enforces ManagementURL only; user request touches both the
 	// enforced field AND a non-enforced field (RosenpassEnabled).

client/server/setconfig_test.go

@@ -61,6 +61,7 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
 	rosenpassEnabled := true
 	rosenpassPermissive := true
 	serverSSHAllowed := true
+	remoteJobsAllowed := true
 	interfaceName := "utun100"
 	wireguardPort := int64(51820)
 	preSharedKey := "test-psk"
@@ -76,6 +77,8 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
 	disableIPv6 := true
 	mtu := int64(1280)
 	sshJWTCacheTTL := int32(300)
+	enableLocalMetrics := true
+	localMetricsAddress := "127.0.0.1:9292"
 
 	req := &proto.SetConfigRequest{
 		ProfileName:          profName,
@@ -85,6 +88,7 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
 		RosenpassEnabled:     &rosenpassEnabled,
 		RosenpassPermissive:  &rosenpassPermissive,
 		ServerSSHAllowed:     &serverSSHAllowed,
+		RemoteJobsAllowed:    &remoteJobsAllowed,
 		InterfaceName:        &interfaceName,
 		WireguardPort:        &wireguardPort,
 		OptionalPreSharedKey: &preSharedKey,
@@ -107,6 +111,8 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
 		DnsRouteInterval:     durationpb.New(2 * time.Minute),
 		Mtu:                  &mtu,
 		SshJWTCacheTTL:       &sshJWTCacheTTL,
+		EnableLocalMetrics:   &enableLocalMetrics,
+		LocalMetricsAddress:  &localMetricsAddress,
 	}
 
 	_, err = s.SetConfig(ctx, req)
@@ -128,6 +134,8 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
 	require.Equal(t, rosenpassPermissive, cfg.RosenpassPermissive)
 	require.NotNil(t, cfg.ServerSSHAllowed)
 	require.Equal(t, serverSSHAllowed, *cfg.ServerSSHAllowed)
+	require.NotNil(t, cfg.RemoteJobsAllowed)
+	require.Equal(t, remoteJobsAllowed, *cfg.RemoteJobsAllowed)
 	require.Equal(t, interfaceName, cfg.WgIface)
 	require.Equal(t, int(wireguardPort), cfg.WgPort)
 	require.Equal(t, preSharedKey, cfg.PreSharedKey)
@@ -153,6 +161,8 @@ func TestSetConfig_AllFieldsSaved(t *testing.T) {
 	require.Equal(t, uint16(mtu), cfg.MTU)
 	require.NotNil(t, cfg.SSHJWTCacheTTL)
 	require.Equal(t, int(sshJWTCacheTTL), *cfg.SSHJWTCacheTTL)
+	require.Equal(t, enableLocalMetrics, cfg.LocalMetricsEnabled)
+	require.Equal(t, localMetricsAddress, cfg.LocalMetricsAddress)
 
 	verifyAllFieldsCovered(t, req)
 }
@@ -180,6 +190,7 @@ func verifyAllFieldsCovered(t *testing.T, req *proto.SetConfigRequest) {
 		"RosenpassEnabled":              true,
 		"RosenpassPermissive":           true,
 		"ServerSSHAllowed":              true,
+		"RemoteJobsAllowed":             true,
 		"InterfaceName":                 true,
 		"WireguardPort":                 true,
 		"OptionalPreSharedKey":          true,
@@ -205,6 +216,8 @@ func verifyAllFieldsCovered(t *testing.T, req *proto.SetConfigRequest) {
 		"EnableSSHRemotePortForwarding": true,
 		"DisableSSHAuth":                true,
 		"SshJWTCacheTTL":                true,
+		"EnableLocalMetrics":            true,
+		"LocalMetricsAddress":           true,
 	}
 
 	val := reflect.ValueOf(req).Elem()
@@ -240,6 +253,7 @@ func TestCLIFlags_MappedToSetConfig(t *testing.T) {
 		"enable-rosenpass":                  "RosenpassEnabled",
 		"rosenpass-permissive":              "RosenpassPermissive",
 		"allow-server-ssh":                  "ServerSSHAllowed",
+		"allow-remote-jobs":                 "RemoteJobsAllowed",
 		"interface-name":                    "InterfaceName",
 		"wireguard-port":                    "WireguardPort",
 		"preshared-key":                     "OptionalPreSharedKey",
@@ -264,6 +278,8 @@ func TestCLIFlags_MappedToSetConfig(t *testing.T) {
 		"enable-ssh-remote-port-forwarding": "EnableSSHRemotePortForwarding",
 		"disable-ssh-auth":                  "DisableSSHAuth",
 		"ssh-jwt-cache-ttl":                 "SshJWTCacheTTL",
+		"enable-local-metrics":              "EnableLocalMetrics",
+		"local-metrics-address":             "LocalMetricsAddress",
 	}
 
 	// SetConfigRequest fields that don't have CLI flags (settable only via UI or other means).

client/server/ssh_gate.go

@@ -14,6 +14,7 @@ import (
 
 	"github.com/netbirdio/netbird/client/internal/daemonaddr"
 	"github.com/netbirdio/netbird/client/internal/ipcauth"
+	"github.com/netbirdio/netbird/client/internal/localmetrics"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/proto"
 	"github.com/netbirdio/netbird/util"
@@ -30,6 +31,8 @@ import (
 //     management identity hands SSH authorization decisions, including which
 //     keys and users are accepted, to whoever controls that identity. Changing
 //     the management URL and deregistering the peer are both ways to do that.
+//   - Binding the local metrics endpoint to a non-loopback address publishes
+//     peer names and connectivity state to the network without authentication.
 //
 // Everything else stays unauthenticated, so this is not an authorization model:
 // it only refuses the changes that would let a local user become root. A caller
@@ -39,27 +42,36 @@ import (
 // user-to-root boundary. Fields are nil or empty when the request leaves them
 // untouched.
 type privilegedConfigChange struct {
-	managementURL    string
-	serverSSHAllowed *bool
-	enableSSHRoot    *bool
-	disableSSHAuth   *bool
+	managementURL       string
+	serverSSHAllowed    *bool
+	remoteJobsAllowed   *bool
+	enableSSHRoot       *bool
+	disableSSHAuth      *bool
+	enableLocalMetrics  *bool
+	localMetricsAddress *string
 }
 
 func privilegedChangeFromSetConfig(msg *proto.SetConfigRequest) privilegedConfigChange {
 	return privilegedConfigChange{
-		managementURL:    msg.GetManagementUrl(),
-		serverSSHAllowed: msg.ServerSSHAllowed,
-		enableSSHRoot:    msg.EnableSSHRoot,
-		disableSSHAuth:   msg.DisableSSHAuth,
+		managementURL:       msg.GetManagementUrl(),
+		serverSSHAllowed:    msg.ServerSSHAllowed,
+		remoteJobsAllowed:   msg.RemoteJobsAllowed,
+		enableSSHRoot:       msg.EnableSSHRoot,
+		disableSSHAuth:      msg.DisableSSHAuth,
+		enableLocalMetrics:  msg.EnableLocalMetrics,
+		localMetricsAddress: msg.LocalMetricsAddress,
 	}
 }
 
 func privilegedChangeFromLogin(msg *proto.LoginRequest) privilegedConfigChange {
 	return privilegedConfigChange{
-		managementURL:    msg.GetManagementUrl(),
-		serverSSHAllowed: msg.ServerSSHAllowed,
-		enableSSHRoot:    msg.EnableSSHRoot,
-		disableSSHAuth:   msg.DisableSSHAuth,
+		managementURL:       msg.GetManagementUrl(),
+		serverSSHAllowed:    msg.ServerSSHAllowed,
+		remoteJobsAllowed:   msg.RemoteJobsAllowed,
+		enableSSHRoot:       msg.EnableSSHRoot,
+		disableSSHAuth:      msg.DisableSSHAuth,
+		enableLocalMetrics:  msg.EnableLocalMetrics,
+		localMetricsAddress: msg.LocalMetricsAddress,
 	}
 }
 
@@ -83,6 +95,21 @@ func requirePrivilegeForConfigChange(ctx context.Context, stored *profilemanager
 		return denyPrivileged(ctx, "enabling the NetBird SSH server", ipcauth.UpCommand("--allow-server-ssh"))
 	}
 
+	// Enabling remote jobs lets the management server run jobs (e.g. debug
+	// bundles) on this host, so turning it on crosses the user-to-root
+	// boundary the same way enabling the SSH server does. The stored value
+	// defaults to off (nil = off), so a legacy config is correctly seen as
+	// off and turning it on requires privilege.
+	if enables(storedFlag(stored, func(c *profilemanager.Config) *bool { return c.RemoteJobsAllowed }), change.remoteJobsAllowed) {
+		return denyPrivileged(ctx, "enabling remote jobs", ipcauth.UpCommand("--allow-remote-jobs"))
+	}
+
+	if addr, exposes := exposesLocalMetrics(stored, change); exposes {
+		return denyPrivileged(ctx,
+			"exposing the local metrics endpoint on a non-loopback address",
+			ipcauth.UpCommand("--enable-local-metrics --local-metrics-address "+addr))
+	}
+
 	// Only guard the management binding while the SSH server is enabled: that is
 	// when the management identity decides who may open a shell here.
 	if !sshServerEnabled(stored) {
@@ -245,6 +272,48 @@ func sshServerCurrentlyAllowed(cfg *profilemanager.Config) *bool {
 	return &enabled
 }
 
+// exposesLocalMetrics reports whether the change would leave the metrics
+// endpoint enabled on an address that is not confirmed loopback, and returns
+// that address. A request that restates the stored state is not a change, so a
+// settings form resubmitted after an administrator opened the endpoint is not
+// refused.
+func exposesLocalMetrics(stored *profilemanager.Config, change privilegedConfigChange) (string, bool) {
+	storedEnabled, storedAddr := storedLocalMetrics(stored)
+
+	enabled := storedEnabled
+	if change.enableLocalMetrics != nil {
+		enabled = *change.enableLocalMetrics
+	}
+	addr := storedAddr
+	if change.localMetricsAddress != nil {
+		addr = metricsAddrOrDefault(*change.localMetricsAddress)
+	}
+
+	if !enabled || localmetrics.IsLoopback(addr) {
+		return "", false
+	}
+	if storedEnabled && storedAddr == addr {
+		return "", false
+	}
+	return addr, true
+}
+
+// storedLocalMetrics reads the metrics settings from the stored config,
+// tolerating a config that does not exist yet.
+func storedLocalMetrics(cfg *profilemanager.Config) (bool, string) {
+	if cfg == nil {
+		return false, localmetrics.DefaultListenAddress
+	}
+	return cfg.LocalMetricsEnabled, metricsAddrOrDefault(cfg.LocalMetricsAddress)
+}
+
+func metricsAddrOrDefault(addr string) string {
+	if addr == "" {
+		return localmetrics.DefaultListenAddress
+	}
+	return addr
+}
+
 // sameManagementURL reports whether requested addresses the same management
 // server as stored, comparing scheme, host and effective port so that an
 // equivalent spelling ("https://api.netbird.io" for a stored

client/server/ssh_gate_test.go

@@ -61,6 +61,8 @@ func noIdentityCtx() context.Context { return context.Background() }
 
 func boolPtr(v bool) *bool { return &v }
 
+func strPtr(v string) *string { return &v }
+
 func mustURL(t *testing.T, raw string) *url.URL {
 	t.Helper()
 	u, err := url.Parse(raw)
@@ -171,6 +173,34 @@ func TestRequirePrivilegeForConfigChange_SSHFlags(t *testing.T) {
 			stored: &profilemanager.Config{DisableSSHAuth: boolPtr(true)},
 			change: privilegedConfigChange{disableSSHAuth: boolPtr(false)},
 		},
+		{
+			name:     "enabling remote jobs unprivileged is refused",
+			stored:   &profilemanager.Config{RemoteJobsAllowed: boolPtr(false)},
+			change:   privilegedConfigChange{remoteJobsAllowed: boolPtr(true)},
+			wantDeny: true,
+		},
+		{
+			name:       "enabling remote jobs as root is allowed",
+			stored:     &profilemanager.Config{RemoteJobsAllowed: boolPtr(false)},
+			change:     privilegedConfigChange{remoteJobsAllowed: boolPtr(true)},
+			privileged: true,
+		},
+		{
+			name:     "a profile with no config yet counts as off, so enabling remote jobs is refused",
+			stored:   nil,
+			change:   privilegedConfigChange{remoteJobsAllowed: boolPtr(true)},
+			wantDeny: true,
+		},
+		{
+			name:   "restating already-enabled remote jobs is not a change",
+			stored: &profilemanager.Config{RemoteJobsAllowed: boolPtr(true)},
+			change: privilegedConfigChange{remoteJobsAllowed: boolPtr(true)},
+		},
+		{
+			name:   "turning remote jobs off is not guarded",
+			stored: &profilemanager.Config{RemoteJobsAllowed: boolPtr(true)},
+			change: privilegedConfigChange{remoteJobsAllowed: boolPtr(false)},
+		},
 		{
 			name:   "a request that touches none of the guarded fields is allowed",
 			stored: &profilemanager.Config{ServerSSHAllowed: boolPtr(false)},
@@ -194,6 +224,102 @@ func TestRequirePrivilegeForConfigChange_SSHFlags(t *testing.T) {
 	}
 }
 
+func TestRequirePrivilegeForConfigChange_LocalMetrics(t *testing.T) {
+	exposed := &profilemanager.Config{LocalMetricsEnabled: true, LocalMetricsAddress: "0.0.0.0:9191"}
+
+	tests := []struct {
+		name       string
+		stored     *profilemanager.Config
+		change     privilegedConfigChange
+		privileged bool
+		wantDeny   bool
+	}{
+		{
+			name:     "binding a non-loopback address unprivileged is refused",
+			stored:   &profilemanager.Config{},
+			change:   privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("0.0.0.0:9191")},
+			wantDeny: true,
+		},
+		{
+			name:       "binding a non-loopback address as root is allowed",
+			stored:     &profilemanager.Config{},
+			change:     privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("0.0.0.0:9191")},
+			privileged: true,
+		},
+		{
+			name:   "enabling on the default loopback address is not guarded",
+			stored: &profilemanager.Config{},
+			change: privilegedConfigChange{enableLocalMetrics: boolPtr(true)},
+		},
+		{
+			name:   "enabling on an explicit loopback address is not guarded",
+			stored: &profilemanager.Config{},
+			change: privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("127.0.0.1:9999")},
+		},
+		{
+			name:   "enabling on the IPv6 loopback address is not guarded",
+			stored: &profilemanager.Config{},
+			change: privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("[::1]:9191")},
+		},
+		{
+			// The address alone does nothing while the endpoint stays off.
+			name:   "a non-loopback address without enabling is not guarded",
+			stored: &profilemanager.Config{},
+			change: privilegedConfigChange{localMetricsAddress: strPtr("0.0.0.0:9191")},
+		},
+		{
+			name:     "widening an already enabled loopback endpoint is refused",
+			stored:   &profilemanager.Config{LocalMetricsEnabled: true, LocalMetricsAddress: "127.0.0.1:9191"},
+			change:   privilegedConfigChange{localMetricsAddress: strPtr("0.0.0.0:9191")},
+			wantDeny: true,
+		},
+		{
+			name:   "restating an already exposed endpoint is not a change",
+			stored: exposed,
+			change: privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("0.0.0.0:9191")},
+		},
+		{
+			name:   "turning an exposed endpoint off is not guarded",
+			stored: exposed,
+			change: privilegedConfigChange{enableLocalMetrics: boolPtr(false)},
+		},
+		{
+			name:     "re-enabling an exposed endpoint that was turned off is refused",
+			stored:   &profilemanager.Config{LocalMetricsEnabled: false, LocalMetricsAddress: "0.0.0.0:9191"},
+			change:   privilegedConfigChange{enableLocalMetrics: boolPtr(true)},
+			wantDeny: true,
+		},
+		{
+			// Fail closed: an address that cannot be parsed is not confirmed loopback.
+			name:     "an unparseable address is refused",
+			stored:   &profilemanager.Config{},
+			change:   privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("not-an-address")},
+			wantDeny: true,
+		},
+		{
+			name:     "a profile with no config yet counts as off, so exposing is refused",
+			stored:   nil,
+			change:   privilegedConfigChange{enableLocalMetrics: boolPtr(true), localMetricsAddress: strPtr("0.0.0.0:9191")},
+			wantDeny: true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ctx := userCtx()
+			if tt.privileged {
+				ctx = rootCtx()
+			}
+			err := requirePrivilegeForConfigChange(ctx, tt.stored, tt.change)
+			if tt.wantDeny {
+				assertDenied(t, err)
+				return
+			}
+			assertAllowed(t, err)
+		})
+	}
+}
+
 func TestRequirePrivilegeForConfigChange_ManagementURL(t *testing.T) {
 	sshOn := func(raw string) *profilemanager.Config {
 		return &profilemanager.Config{ServerSSHAllowed: boolPtr(true), ManagementURL: mustURL(t, raw)}

client/system/info.go

@@ -65,6 +65,7 @@ type Info struct {
 	RosenpassEnabled    bool
 	RosenpassPermissive bool
 	ServerSSHAllowed    bool
+	RemoteJobsAllowed   bool
 
 	DisableClientRoutes bool
 	DisableServerRoutes bool
@@ -90,12 +91,16 @@ func (i *Info) SetFlags(
 	disableDNS, disableFirewall, blockLANAccess, blockInbound, disableIPv6 bool, syncMessageVersion *int,
 	enableSSHRoot, enableSSHSFTP, enableSSHLocalPortForwarding, enableSSHRemotePortForwarding *bool,
 	disableSSHAuth *bool,
+	remoteJobsAllowed *bool,
 ) {
 	i.RosenpassEnabled = rosenpassEnabled
 	i.RosenpassPermissive = rosenpassPermissive
 	if serverSSHAllowed != nil {
 		i.ServerSSHAllowed = *serverSSHAllowed
 	}
+	if remoteJobsAllowed != nil {
+		i.RemoteJobsAllowed = *remoteJobsAllowed
+	}
 
 	i.DisableClientRoutes = disableClientRoutes
 	i.DisableServerRoutes = disableServerRoutes

client/testutil/privileged/runner_test.go

@@ -25,7 +25,7 @@ import (
 // (.github/workflows/golang-test-linux.yml, test_client_on_docker).
 const (
 	containerImage = "golang"
-	containerTag   = "1.25-alpine"
+	containerTag   = "1.26.7-alpine"
 )
 
 const (

client/ui/build/docker/Dockerfile.cross

@@ -13,7 +13,7 @@
 #   docker run --rm -v $(pwd):/app wails-cross windows amd64
 #   docker run --rm -v $(pwd):/app wails-cross windows arm64
 
-FROM golang:1.25-bookworm
+FROM golang:1.26.7-bookworm
 
 ARG TARGETARCH
 

client/ui/build/docker/Dockerfile.server

@@ -2,7 +2,7 @@
 # Multi-stage build for minimal image size
 
 # Build stage
-FROM golang:alpine AS builder
+FROM golang:1.26.7-alpine AS builder
 
 WORKDIR /app
 

client/ui/frontend/src/modules/session/SessionExpirationDialog.tsx

@@ -18,6 +18,11 @@ import { formatRemaining } from "@/lib/formatters";
 const DEFAULT_SECONDS = 360;
 const WINDOW_WIDTH = 360;
 const SOON_THRESHOLD_SECONDS = 60 * 60;
+const DEADLINE_TOLERANCE_MS = 5 * 1000;
+// The final-warning deadline reaches the Go side as RFC3339 truncated to whole
+// seconds, while the status snapshot carries millisecond precision, so an
+// unchanged deadline can look up to 999 ms newer than the exact URL value.
+const EXACT_DEADLINE_TOLERANCE_MS = 999;
 
 export default function SessionExpirationDialog() {
     const { t } = useTranslation();
@@ -29,11 +34,19 @@ export default function SessionExpirationDialog() {
         const n = Number.parseInt(raw, 10);
         return Number.isFinite(n) && n > 0 ? n : DEFAULT_SECONDS;
     }, [params]);
+    const initialDeadline = useMemo(() => {
+        const raw = params.get("deadline");
+        if (!raw) return null;
+        const n = Number.parseInt(raw, 10);
+        return Number.isFinite(n) && n > 0 ? n : null;
+    }, [params]);
 
     const [remaining, setRemaining] = useState(initialSeconds);
     const [busy, setBusy] = useState(false);
     const busyRef = useRef(busy);
     busyRef.current = busy;
+    const openedDeadlineRef = useRef(initialDeadline ?? Date.now() + initialSeconds * 1000);
+    const exactDeadlineRef = useRef(initialDeadline !== null);
     const expired = remaining <= 0;
     const expiredRef = useRef(expired);
     expiredRef.current = expired;
@@ -45,23 +58,45 @@ export default function SessionExpirationDialog() {
 
     useEffect(() => {
         setRemaining(initialSeconds);
-    }, [initialSeconds]);
+        openedDeadlineRef.current = initialDeadline ?? Date.now() + initialSeconds * 1000;
+        exactDeadlineRef.current = initialDeadline !== null;
+    }, [initialSeconds, initialDeadline]);
 
+    // Recompute from the absolute deadline instead of decrementing per tick: webview
+    // timers get suspended for tens of seconds (App Nap / hidden-window throttling),
+    // so a tick counter drifts behind the wall clock by the suspended time.
     useEffect(() => {
         const id = globalThis.setInterval(() => {
-            setRemaining((s) => (s <= 1 ? 0 : s - 1));
+            setRemaining(Math.max(0, Math.ceil((openedDeadlineRef.current - Date.now()) / 1000)));
         }, 1000);
         return () => globalThis.clearInterval(id);
     }, [initialSeconds]);
 
+    // Auto-close only when the session was actually renewed elsewhere (tray action, CLI,
+    // main window): the daemon keeps emitting Connected snapshots regardless of session
+    // state, so the signal is the deadline jumping past the one this dialog was opened for.
+    // With the exact deadline from the URL any jump past its sub-second precision loss
+    // counts; the seconds-derived fallback needs a wider tolerance for the Go-side
+    // truncation and mount latency.
     // Don't auto-close while busy (aborts our WaitExtend) or expired (hides the state).
     useEffect(() => {
-        const off = Events.On("netbird:status", (ev: { data: { status?: string } }) => {
-            if (busyRef.current || expiredRef.current) return;
-            if (ev?.data?.status === "Connected") {
-                WindowManager.CloseSessionExpiration().catch(console.error);
-            }
-        });
+        const off = Events.On(
+            "netbird:status",
+            (ev: { data: { status?: string; sessionExpiresAt?: string | null } }) => {
+                if (busyRef.current || expiredRef.current) return;
+                if (ev?.data?.status !== "Connected") return;
+                const raw = ev?.data?.sessionExpiresAt;
+                if (!raw) return;
+                const renewed = Date.parse(raw);
+                if (!Number.isFinite(renewed)) return;
+                const tolerance = exactDeadlineRef.current
+                    ? EXACT_DEADLINE_TOLERANCE_MS
+                    : DEADLINE_TOLERANCE_MS;
+                if (renewed - openedDeadlineRef.current > tolerance) {
+                    WindowManager.CloseSessionExpiration().catch(console.error);
+                }
+            },
+        );
         return () => {
             off();
         };

client/ui/i18n/locales/_index.json

@@ -1,6 +1,7 @@
 {
     "languages": [
         {"code": "en", "displayName": "English (US)", "englishName": "English (US)"},
+        {"code": "uk", "displayName": "Українська", "englishName": "Ukrainian"},
         {"code": "de", "displayName": "Deutsch", "englishName": "German"},
         {"code": "hu", "displayName": "Magyar", "englishName": "Hungarian"},
         {"code": "ru", "displayName": "Русский", "englishName": "Russian"},

client/ui/services/windowmanager.go

@@ -292,11 +292,15 @@ func (s *WindowManager) CloseBrowserLogin() {
 }
 
 // OpenSessionExpiration shows the countdown warning on the cursor's display; seconds seeds
-// the countdown. Singleton, destroyed on close.
-func (s *WindowManager) OpenSessionExpiration(seconds int) {
+// the countdown and deadlineUnixMilli (0 when unknown) is the absolute deadline the dialog
+// compares renewal snapshots against. Singleton, destroyed on close.
+func (s *WindowManager) OpenSessionExpiration(seconds int, deadlineUnixMilli int64) {
 	s.mu.Lock()
 	defer s.mu.Unlock()
 	startURL := "/#/dialog/session-expiration?seconds=" + strconv.Itoa(seconds)
+	if deadlineUnixMilli > 0 {
+		startURL += "&deadline=" + strconv.FormatInt(deadlineUnixMilli, 10)
+	}
 	if s.sessionExpiration == nil {
 		opts := DialogWindowOptions("session-expiration", s.title("window.title.sessionExpiration"), startURL, s.linuxIcon)
 		opts.Screen = s.getScreenBasedOnCursorPosition()

client/ui/tray_events.go

@@ -81,7 +81,8 @@ func (t *Tray) onSystemEvent(ev *application.CustomEvent) {
 
 	if se.Metadata != nil && se.Metadata[authsession.MetaWarning] == "true" {
 		if se.Metadata[authsession.MetaFinal] == "true" {
-			t.openSessionExpiration()
+			deadline, _ := authsession.ParseExpiresAt(se.Metadata[authsession.MetaExpiresAt])
+			t.openSessionExpiration(deadline)
 			return
 		}
 		t.notifySessionWarning(

client/ui/tray_session.go

@@ -278,12 +278,23 @@ func (t *Tray) dismissSessionWarning() {
 }
 
 // openSessionExpiration fires the fallback dialog when the earlier warning notification wasn't dismissed.
-// Idempotent on the WindowManager side.
-func (t *Tray) openSessionExpiration() {
+// deadline is the absolute expiry from the warning event's metadata; when zero (older daemon,
+// malformed metadata) the cached status-snapshot deadline fills in. Idempotent on the
+// WindowManager side.
+func (t *Tray) openSessionExpiration(deadline time.Time) {
 	if t.svc.WindowManager == nil {
 		return
 	}
-	t.svc.WindowManager.OpenSessionExpiration(finalWarningCountdownSeconds)
+	if deadline.IsZero() {
+		t.sessionMu.Lock()
+		deadline = t.sessionExpiresAt
+		t.sessionMu.Unlock()
+	}
+	var deadlineMs int64
+	if !deadline.IsZero() {
+		deadlineMs = deadline.UnixMilli()
+	}
+	t.svc.WindowManager.OpenSessionExpiration(finalWarningCountdownSeconds, deadlineMs)
 }
 
 // openSessionExtendFlow opens the SessionExpiration window seeded with the cached deadline's remaining time,
@@ -304,5 +315,5 @@ func (t *Tray) openSessionExtendFlow() {
 	if t.svc.WindowManager == nil {
 		return
 	}
-	t.svc.WindowManager.OpenSessionExpiration(seconds)
+	t.svc.WindowManager.OpenSessionExpiration(seconds, deadline.UnixMilli())
 }

combined/Dockerfile.multistage

@@ -1,4 +1,4 @@
-FROM golang:1.25-bookworm AS builder
+FROM golang:1.26.7-bookworm AS builder
 WORKDIR /app
 
 # Install build dependencies

docs/io.netbird.client.plist

@@ -85,6 +85,21 @@
     <false/>
     -->
 
+    <!-- ===== Remote jobs (debug bundles) =====
+         allowRemoteJobs      : opt this device into management-requested
+                                remote jobs (e.g. debug bundles). Off by
+                                default; enabling is a privileged change.
+         debugBundleUploadURL : override the debug-bundle upload service URL
+                                for remote jobs (https URL with a host). Takes
+                                precedence over the management-supplied value. -->
+    <!--
+    <key>allowRemoteJobs</key>
+    <true/>
+
+    <key>debugBundleUploadURL</key>
+    <string>https://upload.example.com</string>
+    -->
+
     <!-- ===== WireGuard UDP port =====
          Range 1-65535. Omit to keep the daemon default. -->
     <!--

docs/netbird-macos.mobileconfig

@@ -121,6 +121,19 @@
                                 <false/>
                                 -->
 
+                                <!-- ===== Remote jobs (debug bundles) =====
+                                     allowRemoteJobs      : opt into management-requested
+                                                            remote jobs. Off by default.
+                                     debugBundleUploadURL : override the debug-bundle upload
+                                                            service (https URL with a host);
+                                                            precedence over the management value. -->
+                                <!--
+                                <key>allowRemoteJobs</key>
+                                <true/>
+                                <key>debugBundleUploadURL</key>
+                                <string>https://upload.example.com</string>
+                                -->
+
                                 <!-- ===== WireGuard UDP port (int) =====
                                      Range 1-65535. Omit to keep the default. -->
                                 <!--

docs/netbird-macos.sh

@@ -36,7 +36,9 @@
 # IDEMPOTENCY: re-running with the same values is a no-op from the
 # daemon's point of view (the 1-minute reload ticker diff returns empty).
 #
-# SECURITY: PreSharedKey is redacted in this script's log output.
+# SECURITY: PreSharedKey (and any secret-bearing debugBundleUploadURL) is
+#   redacted in this script's log output, and the installed plist is 0600
+#   root:wheel so its values are not readable by local non-root users.
 
 set -euo pipefail
 
@@ -56,6 +58,8 @@ NULL='__UNSET__'
 managementURL='https://api.netbird.io:443'
 preSharedKey="$NULL"                       # secret; redacted in log
 allowServerSSH='true'
+allowRemoteJobs="$NULL"
+debugBundleUploadURL="$NULL"                # HTTPS URL with a host; overrides management
 blockInbound="$NULL"
 disableAutoConnect="$NULL"
 disableAutostart="$NULL"
@@ -107,21 +111,35 @@ end_plist() {
 EOF
 }
 
-# emit_string appends a plist `<key>`/`<string>` entry for the given key and value to "$PLIST_PATH.tmp", XML-escaping `&`, `<`, and `>`, and logs the assignment (masking the logged value as `********** (secret)` when the key is `preSharedKey`).
+# emit_string appends a plist `<key>`/`<string>` entry for the given key and value to "$PLIST_PATH.tmp", XML-escaping `&`, `<`, and `>`, and logs the assignment (masking the logged value as `********** (secret)` for secret keys — `preSharedKey` and `debugBundleUploadURL`, which can embed credentials or a signed query token).
 emit_string() {
   local key="$1" value="$2" log_value="$2"
   # Escape XML entities in the value
   local escaped
   escaped="$(printf '%s' "$value" | sed -e 's/&/\&amp;/g' -e 's/</\&lt;/g' -e 's/>/\&gt;/g')"
   printf '    <key>%s</key>\n    <string>%s</string>\n' "$key" "$escaped" >> "$PLIST_PATH.tmp"
-  if [[ "$key" == "preSharedKey" ]]; then
-    log_value='********** (secret)'
-  fi
+  case "$key" in
+    preSharedKey|debugBundleUploadURL) log_value='********** (secret)' ;;
+    *) ;;
+  esac
   log "set $key = $log_value"
 }
 
-# emit_bool writes a boolean plist entry for a given key into the temporary plist file.
-# emit_bool writes a boolean plist entry for a key when the provided value matches an accepted boolean token; logs an error and skips the key on invalid input.
+# is_bool returns success if the value is an accepted boolean token.
+is_bool() {
+  local value="$1"
+  case "$value" in
+    true|True|TRUE|1|yes|false|False|FALSE|0|no) return 0 ;;
+    *) return 1 ;;
+  esac
+}
+
+# emit_bool writes a boolean plist entry for a key when the provided value matches
+# an accepted boolean token; logs an error and skips the key on invalid input.
+# It returns success even on invalid input (like emit_int) so a single typo in one
+# boolean does not abort the whole policy push under `set -euo pipefail`. Callers
+# that must fail closed on an invalid value (e.g. allowRemoteJobs) validate with
+# is_bool before calling and substitute a safe default themselves.
 emit_bool() {
   local key="$1" value="$2"
   local xml_bool
@@ -145,15 +163,35 @@ emit_int() {
   log "set $key = $value"
 }
 
-# main builds the NetBird MDM plist from configured policy variables, validates and installs it to /Library/Managed Preferences/io.netbird.client.plist (root:wheel, 644) and optionally triggers the NetBird daemon to reload.
+# main builds the NetBird MDM plist from configured policy variables, validates and installs it to /Library/Managed Preferences/io.netbird.client.plist (root:wheel, 600 — the daemon reads it directly as root, so it need not be world-readable) and optionally triggers the NetBird daemon to reload.
 main() {
   log "applying NetBird MDM policy to $PLIST_PATH"
+  # Restrict the temp plist while it is being built: it carries the same
+  # secret-bearing values as the final file, which is installed 0600 below.
+  umask 077
   /bin/mkdir -p "$PLIST_DIR"
   start_plist
+  # Force 0600 on the temp file explicitly: start_plist writes it with a
+  # truncating redirect, which keeps an existing file's mode, so a leftover
+  # 0644 tmp from an interrupted run would not be tightened by umask alone.
+  # start_plist only wrote the header so far — the secret-bearing values are
+  # appended after this point.
+  /bin/chmod 600 "$PLIST_PATH.tmp"
 
   is_set "$managementURL"             && emit_string  managementURL             "$managementURL"
   is_set "$preSharedKey"              && emit_string  preSharedKey              "$preSharedKey"
   is_set "$allowServerSSH"            && emit_bool    allowServerSSH            "$allowServerSSH"
+  # Fail closed: an invalid allowRemoteJobs value must not drop the key and
+  # leave a conflicting local opt-in active — enforce the safe default (false).
+  if is_set "$allowRemoteJobs"; then
+    if is_bool "$allowRemoteJobs"; then
+      emit_bool allowRemoteJobs "$allowRemoteJobs"
+    else
+      log "invalid boolean for allowRemoteJobs: $allowRemoteJobs; enforcing safe default (false)"
+      emit_bool allowRemoteJobs false
+    fi
+  fi
+  is_set "$debugBundleUploadURL"      && emit_string  debugBundleUploadURL      "$debugBundleUploadURL"
   is_set "$blockInbound"              && emit_bool    blockInbound              "$blockInbound"
   is_set "$disableAutoConnect"        && emit_bool    disableAutoConnect        "$disableAutoConnect"
   is_set "$disableAutostart"          && emit_bool    disableAutostart          "$disableAutostart"
@@ -181,7 +219,12 @@ main() {
 
   /bin/mv -f "$PLIST_PATH.tmp" "$PLIST_PATH"
   /usr/sbin/chown root:wheel "$PLIST_PATH"
-  /bin/chmod 644 "$PLIST_PATH"
+  # 0600, not 0644: the daemon's loader (client/mdm/policy_darwin.go) opens the
+  # plist directly as root, so it does not need to be world-readable. Restricting
+  # it keeps secret-bearing values (preSharedKey, a signed debugBundleUploadURL)
+  # from any local non-root user. The loader's only mode check refuses a
+  # world-writable file, which 0600 satisfies.
+  /bin/chmod 600 "$PLIST_PATH"
 
   log "policy installed; NetBird daemon will pick it up within the next 1-minute reload tick"
 

docs/netbird.adml

@@ -39,6 +39,12 @@
       <string id="AllowServerSSH_Name">Allow server SSH</string>
       <string id="AllowServerSSH_Help">When enabled, this client accepts incoming SSH sessions via NetBird SSH. Equivalent to --allow-server-ssh.</string>
 
+      <string id="AllowRemoteJobs_Name">Allow remote jobs</string>
+      <string id="AllowRemoteJobs_Help">When enabled, this client accepts management-requested remote jobs (e.g. debug bundles). Off by default. Equivalent to --allow-remote-jobs.</string>
+
+      <string id="DebugBundleUploadURL_Name">Debug bundle upload URL</string>
+      <string id="DebugBundleUploadURL_Help">Overrides the upload service used for debug bundles produced by remote jobs, taking precedence over the value requested by management. Must be an https URL with a host.</string>
+
       <string id="RosenpassEnabled_Name">Enable Rosenpass</string>
       <string id="RosenpassEnabled_Help">Enables Rosenpass post-quantum key exchange on WireGuard tunnels. Both peers must support it.</string>
 
@@ -79,6 +85,12 @@
         </textBox>
       </presentation>
 
+      <presentation id="DebugBundleUploadURL_Pres">
+        <textBox refId="DebugBundleUploadURL_Text">
+          <label>Debug bundle upload URL:</label>
+        </textBox>
+      </presentation>
+
       <presentation id="PreSharedKey_Pres">
         <textBox refId="PreSharedKey_Text">
           <label>Pre-shared key:</label>

docs/netbird.admx

@@ -124,6 +124,31 @@
       <disabledValue><decimal value="0" /></disabledValue>
     </policy>
 
+    <policy name="AllowRemoteJobs"
+            class="Machine"
+            displayName="$(string.AllowRemoteJobs_Name)"
+            explainText="$(string.AllowRemoteJobs_Help)"
+            key="Software\Policies\NetBird"
+            valueName="AllowRemoteJobs">
+      <parentCategory ref="NetBird" />
+      <supportedOn ref="SUPPORTED_NetBird_All" />
+      <enabledValue><decimal value="1" /></enabledValue>
+      <disabledValue><decimal value="0" /></disabledValue>
+    </policy>
+
+    <policy name="DebugBundleUploadURL"
+            class="Machine"
+            displayName="$(string.DebugBundleUploadURL_Name)"
+            explainText="$(string.DebugBundleUploadURL_Help)"
+            key="Software\Policies\NetBird"
+            presentation="$(presentation.DebugBundleUploadURL_Pres)">
+      <parentCategory ref="NetBird" />
+      <supportedOn ref="SUPPORTED_NetBird_All" />
+      <elements>
+        <text id="DebugBundleUploadURL_Text" valueName="DebugBundleUploadURL" required="false" />
+      </elements>
+    </policy>
+
     <policy name="RosenpassEnabled"
             class="Machine"
             displayName="$(string.RosenpassEnabled_Name)"

docs/testing-privileged.md

@@ -32,7 +32,7 @@ list; both are optional and default to the full privileged suite.
 
 1. Skips immediately when it detects it is already inside the container
    (`DOCKER_CI=true`), so the privileged tests run in place instead of recursing.
-2. Otherwise spins up a `golang:1.25-alpine` container (matching CI),
+2. Otherwise spins up a `golang:1.26.7-alpine` container (matching CI),
    bind-mounts the repo and the host Go build/module caches, installs the
    required packages, and runs `go test -tags 'devcert privileged'` over the
    client packages.

e2e/harness/Dockerfile.client

@@ -3,7 +3,7 @@
 # artifact), so this mirrors its alpine runtime + entrypoint while compiling the
 # CGO-free client inline. BuildKit cache mounts keep rebuilds incremental.
 
-FROM golang:1.25-bookworm AS builder
+FROM golang:1.26.7-bookworm AS builder
 WORKDIR /src
 COPY go.mod go.sum ./
 RUN --mount=type=cache,target=/go/pkg/mod go mod download

e2e/harness/client.go

@@ -31,6 +31,9 @@ const (
 // Client is a running NetBird client container joined to the combined server.
 type Client struct {
 	container testcontainers.Container
+	// name is the container hostname the agent reports to management at
+	// registration — the name the peer appears under in the peers API.
+	name string
 }
 
 // clientOptions is what the ClientOption values assemble.
@@ -99,24 +102,38 @@ func StartClient(ctx context.Context, c *Combined, setupKey string, opts ...Clie
 	if err != nil {
 		return nil, fmt.Errorf("start client container: %w", err)
 	}
-	return &Client{container: ctr}, nil
+	return &Client{container: ctr, name: o.name}, nil
+}
+
+// Hostname returns the container hostname the agent reports to management —
+// the name the registered peer appears under in the peers API.
+func (cl *Client) Hostname() string {
+	return cl.name
 }
 
 // Restart bounces the client connection (netbird down/up) so it pulls a fresh
 // network map — the documented workaround for a freshly-joined client not yet
 // seeing a synthesized agent-network service.
 func (cl *Client) Restart(ctx context.Context) error {
+	return cl.Up(ctx)
+}
+
+// Up re-runs `netbird up` inside the client with the given extra flags (e.g.
+// "--allow-remote-jobs"), bouncing the connection first so the new config is
+// picked up and re-synced to management. Used to toggle peer options that ride
+// on the login/sync request without recreating the container.
+func (cl *Client) Up(ctx context.Context, extraArgs ...string) error {
 	if _, _, err := cl.container.Exec(ctx, []string{"netbird", "down"}, tcexec.Multiplexed()); err != nil {
 		return fmt.Errorf("netbird down: %w", err)
 	}
 	time.Sleep(2 * time.Second)
-	code, reader, err := cl.container.Exec(ctx, []string{"netbird", "up"}, tcexec.Multiplexed())
+	code, reader, err := cl.container.Exec(ctx, append([]string{"netbird", "up"}, extraArgs...), tcexec.Multiplexed())
 	if err != nil {
 		return fmt.Errorf("netbird up: %w", err)
 	}
 	if code != 0 {
 		out, _ := io.ReadAll(reader)
-		return fmt.Errorf("netbird up exited %d: %s", code, string(out))
+		return fmt.Errorf("netbird up %v exited %d: %s", extraArgs, code, string(out))
 	}
 	return nil
 }

e2e/remotejobs/main_test.go

@@ -0,0 +1,47 @@
+//go:build e2e
+
+// Package remotejobs holds the container-based e2e suite for the remote-jobs
+// opt-in (PR #7153) and the debug-bundle job parameters anonymize_level /
+// upload_url (PR #7147). A combined server is built and bootstrapped once per
+// package run (TestMain) and shared via srv; each test registers its own client
+// and cleans it up.
+package remotejobs
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"testing"
+	"time"
+
+	"github.com/netbirdio/netbird/e2e/harness"
+)
+
+// srv is the shared combined server for the package, PAT-authenticated by the
+// time any Test runs.
+var srv *harness.Combined
+
+func TestMain(m *testing.M) {
+	os.Exit(run(m))
+}
+
+func run(m *testing.M) int {
+	// Generous timeout to cover a cold image build on first run.
+	ctx, cancel := context.WithTimeout(context.Background(), 15*time.Minute)
+	defer cancel()
+
+	var err error
+	srv, err = harness.StartCombined(ctx)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "e2e: start combined server: %v\n", err)
+		return 1
+	}
+	defer func() { _ = srv.Terminate(context.Background()) }()
+
+	if _, err := srv.Bootstrap(ctx); err != nil {
+		fmt.Fprintf(os.Stderr, "e2e: bootstrap admin PAT: %v\n", err)
+		return 1
+	}
+
+	return m.Run()
+}

e2e/remotejobs/remotejobs_test.go

@@ -0,0 +1,197 @@
+//go:build e2e
+
+package remotejobs
+
+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"
+)
+
+const (
+	refusedReason = "remote jobs are not enabled on this peer"
+	// testUploadURL is the debug-bundle upload URL the job subtests pass; it only
+	// needs to be a well-formed https URL with a host (see ValidateBundleUploadURL).
+	testUploadURL = "https://uploads.example.com/bundle"
+)
+
+// TestRemoteJobsOptInAndBundleParams exercises the two PRs end-to-end against a
+// live management server and a real client:
+//
+//   - #7153: the peer's remote-jobs opt-in defaults off, is reported to
+//     management (visible via the peers API as remote_jobs_allowed), and gates
+//     job execution on the client — a streamed job is refused until the peer
+//     opts in with `netbird up --allow-remote-jobs`, after which it runs.
+//   - #7147: the debug-bundle job's anonymize_level is validated (an unknown
+//     value is rejected at creation) and normalized (trimmed + lowercased) in
+//     the stored job the API returns.
+func TestRemoteJobsOptInAndBundleParams(t *testing.T) {
+	ctx := context.Background()
+
+	// A group for the setup key to auto-assign; peers must land in some group.
+	grp, err := srv.API().Groups.Create(ctx, api.PostApiGroupsJSONRequestBody{Name: "e2e-remotejobs"})
+	require.NoError(t, err, "create group")
+	t.Cleanup(func() { _ = srv.API().Groups.Delete(context.Background(), grp.Id) })
+
+	sk, err := srv.API().SetupKeys.Create(ctx, api.PostApiSetupKeysJSONRequestBody{
+		Name:       "e2e-remotejobs",
+		Type:       "reusable",
+		ExpiresIn:  86400,
+		UsageLimit: 0,
+		AutoGroups: []string{grp.Id},
+	})
+	require.NoError(t, err, "mint setup key")
+	require.NotEmpty(t, sk.Key, "setup key plaintext")
+	t.Cleanup(func() { _ = srv.API().SetupKeys.Delete(context.Background(), sk.Id) })
+
+	// Start the client with a plain `netbird up` (remote jobs NOT enabled).
+	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")
+
+	peerID := waitForPeer(ctx, t, cl.Hostname())
+
+	t.Run("opt-in flag defaults to false and is reported to management (#7153)", func(t *testing.T) {
+		p, err := srv.API().Peers.Get(ctx, peerID)
+		require.NoError(t, err)
+		allowed := remoteJobsAllowed(p)
+		require.NotNil(t, allowed, "remote_jobs_allowed must be present on the peer API")
+		assert.False(t, *allowed, "a peer that ran plain `netbird up` must default to opt-out")
+	})
+
+	t.Run("anonymize_level is validated and normalized (#7147)", func(t *testing.T) {
+		// Unknown level is rejected at job creation.
+		_, err := srv.API().Peers.Jobs(peerID).Create(ctx, bundleJob("bogus", testUploadURL))
+		require.Error(t, err, "an unknown anonymize_level must be rejected")
+		assert.Contains(t, strings.ToLower(err.Error()), "anonymize_level",
+			"the rejection must name the offending field")
+
+		// A messy but valid level is normalized (trimmed + lowercased) in the
+		// stored job the API echoes back.
+		job, err := srv.API().Peers.Jobs(peerID).Create(ctx, bundleJob("  Strict  ", testUploadURL))
+		require.NoError(t, err, "a valid anonymize_level must be accepted")
+		bw, err := job.Workload.AsBundleWorkloadResponse()
+		require.NoError(t, err, "job workload must be a bundle")
+		require.NotNil(t, bw.Parameters.AnonymizeLevel)
+		assert.Equal(t, "strict", *bw.Parameters.AnonymizeLevel,
+			"anonymize_level must be normalized to trimmed lowercase")
+		waitForJobTerminal(ctx, t, peerID, job.Id) // let it settle before the next create
+	})
+
+	t.Run("a job is refused while the peer has not opted in (#7153 enforcement)", func(t *testing.T) {
+		job, err := srv.API().Peers.Jobs(peerID).Create(ctx, bundleJob("default", testUploadURL))
+		require.NoError(t, err, "job creation itself is allowed; enforcement is on the client")
+		final := waitForJobTerminal(ctx, t, peerID, job.Id)
+		assert.Equal(t, api.JobResponseStatusFailed, final.Status, "the client must refuse the job")
+		require.NotNil(t, final.FailedReason)
+		assert.Contains(t, *final.FailedReason, refusedReason,
+			"the failure must be the opt-out refusal, not some other error")
+	})
+
+	t.Run("opting in flips the flag and lets the job run (#7153)", func(t *testing.T) {
+		require.NoError(t, cl.Up(ctx, "--allow-remote-jobs"), "re-run up with --allow-remote-jobs")
+		require.NoError(t, cl.WaitConnected(ctx, 90*time.Second), "client must reconnect")
+
+		// The new opt-in must round-trip to management and surface on the API.
+		require.Eventually(t, func() bool {
+			p, err := srv.API().Peers.Get(ctx, peerID)
+			if err != nil {
+				return false
+			}
+			allowed := remoteJobsAllowed(p)
+			return allowed != nil && *allowed
+		}, 60*time.Second, 2*time.Second, "remote_jobs_allowed must become true after opt-in")
+
+		// The same job that was refused before must now be accepted for
+		// execution: whatever its outcome, it must NOT be the opt-out refusal.
+		job, err := srv.API().Peers.Jobs(peerID).Create(ctx, bundleJob("default", testUploadURL))
+		require.NoError(t, err)
+		final := waitForJobTerminal(ctx, t, peerID, job.Id)
+		if final.Status == api.JobResponseStatusFailed && final.FailedReason != nil {
+			assert.NotContains(t, *final.FailedReason, refusedReason,
+				"once opted in, the job must not be refused for opt-out; any failure must be for another reason (e.g. upload)")
+		}
+	})
+}
+
+// remoteJobsAllowed returns the peer's remote-jobs opt-in flag from the API
+// response (nil if the peer or its local flags are absent).
+func remoteJobsAllowed(p *api.Peer) *bool {
+	if p == nil || p.LocalFlags == nil {
+		return nil
+	}
+	return p.LocalFlags.RemoteJobsAllowed
+}
+
+// bundleJob builds a debug-bundle job request with the given anonymize_level
+// (omitted when empty) and upload_url (omitted when empty).
+func bundleJob(anonymizeLevel, uploadURL string) api.JobRequest {
+	params := api.BundleParameters{
+		Anonymize:    true,
+		LogFileCount: 1,
+	}
+	if anonymizeLevel != "" {
+		params.AnonymizeLevel = &anonymizeLevel
+	}
+	if uploadURL != "" {
+		params.UploadUrl = &uploadURL
+	}
+	var wl api.WorkloadRequest
+	// FromBundleWorkloadRequest cannot fail for a well-formed value.
+	_ = wl.FromBundleWorkloadRequest(api.BundleWorkloadRequest{
+		Type:       api.WorkloadTypeBundle,
+		Parameters: params,
+	})
+	return api.JobRequest{Workload: wl}
+}
+
+// waitForPeer polls the peers API until the client that registered under the
+// given hostname appears and returns its ID. Matching by hostname rather than
+// taking the first list entry keeps the test correct if the account ever holds
+// more than one peer (a shared bootstrap account, or a second client added to
+// the package).
+func waitForPeer(ctx context.Context, t *testing.T, hostname string) string {
+	t.Helper()
+	var peerID string
+	require.Eventually(t, func() bool {
+		peers, err := srv.API().Peers.List(ctx)
+		if err != nil {
+			return false
+		}
+		for _, p := range peers {
+			if p.Hostname == hostname {
+				peerID = p.Id
+				return true
+			}
+		}
+		return false
+	}, 60*time.Second, 2*time.Second, "the client peer must register with management")
+	return peerID
+}
+
+// waitForJobTerminal polls a job until it leaves the pending state, then returns
+// the final response.
+func waitForJobTerminal(ctx context.Context, t *testing.T, peerID, jobID string) *api.JobResponse {
+	t.Helper()
+	var final *api.JobResponse
+	require.Eventually(t, func() bool {
+		j, err := srv.API().Peers.Jobs(peerID).Get(ctx, jobID)
+		if err != nil || j == nil {
+			return false
+		}
+		if j.Status == api.JobResponseStatusPending {
+			return false
+		}
+		final = j
+		return true
+	}, 120*time.Second, 2*time.Second, "job must reach a terminal state")
+	return final
+}

go.mod

@@ -1,8 +1,10 @@
 module github.com/netbirdio/netbird
 
-go 1.25.5
+go 1.26.0
 
-toolchain go1.25.12
+// Pin the toolchain to a patch release >= go1.26.2
+// See https://go.dev/issue/77875.
+toolchain go1.26.7
 
 require (
 	cunicu.li/go-rosenpass v0.5.42
@@ -71,17 +73,18 @@ require (
 	github.com/hashicorp/go-multierror v1.1.1
 	github.com/hashicorp/go-secure-stdlib/base62 v0.1.2
 	github.com/hashicorp/go-version v1.7.0
-	github.com/jackc/pgx/v5 v5.5.5
+	github.com/jackc/pgx/v5 v5.10.0
 	github.com/libdns/route53 v1.5.0
 	github.com/libp2p/go-netroute v0.4.0
 	github.com/lrh3321/ipset-go v0.0.0-20250619021614-54a0a98ace81
+	github.com/magefile/mage v1.17.2
 	github.com/mdlayher/socket v0.5.1
 	github.com/mdp/qrterminal/v3 v3.2.1
 	github.com/miekg/dns v1.1.72
 	github.com/mitchellh/hashstructure/v2 v2.0.2
 	github.com/moby/moby/api v1.54.1
 	github.com/netbirdio/go-nat v0.0.0-20260821095157-6b2c8c5c74e8
-	github.com/netbirdio/management-integrations/integrations v0.0.0-20260416123949-2355d972be42
+	github.com/netbirdio/management-integrations/integrations v0.0.0-20260803100840-78e79ba20f87
 	github.com/netbirdio/signal-dispatcher/dispatcher v0.0.0-20250805121659-6b4ac470ca45
 	github.com/oapi-codegen/runtime v1.1.2
 	github.com/okta/okta-sdk-golang/v2 v2.18.0
@@ -99,13 +102,14 @@ require (
 	github.com/pires/go-proxyproto v0.11.0
 	github.com/pkg/sftp v1.13.9
 	github.com/prometheus/client_golang v1.23.2
-	github.com/quic-go/quic-go v0.59.1
+	github.com/prometheus/client_model v0.6.2
+	github.com/quic-go/quic-go v0.62.0
 	github.com/redis/go-redis/v9 v9.7.3
 	github.com/rs/xid v1.3.0
 	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
+	github.com/stretchr/testify v1.12.1
 	github.com/testcontainers/testcontainers-go v0.37.0
 	github.com/testcontainers/testcontainers-go/modules/mysql v0.37.0
 	github.com/testcontainers/testcontainers-go/modules/postgres v0.37.0
@@ -236,8 +240,8 @@ require (
 	github.com/huin/goupnp v1.2.0 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
 	github.com/jackc/pgpassfile v1.0.0 // indirect
-	github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a // indirect
-	github.com/jackc/puddle/v2 v2.2.1 // indirect
+	github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 // indirect
+	github.com/jackc/puddle/v2 v2.2.2 // indirect
 	github.com/jackpal/go-nat-pmp v1.0.2 // indirect
 	github.com/jinzhu/inflection v1.0.0 // indirect
 	github.com/jinzhu/now v1.1.5 // indirect
@@ -249,6 +253,7 @@ require (
 	github.com/klauspost/cpuid/v2 v2.3.0 // indirect
 	github.com/koron/go-ssdp v0.0.4 // indirect
 	github.com/kr/fs v0.1.0 // indirect
+	github.com/kylelemons/godebug v1.1.0 // indirect
 	github.com/lib/pq v1.12.3 // indirect
 	github.com/libdns/libdns v0.2.2 // indirect
 	github.com/lufia/plan9stats v0.0.0-20240513124658-fba389f38bae // indirect
@@ -286,10 +291,8 @@ require (
 	github.com/pion/transport/v2 v2.2.4 // indirect
 	github.com/pion/turn/v4 v4.1.1 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
-	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
 	github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55 // indirect
 	github.com/pquerna/otp v1.5.0 // indirect
-	github.com/prometheus/client_model v0.6.2 // indirect
 	github.com/prometheus/common v0.67.5 // indirect
 	github.com/prometheus/otlptranslator v1.0.0 // indirect
 	github.com/prometheus/procfs v0.19.2 // indirect
@@ -297,7 +300,7 @@ require (
 	github.com/ryanuber/go-glob v1.0.0 // 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
+	github.com/stretchr/objx v0.5.3 // indirect
 	github.com/tinylib/msgp v1.6.3 // indirect
 	github.com/tklauser/go-sysconf v0.3.15 // indirect
 	github.com/tklauser/numcpus v0.10.0 // indirect
@@ -313,6 +316,7 @@ 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
+	go.yaml.in/yaml/v3 v3.0.5 // indirect
 	golang.org/x/text v0.41.0 // indirect
 	golang.org/x/tools v0.49.0 // indirect
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect

go.sum

@@ -341,12 +341,12 @@ github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
 github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM=
 github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg=
-github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a h1:bbPeKD0xmW/Y25WS6cokEszi5g+S0QxI/d45PkRi7Nk=
-github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
-github.com/jackc/pgx/v5 v5.5.5 h1:amBjrZVmksIdNjxGW/IiIMzxMKZFelXbUoPNb+8sjQw=
-github.com/jackc/pgx/v5 v5.5.5/go.mod h1:ez9gk+OAat140fv9ErkZDYFWmXLfV+++K0uAOiwgm1A=
-github.com/jackc/puddle/v2 v2.2.1 h1:RhxXJtFG022u4ibrCSMSiu5aOq1i77R3OHKNJj77OAk=
-github.com/jackc/puddle/v2 v2.2.1/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7UlwOQYpKFhBabPMi4aNAfoODPEFNiAnClxo=
+github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761/go.mod h1:5TJZWKEWniPve33vlWYSoGYefn3gLQRzjfDlhSJ9ZKM=
+github.com/jackc/pgx/v5 v5.10.0 h1:VhSvgU2jSli8o3AqIEOTJr7rZwAEUVo4E4XhR94Zfr0=
+github.com/jackc/pgx/v5 v5.10.0/go.mod h1:mal1tBGAFfLHvZzaYh77YS/eC6IX9OWbRV1QIIM0Jn4=
+github.com/jackc/puddle/v2 v2.2.2 h1:PR8nw+E/1w0GLuRFSmiioY6UooMp6KJv0/61nB7icHo=
+github.com/jackc/puddle/v2 v2.2.2/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
 github.com/jackpal/go-nat-pmp v1.0.2 h1:KzKSgb7qkJvOUTqYl9/Hg/me3pWgBmERKrTGD7BdWus=
 github.com/jackpal/go-nat-pmp v1.0.2/go.mod h1:QPH045xvCAeXUZOxsnwmrtiCoxIr9eob+4orBN1SBKc=
 github.com/jcmturner/aescts/v2 v2.0.0 h1:9YKLH6ey7H4eDBXW8khjYslgyqG2xZikXP0EQFKrle8=
@@ -413,6 +413,8 @@ github.com/lrh3321/ipset-go v0.0.0-20250619021614-54a0a98ace81 h1:J56rFEfUTFT9j9
 github.com/lrh3321/ipset-go v0.0.0-20250619021614-54a0a98ace81/go.mod h1:RD8ML/YdXctQ7qbcizZkw5mZ6l8Ogrl1dodBzVJduwI=
 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/magefile/mage v1.17.2 h1:fyXVu1eadI8Ap1HCCNgEhJ5McIWiYhLR8uol64ZZc40=
+github.com/magefile/mage v1.17.2/go.mod h1:Yj51kqllmsgFpvvSzgrZPK9WtluG3kUhFaBUVLo4feA=
 github.com/magiconair/properties v1.8.10 h1:s31yESBquKXCV9a/ScB3ESkOjUYYv+X0rg8SYxI99mE=
 github.com/magiconair/properties v1.8.10/go.mod h1:Dhd985XPs7jluiymwWYZ0G4Z61jb3vdS329zhj2hYo0=
 github.com/matryer/is v1.4.1 h1:55ehd8zaGABKLXQUe2awZ99BD/PTc2ls+KV/dXphgEQ=
@@ -482,8 +484,8 @@ github.com/netbirdio/go-nat v0.0.0-20260821095157-6b2c8c5c74e8 h1:pBxXEsxcsO3qVU
 github.com/netbirdio/go-nat v0.0.0-20260821095157-6b2c8c5c74e8/go.mod h1:mFViabv4PpnoDw9w7W21a7xux6APA4q7KQZRsv4BCl8=
 github.com/netbirdio/ice/v4 v4.0.0-20250908184934-6202be846b51 h1:Ov4qdafATOgGMB1wbSuh+0aAHcwz9hdvB6VZjh1mVMI=
 github.com/netbirdio/ice/v4 v4.0.0-20250908184934-6202be846b51/go.mod h1:ZSIbPdBn5hePO8CpF1PekH2SfpTxg1PDhEwtbqZS7R8=
-github.com/netbirdio/management-integrations/integrations v0.0.0-20260416123949-2355d972be42 h1:F3zS5fT9xzD1OFLfcdAE+3FfyiwjGukF1hvj0jErgs8=
-github.com/netbirdio/management-integrations/integrations v0.0.0-20260416123949-2355d972be42/go.mod h1:n47r67ZSPgwSmT/Z1o48JjZQW9YJ6m/6Bd/uAXkL3Pg=
+github.com/netbirdio/management-integrations/integrations v0.0.0-20260803100840-78e79ba20f87 h1:iJeUvSMC0BTpkw7u4JyWcY4/3dl7fEL9DR/TpKf2+1w=
+github.com/netbirdio/management-integrations/integrations v0.0.0-20260803100840-78e79ba20f87/go.mod h1:pmsCPx1S0nuZRxCextGpc9AV4hLgGSuTsc4NMuwGeCo=
 github.com/netbirdio/service v0.0.0-20240911161631-f62744f42502 h1:3tHlFmhTdX9axERMVN63dqyFqnvuD+EMJHzM7mNGON8=
 github.com/netbirdio/service v0.0.0-20240911161631-f62744f42502/go.mod h1:CIMRFEJVL+0DS1a3Nx06NaMn4Dz63Ng6O7dl0qH0zVM=
 github.com/netbirdio/signal-dispatcher/dispatcher v0.0.0-20250805121659-6b4ac470ca45 h1:ujgviVYmx243Ksy7NdSwrdGPSRNE3pb8kEDSpH0QuAQ=
@@ -580,8 +582,10 @@ github.com/prometheus/otlptranslator v1.0.0 h1:s0LJW/iN9dkIH+EnhiD3BlkkP5QVIUVEo
 github.com/prometheus/otlptranslator v1.0.0/go.mod h1:vRYWnXvI6aWGpsdY/mOT/cbeVRBlPWtBNDb7kGR3uKM=
 github.com/prometheus/procfs v0.19.2 h1:zUMhqEW66Ex7OXIiDkll3tl9a1ZdilUOd/F6ZXw4Vws=
 github.com/prometheus/procfs v0.19.2/go.mod h1:M0aotyiemPhBCM0z5w87kL22CxfcH05ZpYlu+b4J7mw=
-github.com/quic-go/quic-go v0.59.1 h1:0Gmua0HW1Tv7ANR7hUYwRyD0MG5OJfgvYSZasGZzBic=
-github.com/quic-go/quic-go v0.59.1/go.mod h1:upnsH4Ju1YkqpLXC305eW3yDZ4NfnNbmQRCMWS58IKU=
+github.com/quic-go/go-ossfuzz-seeds v0.1.0 h1:APacT+iIaNF6fd8AGEiN3bT/Jtkd2jz4v4TzM7MFjy0=
+github.com/quic-go/go-ossfuzz-seeds v0.1.0/go.mod h1:3IOHRbJIc+L6YKMwfDtJAM9Vj9k0YY4muhuyUYk5tbk=
+github.com/quic-go/quic-go v0.62.0 h1:ZHDjCk5OacATwGvs8PWE97CTvX7AqZiVoW7++ZOXTf8=
+github.com/quic-go/quic-go v0.62.0/go.mod h1:RAro2j2yN9a9EiPACLHT9IB2NXCvGQmmo/alT0yYI0w=
 github.com/redis/go-redis/v9 v9.7.3 h1:YpPyAayJV+XErNsatSElgRZZVCwXX9QzkKYNvO7x0wM=
 github.com/redis/go-redis/v9 v9.7.3/go.mod h1:bGUrSggJ9X9GUmZpZNEOQKaANxSGgOEBRltRTZHSvrA=
 github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
@@ -617,8 +621,8 @@ github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+
 github.com/stretchr/objx v0.2.0/go.mod h1:qt09Ya8vawLte6SNmTgCsAVtYtaKzEcn8ATUoHMkEqE=
 github.com/stretchr/objx v0.4.0/go.mod h1:YvHI0jy2hoMjB+UWwv71VJQ9isScKT/TqJzVSSt89Yw=
 github.com/stretchr/objx v0.5.0/go.mod h1:Yh+to48EsGEfYuaHDzXPcE3xhTkx73EhmCGUpEOglKo=
-github.com/stretchr/objx v0.5.2 h1:xuMeJ0Sdp5ZMRXx/aWO6RZxdr3beISkG5/G/aIRr3pY=
-github.com/stretchr/objx v0.5.2/go.mod h1:FRsXN1f5AsAjCGJKqEizvkpNtU+EGNCLh3NxZ/8L+MA=
+github.com/stretchr/objx v0.5.3 h1:jmXUvGomnU1o3W/V5h2VEradbpJDwGrzugQQvL0POH4=
+github.com/stretchr/objx v0.5.3/go.mod h1:rDQraq+vQZU7Fde9LOZLr8Tax6zZvy4kuNKF+QYS+U0=
 github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
 github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
 github.com/stretchr/testify v1.5.1/go.mod h1:5W2xD1RspED5o8YsWQXVCued0rvSQ+mT+I5cxcmMvtA=
@@ -628,8 +632,8 @@ 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 v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
-github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
+github.com/stretchr/testify v1.12.1 h1:EuwCh5fleGS7H32xRwO3wRGT7DxrDhLAT6FF8MpWDWE=
+github.com/stretchr/testify v1.12.1/go.mod h1:MDEgiDPPsNp5cuIrHPPCyornHKgEVbtFUmoNlxoYthg=
 github.com/testcontainers/testcontainers-go v0.37.0 h1:L2Qc0vkTw2EHWQ08djon0D2uw7Z/PtHS/QzZZ5Ra/hg=
 github.com/testcontainers/testcontainers-go v0.37.0/go.mod h1:QPzbxZhQ6Bclip9igjLFj6z0hs01bU8lrl2dHQmgFGM=
 github.com/testcontainers/testcontainers-go/modules/mysql v0.37.0 h1:LqUos1oR5iuuzorFnSvxsHNdYdCHB/DfI82CuT58wbI=
@@ -715,6 +719,8 @@ go.uber.org/zap v1.27.0/go.mod h1:GB2qFLM7cTU87MWRP2mPIjqfIDnGu+VIO4V/SdhGo2E=
 go.yaml.in/yaml/v2 v2.4.3 h1:6gvOSjQoTB3vt1l+CU+tSyi/HOjfOjRLJ4YwYZGwRO0=
 go.yaml.in/yaml/v2 v2.4.3/go.mod h1:zSxWcmIDjOzPXpjlTTbAsKokqkDNAVtZO0WOMiT90s8=
 go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
+go.yaml.in/yaml/v3 v3.0.5 h1:N6y/pJk8buWs9NY5ERU2HSMfm+IuD/OtfdAnq6kESPw=
+go.yaml.in/yaml/v3 v3.0.5/go.mod h1:HVTZu1O7/Vkt2N+BFy8Zza+lnLsABggaTM2ZpNIGuKg=
 goauthentik.io/api/v3 v3.2023051.3 h1:NebAhD/TeTWNo/9X3/Uj+rM5fG1HaiLOlKTNLQv9Qq4=
 goauthentik.io/api/v3 v3.2023051.3/go.mod h1:nYECml4jGbp/541hj8GcylKQG1gVBsKppHy4+7G8u4U=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=

infrastructure_files/getting-started-enterprise.sh

@@ -15,16 +15,25 @@ NETBIRD_EULA_URL="https://netbird.io/self-hosted-EULA"
 # server trusts X-Forwarded-* headers from this address only.
 TRAEFIK_IP="172.30.0.10"
 
+LICENSE_VERDICT="unknown"
+LICENSE_LOG_LINES=""
+
 check_docker_compose() {
-  if command -v docker-compose &> /dev/null; then
-    echo "docker-compose"
-    return
+  if ! command -v docker &> /dev/null && ! command -v docker-compose &> /dev/null; then
+    echo "Docker is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/engine/install/" > /dev/stderr
+    exit 1
   fi
-  if docker compose --help &> /dev/null; then
+
+  if docker compose version &> /dev/null; then
     echo "docker compose"
     return
   fi
-  echo "docker-compose is not installed or not in PATH. See https://docs.docker.com/engine/install/" > /dev/stderr
+  if command -v docker-compose &> /dev/null && docker-compose version &> /dev/null; then
+    echo "docker-compose"
+    return
+  fi
+
+  echo "Docker Compose is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/compose/install/" > /dev/stderr
   exit 1
 }
 
@@ -221,6 +230,90 @@ wait_postgres() {
   set -e
 }
 
+wait_for_license_verdict() {
+  local counter=0
+  local logs=""
+
+  echo -n "Waiting for the server to validate the license"
+  while [[ $counter -lt 60 ]]; do
+    logs=$($DOCKER_COMPOSE_COMMAND logs --no-color --tail=all netbird-server 2>/dev/null || true)
+
+    if grep -qi "license invalidated" <<< "$logs"; then
+      echo " rejected"
+      LICENSE_VERDICT="rejected"
+      LICENSE_LOG_LINES=$(grep -i "license" <<< "$logs" | tail -n 5 || true)
+      return 0
+    fi
+
+    if grep -qi "license validated" <<< "$logs"; then
+      echo " ok"
+      LICENSE_VERDICT="ok"
+      return 0
+    fi
+
+    echo -n " ."
+    sleep 2
+    counter=$((counter + 1))
+  done
+
+  echo " no verdict in 120s"
+  LICENSE_VERDICT="unknown"
+  LICENSE_LOG_LINES=$(grep -iE "failed to validate license|error validating license" <<< "$logs" | tail -n 3 || true)
+  return 0
+}
+
+report_license_verdict() {
+  if [[ "$LICENSE_VERDICT" == "ok" ]]; then
+    return 0
+  fi
+
+  if [[ "$LICENSE_VERDICT" == "unknown" ]]; then
+    echo ""
+    echo "  ⚠  The server logged no license verdict within 120s."
+    if [[ -n "$LICENSE_LOG_LINES" ]]; then
+      echo "     It was still reporting validation errors:"
+      while IFS= read -r line; do
+        [[ -n "$line" ]] && echo "     $line"
+      done <<< "$LICENSE_LOG_LINES"
+    fi
+    echo ""
+    echo "     Check the verdict with:"
+    echo ""
+    echo "       $DOCKER_COMPOSE_COMMAND logs netbird-server | grep -i license"
+    return 0
+  fi
+
+  local unreachable="false"
+  if grep -qi "couldn't be validated with the license server" <<< "$LICENSE_LOG_LINES"; then
+    unreachable="true"
+  fi
+
+  echo ""
+  if [[ "$unreachable" == "true" ]]; then
+    echo "  ⚠  The server could not validate the license:"
+  else
+    echo "  ⚠  The server rejected the license key:"
+  fi
+  while IFS= read -r line; do
+    [[ -n "$line" ]] && echo "     $line"
+  done <<< "$LICENSE_LOG_LINES"
+  echo ""
+  echo "     The stack is up, and only the license check did not pass."
+  echo ""
+  if [[ "$unreachable" == "true" ]]; then
+    echo "     The license server could not be reached, so the key itself was"
+    echo "     never checked. Confirm this host has outbound access to the"
+    echo "     license server, then restart:"
+  else
+    echo "     Check the reason the server gave above, verify that"
+    echo "     NETBIRD_LICENSE_KEY in .env matches the key you were issued,"
+    echo "     then restart:"
+  fi
+  echo ""
+  echo "       $DOCKER_COMPOSE_COMMAND up -d"
+  return 0
+}
+
 init_environment() {
   check_openssl
   DOCKER_COMPOSE_COMMAND=$(check_docker_compose)
@@ -299,6 +392,9 @@ init_environment() {
   echo "Starting remaining services ..."
   $DOCKER_COMPOSE_COMMAND up -d
 
+  echo ""
+  wait_for_license_verdict
+
   echo ""
   echo "Done."
   echo ""
@@ -309,6 +405,12 @@ init_environment() {
   echo ""
   echo "Tail logs:"
   echo "  cd $(pwd) && $DOCKER_COMPOSE_COMMAND logs -f netbird-server traefik"
+
+  report_license_verdict
+
+  if [[ "$LICENSE_VERDICT" == "rejected" ]]; then
+    exit 1
+  fi
 }
 
 # ------------------------------------------------------------------

infrastructure_files/getting-started.sh

@@ -60,18 +60,21 @@ check_docker_sock_perms() {
 }
 
 check_docker_compose() {
-  if command -v docker-compose &> /dev/null
-  then
-      echo "docker-compose"
-      return
-  fi
-  if docker compose --help &> /dev/null
-  then
-      echo "docker compose"
-      return
+  if ! command -v docker &> /dev/null && ! command -v docker-compose &> /dev/null; then
+    echo "Docker is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/engine/install/" > /dev/stderr
+    exit 1
   fi
 
-  echo "docker-compose is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/engine/install/" > /dev/stderr
+  if docker compose version &> /dev/null; then
+    echo "docker compose"
+    return
+  fi
+  if command -v docker-compose &> /dev/null && docker-compose version &> /dev/null; then
+    echo "docker-compose"
+    return
+  fi
+
+  echo "Docker Compose is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/compose/install/" > /dev/stderr
   exit 1
 }
 
@@ -98,19 +101,39 @@ get_main_ip_address() {
 }
 
 check_nb_domain() {
-  DOMAIN=$1
-  if [[ "$DOMAIN-x" == "-x" ]]; then
+  local domain="$1"
+
+  if [[ -z "$domain" ]]; then
     echo "The NETBIRD_DOMAIN variable cannot be empty." > /dev/stderr
     return 1
   fi
-
-  if [[ "$DOMAIN" == "netbird.example.com" ]]; then
+  if [[ "$domain" == "use-ip" ]]; then
+    return 0
+  fi
+  if [[ "$domain" == "netbird.example.com" ]]; then
     echo "The NETBIRD_DOMAIN cannot be netbird.example.com" > /dev/stderr
     return 1
   fi
+  if [[ "$domain" =~ ^[0-9.]+$ ]]; then
+    echo "'$domain' is an IP address. Use 'use-ip' to install on this host's IP over HTTP, or an FQDN to get a TLS certificate." > /dev/stderr
+    return 1
+  fi
+  if [[ ! "$domain" =~ ^[A-Za-z0-9]([A-Za-z0-9-]*[A-Za-z0-9])?(\.[A-Za-z0-9]([A-Za-z0-9-]*[A-Za-z0-9])?)+$ ]]; then
+    echo "'$domain' is not a valid FQDN. It needs at least one dot (e.g. netbird.my-domain.com), with no scheme, port or trailing dot." > /dev/stderr
+    return 1
+  fi
   return 0
 }
 
+check_domain_resolves() {
+  local domain="$1"
+  if command -v getent &> /dev/null && getent hosts "$domain" &> /dev/null; then return 0; fi
+  if command -v host &> /dev/null && host "$domain" &> /dev/null; then return 0; fi
+  if command -v dig &> /dev/null && [[ -n "$(dig +short "$domain" 2>/dev/null)" ]]; then return 0; fi
+  if command -v nslookup &> /dev/null && nslookup "$domain" &> /dev/null; then return 0; fi
+  return 1
+}
+
 # Non-interactive configuration
 # ------------------------------
 # Every prompt below can be pre-answered with an environment variable, so the
@@ -170,7 +193,22 @@ read_nb_domain() {
   read -r READ_NETBIRD_DOMAIN < /dev/tty
   if ! check_nb_domain "$READ_NETBIRD_DOMAIN"; then
     read_nb_domain
+    return
   fi
+
+  if [[ "$READ_NETBIRD_DOMAIN" != "use-ip" ]] && ! check_domain_resolves "$READ_NETBIRD_DOMAIN"; then
+    local confirm=""
+    echo "" > /dev/stderr
+    echo "Warning: '$READ_NETBIRD_DOMAIN' does not resolve via DNS from this host." > /dev/stderr
+    echo "TLS certificate issuance and client connections will fail until it does." > /dev/stderr
+    echo -n "Continue anyway? [y/N]: " > /dev/stderr
+    read -r confirm < /dev/tty
+    if [[ ! "$confirm" =~ ^[Yy]$ ]]; then
+      read_nb_domain
+      return
+    fi
+  fi
+
   echo "$READ_NETBIRD_DOMAIN"
   return 0
 }
@@ -439,12 +477,23 @@ configure_domain() {
   # Domain is validated (not a free-form value), so it keeps its own guard
   # rather than going through resolve(): a valid NETBIRD_DOMAIN is used as-is,
   # otherwise we prompt, or abort when there is no terminal to prompt on.
+  local prompted="false"
   if ! check_nb_domain "$NETBIRD_DOMAIN"; then
     if ! tty_available; then
-      echo "NETBIRD_DOMAIN is required for a non-interactive install." > /dev/stderr
+      if [[ -n "$NETBIRD_DOMAIN" ]]; then
+        echo "NETBIRD_DOMAIN='$NETBIRD_DOMAIN' cannot be used for a non-interactive install." > /dev/stderr
+      else
+        echo "NETBIRD_DOMAIN is required for a non-interactive install." > /dev/stderr
+      fi
       exit 1
     fi
     NETBIRD_DOMAIN=$(read_nb_domain)
+    prompted="true"
+  fi
+
+  if [[ "$prompted" == "false" && "$NETBIRD_DOMAIN" != "use-ip" ]] && ! check_domain_resolves "$NETBIRD_DOMAIN"; then
+    echo "Warning: '$NETBIRD_DOMAIN' does not resolve via DNS from this host." > /dev/stderr
+    echo "TLS certificate issuance and client connections will fail until it does." > /dev/stderr
   fi
 
   if [[ "$NETBIRD_DOMAIN" == "use-ip" ]]; then

infrastructure_files/migrate-to-enterprise.sh

@@ -40,6 +40,10 @@ ENTERPRISE_CONFIG_FILE="config.yaml.enterprise"
 # completed successfully.
 ROLLBACK_STATE="disarmed"
 ENV_EXISTED="unknown"
+# Verdict the server logs about the license key on startup: ok, rejected, or
+# unknown when neither line appeared before the timeout.
+LICENSE_VERDICT="unknown"
+LICENSE_LOG_LINES=""
 ENV_BACKUP=""
 PG_VOLUME_NAME=""
 BACKUP_DIR=""
@@ -59,15 +63,21 @@ ENTERPRISE_CONFIG="no"
 NETBIRD_EULA_URL="https://netbird.io/self-hosted-EULA"
 
 check_docker_compose() {
-  if command -v docker-compose &> /dev/null; then
-    echo "docker-compose"
-    return
+  if ! command -v docker &> /dev/null && ! command -v docker-compose &> /dev/null; then
+    echo "Docker is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/engine/install/" > /dev/stderr
+    exit 1
   fi
-  if docker compose --help &> /dev/null; then
+
+  if docker compose version &> /dev/null; then
     echo "docker compose"
     return
   fi
-  echo "docker-compose is not installed or not in PATH." > /dev/stderr
+  if command -v docker-compose &> /dev/null && docker-compose version &> /dev/null; then
+    echo "docker-compose"
+    return
+  fi
+
+  echo "Docker Compose is not installed or not in PATH. Please follow the steps from the official guide: https://docs.docker.com/compose/install/" > /dev/stderr
   exit 1
 }
 
@@ -1000,6 +1010,39 @@ init_migration() {
   check_stale_postgres_volume
 }
 
+wait_for_license_verdict() {
+  local counter=0
+  local logs=""
+
+  echo -n "Waiting for the server to validate the license"
+  while [[ $counter -lt 60 ]]; do
+
+    logs=$($DOCKER_COMPOSE_COMMAND logs --no-color --tail=all "$COMBINED_SERVICE" 2>/dev/null || true)
+
+    if grep -qi "license invalidated" <<< "$logs"; then
+      echo " rejected"
+      LICENSE_VERDICT="rejected"
+      LICENSE_LOG_LINES=$(grep -i "license" <<< "$logs" | tail -n 5 || true)
+      return 0
+    fi
+
+    if grep -qi "license validated" <<< "$logs"; then
+      echo " ok"
+      LICENSE_VERDICT="ok"
+      return 0
+    fi
+
+    echo -n " ."
+    sleep 2
+    counter=$((counter + 1))
+  done
+
+  echo " no verdict in 120s"
+  LICENSE_VERDICT="unknown"
+  LICENSE_LOG_LINES=$(grep -iE "failed to validate license|error validating license" <<< "$logs" | tail -n 3 || true)
+  return 0
+}
+
 apply_changes() {
   # From here on a failure must roll the deployment back.
   ROLLBACK_STATE="armed"
@@ -1100,9 +1143,57 @@ apply_changes() {
   echo "Bringing up all services ..."
   $DOCKER_COMPOSE_COMMAND up -d
 
+  echo ""
+  wait_for_license_verdict
+
   echo ""
   echo "Migration complete."
 
+  if [[ "$LICENSE_VERDICT" == "rejected" ]]; then
+    local unreachable="false"
+    if grep -qi "couldn't be validated with the license server" <<< "$LICENSE_LOG_LINES"; then
+      unreachable="true"
+    fi
+
+    echo ""
+    if [[ "$unreachable" == "true" ]]; then
+      echo "  ⚠  The server could not validate the license:"
+    else
+      echo "  ⚠  The server rejected the license key:"
+    fi
+    while IFS= read -r line; do
+      [[ -n "$line" ]] && echo "     $line"
+    done <<< "$LICENSE_LOG_LINES"
+    echo ""
+    echo "     The migration itself completed: the images and any migrated data"
+    echo "     are in place, and only the license check did not pass."
+    echo ""
+    if [[ "$unreachable" == "true" ]]; then
+      echo "     The license server could not be reached, so the key itself was"
+      echo "     never checked. Confirm this host has outbound access to the"
+      echo "     license server, then restart:"
+    else
+      echo "     Check the reason the server gave above, verify that"
+      echo "     NB_LICENSE_KEY in .env matches the key you were issued, then"
+      echo "     restart:"
+    fi
+    echo ""
+    echo "       $DOCKER_COMPOSE_COMMAND up -d"
+  elif [[ "$LICENSE_VERDICT" == "unknown" ]]; then
+    echo ""
+    echo "  ⚠  The server logged no license verdict within 120s."
+    if [[ -n "$LICENSE_LOG_LINES" ]]; then
+      echo "     It was still reporting validation errors:"
+      while IFS= read -r line; do
+        [[ -n "$line" ]] && echo "     $line"
+      done <<< "$LICENSE_LOG_LINES"
+    fi
+    echo ""
+    echo "     Check the verdict with:"
+    echo ""
+    echo "       $DOCKER_COMPOSE_COMMAND logs $COMBINED_SERVICE | grep -i license"
+  fi
+
   # Nothing left to undo.
   ROLLBACK_STATE="disarmed"
 }
@@ -1122,6 +1213,11 @@ print_summary() {
   fi
   [[ "$ENABLE_FLOW" == "yes" ]] && echo "  Traffic flow:     enabled"
   [[ "$ENABLE_FLOW" != "yes" ]] && echo "  Traffic flow:     disabled"
+  case "$LICENSE_VERDICT" in
+    ok)       echo "  License:          validated by the server" ;;
+    rejected) echo "  License:          REJECTED - see above, the install is not usable yet" ;;
+    *)        echo "  License:          not confirmed (no verdict in the logs yet)" ;;
+  esac
   echo ""
   echo "  Generated files (next to your docker-compose.yml):"
   echo "    $OVERRIDE_FILE"
@@ -1176,3 +1272,10 @@ trap 'exit 130' INT TERM
 init_migration
 apply_changes
 print_summary
+
+# A rejected license leaves a migrated but unusable install. Say so in the exit
+# code too, or a wrapper script reads this run as a clean success.
+if [[ "$LICENSE_VERDICT" == "rejected" ]]; then
+  exit 1
+fi
+exit 0

integration_tests/management/network_map_db/account_settings_test.go

@@ -0,0 +1,58 @@
+//go:build integration
+
+package networkmap_pgsql
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestGetAccountSettings(t *testing.T) {
+	ctx := context.TODO()
+
+	execQuery(t, ctx,
+		`insert into accounts (id, settings_peer_login_expiration_enabled, settings_peer_login_expiration, settings_peer_inactivity_expiration_enabled,
+                      settings_peer_inactivity_expiration, settings_dns_domain, settings_ipv6_enabled_groups, settings_routing_peer_dns_resolution_enabled,
+                      settings_lazy_connection_enabled, settings_auto_update_version, settings_auto_update_always, settings_metrics_push_enabled)
+		 values('account-3',null,null,null,null,null,null,null,null,null,null,null)`)
+
+	accountSettings, err := conn(t, ctx).GetAccountSettings(ctx, "account-1")
+	assert.NoError(t, err)
+	assert.Equal(t, accountSettings, nmdata.AccountSettingsInfo{
+		PeerLoginExpirationEnabled:      true,
+		PeerLoginExpiration:             86400000000000 * time.Nanosecond,
+		PeerInactivityExpirationEnabled: false,
+		PeerInactivityExpiration:        86400000000000 * time.Nanosecond,
+		DNSDomain:                       "",
+		IPv6EnabledGroups:               []string{"group-one-resource-id"},
+		RoutingPeerDNSResolutionEnabled: false,
+		LazyConnectionEnabled:           false,
+		AutoUpdateVersion:               "disabled",
+		AutoUpdateAlways:                false,
+		MetricsPushEnabled:              false,
+	})
+
+	accountSettings, err = conn(t, ctx).GetAccountSettings(ctx, "account-2")
+	assert.NoError(t, err)
+	assert.Equal(t, accountSettings, nmdata.AccountSettingsInfo{
+		PeerLoginExpirationEnabled:      true,
+		PeerLoginExpiration:             86400000000000 * time.Nanosecond,
+		PeerInactivityExpirationEnabled: false,
+		PeerInactivityExpiration:        86400000000000 * time.Nanosecond,
+		DNSDomain:                       "",
+		IPv6EnabledGroups:               []string{"group-two-resources-id"},
+		RoutingPeerDNSResolutionEnabled: false,
+		LazyConnectionEnabled:           false,
+		AutoUpdateVersion:               "disabled",
+		AutoUpdateAlways:                false,
+		MetricsPushEnabled:              false,
+	})
+
+	accountSettings, err = conn(t, ctx).GetAccountSettings(ctx, "account-3")
+	assert.NoError(t, err)
+	assert.Equal(t, accountSettings, nmdata.AccountSettingsInfo{})
+}

integration_tests/management/network_map_db/base_data.sql

@@ -0,0 +1,53 @@
+insert into accounts (id, network_identifier, network_net, network_net_v6, network_dns, network_serial,dns_settings_disabled_management_groups,
+                      settings_peer_login_expiration_enabled, settings_peer_login_expiration, settings_peer_inactivity_expiration_enabled,
+                      settings_peer_inactivity_expiration, settings_dns_domain, settings_ipv6_enabled_groups, settings_routing_peer_dns_resolution_enabled,
+                      settings_lazy_connection_enabled, settings_auto_update_version, settings_auto_update_always, settings_metrics_push_enabled)
+VALUES('account-1','network-1','{"IP":"100.103.0.0","Mask":"//8AAA=="}','{"IP":"fdde:e995:fd38:a465::","Mask":"//////////8AAAAAAAAAAA=="}','',1,'["disabled-group-1","disabled-group-2"]',
+       true, 86400000000000, false,
+       86400000000000, null, '["group-one-resource-id"]', false,
+       false, 'disabled', false, false);
+insert into accounts (id, network_identifier, network_net, network_net_v6, network_dns, network_serial,dns_settings_disabled_management_groups,
+                      settings_peer_login_expiration_enabled, settings_peer_login_expiration, settings_peer_inactivity_expiration_enabled,
+                      settings_peer_inactivity_expiration, settings_dns_domain, settings_ipv6_enabled_groups, settings_routing_peer_dns_resolution_enabled,
+                      settings_lazy_connection_enabled, settings_auto_update_version, settings_auto_update_always, settings_metrics_push_enabled)
+VALUES('account-2','network-2','{"IP":"110.0.0.0","Mask":"//8AAA=="}','{"IP":"fddf:e995:fd38:a465::","Mask":"//////////8AAAAAAAAAAA=="}','',2,null,
+       true, 86400000000000, false,
+       86400000000000, null, '["group-two-resources-id"]', false,
+       false, 'disabled', false, false);
+insert into groups (id, account_id, name, resources, public_id) VALUES('group-one-resource-id','account-1','group-1-name', '[{"ID":"host-id-1","Type":"host"}]','group-one-resource-id-public');
+insert into groups (id, account_id, name, resources, public_id) VALUES('group-two-resources-id','account-1','group-2-name', '[{"ID":"subnet-id-1","Type":"subnet"}, {"ID":"host-id-2","Type":"host"}]','group-two-resources-id-public');
+insert into groups (id, account_id, name, resources, public_id) VALUES('group-no-resources-id','account-1','group-3-name', null,'group-no-resources-id-public');
+insert into group_peers (account_id, peer_id, group_id) VALUES('account-1','peer-id-1','group-one-resource-id');
+insert into group_peers (account_id, peer_id, group_id) VALUES('account-1','peer-id-2','group-two-resources-id');
+insert into group_peers (account_id, peer_id, group_id) VALUES('account-1','peer-id-3','group-two-resources-id');
+insert into peers (id, account_id, "key", ssh_key, dns_label, extra_dns_labels, user_id, ssh_enabled, login_expiration_enabled, last_login, ip, ipv6,
+	            peer_status_requires_approval, peer_status_connected, proxy_meta_embedded, proxy_meta_cluster,
+	            meta_wt_version, meta_go_os, meta_os_version, meta_kernel_version, meta_network_addresses, meta_files,
+                   meta_capabilities, meta_flags, meta_sync_message_version,
+	            location_country_code, location_city_name, location_connection_ip)
+            values('peer-id-1','account-1','key-1','ssh-key-1','peer-1','["extra-peer-1"]','user-id-1',true,true,'2026-08-06 13:25:59.12999','"10.10.10.1"','"fdf4:ba80:6aa5:89f1:44d7:8701:8699:4940"',
+                   false,true,true,'cluster-1.netbird.services',
+                   '0.76.0','linux','26.4.1','6.8.0-134-generic','[{"NetIP":"fe80::8b4c:973f:a76b:3771/64","Mac":"00:15:5d:24:0c:ac"},{"NetIP":"192.168.16.1/20","Mac":"00:15:5d:24:0c:ac"}]','[{"Path":"/usr/bin/netbird","Exist":false,"ProcessIsRunning":false}]',
+                   '[1,2]','{"RosenpassEnabled":false,"RosenpassPermissive":false,"ServerSSHAllowed":true,"DisableClientRoutes":false,"DisableServerRoutes":false,"DisableDNS":false,"DisableFirewall":false,"BlockLANAccess":false,"BlockInbound":false,"DisableIPv6":false,"LazyConnectionEnabled":false}',1,
+                   'DE','Berlin','"46.201.148.187"');
+insert into peers (id,account_id,"key", ssh_key, dns_label, extra_dns_labels, user_id, ssh_enabled, login_expiration_enabled, last_login, ip, ipv6,
+	            peer_status_requires_approval, peer_status_connected, proxy_meta_embedded, proxy_meta_cluster,
+	            meta_wt_version, meta_go_os, meta_os_version, meta_kernel_version, meta_network_addresses, meta_files,
+                   meta_capabilities, meta_flags, meta_sync_message_version,
+	            location_country_code, location_city_name, location_connection_ip)
+            values('peer-id-2','account-1','key-2','ssh-key-2','peer-2','["extra-peer-2"]','user-id-2',true,true,'2026-08-06 14:25:59.12999','"10.10.100.1"','"fdf5:ba80:6aa5:89f1:44d7:8701:8699:4940"',
+                   false,true,true,'cluster-2.netbird.services',
+                   '0.76.1','linux','26.4.2','6.8.0-135-generic','[{"NetIP":"fe81::8b4c:973f:a76b:3771/64","Mac":"00:15:5d:24:0c:ad"},{"NetIP":"192.168.17.1/20","Mac":"00:15:5d:24:0c:ad"}]','[{"Path":"/usr/bin/netbird","Exist":false,"ProcessIsRunning":false}]',
+                   '[1,2]','{"RosenpassEnabled":false,"RosenpassPermissive":false,"ServerSSHAllowed":true,"DisableClientRoutes":false,"DisableServerRoutes":false,"DisableDNS":false,"DisableFirewall":false,"BlockLANAccess":false,"BlockInbound":false,"DisableIPv6":false,"LazyConnectionEnabled":false}',0,
+                   'DE','Berlin','"46.201.149.187"');
+insert into peers (id,account_id,"key", ssh_key, dns_label, extra_dns_labels, user_id, ssh_enabled, login_expiration_enabled, last_login, ip, ipv6,
+	            peer_status_requires_approval, peer_status_connected, proxy_meta_embedded, proxy_meta_cluster,
+	            meta_wt_version, meta_go_os, meta_os_version, meta_kernel_version, meta_network_addresses, meta_files,
+                   meta_capabilities, meta_flags, meta_sync_message_version,
+	            location_country_code, location_city_name, location_connection_ip)
+            values('peer-id-3','account-1','key-3','ssh-key-3','peer-3','["extra-peer-3"]','user-id-3',true,true,'2026-08-06 12:25:59.12999','"10.10.200.1"','"fdf6:ba80:6aa5:89f1:44d7:8701:8699:4940"',
+                   false,true,true,'cluster-3.netbird.services',
+                   '0.76.2','linux','26.4.3','6.8.0-136-generic','[{"NetIP":"fe82::8b4c:973f:a76b:3771/64","Mac":"00:15:5d:24:0c:ae"},{"NetIP":"192.168.18.1/20","Mac":"00:15:5d:24:0c:ae"}]','[{"Path":"/usr/bin/netbird","Exist":false,"ProcessIsRunning":false}]',
+                   '[1,2]','{"RosenpassEnabled":false,"RosenpassPermissive":false,"ServerSSHAllowed":true,"DisableClientRoutes":false,"DisableServerRoutes":false,"DisableDNS":false,"DisableFirewall":false,"BlockLANAccess":false,"BlockInbound":false,"DisableIPv6":false,"LazyConnectionEnabled":false}',1,
+                   'DE','Berlin','"46.201.150.187"');
+

integration_tests/management/network_map_db/dns_settings_test.go

@@ -0,0 +1,25 @@
+//go:build integration
+
+package networkmap_pgsql
+
+import (
+	"context"
+	"testing"
+
+	"github.com/netbirdio/netbird/shared/management/networkmap/nmdata"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestGetDnsSettings(t *testing.T) {
+	ctx := context.TODO()
+
+	settings, err := conn(t, ctx).GetDnsSettings(ctx, "account-1")
+	assert.NoError(t, err)
+	assert.Equal(t, settings, nmdata.DNSSettings{
+		DisabledManagementGroups: []string{"disabled-group-1", "disabled-group-2"},
+	})
+
+	settings, err = conn(t, ctx).GetDnsSettings(ctx, "account-2")
+	assert.NoError(t, err)
+	assert.Equal(t, settings, nmdata.DNSSettings{})
+}