Subnet docker conflict check; enterprice domain alias.

## Describe your changes

The relay client's reconnect backoff was constructed without a
`RandomizationFactor`
([guard.go:156-165](https://github.com/netbirdio/netbird/blob/main/shared/relay/client/guard.go#L156-L165)),
so it kept the zero value: every client that lost the same relay server
retried on the identical 2/4/8/16/32/60, potentially in waves.

It was the only exponential backoff in the codebase without a
randomization factor.

Use `backoff.DefaultRandomizationFactor`, as random factor.

## Issue ticket number and link

No public issue. Found while reviewing the relay reconnect path for the
client-metrics review:
22k relay reconnection events in 24h, and the shared transport's own
retry schedule was
identical across all clients.

## Stack

<!-- branch-stack -->

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

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

## Documentation
Select exactly one:

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

Internal retry-timing change with no user-visible surface: no CLI flag,
configuration option
or API field is added or altered, and the mean reconnect delay is
unchanged.

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

N/A


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

* **Bug Fixes**
* Improved reconnect timing to distribute repeated connection attempts
more evenly and reduce synchronized retry spikes.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

client/android/profile_state_test.go

@@ -16,17 +16,17 @@ func TestProfileAccountPathFor(t *testing.T) {
 		{
 			name:       "default profile",
 			configPath: "/data/data/io.netbird.client/files/netbird.cfg",
-			want:       "/data/data/io.netbird.client/files/netbird.account.json",
+			want:       filepath.FromSlash("/data/data/io.netbird.client/files/netbird.account.json"),
 		},
 		{
 			name:       "id profile",
 			configPath: "/data/data/io.netbird.client/files/profiles/4c5f5c8198c3989cffb5b5394f5a7ae0.json",
-			want:       "/data/data/io.netbird.client/files/profiles/4c5f5c8198c3989cffb5b5394f5a7ae0.account.json",
+			want:       filepath.FromSlash("/data/data/io.netbird.client/files/profiles/4c5f5c8198c3989cffb5b5394f5a7ae0.account.json"),
 		},
 		{
 			name:       "legacy name-keyed profile is handled the same way",
 			configPath: "/data/data/io.netbird.client/files/profiles/work.json",
-			want:       "/data/data/io.netbird.client/files/profiles/work.account.json",
+			want:       filepath.FromSlash("/data/data/io.netbird.client/files/profiles/work.account.json"),
 		},
 		{
 			name:       "empty path is rejected",

client/server/login_outcome_test.go

@@ -8,8 +8,6 @@ import (
 	"testing"
 
 	"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/proto"
@@ -27,9 +25,9 @@ func TestLogin_ManagementUnreachableIsReturnedInsteadOfDemandingSSO(t *testing.T
 
 	unreachable := errors.New("create connection: dial context: context deadline exceeded")
 	attempts := 0
-	s.loginAttemptFn = func(context.Context, string, string) (internal.StatusType, error) {
+	s.isLoginRequiredFn = func(context.Context) (bool, error) {
 		attempts++
-		return internal.StatusLoginFailed, unreachable
+		return false, unreachable
 	}
 
 	resp, err := s.Login(userCtx(), &proto.LoginRequest{Username: &username})
@@ -55,15 +53,12 @@ func TestLogin_AuthRefusalStartsSSOFlow(t *testing.T) {
 	s.rootCtx = internal.CtxInitState(context.Background())
 	breakProfilePrivateKey(t, cfgPath)
 
-	refused := gstatus.Error(codes.PermissionDenied, "peer is not registered")
-	s.loginAttemptFn = func(context.Context, string, string) (internal.StatusType, error) {
-		return internal.StatusNeedsLogin, refused
+	s.isLoginRequiredFn = func(context.Context) (bool, error) {
+		return true, nil
 	}
 
 	_, err := s.Login(userCtx(), &proto.LoginRequest{Username: &username})
 	require.Error(t, err)
-	require.NotErrorIs(t, err, refused,
-		"the refusal was handed back to the caller instead of starting the SSO flow")
 
 	status, stateErr := internal.CtxGetState(s.rootCtx).Status()
 	require.NoError(t, stateErr)
@@ -71,6 +66,32 @@ func TestLogin_AuthRefusalStartsSSOFlow(t *testing.T) {
 		"the SSO flow setup was never reached with the broken key")
 }
 
+func TestLogin_SetupKeyStillRunsWhenPeerNeedsLogin(t *testing.T) {
+	s, _, _, username, _ := setupServerWithProfile(t)
+	s.rootCtx = internal.CtxInitState(context.Background())
+
+	s.isLoginRequiredFn = func(context.Context) (bool, error) {
+		return true, nil
+	}
+
+	var keysTried []string
+	s.loginAttemptFn = func(_ context.Context, setupKey, _ string) (internal.StatusType, error) {
+		keysTried = append(keysTried, setupKey)
+		return "", nil
+	}
+
+	setupKey := "A2C8E32F-AEB2-4B45-8FD3-8A0C1B2D3E4F"
+	resp, err := s.Login(userCtx(), &proto.LoginRequest{Username: &username, SetupKey: setupKey})
+	require.NoError(t, err, "the probe's outcome leaked out as the login result")
+	require.NotNil(t, resp)
+	require.Equal(t, []string{setupKey}, keysTried, "the setup key never reached the login attempt")
+	require.Nil(t, s.oauthAuthFlow.flow, "a setup-key login started an SSO flow")
+
+	status, err := internal.CtxGetState(s.rootCtx).Status()
+	require.NoError(t, err)
+	require.Equal(t, internal.StatusIdle, status)
+}
+
 // breakProfilePrivateKey replaces the profile's private key with an unparseable
 // one, which makes any attempt to build a Management client fail on the spot.
 func breakProfilePrivateKey(t *testing.T, cfgPath string) {

client/server/server.go

@@ -140,6 +140,8 @@ type Server struct {
 	// it to drive the login outcomes that need a server on the other end;
 	// production leaves it nil, and every login goes through loginAttempt.
 	loginAttemptFn func(ctx context.Context, setupKey, jwtToken string) (internal.StatusType, error)
+
+	isLoginRequiredFn func(ctx context.Context) (bool, error)
 }
 
 type oauthAuthFlow struct {
@@ -384,6 +386,21 @@ func (s *Server) attemptLogin(ctx context.Context, setupKey, jwtToken string) (i
 	return s.loginAttempt(ctx, setupKey, jwtToken)
 }
 
+func (s *Server) isLoginRequired(ctx context.Context) (bool, error) {
+	if s.isLoginRequiredFn != nil {
+		return s.isLoginRequiredFn(ctx)
+	}
+
+	authClient, err := auth.NewAuth(ctx, s.config.PrivateKey, s.config.ManagementURL, s.config)
+	if err != nil {
+		log.Errorf("failed to create auth client: %v", err)
+		return false, err
+	}
+	defer authClient.Close()
+
+	return authClient.IsLoginRequired(ctx)
+}
+
 // loginAttempt attempts to login using the provided information. It returns
 // StatusNeedsLogin when Management refused the peer's credentials and
 // StatusLoginFailed for every other failure, so callers can tell an
@@ -640,22 +657,22 @@ func (s *Server) Login(callerCtx context.Context, msg *proto.LoginRequest) (*pro
 	s.config = config
 	s.mutex.Unlock()
 
-	loginStatus, err := s.attemptLogin(ctx, "", "")
-	if err == nil {
-		state.Set(internal.StatusIdle)
-		return &proto.LoginResponse{}, nil
-	}
-
-	// Only an authentication refusal means the peer has to (re-)authenticate.
-	// Any other failure leaves the login undecided: Management unreachable, a
+	// 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
 	// something that is not theirs to solve, and a browser login cannot succeed
-	// while Management is unreachable anyway.
-	if loginStatus != internal.StatusNeedsLogin {
-		state.Set(loginStatus)
+	// while Management is unreachable anyway. Only Management refusing the
+	// peer's key is a decision, and IsLoginRequired reports that as
+	// needsLogin=true rather than an error.
+	needsLogin, err := s.isLoginRequired(ctx)
+	if err != nil {
+		state.Set(internal.StatusLoginFailed)
 		return nil, err
 	}
+	if !needsLogin {
+		state.Set(internal.StatusIdle)
+		return &proto.LoginResponse{}, nil
+	}
 
 	if msg.SetupKey == "" {
 		hint := ""
@@ -1798,6 +1815,9 @@ func (s *Server) RequestExtendAuthSession(
 	if connectClient == nil {
 		return nil, gstatus.Errorf(codes.FailedPrecondition, "client is not running")
 	}
+	if connectClient.Engine() == nil {
+		return nil, gstatus.Errorf(codes.FailedPrecondition, "session can no longer be extended, log in again to reconnect")
+	}
 
 	hint := ""
 	if msg.Hint != nil {

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

@@ -11,7 +11,7 @@ import { DialogHeading } from "@/components/dialog/DialogHeading";
 import { SquareIcon } from "@/components/SquareIcon";
 import { Connection, Profiles as ProfilesSvc, Session, WindowManager } from "@bindings/services";
 import { useAutoSizeWindow } from "@/hooks/useAutoSizeWindow";
-import { EVENT_BROWSER_LOGIN_CANCEL } from "@/lib/connection";
+import { EVENT_BROWSER_LOGIN_CANCEL, EVENT_TRIGGER_LOGIN } from "@/lib/connection";
 import { errorDialog, formatErrorMessage } from "@/lib/errors.ts";
 import { formatRemaining } from "@/lib/formatters";
 
@@ -131,6 +131,21 @@ export default function SessionExpirationDialog() {
         }
     }, [busy, t]);
 
+    const authenticate = useCallback(async () => {
+        if (busy) return;
+        setBusy(true);
+        try {
+            await Events.Emit(EVENT_TRIGGER_LOGIN);
+            await WindowManager.CloseSessionExpiration();
+        } catch (e) {
+            setBusy(false);
+            await errorDialog({
+                Title: t("connect.error.loginTitle"),
+                Message: formatErrorMessage(e),
+            });
+        }
+    }, [busy, t]);
+
     const logout = useCallback(async () => {
         if (busy) return;
         setBusy(true);
@@ -185,7 +200,7 @@ export default function SessionExpirationDialog() {
                     variant={"primary"}
                     size={"md"}
                     className={"w-full"}
-                    onClick={stay}
+                    onClick={expired ? authenticate : stay}
                     disabled={busy}
                 >
                     {expired ? t("sessionExpiration.authenticate") : t("sessionExpiration.stay")}

client/ui/tray_click_linux.go

@@ -4,17 +4,26 @@ package main
 
 // bindTrayClick wires the tray icon's left-click handler on Linux.
 //
-// Both Linux click paths converge on Wails' linuxSystemTray.Activate, which
-// fires the registered clickHandler:
-//   - Real SNI hosts (KDE Plasma, Waybar, GNOME Shell + AppIndicator) invoke
-//     org.kde.StatusNotifierItem.Activate over D-Bus on left-click.
-//   - The in-process StatusNotifierWatcher + XEmbed host used on minimal WMs
-//     (Fluxbox, i3, dwm, OpenBox) maps a Button1 press to that same Activate
-//     call itself (xembed_host_linux.go), so it routes through the same hook.
-// Registering OnClick here therefore covers both paths with one handler — no
-// changes to the watcher or XEmbed C code are needed. Left-click now opens the
-// main window; right-click still opens the menu via Wails' default
-// SecondaryActivate→OpenMenu handler (and the XEmbed GTK popup on minimal WMs).
+// Expected behaviour per tray host:
+//
+//	Host                           Left click              Right click
+//	KDE Plasma, Waybar             main window (Activate)  menu (host-rendered)
+//	GNOME Shell + AppIndicator     menu only               menu only
+//	Minimal WMs via XEmbed host    main window (Activate)  XEmbed GTK popup
+//
+// OnClick fires only on org.kde.StatusNotifierItem.Activate — a real left
+// click. KDE/Waybar send it over D-Bus; the in-process XEmbed host
+// (xembed_host_linux.go) maps a Button1 press to the same Activate call.
+//
+// GNOME Shell + AppIndicator never sends Activate: it renders the dbusmenu
+// on ANY click and only reports the menu opening via dbusmenu
+// Event("opened"). Upstream Wails treated that event as a click, so on GNOME
+// both buttons raised the main window on top of the menu, and on KDE/Waybar
+// a right click raised it over the freshly opened menu. The netbirdio/wails
+// fork (go.mod replace) drops that heuristic: a menu open never fires
+// OnClick. On GNOME the main window is reached via the "Open NetBird" menu
+// entry; left-click-opens-window is not achievable there anyway, since the
+// host always opens the menu itself.
 //
 // We do NOT register OnDoubleClick: Wails' Linux SNI backend never fires it
 // (unlike Windows). And we deliberately skip AttachWindow — it plus Wails3's

client/ui/tray_session.go

@@ -27,11 +27,10 @@ const (
 	finalWarningCountdownSeconds = 120
 )
 
-// handleSessionExpired notifies and brings the window forward so the frontend's /login route drives renewal.
+// handleSessionExpired notifies and brings the window forward so the user can reconnect.
 func (t *Tray) handleSessionExpired() {
 	t.notify(t.loc.T("notify.sessionExpired.title"), t.loc.T("notify.sessionExpired.body"), notifyIDSessionExpired)
 	if t.window != nil {
-		t.window.SetURL("/#/login")
 		t.window.Show()
 		t.window.Focus()
 	}
@@ -308,11 +307,7 @@ func (t *Tray) openSessionExtendFlow() {
 	}
 	seconds := int(time.Until(deadline).Seconds())
 	if seconds <= 0 {
-		if t.window != nil {
-			t.window.SetURL("/#/login")
-			t.window.Show()
-			t.window.Focus()
-		}
+		t.app.Event.Emit(services.EventTriggerLogin)
 		return
 	}
 	if t.svc.WindowManager == nil {

go.mod

@@ -339,3 +339,5 @@ replace github.com/dexidp/dex => github.com/netbirdio/dex v0.244.1-0.20260716205
 replace github.com/dexidp/dex/api/v2 => github.com/netbirdio/dex/api/v2 v2.0.0-20260512110716-8d70ad8647c1
 
 replace github.com/mailru/easyjson => github.com/netbirdio/easyjson v0.9.0
+
+replace github.com/wailsapp/wails/v3 => github.com/netbirdio/wails/v3 v3.0.0-beta.3.0.20260803205919-ad21e92381f4

go.sum

@@ -490,6 +490,8 @@ github.com/netbirdio/service v0.0.0-20240911161631-f62744f42502 h1:3tHlFmhTdX9ax
 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=
 github.com/netbirdio/signal-dispatcher/dispatcher v0.0.0-20250805121659-6b4ac470ca45/go.mod h1:5/sjFmLb8O96B5737VCqhHyGRzNFIaN/Bu7ZodXc3qQ=
+github.com/netbirdio/wails/v3 v3.0.0-beta.3.0.20260803205919-ad21e92381f4 h1:UKztc3QjWvzU5DZk+uYaOWN0x62NSe/pkxuPvzqZIy4=
+github.com/netbirdio/wails/v3 v3.0.0-beta.3.0.20260803205919-ad21e92381f4/go.mod h1:BzATbK71VFikMMMCo434wAi0QcaI03P+xeaWgDvQvjw=
 github.com/netbirdio/wireguard-go v0.0.0-20260628102922-2834bebf6c1a h1:3CWK+yTvRKOcC0Q8VCTGy4l60TEb27CQVS7LkMxwjmw=
 github.com/netbirdio/wireguard-go v0.0.0-20260628102922-2834bebf6c1a/go.mod h1:rpwXGsirqLqN2L0JDJQlwOboGHmptD5ZD6T2VmcqhTw=
 github.com/nxadm/tail v1.4.4/go.mod h1:kenIhsEOeOJmVchQTgglprH7qJGnHDVpk1VPCcaMI8A=
@@ -660,8 +662,6 @@ github.com/vmihailenco/msgpack/v5 v5.4.1 h1:cQriyiUvjTwOHg8QZaPihLWeRAAVoCpE00IU
 github.com/vmihailenco/msgpack/v5 v5.4.1/go.mod h1:GaZTsDaehaPpQVyxrf5mtQlH+pc21PIudVV/E3rRQok=
 github.com/vmihailenco/tagparser/v2 v2.0.0 h1:y09buUbR+b5aycVFQs/g70pqKVZNBmxwAhO7/IwNM9g=
 github.com/vmihailenco/tagparser/v2 v2.0.0/go.mod h1:Wri+At7QHww0WTrCBeu4J6bNtoV6mEfg5OIWRZA9qds=
-github.com/wailsapp/wails/v3 v3.0.0-beta.3 h1:BrcZunEBVucncRx+xgkk9TzlXU4qc0ygJuEhKAAGaeA=
-github.com/wailsapp/wails/v3 v3.0.0-beta.3/go.mod h1:BzATbK71VFikMMMCo434wAi0QcaI03P+xeaWgDvQvjw=
 github.com/wlynxg/anet v0.0.5 h1:J3VJGi1gvo0JwZ/P1/Yc/8p63SoW98B5dHkYDmpgvvU=
 github.com/wlynxg/anet v0.0.5/go.mod h1:eay5PRQr7fIVAMbTbchTnO9gG65Hg/uYGdc7mguHxoA=
 github.com/x448/float16 v0.8.4 h1:qLwI1I70+NjRFUR3zs1JPUCgaCXSh3SW62uAKT1mSBM=

infrastructure_files/getting-started-enterprise.sh

@@ -11,6 +11,13 @@ SED_STRIP_PADDING='s/=//g'
 
 NETBIRD_EULA_URL="https://netbird.io/self-hosted-EULA"
 
+# Static IP for Traefik inside the compose bridge network. The management
+# server trusts X-Forwarded-* headers from this address only, so all three
+# values derive from the same /24. Override with NETBIRD_DOCKER_SUBNET.
+DOCKER_SUBNET="172.30.0.0/24"
+DOCKER_GATEWAY="172.30.0.1"
+TRAEFIK_IP="172.30.0.10"
+
 check_docker_compose() {
   if command -v docker-compose &> /dev/null; then
     echo "docker-compose"
@@ -39,6 +46,90 @@ rand_b64_key() {
   openssl rand -base64 32
 }
 
+# ------------------------------------------------------------------
+# Docker network subnet override and conflict check
+# (kept in sync with getting-started.sh; only the compose network
+# name differs)
+# ------------------------------------------------------------------
+
+ip_to_int() {
+  local a b c d
+  IFS=. read -r a b c d <<< "$1"
+  echo $(( (10#$a << 24) + (10#$b << 16) + (10#$c << 8) + 10#$d ))
+}
+
+# cidrs_overlap <cidr> <cidr> — succeeds if the networks overlap
+cidrs_overlap() {
+  local net1="${1%/*}" len1="${1#*/}" net2="${2%/*}" len2="${2#*/}"
+  local min_len=$(( len1 < len2 ? len1 : len2 ))
+  local mask=0
+  if [[ "$min_len" -gt 0 ]]; then
+    mask=$(( (0xFFFFFFFF << (32 - min_len)) & 0xFFFFFFFF ))
+  fi
+  [[ $(( $(ip_to_int "$net1") & mask )) -eq $(( $(ip_to_int "$net2") & mask )) ]]
+}
+
+# valid_ipv4_slash24 <cidr> — accepts a unicast IPv4 /24 like 10.123.45.0/24
+valid_ipv4_slash24() {
+  local octet='(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|[1-9][0-9]|[0-9])'
+  local re="^${octet}\.${octet}\.${octet}\.0/24$"
+  [[ "$1" =~ $re ]] || return 1
+  # Reject non-unicast/reserved ranges: 0/8, loopback, link-local, 224+
+  case "$1" in
+    0.*|127.*|169.254.*|22[4-9].*|2[34][0-9].*|25[0-5].*) return 1 ;;
+  esac
+  return 0
+}
+
+# Apply NETBIRD_DOCKER_SUBNET and derive the gateway (.1) and Traefik IP (.10)
+apply_docker_subnet_override() {
+  if [[ -n "${NETBIRD_DOCKER_SUBNET:-}" ]]; then
+    if ! valid_ipv4_slash24 "$NETBIRD_DOCKER_SUBNET"; then
+      echo "NETBIRD_DOCKER_SUBNET must be a unicast IPv4 /24 network like 10.123.45.0/24 (0/8, 127/8, 169.254/16, and 224+ are not allowed), got: $NETBIRD_DOCKER_SUBNET" > /dev/stderr
+      exit 1
+    fi
+    DOCKER_SUBNET="$NETBIRD_DOCKER_SUBNET"
+  fi
+  local base="${DOCKER_SUBNET%.0/24}"
+  DOCKER_GATEWAY="${base}.1"
+  TRAEFIK_IP="${base}.10"
+  return 0
+}
+
+# check_docker_subnet_conflicts <compose network name>
+# Fail early if an existing Docker network overlaps DOCKER_SUBNET, instead
+# of letting "docker compose up" fail later. Host routes are not checked;
+# NETBIRD_DOCKER_SUBNET covers those cases.
+check_docker_subnet_conflicts() {
+  local expected_network="$1"
+  command -v docker &> /dev/null || return 0
+
+  local name subnets subnet
+  while IFS='|' read -r name subnets; do
+    for subnet in $subnets; do
+      [[ "$subnet" =~ ^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+/[0-9]+$ ]] || continue
+      if [[ "$name" == "$expected_network" ]]; then
+        # Our own leftover network: compose reuses it as-is, so its subnet
+        # must match the one we render
+        if [[ "$subnet" != "$DOCKER_SUBNET" ]]; then
+          echo "ERROR: the Docker network '$name', left over from a previous NetBird install, uses $subnet instead of $DOCKER_SUBNET." > /dev/stderr
+          echo "docker compose would reuse it as-is, and the generated configuration would not match it." > /dev/stderr
+          echo "Remove it and run this script again:" > /dev/stderr
+          echo "  docker network rm $name" > /dev/stderr
+          exit 1
+        fi
+      elif cidrs_overlap "$DOCKER_SUBNET" "$subnet"; then
+        echo "ERROR: the existing Docker network '$name' ($subnet) overlaps $DOCKER_SUBNET, the subnet NetBird would use." > /dev/stderr
+        echo "That network is not managed by this script and is left untouched." > /dev/stderr
+        echo "Pick a free /24 for NetBird instead and run this script again:" > /dev/stderr
+        echo "  NETBIRD_DOCKER_SUBNET=10.123.45.0/24 ./getting-started-enterprise.sh" > /dev/stderr
+        exit 1
+      fi
+    done
+  done < <(docker network inspect --format '{{.Name}}|{{range .IPAM.Config}}{{.Subnet}} {{end}}' $(docker network ls -q 2>/dev/null) 2>/dev/null || true)
+  return 0
+}
+
 check_nb_domain() {
   local domain="$1"
   if [[ -z "$domain" ]]; then
@@ -80,7 +171,7 @@ read_nb_domain() {
   if ! check_domain_resolves "$value"; then
     echo "" > /dev/stderr
     echo "Warning: '$value' does not resolve via DNS from this host." > /dev/stderr
-    echo "Caddy will not be able to issue TLS certificates until it does." > /dev/stderr
+    echo "Traefik will not be able to issue TLS certificates until it does." > /dev/stderr
     local confirm=""
     echo -n "Continue anyway? [y/N]: " > /dev/stderr
     read -r confirm < /dev/tty
@@ -92,6 +183,23 @@ read_nb_domain() {
   echo "$value"
 }
 
+read_letsencrypt_email() {
+  if [[ -n "${NETBIRD_LETSENCRYPT_EMAIL:-}" ]]; then
+    echo "$NETBIRD_LETSENCRYPT_EMAIL"
+    return
+  fi
+  local value=""
+  echo "Enter your email for Let's Encrypt certificate notifications." > /dev/stderr
+  echo -n "Email address: " > /dev/stderr
+  read -r value < /dev/tty
+  if [[ -z "$value" ]]; then
+    echo "Email is required for Let's Encrypt." > /dev/stderr
+    read_letsencrypt_email
+    return
+  fi
+  echo "$value"
+}
+
 read_required() {
   local prompt="$1"
   local value=""
@@ -203,12 +311,15 @@ wait_postgres() {
 init_environment() {
   check_openssl
   DOCKER_COMPOSE_COMMAND=$(check_docker_compose)
+  # Settle the subnet (and fail on conflicts) before the EULA and prompts
+  apply_docker_subnet_override
+  check_docker_subnet_conflicts "netbird"
 
-  if [[ -f .env ]] || [[ -f docker-compose.yml ]] || [[ -f config.yaml ]] || [[ -f Caddyfile ]]; then
+  if [[ -f .env ]] || [[ -f docker-compose.yml ]] || [[ -f config.yaml ]]; then
     echo "Generated files already exist in $(pwd)."
     echo "If you want to reinitialize the environment, please remove them first:"
     echo "  $DOCKER_COMPOSE_COMMAND down --volumes # removes all containers and volumes"
-    echo "  rm -f .env docker-compose.yml Caddyfile config.yaml"
+    echo "  rm -f .env docker-compose.yml config.yaml"
     echo "Be aware this will remove all data from the database."
     exit 1
   fi
@@ -230,6 +341,9 @@ init_environment() {
   echo ""
   NETBIRD_DOMAIN=$(read_nb_domain)
 
+  echo ""
+  NETBIRD_LETSENCRYPT_EMAIL=$(read_letsencrypt_email)
+
   echo ""
 
   NETBIRD_LICENSE_KEY=$(read_secret "Enter license key (input hidden)")
@@ -247,6 +361,8 @@ init_environment() {
   echo "Selected:"
   echo "  Traffic flow: ${NETBIRD_TRAFFIC_FLOW}"
   echo "  Domain:       ${NETBIRD_DOMAIN}"
+  echo "  ACME email:   ${NETBIRD_LETSENCRYPT_EMAIL}"
+  echo "  Subnet:       ${DOCKER_SUBNET} (Traefik at ${TRAEFIK_IP})"
   echo ""
   echo "Rendering files into $(pwd) ..."
   install -m 600 /dev/null .env
@@ -256,7 +372,6 @@ init_environment() {
   if [[ -z "${NETBIRD_LICENSE_SERVER_BASE_URL:-}" ]]; then
     sed -i.bak '/NETBIRD_LICENSE_SERVER_BASE_URL/d' docker-compose.yml && rm -f docker-compose.yml.bak
   fi
-  render_caddyfile > Caddyfile
   install -m 600 /dev/null config.yaml
   render_config_yaml >> config.yaml
 
@@ -283,7 +398,7 @@ init_environment() {
   echo "All configuration and secrets are stored (mode 600) in $(pwd)/.env"
   echo ""
   echo "Tail logs:"
-  echo "  cd $(pwd) && $DOCKER_COMPOSE_COMMAND logs -f netbird-server caddy"
+  echo "  cd $(pwd) && $DOCKER_COMPOSE_COMMAND logs -f netbird-server traefik"
 }
 
 # ------------------------------------------------------------------
@@ -306,6 +421,16 @@ NETBIRD_TRAFFIC_FLOW_ENABLED=${NETBIRD_TRAFFIC_FLOW}
 # Domain
 NETBIRD_DOMAIN=${NETBIRD_DOMAIN}
 
+# Reverse proxy (Traefik)
+NETBIRD_LETSENCRYPT_EMAIL=${NETBIRD_LETSENCRYPT_EMAIL}
+NETBIRD_TRAEFIK_TAG=${NETBIRD_TRAEFIK_TAG:-v3.6}
+# These three must stay in step with the /32 trust pins in config.yaml
+# (reverseProxy.trustedPeers/trustedHTTPProxies). Shell env vars override
+# this file at compose time.
+NETBIRD_TRAEFIK_IP=${TRAEFIK_IP}
+NETBIRD_NETWORK_SUBNET=${DOCKER_SUBNET}
+NETBIRD_NETWORK_GATEWAY=${DOCKER_GATEWAY}
+
 # Image tags. Default to "latest"
 NETBIRD_DASHBOARD_TAG=${NETBIRD_DASHBOARD_TAG:-latest}
 NETBIRD_SERVER_TAG=${NETBIRD_SERVER_TAG:-latest}
@@ -378,26 +503,81 @@ EOF
 
 render_compose_common() {
   cat <<'EOF'
-  caddy:
+  # Reverse proxy with automatic TLS via Let's Encrypt. Routes are declared as
+  # labels on the services below and picked up through the Docker provider.
+  traefik:
     <<: *default
-    image: caddy:2
-    container_name: netbird-caddy
-    networks: [netbird]
-    environment:
-      - CADDY_SECURE_DOMAIN=${NETBIRD_DOMAIN}
+    image: traefik:${NETBIRD_TRAEFIK_TAG}
+    container_name: netbird-traefik
+    networks:
+      netbird:
+        ipv4_address: ${NETBIRD_TRAEFIK_IP}
+        # Resolve the public domain inside this network (avoids hairpin NAT)
+        aliases:
+          - "${NETBIRD_DOMAIN}"
+    command:
+      # Logging
+      - "--log.level=INFO"
+      - "--accesslog=true"
+      # Docker provider
+      - "--providers.docker=true"
+      - "--providers.docker.exposedbydefault=false"
+      - "--providers.docker.network=netbird"
+      # Entrypoints
+      - "--entrypoints.web.address=:80"
+      - "--entrypoints.websecure.address=:443"
+      - "--entrypoints.websecure.allowACMEByPass=true"
+      # readTimeout bounds the whole request, and gRPC streams / relay WebSockets
+      # never end one; idleTimeout would close the keep-alive connection they
+      # are reused over. Entrypoint-wide is the only scope Traefik offers here.
+      # writeTimeout is left alone: it already defaults to 0.
+      - "--entrypoints.websecure.transport.respondingTimeouts.readTimeout=0"
+      - "--entrypoints.websecure.transport.respondingTimeouts.idleTimeout=0"
+      # HTTP to HTTPS redirect
+      - "--entrypoints.web.http.redirections.entrypoint.to=websecure"
+      - "--entrypoints.web.http.redirections.entrypoint.scheme=https"
+      # Let's Encrypt ACME
+      - "--certificatesresolvers.letsencrypt.acme.email=${NETBIRD_LETSENCRYPT_EMAIL}"
+      - "--certificatesresolvers.letsencrypt.acme.storage=/letsencrypt/acme.json"
+      - "--certificatesresolvers.letsencrypt.acme.tlschallenge=true"
     ports:
       - '443:443'
-      - '443:443/udp'
       - '80:80'
     volumes:
-      - netbird_caddy_data:/data
-      - ./Caddyfile:/etc/caddy/Caddyfile
+      - /var/run/docker.sock:/var/run/docker.sock:ro
+      - netbird_traefik_letsencrypt:/letsencrypt
+    labels:
+      - traefik.enable=true
+      # Shared security headers, referenced by every NetBird router below. A
+      # label-declared middleware only exists while its container runs, so this
+      # lives on Traefik itself: declaring it on an app container would drop
+      # every router referencing it whenever that container restarts.
+      - traefik.http.middlewares.nb-security.headers.stsSeconds=3600
+      - traefik.http.middlewares.nb-security.headers.stsIncludeSubdomains=true
+      - traefik.http.middlewares.nb-security.headers.contentTypeNosniff=true
+      - traefik.http.middlewares.nb-security.headers.browserXssFilter=true
+      - traefik.http.middlewares.nb-security.headers.referrerPolicy=strict-origin-when-cross-origin
+      - traefik.http.middlewares.nb-security.headers.customResponseHeaders.X-Frame-Options=SAMEORIGIN
+      # Empty value strips the header. Only the dashboard's nginx sets one; the
+      # server emits none. Do not quote it — "" would send a literal Server: "".
+      - traefik.http.middlewares.nb-security.headers.customResponseHeaders.Server=
 
   dashboard:
     <<: *default
     image: ghcr.io/netbirdio/dashboard-cloud:${NETBIRD_DASHBOARD_TAG}
     container_name: netbird-dashboard
     networks: [netbird]
+    labels:
+      - traefik.enable=true
+      # Dashboard catch-all: lowest priority so every route below wins
+      - traefik.http.routers.netbird-dashboard.rule=Host(`${NETBIRD_DOMAIN}`)
+      - traefik.http.routers.netbird-dashboard.entrypoints=websecure
+      - traefik.http.routers.netbird-dashboard.tls=true
+      - traefik.http.routers.netbird-dashboard.tls.certresolver=letsencrypt
+      - traefik.http.routers.netbird-dashboard.middlewares=nb-security@docker
+      - traefik.http.routers.netbird-dashboard.service=dashboard
+      - traefik.http.routers.netbird-dashboard.priority=1
+      - traefik.http.services.dashboard.loadbalancer.server.port=80
     environment:
       - NETBIRD_MGMT_API_ENDPOINT=https://${NETBIRD_DOMAIN}
       - NETBIRD_MGMT_GRPC_API_ENDPOINT=https://${NETBIRD_DOMAIN}
@@ -435,6 +615,28 @@ render_compose_server() {
       - netbird_data:/var/lib/netbird
       - ./config.yaml:/etc/netbird/config.yaml
     command: ["--config", "/etc/netbird/config.yaml"]
+    labels:
+      - traefik.enable=true
+      # Signal + Management gRPC (needs an h2c backend for HTTP/2 cleartext)
+      - traefik.http.routers.netbird-grpc.rule=Host(`${NETBIRD_DOMAIN}`) && (PathPrefix(`/signalexchange.SignalExchange/`) || PathPrefix(`/management.ManagementService/`) || PathPrefix(`/management.ProxyService/`))
+      - traefik.http.routers.netbird-grpc.entrypoints=websecure
+      - traefik.http.routers.netbird-grpc.tls=true
+      - traefik.http.routers.netbird-grpc.tls.certresolver=letsencrypt
+      - traefik.http.routers.netbird-grpc.middlewares=nb-security@docker
+      - traefik.http.routers.netbird-grpc.service=netbird-server-h2c
+      - traefik.http.routers.netbird-grpc.priority=100
+      # Relay WebSocket, management API, and the embedded IdP
+      - traefik.http.routers.netbird-backend.rule=Host(`${NETBIRD_DOMAIN}`) && (PathPrefix(`/relay`) || PathPrefix(`/ws-proxy/`) || PathPrefix(`/api`) || PathPrefix(`/oauth2`))
+      - traefik.http.routers.netbird-backend.entrypoints=websecure
+      - traefik.http.routers.netbird-backend.tls=true
+      - traefik.http.routers.netbird-backend.tls.certresolver=letsencrypt
+      - traefik.http.routers.netbird-backend.middlewares=nb-security@docker
+      - traefik.http.routers.netbird-backend.service=netbird-server
+      - traefik.http.routers.netbird-backend.priority=100
+      # Services
+      - traefik.http.services.netbird-server.loadbalancer.server.port=80
+      - traefik.http.services.netbird-server-h2c.loadbalancer.server.port=80
+      - traefik.http.services.netbird-server-h2c.loadbalancer.server.scheme=h2c
     environment:
       - NB_LICENSE_KEY=${NETBIRD_LICENSE_KEY}
       - NETBIRD_LICENSE_SERVER_BASE_URL=${NETBIRD_LICENSE_SERVER_BASE_URL}
@@ -497,6 +699,18 @@ render_compose_flow() {
       - NB_FLOW_NATS_ENDPOINTS=nats://nats:4222
       - NB_FLOW_NATS_STREAM=traffic-events
       - NB_FLOW_AUTH_SECRET=${NETBIRD_RELAY_AUTH_SECRET}
+    labels:
+      - traefik.enable=true
+      # Flow receiver gRPC (h2c backend)
+      - traefik.http.routers.netbird-flow.rule=Host(`${NETBIRD_DOMAIN}`) && PathPrefix(`/flow.FlowService/`)
+      - traefik.http.routers.netbird-flow.entrypoints=websecure
+      - traefik.http.routers.netbird-flow.tls=true
+      - traefik.http.routers.netbird-flow.tls.certresolver=letsencrypt
+      - traefik.http.routers.netbird-flow.middlewares=nb-security@docker
+      - traefik.http.routers.netbird-flow.service=netbird-flow-h2c
+      - traefik.http.routers.netbird-flow.priority=100
+      - traefik.http.services.netbird-flow-h2c.loadbalancer.server.port=80
+      - traefik.http.services.netbird-flow-h2c.loadbalancer.server.scheme=h2c
 
 EOF
 }
@@ -536,61 +750,16 @@ EOF
   fi
   cat <<'EOF'
   netbird_postgres:
-  netbird_caddy_data:
+  netbird_traefik_letsencrypt:
 
 networks:
   netbird:
-EOF
-}
-
-render_caddyfile() {
-  cat <<'EOF'
-{
-  servers :80,:443 {
-    protocols h1 h2c h2 h3
-  }
-}
-
-(security_headers) {
-    header * {
-        Strict-Transport-Security "max-age=3600; includeSubDomains; preload"
-        X-Content-Type-Options "nosniff"
-        X-Frame-Options "SAMEORIGIN"
-        X-XSS-Protection "1; mode=block"
-        -Server
-        Referrer-Policy strict-origin-when-cross-origin
-    }
-}
-
-:80 {
-    redir https://{$CADDY_SECURE_DOMAIN}{uri} permanent
-}
-
-{$CADDY_SECURE_DOMAIN}:443 {
-    import security_headers
-    # Signal (gRPC over h2c)
-    reverse_proxy /signalexchange.SignalExchange/* h2c://netbird-server:80
-    # Management (gRPC over h2c + HTTP)
-    reverse_proxy /management.ManagementService/* h2c://netbird-server:80
-    reverse_proxy /api/* netbird-server:80
-    reverse_proxy /ws-proxy/* netbird-server:80
-    # Embedded IdP (OAuth2 endpoints served by netbird server)
-    reverse_proxy /oauth2/* netbird-server:80
-    # Relay (WebSocket multiplexed on the same port)
-    reverse_proxy /relay* netbird-server:80
-EOF
-
-  if [[ "$NETBIRD_TRAFFIC_FLOW" == "yes" ]]; then
-    cat <<'EOF'
-    # Flow receiver (gRPC over h2c)
-    reverse_proxy /flow.FlowService/* h2c://receiver:80
-EOF
-  fi
-
-  cat <<'EOF'
-    # Dashboard
-    reverse_proxy /* dashboard:80
-}
+    name: netbird
+    driver: bridge
+    ipam:
+      config:
+        - subnet: ${NETBIRD_NETWORK_SUBNET}
+          gateway: ${NETBIRD_NETWORK_GATEWAY}
 EOF
 }
 
@@ -609,7 +778,7 @@ server:
   logLevel: "info"
   logFile: "console"
 
-  # TLS is terminated by Caddy in front; leave this block empty.
+  # TLS is terminated by Traefik in front; leave this block empty.
   tls:
     certFile: ""
     keyFile: ""
@@ -632,6 +801,16 @@ server:
     cliRedirectURIs:
       - "http://localhost:53000/"
 
+  # Trust X-Forwarded-* only from the Traefik container's static address. Both
+  # keys must stay in step with the ipv4_address pinned in docker-compose.yml:
+  # trustedPeers decides whether forwarded headers are read at all, and leaving
+  # it unset falls back to 0.0.0.0/0.
+  reverseProxy:
+    trustedPeers:
+      - "${TRAEFIK_IP}/32"
+    trustedHTTPProxies:
+      - "${TRAEFIK_IP}/32"
+
   store:
     engine: "postgres"
     dsn: "${POSTGRES_DSN}"

infrastructure_files/getting-started.sh

@@ -108,6 +108,14 @@ check_nb_domain() {
     echo "The NETBIRD_DOMAIN cannot be netbird.example.com" > /dev/stderr
     return 1
   fi
+
+  # Letters, digits, dots, and hyphens only; the domain is embedded in
+  # generated YAML and env files
+  local re='^[A-Za-z0-9]([A-Za-z0-9.-]*[A-Za-z0-9])?$'
+  if [[ "$DOMAIN" != "use-ip" ]] && [[ ! "$DOMAIN" =~ $re ]]; then
+    echo "The NETBIRD_DOMAIN may only contain letters, digits, dots, and hyphens." > /dev/stderr
+    return 1
+  fi
   return 0
 }
 
@@ -337,6 +345,103 @@ wait_management_direct() {
   return 0
 }
 
+############################################
+# Docker Network Subnet Override and Conflict Check
+############################################
+
+ip_to_int() {
+  local a b c d
+  IFS=. read -r a b c d <<< "$1"
+  echo $(( (10#$a << 24) + (10#$b << 16) + (10#$c << 8) + 10#$d ))
+}
+
+# cidrs_overlap <cidr> <cidr> — succeeds if the networks overlap
+cidrs_overlap() {
+  local net1="${1%/*}" len1="${1#*/}" net2="${2%/*}" len2="${2#*/}"
+  local min_len=$(( len1 < len2 ? len1 : len2 ))
+  local mask=0
+  if [[ "$min_len" -gt 0 ]]; then
+    mask=$(( (0xFFFFFFFF << (32 - min_len)) & 0xFFFFFFFF ))
+  fi
+  [[ $(( $(ip_to_int "$net1") & mask )) -eq $(( $(ip_to_int "$net2") & mask )) ]]
+}
+
+# valid_ipv4_slash24 <cidr> — accepts a unicast IPv4 /24 like 10.123.45.0/24
+valid_ipv4_slash24() {
+  local octet='(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|[1-9][0-9]|[0-9])'
+  local re="^${octet}\.${octet}\.${octet}\.0/24$"
+  [[ "$1" =~ $re ]] || return 1
+  # Reject non-unicast/reserved ranges: 0/8, loopback, link-local, 224+
+  case "$1" in
+    0.*|127.*|169.254.*|22[4-9].*|2[34][0-9].*|25[0-5].*) return 1 ;;
+  esac
+  return 0
+}
+
+# Apply NETBIRD_DOCKER_SUBNET and derive the gateway (.1) and Traefik IP (.10)
+apply_docker_subnet_override() {
+  if [[ -n "${NETBIRD_DOCKER_SUBNET:-}" ]]; then
+    if ! valid_ipv4_slash24 "$NETBIRD_DOCKER_SUBNET"; then
+      echo "NETBIRD_DOCKER_SUBNET must be a unicast IPv4 /24 network like 10.123.45.0/24 (0/8, 127/8, 169.254/16, and 224+ are not allowed), got: $NETBIRD_DOCKER_SUBNET" > /dev/stderr
+      exit 1
+    fi
+    DOCKER_SUBNET="$NETBIRD_DOCKER_SUBNET"
+  fi
+  local base="${DOCKER_SUBNET%.0/24}"
+  DOCKER_GATEWAY="${base}.1"
+  TRAEFIK_IP="${base}.10"
+  return 0
+}
+
+# check_docker_subnet_conflicts <compose network name>
+# Fail early if an existing Docker network overlaps DOCKER_SUBNET, instead
+# of letting "docker compose up" fail later. Host routes are not checked;
+# NETBIRD_DOCKER_SUBNET covers those cases.
+check_docker_subnet_conflicts() {
+  local expected_network="$1"
+  command -v docker &> /dev/null || return 0
+
+  local name subnets subnet
+  while IFS='|' read -r name subnets; do
+    for subnet in $subnets; do
+      [[ "$subnet" =~ ^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+/[0-9]+$ ]] || continue
+      if [[ "$name" == "$expected_network" ]]; then
+        # Our own leftover network: compose reuses it as-is, so its subnet
+        # must match the one we render
+        if [[ "$subnet" != "$DOCKER_SUBNET" ]]; then
+          echo "ERROR: the Docker network '$name', left over from a previous NetBird install, uses $subnet instead of $DOCKER_SUBNET." > /dev/stderr
+          echo "docker compose would reuse it as-is, and the generated configuration would not match it." > /dev/stderr
+          echo "Remove it and run this script again:" > /dev/stderr
+          echo "  docker network rm $name" > /dev/stderr
+          exit 1
+        fi
+      elif cidrs_overlap "$DOCKER_SUBNET" "$subnet"; then
+        echo "ERROR: the existing Docker network '$name' ($subnet) overlaps $DOCKER_SUBNET, the subnet NetBird would use." > /dev/stderr
+        echo "That network is not managed by this script and is left untouched." > /dev/stderr
+        echo "Pick a free /24 for NetBird instead and run this script again:" > /dev/stderr
+        echo "  NETBIRD_DOCKER_SUBNET=10.123.45.0/24 ./getting-started.sh" > /dev/stderr
+        exit 1
+      fi
+    done
+  done < <(docker network inspect --format '{{.Name}}|{{range .IPAM.Config}}{{.Subnet}} {{end}}' $(docker network ls -q 2>/dev/null) 2>/dev/null || true)
+  return 0
+}
+
+configure_docker_subnet() {
+  # Only the built-in Traefik mode pins a subnet; other modes let Docker pick
+  if [[ "$REVERSE_PROXY_TYPE" != "0" ]]; then
+    return 0
+  fi
+
+  # Skip our own network (<project>_netbird) in the conflict check;
+  # normalization matches compose-go's NormalizeProjectName
+  local project
+  project="${COMPOSE_PROJECT_NAME:-$(basename "$PWD")}"
+  project=$(echo "$project" | tr '[:upper:]' '[:lower:]' | sed 's/[^a-z0-9_-]//g; s/^[_-]*//')
+  check_docker_subnet_conflicts "${project}_netbird"
+  return 0
+}
+
 ############################################
 # Initialization and Configuration
 ############################################
@@ -368,7 +473,11 @@ initialize_default_values() {
   BIND_LOCALHOST_ONLY="true"
   EXTERNAL_PROXY_NETWORK=""
 
-  # Traefik static IP within the internal bridge network
+  # Internal bridge network. Management and proxy trust forwarded headers
+  # from TRAEFIK_IP only, so all three values derive from the same /24.
+  # Override with NETBIRD_DOCKER_SUBNET.
+  DOCKER_SUBNET="172.30.0.0/24"
+  DOCKER_GATEWAY="172.30.0.1"
   TRAEFIK_IP="172.30.0.10"
 
   # NetBird Proxy configuration
@@ -663,8 +772,10 @@ init_environment() {
   check_docker_sock_perms
 
   initialize_default_values
+  apply_docker_subnet_override
   configure_domain
   configure_reverse_proxy
+  configure_docker_subnet
 
   check_jq
 
@@ -686,6 +797,7 @@ render_docker_compose_traefik_builtin() {
   local crowdsec_volumes=""
   local traefik_file_provider=""
   local traefik_dynamic_volume=""
+
   if [[ "$ENABLE_PROXY" == "true" ]]; then
     traefik_file_provider='      - "--providers.file.filename=/etc/traefik/dynamic.yaml"'
     traefik_dynamic_volume="      - ./traefik-dynamic.yaml:/etc/traefik/dynamic.yaml:ro"
@@ -884,8 +996,8 @@ networks:
     driver: bridge
     ipam:
       config:
-        - subnet: 172.30.0.0/24
-          gateway: 172.30.0.1
+        - subnet: $DOCKER_SUBNET
+          gateway: $DOCKER_GATEWAY
 EOF
   return 0
 }

infrastructure_files/migrate-to-enterprise.sh

@@ -15,7 +15,11 @@ set -o pipefail
 #   2. Postgres migration — add Postgres, migrate SQLite data via migrate-store.
 #   3. Traffic flow       — add NATS + flow-enricher + flow-receiver.
 #
-# To revert:
+# If any step fails once the stack has been touched, the script rolls itself
+# back automatically: generated files are removed, the Postgres volume this run
+# created is dropped, and the original deployment is started again.
+#
+# To revert a successful migration:
 #   docker compose down
 #   rm -f docker-compose.override.yml config.yaml.enterprise
 #   # If Postgres migration was done, also restore the SQLite backup printed
@@ -25,6 +29,15 @@ set -o pipefail
 OVERRIDE_FILE="docker-compose.override.yml"
 ENTERPRISE_CONFIG_FILE="config.yaml.enterprise"
 
+# Rollback bookkeeping. ROLLBACK_STATE flips to "armed" the moment the script
+# starts mutating the deployment, and back to "disarmed" once the migration has
+# completed successfully.
+ROLLBACK_STATE="disarmed"
+ENV_EXISTED="unknown"
+ENV_BACKUP=""
+PG_VOLUME_NAME=""
+BACKUP_DIR=""
+
 NETBIRD_EULA_URL="https://netbird.io/self-hosted-EULA"
 
 check_docker_compose() {
@@ -361,7 +374,77 @@ render_enterprise_config() {
 # Execution steps
 # ---------------------------------------------------------------------------
 
-resolve_data_volume() {
+combined_container_id() {
+  $DOCKER_COMPOSE_COMMAND ps -aq "$COMBINED_SERVICE" 2>/dev/null | head -1
+}
+
+container_data_mount() {
+  local container="$1"
+  [[ -n "$container" ]] || return 0
+  docker inspect "$container" --format \
+    '{{range .Mounts}}{{if eq .Destination "/var/lib/netbird"}}{{if .Name}}{{.Name}}{{else}}{{.Source}}{{end}}{{end}}{{end}}' 2>/dev/null
+}
+
+# The name comes from the container, so `-v` cannot invent an empty volume here.
+# 0 = empty, 1 = holds data, 2 = could not determine. A failed listing must not
+# be reported as empty: that would abort a healthy migration over a pull error
+# or an unreadable bind mount.
+data_dir_state() {
+  local src="$1" out
+  if [[ "$src" == /* ]]; then
+    [[ -d "$src" ]] || return 2
+    out=$(ls -A "$src" 2>/dev/null) || return 2
+  else
+    docker volume inspect "$src" &> /dev/null || return 0
+    out=$(docker run --rm -v "${src}:/d:ro" busybox sh -c 'ls -A /d' 2>/dev/null) || return 2
+  fi
+  [[ -z "$out" ]] && return 0
+  return 1
+}
+
+check_data_directory() {
+  [[ "$MIGRATE_POSTGRES" == "yes" ]] || return 0
+
+  local container
+  container=$(combined_container_id)
+  if [[ -z "$container" ]]; then
+    echo "" > /dev/stderr
+    echo "No container found for service '$COMBINED_SERVICE'." > /dev/stderr
+    echo "The migration backs up the store by copying it out of that container," > /dev/stderr
+    echo "so it has to exist. Start the deployment and re-run:" > /dev/stderr
+    echo "  $DOCKER_COMPOSE_COMMAND up -d" > /dev/stderr
+    exit 1
+  fi
+
+  local src
+  src=$(container_data_mount "$container")
+  if [[ -z "$src" ]]; then
+    echo "" > /dev/stderr
+    echo "The '$COMBINED_SERVICE' container has nothing mounted at /var/lib/netbird." > /dev/stderr
+    echo "Cannot locate the NetBird store to back it up." > /dev/stderr
+    exit 1
+  fi
+
+  local state=0
+  data_dir_state "$src" || state=$?
+  if [[ $state -eq 0 ]]; then
+    echo "" > /dev/stderr
+    echo "The NetBird data directory is empty:" > /dev/stderr
+    echo "  $src" > /dev/stderr
+    echo "There is nothing to migrate. Check that you are running this from the" > /dev/stderr
+    echo "deployment directory of the NetBird install you mean to migrate." > /dev/stderr
+    exit 1
+  fi
+  if [[ $state -eq 2 ]]; then
+    echo "  ⚠ Could not read $src to confirm it holds data — continuing." > /dev/stderr
+    echo "    The backup step still fails loudly if it turns out to be empty." > /dev/stderr
+  fi
+
+  echo "  Data directory:   $src"
+}
+
+# Only for the Postgres volume, which has no container to read it off yet.
+resolve_compose_volume() {
   local short="$1"
   local actual
   # Resolve project-prefixed volume name from Docker Compose config first.
@@ -391,18 +474,21 @@ resolve_data_volume() {
 backup_sqlite() {
   BACKUP_DIR="$(pwd)/backups/sqlite-pre-enterprise-$(date +%Y%m%d-%H%M%S)"
   mkdir -p "$BACKUP_DIR"
-  local data_volume_actual
-  data_volume_actual=$(resolve_data_volume "$DATA_VOLUME")
-  echo "Backing up SQLite store from volume '$data_volume_actual' to $BACKUP_DIR ..."
-  docker run --rm \
-    -v "${data_volume_actual}:/var/lib/netbird:ro" \
-    -v "${BACKUP_DIR}:/backup" \
-    busybox \
-    sh -c 'cp -a /var/lib/netbird/. /backup/ 2>/dev/null || true'
+
+  local container
+  container=$(combined_container_id)
+  if [[ -z "$container" ]]; then
+    echo "  ⚠ No container found for '$COMBINED_SERVICE' — cannot back up the store." > /dev/stderr
+    exit 1
+  fi
+
+  echo "Backing up the NetBird store to $BACKUP_DIR ..."
+  docker cp "${container}:/var/lib/netbird/." "$BACKUP_DIR/"
+
   local copied
   copied=$(find "$BACKUP_DIR" -mindepth 1 | head -1)
   if [[ -z "$copied" ]]; then
-    echo "  ⚠ Backup directory is empty — the volume '$data_volume_actual' didn't contain data. Aborting." > /dev/stderr
+    echo "  ⚠ Backup directory is empty — /var/lib/netbird held no data. Aborting." > /dev/stderr
     exit 1
   fi
   echo "  done"
@@ -414,6 +500,135 @@ run_migrate_store() {
   echo "  done"
 }
 
+# ---------------------------------------------------------------------------
+# Rollback — a failed run must not leave the operator with a stopped stack and
+# half-written artifacts.
+# ---------------------------------------------------------------------------
+
+# Resolve the name Compose would give the Postgres volume before the override
+# exists, so a leftover volume can be spotted up front.
+compose_project_name() {
+  local container project
+  container=$($DOCKER_COMPOSE_COMMAND ps -aq 2>/dev/null | head -1)
+  if [[ -n "$container" ]]; then
+    project=$(docker inspect "$container" \
+      --format '{{index .Config.Labels "com.docker.compose.project"}}' 2>/dev/null)
+    if [[ -n "$project" ]]; then
+      echo "$project"
+      return 0
+    fi
+  fi
+  project=$($DOCKER_COMPOSE_COMMAND config 2>/dev/null | yq eval '.name // ""' - 2>/dev/null)
+  if [[ -n "$project" ]] && [[ "$project" != "null" ]]; then
+    echo "$project"
+  fi
+  return 0
+}
+
+postgres_volume_name() {
+  local project
+  project=$(compose_project_name)
+  if [[ -n "$project" ]]; then
+    echo "${project}_netbird_postgres"
+  fi
+  return 0
+}
+
+# Postgres skips initdb when its data directory is non-empty, so a volume left
+# behind by an interrupted run would keep the old password and old contents,
+# and migrate-store would fail against it.
+check_stale_postgres_volume() {
+  [[ "$MIGRATE_POSTGRES" == "yes" ]] || return 0
+
+  PG_VOLUME_NAME=$(postgres_volume_name)
+  if [[ -z "$PG_VOLUME_NAME" ]]; then
+    echo ""
+    echo "  ⚠ Could not determine the Compose project name, so a Postgres volume"
+    echo "    left over from an earlier attempt cannot be checked for. If a"
+    echo "    previous run failed, remove it before continuing:"
+    echo "      docker volume ls | grep netbird_postgres"
+    return 0
+  fi
+  docker volume inspect "$PG_VOLUME_NAME" &> /dev/null || return 0
+
+  echo ""
+  echo "  ⚠  A Postgres volume from an earlier attempt already exists:"
+  echo "       $PG_VOLUME_NAME"
+  echo "     Postgres does not re-initialise a non-empty data directory, so the"
+  echo "     migration would run against stale credentials and stale data."
+  local remove
+  remove=$(read_yes_no "  Remove it and continue?" "y")
+  if [[ "$remove" != "yes" ]]; then
+    echo "" > /dev/stderr
+    echo "Aborted. Remove it manually with: docker volume rm $PG_VOLUME_NAME" > /dev/stderr
+    exit 1
+  fi
+  docker volume rm "$PG_VOLUME_NAME" > /dev/null
+  echo "  Removed."
+}
+
+# Undo whatever this run changed and start the previous deployment again.
+rollback() {
+  ROLLBACK_STATE="done"
+
+  echo ""
+  echo "──────────────────────────────────────────────────────────────────────"
+  echo " Migration failed — restoring the previous deployment"
+  echo "──────────────────────────────────────────────────────────────────────"
+
+  # Resolve while the override is still present; without it Compose no longer
+  # knows about the Postgres volume.
+  local pg_volume="$PG_VOLUME_NAME"
+  if [[ -z "$pg_volume" ]] && [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
+    pg_volume=$(postgres_volume_name)
+  fi
+
+  echo ""
+  echo "Stopping services ..."
+  $DOCKER_COMPOSE_COMMAND down || true
+
+  echo "Removing generated files ..."
+  rm -f "$OVERRIDE_FILE" "$ENTERPRISE_CONFIG_FILE"
+
+  # Restore .env to exactly what it was, or remove it if this run created it.
+  if [[ "$ENV_EXISTED" == "yes" ]] && [[ -f "$ENV_BACKUP" ]]; then
+    mv -f "$ENV_BACKUP" .env || echo "  ⚠ Could not restore .env from $ENV_BACKUP." > /dev/stderr
+  elif [[ "$ENV_EXISTED" == "no" ]]; then
+    rm -f .env || true
+  fi
+
+  # Only ever the volume this run created — never the NetBird data volume.
+  if [[ -n "$pg_volume" ]] && [[ "$pg_volume" != "null" ]]; then
+    echo "Removing Postgres volume $pg_volume ..."
+    docker volume rm "$pg_volume" &> /dev/null || true
+  fi
+
+  echo "Starting the previous deployment ..."
+  if ! $DOCKER_COMPOSE_COMMAND up -d; then
+    echo ""
+    echo "  ⚠ Could not start the previous deployment automatically." > /dev/stderr
+    echo "    Run: $DOCKER_COMPOSE_COMMAND up -d" > /dev/stderr
+  fi
+
+  echo ""
+  echo "Rolled back. Your docker-compose.yml, config.yaml and the NetBird data"
+  echo "volume were never modified."
+  if [[ -n "$BACKUP_DIR" ]] && [[ -d "$BACKUP_DIR" ]]; then
+    echo "The SQLite backup taken during this run is kept at:"
+    echo "  $BACKUP_DIR"
+  fi
+  echo "──────────────────────────────────────────────────────────────────────"
+}
+
+on_exit() {
+  local code=$?
+  trap - EXIT
+  if [[ $code -ne 0 ]] && [[ "$ROLLBACK_STATE" == "armed" ]]; then
+    rollback
+  fi
+  exit $code
+}
+
 # ---------------------------------------------------------------------------
 # Main
 # ---------------------------------------------------------------------------
@@ -541,9 +756,15 @@ init_migration() {
     ENABLE_FLOW="no"
     echo "Step 3 (traffic flow) skipped — requires Postgres."
   fi
+
+  check_data_directory
+  check_stale_postgres_volume
 }
 
 apply_changes() {
+  # From here on a failure must roll the deployment back.
+  ROLLBACK_STATE="armed"
+
   echo ""
   echo "Writing $OVERRIDE_FILE ..."
   install -m 644 /dev/null "$OVERRIDE_FILE"
@@ -564,6 +785,14 @@ apply_changes() {
   # picks it up automatically.
   echo "Writing .env additions (mode 600) ..."
   local ENV_FILE=".env"
+  # Snapshot the operator's .env so a rollback can restore it byte for byte.
+  if [[ -f "$ENV_FILE" ]]; then
+    ENV_EXISTED="yes"
+    ENV_BACKUP="${ENV_FILE}.pre-enterprise-$(date +%Y%m%d-%H%M%S)"
+    cp -p "$ENV_FILE" "$ENV_BACKUP"
+  else
+    ENV_EXISTED="no"
+  fi
   touch "$ENV_FILE"
   chmod 600 "$ENV_FILE"
   {
@@ -592,11 +821,16 @@ apply_changes() {
 
   if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
     echo ""
-    echo "Stopping existing services (volumes preserved) ..."
-    $DOCKER_COMPOSE_COMMAND down
+    # Stop, but keep the containers: the backup reads the store out of one.
+    echo "Stopping services so the store is quiescent ..."
+    $DOCKER_COMPOSE_COMMAND stop
 
     backup_sqlite
 
+    echo ""
+    echo "Removing stopped containers (volumes preserved) ..."
+    $DOCKER_COMPOSE_COMMAND down
+
     echo ""
     echo "Starting Postgres ..."
     $DOCKER_COMPOSE_COMMAND up -d postgres
@@ -626,6 +860,9 @@ apply_changes() {
 
   echo ""
   echo "Migration complete."
+
+  # Nothing left to undo.
+  ROLLBACK_STATE="disarmed"
 }
 
 print_summary() {
@@ -643,6 +880,7 @@ print_summary() {
   echo "    $OVERRIDE_FILE"
   [[ "$MIGRATE_POSTGRES" == "yes" ]] && echo "    $ENTERPRISE_CONFIG_FILE"
   echo "    .env  (license key + secrets, mode 600)"
+  [[ "$ENV_EXISTED" == "yes" ]] && [[ -f "$ENV_BACKUP" ]] && echo "    $ENV_BACKUP  (.env as it was before this run)"
   [[ "$MIGRATE_POSTGRES" == "yes" ]] && echo "    backups/sqlite-pre-enterprise-*/  (SQLite backup)"
   echo ""
   echo " Tail logs:"
@@ -651,19 +889,27 @@ print_summary() {
   echo "──────────────────────────────────────────────────────────────────────"
   echo " To revert"
   echo "──────────────────────────────────────────────────────────────────────"
-  echo "  $DOCKER_COMPOSE_COMMAND down"
   if [[ "$MIGRATE_POSTGRES" == "yes" ]]; then
-    # Resolve project-prefixed volume names now (before override is removed).
-    local pg_volume data_volume_actual
-    pg_volume=$(resolve_data_volume "netbird_postgres")
-    data_volume_actual=$(resolve_data_volume "$DATA_VOLUME")
-    echo "  # Remove the Postgres volume FIRST, before deleting the override file:"
-    echo "  docker volume rm $pg_volume"
+    # Resolve the project-prefixed volume name now, before the override is gone.
+    local pg_volume
+    pg_volume=$(resolve_compose_volume "netbird_postgres")
+    echo "  # Stop, but keep the containers so the store can be copied back in:"
+    echo "  $DOCKER_COMPOSE_COMMAND stop"
     echo "  # Restore SQLite from the backup created during this run:"
-    echo "  docker run --rm -v ${data_volume_actual}:/var/lib/netbird -v ${BACKUP_DIR}:/backup busybox sh -c 'cp -a /backup/. /var/lib/netbird/'"
+    echo "  docker cp ${BACKUP_DIR}/. \$($DOCKER_COMPOSE_COMMAND ps -aq $COMBINED_SERVICE):/var/lib/netbird/"
+    echo "  $DOCKER_COMPOSE_COMMAND down"
+    echo "  docker volume rm $pg_volume"
+  else
+    echo "  $DOCKER_COMPOSE_COMMAND down"
   fi
   echo "  rm -f $OVERRIDE_FILE $ENTERPRISE_CONFIG_FILE"
-  echo "  # Remove migrate-to-enterprise.sh additions from .env (search for the timestamp marker)"
+  if [[ "$ENV_EXISTED" == "yes" ]] && [[ -f "$ENV_BACKUP" ]]; then
+    echo "  mv $ENV_BACKUP .env   # restores .env as it was before this run"
+  elif [[ "$ENV_EXISTED" == "no" ]]; then
+    echo "  rm -f .env   # created by this run"
+  else
+    echo "  # Remove migrate-to-enterprise.sh additions from .env (search for the timestamp marker)"
+  fi
   echo "  $DOCKER_COMPOSE_COMMAND up -d"
   echo "──────────────────────────────────────────────────────────────────────"
 }
@@ -672,6 +918,10 @@ print_summary() {
 # Run
 # ---------------------------------------------------------------------------
 
+trap on_exit EXIT
+# Turn signals into a normal exit so the EXIT trap can roll back.
+trap 'exit 130' INT TERM
+
 init_migration
 apply_changes
 print_summary

management/server/group.go

@@ -6,6 +6,8 @@ import (
 	"fmt"
 	"slices"
 
+	agentNetworkTypes "github.com/netbirdio/netbird/management/internals/modules/agentnetwork/types"
+	"github.com/netbirdio/netbird/management/internals/modules/reverseproxy/service"
 	"github.com/rs/xid"
 	log "github.com/sirupsen/logrus"
 
@@ -744,6 +746,14 @@ func validateDeleteGroup(ctx context.Context, transaction store.Store, group *ty
 		return &GroupLinkError{"network router", linkedRouter.ID}
 	}
 
+	if isLinked, linkedService := isGroupLinkedToReverseProxyService(ctx, transaction, group.AccountID, group.ID); isLinked {
+		return &GroupLinkError{"reverse proxy service", linkedService.Domain}
+	}
+
+	if isLinked, linkedPolicy := isGroupLinkedToAgentNetworkPolicy(ctx, transaction, group.AccountID, group.ID); isLinked {
+		return &GroupLinkError{"agent network policy", linkedPolicy.Name}
+	}
+
 	return checkGroupLinkedToSettings(ctx, transaction, group)
 }
 
@@ -875,6 +885,46 @@ func isGroupLinkedToNetworkRouter(ctx context.Context, transaction store.Store,
 	return false, nil
 }
 
+// isGroupLinkedToReverseProxyService checks if a group is used as an access group
+// of a private reverse proxy service or as a bearer-auth distribution group.
+func isGroupLinkedToReverseProxyService(ctx context.Context, transaction store.Store, accountID string, groupID string) (bool, *service.Service) {
+	services, err := transaction.GetAccountServices(ctx, store.LockingStrengthNone, accountID)
+	if err != nil {
+		log.WithContext(ctx).Errorf("error retrieving reverse proxy services while checking group linkage: %v", err)
+		return false, nil
+	}
+
+	for _, svc := range services {
+		if svc.Private && slices.Contains(svc.AccessGroups, groupID) {
+			return true, svc
+		}
+		if svc.Auth.BearerAuth != nil && svc.Auth.BearerAuth.Enabled && slices.Contains(svc.Auth.BearerAuth.DistributionGroups, groupID) {
+			return true, svc
+		}
+	}
+	return false, nil
+}
+
+// isGroupLinkedToAgentNetworkPolicy checks if a group is used as a source group by any
+// agent network policy in the account.
+func isGroupLinkedToAgentNetworkPolicy(ctx context.Context, transaction store.Store, accountID string, groupID string) (bool, *agentNetworkTypes.Policy) {
+	policies, err := transaction.GetAccountAgentNetworkPolicies(ctx, store.LockingStrengthNone, accountID)
+	if err != nil {
+		log.WithContext(ctx).Errorf("error retrieving agent network policies while checking group linkage: %v", err)
+		return false, nil
+	}
+
+	for _, policy := range policies {
+		if policy == nil {
+			continue
+		}
+		if slices.Contains(policy.SourceGroups, groupID) {
+			return true, policy
+		}
+	}
+	return false, nil
+}
+
 // areGroupChangesAffectPeers checks if any changes to the specified groups will affect peers.
 // It fetches each collection once and checks all groupIDs against them in memory.
 func areGroupChangesAffectPeers(ctx context.Context, transaction store.Store, accountID string, groupIDs []string) (bool, error) {

management/server/group_test.go

@@ -18,6 +18,8 @@ import (
 	"golang.org/x/exp/maps"
 
 	nbdns "github.com/netbirdio/netbird/dns"
+	agentNetworkTypes "github.com/netbirdio/netbird/management/internals/modules/agentnetwork/types"
+	rpservice "github.com/netbirdio/netbird/management/internals/modules/reverseproxy/service"
 	"github.com/netbirdio/netbird/management/server/groups"
 	"github.com/netbirdio/netbird/management/server/networks"
 	"github.com/netbirdio/netbird/management/server/networks/resources"
@@ -125,6 +127,21 @@ func TestDefaultAccountManager_DeleteGroup(t *testing.T) {
 			"grp-for-integration",
 			"only service users with admin power can delete integration group",
 		},
+		{
+			"agent network policy",
+			"grp-for-agent-network-policy",
+			"agent network policy",
+		},
+		{
+			"reverse proxy private service access group",
+			"grp-for-rp-private",
+			"reverse proxy service",
+		},
+		{
+			"reverse proxy bearer distribution group",
+			"grp-for-rp-bearer",
+			"reverse proxy service",
+		},
 	}
 
 	for _, testCase := range testCases {
@@ -218,6 +235,17 @@ func TestDefaultAccountManager_DeleteGroups(t *testing.T) {
 			groupIDs:        []string{"grp-for-integration"},
 			expectedReasons: []string{"only service users with admin power can delete integration group"},
 		},
+		{
+			name:            "agent network policy",
+			groupIDs:        []string{"grp-for-agent-network-policy"},
+			expectedReasons: []string{"agent network policy"},
+		},
+		{
+			name:               "reverse proxy services",
+			groupIDs:           []string{"grp-for-rp-private", "grp-for-rp-bearer"},
+			expectedReasons:    []string{"reverse proxy service", "reverse proxy service"},
+			expectedNotDeleted: []string{"grp-for-rp-private", "grp-for-rp-bearer"},
+		},
 		{
 			name:            "successfully delete multiple groups",
 			groupIDs:        []string{"group-1", "group-2"},
@@ -285,6 +313,65 @@ func TestDefaultAccountManager_DeleteGroups(t *testing.T) {
 	}
 }
 
+func TestDefaultAccountManager_DeleteGroupUnlinkedFromReverseProxyService(t *testing.T) {
+	am, _, err := createManager(t)
+	require.NoError(t, err, "Failed to create account manager")
+
+	_, account, err := initTestGroupAccount(am)
+	require.NoError(t, err, "Failed to init testing account")
+
+	deletableGroups := []*types.Group{
+		{
+			ID:        "grp-rp-bearer-disabled",
+			AccountID: account.Id,
+			Name:      "Group only in a disabled bearer auth",
+			Issued:    types.GroupIssuedAPI,
+			Peers:     make([]string, 0),
+		},
+		{
+			ID:        "grp-rp-nonprivate-access",
+			AccountID: account.Id,
+			Name:      "Group only in a non-private service's access groups",
+			Issued:    types.GroupIssuedAPI,
+			Peers:     make([]string, 0),
+		},
+	}
+	for _, group := range deletableGroups {
+		require.NoError(t, am.CreateGroup(context.Background(), account.Id, groupAdminUserID, group))
+	}
+
+	// Disabled bearer auth and stale access groups on a non-private service
+	// are inert configuration and must not block group deletion.
+	services := []*rpservice.Service{
+		{
+			ID:        "rp-svc-bearer-disabled",
+			AccountID: account.Id,
+			Domain:    "bearer-disabled.services.example.com",
+			Auth: rpservice.AuthConfig{
+				BearerAuth: &rpservice.BearerAuthConfig{
+					Enabled:            false,
+					DistributionGroups: []string{"grp-rp-bearer-disabled"},
+				},
+			},
+		},
+		{
+			ID:           "rp-svc-nonprivate-access",
+			AccountID:    account.Id,
+			Domain:       "nonprivate.services.example.com",
+			Private:      false,
+			AccessGroups: []string{"grp-rp-nonprivate-access"},
+		},
+	}
+	for _, svc := range services {
+		require.NoError(t, am.Store.CreateService(context.Background(), svc))
+	}
+
+	for _, group := range deletableGroups {
+		err = am.DeleteGroup(context.Background(), account.Id, groupAdminUserID, group.ID)
+		assert.NoError(t, err, "group %s is not referenced by an active reverse proxy gate and should be deletable", group.ID)
+	}
+}
+
 func TestDefaultAccountManager_DeleteGroupLinkedToFlowGroup(t *testing.T) {
 	am, _, err := createManager(t)
 	require.NoError(t, err)
@@ -406,6 +493,30 @@ func initTestGroupAccount(am *DefaultAccountManager) (*DefaultAccountManager, *t
 		Peers:     make([]string, 0),
 	}
 
+	groupForAgentNetworkPolicy := &types.Group{
+		ID:        "grp-for-agent-network-policy",
+		AccountID: "account-id",
+		Name:      "Group for agent network policies",
+		Issued:    types.GroupIssuedAPI,
+		Peers:     make([]string, 0),
+	}
+
+	groupForRPPrivate := &types.Group{
+		ID:        "grp-for-rp-private",
+		AccountID: "account-id",
+		Name:      "Group for private reverse proxy service",
+		Issued:    types.GroupIssuedAPI,
+		Peers:     make([]string, 0),
+	}
+
+	groupForRPBearer := &types.Group{
+		ID:        "grp-for-rp-bearer",
+		AccountID: "account-id",
+		Name:      "Group for bearer reverse proxy service",
+		Issued:    types.GroupIssuedAPI,
+		Peers:     make([]string, 0),
+	}
+
 	routeResource := &route.Route{
 		ID:     "example route",
 		Groups: []string{groupForRoute.ID},
@@ -461,6 +572,66 @@ func initTestGroupAccount(am *DefaultAccountManager) (*DefaultAccountManager, *t
 	_ = am.CreateGroup(context.Background(), accountID, groupAdminUserID, groupForSetupKeys)
 	_ = am.CreateGroup(context.Background(), accountID, groupAdminUserID, groupForUsers)
 	_ = am.CreateGroup(context.Background(), accountID, groupAdminUserID, groupForIntegration)
+	_ = am.CreateGroup(context.Background(), accountID, groupAdminUserID, groupForAgentNetworkPolicy)
+	_ = am.CreateGroup(context.Background(), accountID, groupAdminUserID, groupForRPPrivate)
+	_ = am.CreateGroup(context.Background(), accountID, groupAdminUserID, groupForRPBearer)
+
+	agentNetworkPolicy := &agentNetworkTypes.Policy{
+		ID:           "example agent network policy",
+		AccountID:    accountID,
+		Name:         "Example agent network policy",
+		Enabled:      true,
+		SourceGroups: []string{groupForAgentNetworkPolicy.ID},
+	}
+	if err := am.Store.SaveAgentNetworkPolicy(context.Background(), agentNetworkPolicy); err != nil {
+		return nil, nil, err
+	}
+
+	// The decoy services are created first so the linkage check has to scan
+	// past services that do not reference the groups under test.
+	rpServices := []*rpservice.Service{
+		{
+			ID:           "rp-svc-private-decoy",
+			AccountID:    accountID,
+			Domain:       "private-decoy.services.example.com",
+			Private:      true,
+			AccessGroups: []string{"unrelated-group"},
+		},
+		{
+			ID:        "rp-svc-bearer-decoy",
+			AccountID: accountID,
+			Domain:    "bearer-decoy.services.example.com",
+			Auth: rpservice.AuthConfig{
+				BearerAuth: &rpservice.BearerAuthConfig{
+					Enabled:            true,
+					DistributionGroups: []string{"unrelated-group"},
+				},
+			},
+		},
+		{
+			ID:           "rp-svc-private",
+			AccountID:    accountID,
+			Domain:       "private.services.example.com",
+			Private:      true,
+			AccessGroups: []string{groupForRPPrivate.ID},
+		},
+		{
+			ID:        "rp-svc-bearer",
+			AccountID: accountID,
+			Domain:    "bearer.services.example.com",
+			Auth: rpservice.AuthConfig{
+				BearerAuth: &rpservice.BearerAuthConfig{
+					Enabled:            true,
+					DistributionGroups: []string{groupForRPBearer.ID},
+				},
+			},
+		},
+	}
+	for _, svc := range rpServices {
+		if err := am.Store.CreateService(context.Background(), svc); err != nil {
+			return nil, nil, err
+		}
+	}
 
 	acc, err := am.Store.GetAccount(context.Background(), account.Id)
 	if err != nil {

management/server/types/account.go

@@ -1707,14 +1707,34 @@ func (a *Account) injectPrivateServicePolicies(svc *service.Service, proxyPeers
 	if len(proxyPeers) == 0 {
 		return
 	}
+	// A service's AccessGroups can name groups that no longer exist — persisted
+	// services and the agent-network synthesiser both carry the ids verbatim from
+	// their own state. An unresolvable source authorises nothing, so drop it here
+	// rather than let the network-map assembly resolve it to a nil group.
+	sources := a.existingGroupIDs(svc.AccessGroups)
+	if len(sources) == 0 {
+		return
+	}
 	for _, proxyPeer := range proxyPeers {
-		a.Policies = append(a.Policies, a.createPrivateServicePolicy(svc, proxyPeer))
+		a.Policies = append(a.Policies, a.createPrivateServicePolicy(svc, proxyPeer, sources))
 	}
 }
 
-func (a *Account) createPrivateServicePolicy(svc *service.Service, proxyPeer *nbpeer.Peer) *Policy {
+// existingGroupIDs returns the subset of groupIDs that resolve to a group in the account,
+// preserving the input order.
+func (a *Account) existingGroupIDs(groupIDs []string) []string {
+	out := make([]string, 0, len(groupIDs))
+	for _, groupID := range groupIDs {
+		if _, ok := a.Groups[groupID]; ok {
+			out = append(out, groupID)
+		}
+	}
+	return out
+}
+
+func (a *Account) createPrivateServicePolicy(svc *service.Service, proxyPeer *nbpeer.Peer, accessGroups []string) *Policy {
 	policyID := fmt.Sprintf("private-access-%s-%s", svc.ID, proxyPeer.ID)
-	sources := append([]string(nil), svc.AccessGroups...)
+	sources := append([]string(nil), accessGroups...)
 	return &Policy{
 		ID:      policyID,
 		Name:    fmt.Sprintf("Private Access to %s", svc.Name),

shared/relay/client/guard.go

@@ -156,9 +156,11 @@ func (g *Guard) notifyReconnected() {
 func (g *Guard) exponentTicker(ctx context.Context) *backoff.Ticker {
 	bo := backoff.WithContext(&backoff.ExponentialBackOff{
 		InitialInterval: 2 * time.Second,
-		Multiplier:      2,
-		MaxInterval:     g.maxBackoffInterval,
-		Clock:           backoff.SystemClock,
+		// Spreads the reconnects of every client that lost the same relay server.
+		RandomizationFactor: backoff.DefaultRandomizationFactor,
+		Multiplier:          2,
+		MaxInterval:         g.maxBackoffInterval,
+		Clock:               backoff.SystemClock,
 	}, ctx)
 
 	return backoff.NewTicker(bo)