Merge branch 'main' into worktree-dns-route-qtype-fallthrough

Co-authored-by: braginini <bangvalo@gmail.com>

.github/workflows/agent-network-e2e.yml

@@ -5,6 +5,13 @@ on:
   schedule:
     - cron: "0 3 * * *"
   workflow_dispatch:
+    inputs:
+      bedrock_model:
+        description: >-
+          Bedrock inference-profile id to drive the matrix with, exactly as
+          AWS issues it. Leave empty for the Sonnet 4.6 default.
+        required: false
+        default: ""
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
@@ -51,6 +58,9 @@ jobs:
           # token (and URL, for gateways) is unset, so partial coverage is fine.
           OPENAI_TOKEN: ${{ secrets.E2E_OPENAI_TOKEN }}
           ANTHROPIC_TOKEN: ${{ secrets.E2E_ANTHROPIC_TOKEN }}
+          # Moonshot AI platform key (platform.kimi.ai); drives both Kimi wire
+          # shapes (OpenAI /v1 and Anthropic /anthropic) through kimi_api.
+          KIMI_TOKEN: ${{ secrets.E2E_KIMI_TOKEN }}
           VERCEL_URL: ${{ secrets.E2E_VERCEL_URL }}
           VERCEL_TOKEN: ${{ secrets.E2E_VERCEL_TOKEN }}
           OPENROUTER_URL: ${{ secrets.E2E_OPENROUTER_URL }}
@@ -59,6 +69,8 @@ jobs:
           CLOUDFLARE_TOKEN: ${{ secrets.E2E_CLOUDFLARE_TOKEN }}
           AWS_BEARER_TOKEN_BEDROCK: ${{ secrets.E2E_AWS_BEARER_TOKEN_BEDROCK }}
           AWS_REGION: ${{ secrets.E2E_AWS_REGION }}
+          # Bedrock model override: dispatch input wins, then the repo variable, else the test default.
+          AWS_BEDROCK_MODEL: ${{ inputs.bedrock_model || vars.E2E_AWS_BEDROCK_MODEL }}
           # Vertex (Anthropic-on-Vertex): SA + project required; region defaults
           # to "global", model to a pinned claude snapshot.
           GOOGLE_VERTEX_SA_BASE64: ${{ secrets.E2E_GOOGLE_VERTEX_SA_BASE64 }}

.github/workflows/frontend-ui.yml

@@ -86,7 +86,7 @@ jobs:
             ${{ runner.os }}-pnpm-
 
       - name: Install dependencies
-        run: pnpm install --frozen-lockfile
+        run: pnpm install --frozen-lockfile --ignore-scripts
 
       - name: Generate Wails bindings
         run: pnpm run bindings

.github/workflows/golangci-lint.yml

@@ -45,7 +45,7 @@ jobs:
             display_name: Linux
     name: ${{ matrix.display_name }}
     runs-on: ${{ matrix.os }}
-    timeout-minutes: 15
+    timeout-minutes: 25
     steps:
       - name: Checkout code
         uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
@@ -79,4 +79,4 @@ jobs:
           skip-cache: true
           skip-save-cache: true
           cache-invalidation-interval: 0
-          args: --timeout=12m
+          args: --timeout=20m

client/android/client.go

@@ -247,6 +247,9 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		deps.SyncResponse = resp
 
 		if e := cc.Engine(); e != nil {
+			deps.RefreshStatus = func() {
+				e.RunHealthProbes(context.Background(), true)
+			}
 			if cm := e.GetClientMetrics(); cm != nil {
 				deps.ClientMetrics = cm
 			}

client/firewall/uspfilter/filter.go

@@ -121,6 +121,7 @@ type Manager struct {
 	udpTracker     *conntrack.UDPTracker
 	icmpTracker    *conntrack.ICMPTracker
 	tcpTracker     *conntrack.TCPTracker
+	fragments      *fragmentTracker
 	forwarder      atomic.Pointer[forwarder.Forwarder]
 	pendingCapture atomic.Pointer[forwarder.PacketCapture]
 	logger         *nblog.Logger
@@ -183,6 +184,41 @@ func (d *decoder) decodePacket(data []byte) error {
 	}
 }
 
+// decodeTransport decodes the transport header of a first fragment (which
+// gopacket leaves undecoded) into the decoder and appends its layer type to
+// decoded, so the ACL pipeline can evaluate it like a normal packet. It returns
+// false if the protocol is unsupported or the header is truncated.
+func (d *decoder) decodeTransport(proto layers.IPProtocol, payload []byte) bool {
+	var l4 gopacket.DecodingLayer
+	var layerType gopacket.LayerType
+	var minLen int
+	switch proto {
+	case layers.IPProtocolTCP:
+		l4, layerType, minLen = &d.tcp, layers.LayerTypeTCP, 20
+	case layers.IPProtocolUDP:
+		l4, layerType, minLen = &d.udp, layers.LayerTypeUDP, 8
+	case layers.IPProtocolICMPv4:
+		l4, layerType, minLen = &d.icmp4, layers.LayerTypeICMPv4, 8
+	case layers.IPProtocolICMPv6:
+		l4, layerType, minLen = &d.icmp6, layers.LayerTypeICMPv6, 8
+	default:
+		return false
+	}
+
+	// Reject a fragment too small to hold the full transport header before
+	// decoding: it can't be ACL-evaluated (tiny-fragment attack), and skipping
+	// the decode avoids gopacket allocating an error on the drop path.
+	if len(payload) < minLen {
+		return false
+	}
+
+	if err := l4.DecodeFromBytes(payload, gopacket.NilDecodeFeedback); err != nil {
+		return false
+	}
+	d.decoded = append(d.decoded, layerType)
+	return true
+}
+
 // Create userspace firewall manager constructor
 func Create(iface common.IFaceMapper, disableServerRoutes bool, flowLogger nftypes.FlowLogger, mtu uint16) (*Manager, error) {
 	return create(iface, nil, disableServerRoutes, flowLogger, mtu)
@@ -286,6 +322,8 @@ func create(iface common.IFaceMapper, nativeFirewall firewall.Manager, disableSe
 	if err := m.localipmanager.UpdateLocalIPs(iface); err != nil {
 		return nil, fmt.Errorf("update local IPs: %w", err)
 	}
+	m.fragments = newFragmentTracker(m.logger)
+
 	if disableConntrack {
 		log.Info("conntrack is disabled")
 	} else {
@@ -299,6 +337,7 @@ func create(iface common.IFaceMapper, nativeFirewall firewall.Manager, disableSe
 		}
 	}
 	if err := iface.SetFilter(m); err != nil {
+		m.fragments.Close()
 		return nil, fmt.Errorf("set filter: %w", err)
 	}
 	return m, nil
@@ -694,6 +733,10 @@ func (m *Manager) resetState() {
 		m.tcpTracker.Close()
 	}
 
+	if m.fragments != nil {
+		m.fragments.Close()
+	}
+
 	if fwder := m.forwarder.Load(); fwder != nil {
 		fwder.SetCapture(nil)
 		fwder.Stop()
@@ -1046,19 +1089,20 @@ func (m *Manager) filterInbound(packetData []byte, size int) bool {
 		return true
 	}
 
-	// TODO: pass fragments of routed packets to forwarder
+	// gopacket does not decode the transport header of any IP fragment, so
+	// fragments take a dedicated path: the first fragment's header is decoded
+	// and ACL-evaluated here, and the remaining fragments inherit its verdict.
 	if fragment {
-		if m.logger.Enabled(nblog.LevelTrace) {
-			if d.decoded[0] == layers.LayerTypeIPv4 {
-				m.logger.Trace4("packet is a fragment: src=%v dst=%v id=%v flags=%v",
-					srcIP, dstIP, d.ip4.Id, d.ip4.Flags)
-			} else {
-				m.logger.Trace2("packet is an IPv6 fragment: src=%v dst=%v", srcIP, dstIP)
-			}
-		}
-		return false
+		return m.filterInboundFragment(d, srcIP, dstIP, size)
 	}
 
+	return m.filterInboundDecoded(d, srcIP, dstIP, packetData, size)
+}
+
+// filterInboundDecoded runs the ACL, DNAT and conntrack pipeline on a fully
+// decoded (non-fragment) inbound packet. It returns true if the packet should
+// be dropped.
+func (m *Manager) filterInboundDecoded(d *decoder, srcIP, dstIP netip.Addr, packetData []byte, size int) bool {
 	// TODO: optimize port DNAT by caching matched rules in conntrack
 	if translated := m.translateInboundPortDNAT(packetData, d, srcIP, dstIP); translated {
 		// Re-decode after port DNAT translation to update port information
@@ -1089,33 +1133,226 @@ func (m *Manager) filterInbound(packetData []byte, size int) bool {
 	return m.handleRoutedTraffic(d, srcIP, dstIP, packetData, size)
 }
 
+// fragmentMeta holds the reassembly identity and layout of an IP fragment,
+// extracted uniformly for IPv4 and IPv6.
+type fragmentMeta struct {
+	key fragmentKey
+	// offset is the fragment offset in 8-byte units (zero for the first
+	// fragment).
+	offset uint16
+	// moreFragments is the More Fragments bit. A first fragment with it unset is
+	// an IPv6 atomic fragment (a complete datagram, RFC 6946): it has no trailing
+	// fragments to inherit a verdict, so it must not be recorded.
+	moreFragments bool
+	proto         layers.IPProtocol
+	// l4payload is the fragmentable payload of this fragment. For the first
+	// fragment it starts with the transport header.
+	l4payload []byte
+	// headerEndOctets is the first fragment's payload length in 8-byte units:
+	// the smallest offset a trailing fragment may start at without overlapping
+	// the inspected transport header.
+	headerEndOctets uint16
+}
+
+// fragmentMetadata extracts the fragment identity and layout from a decoded IP
+// fragment. It returns false for fragments it can't interpret (e.g. an IPv6
+// fragment header shorter than 8 bytes), which are then dropped.
+func fragmentMetadata(d *decoder, srcIP, dstIP netip.Addr) (fragmentMeta, bool) {
+	switch d.decoded[0] {
+	case layers.LayerTypeIPv4:
+		payload := d.ip4.Payload
+		return fragmentMeta{
+			key:             fragmentKey{srcIP: srcIP, dstIP: dstIP, id: uint32(d.ip4.Id), proto: uint8(d.ip4.Protocol)},
+			offset:          d.ip4.FragOffset,
+			moreFragments:   d.ip4.Flags&layers.IPv4MoreFragments != 0,
+			proto:           d.ip4.Protocol,
+			l4payload:       payload,
+			headerEndOctets: octets(len(payload)),
+		}, true
+
+	case layers.LayerTypeIPv6:
+		// IPv6 fragment extension header: 8 bytes, followed by the fragmentable
+		// payload. Layout: next header (1), reserved (1), offset+flags (2), id (4).
+		payload := d.ip6.Payload
+		if len(payload) < 8 {
+			return fragmentMeta{}, false
+		}
+		nextHeader := layers.IPProtocol(payload[0])
+		offsetFlags := binary.BigEndian.Uint16(payload[2:4])
+		id := binary.BigEndian.Uint32(payload[4:8])
+		l4 := payload[8:]
+		return fragmentMeta{
+			key:             fragmentKey{srcIP: srcIP, dstIP: dstIP, id: id, proto: uint8(nextHeader)},
+			offset:          offsetFlags >> 3,
+			moreFragments:   offsetFlags&1 != 0,
+			proto:           nextHeader,
+			l4payload:       l4,
+			headerEndOctets: octets(len(l4)),
+		}, true
+
+	default:
+		return fragmentMeta{}, false
+	}
+}
+
+// octets rounds a byte length up to whole 8-byte units, the granularity of the
+// IP fragment offset field.
+func octets(nbytes int) uint16 {
+	return uint16((nbytes + 7) / 8)
+}
+
+// filterInboundFragment decides the fate of an IP fragment. gopacket stops
+// decoding at the network layer for every fragment, so the first fragment's
+// transport header is decoded and ACL-evaluated here and its verdict recorded;
+// the remaining (headerless) fragments inherit that verdict. Anything that
+// cannot be tied to an allowed, non-overlapping first fragment is dropped.
+func (m *Manager) filterInboundFragment(d *decoder, srcIP, dstIP netip.Addr, size int) bool {
+	meta, ok := fragmentMetadata(d, srcIP, dstIP)
+	if !ok {
+		if m.logger.Enabled(nblog.LevelTrace) {
+			m.logger.Trace2("dropping unsupported fragment: src=%v dst=%v", srcIP, dstIP)
+		}
+		return true
+	}
+
+	if meta.offset != 0 {
+		return m.filterTrailingFragment(meta, srcIP, dstIP)
+	}
+
+	// A new first fragment supersedes any recorded verdict for this datagram, so
+	// a re-sent or overlapping offset-zero fragment can't inherit the old one.
+	m.fragments.poison(meta.key)
+
+	// First fragment: decode its transport header so the ACL can evaluate it. A
+	// decode failure means the fragment is too small to hold the full transport
+	// header (RFC 1858 §3 tiny-fragment attack); it can't be evaluated, so drop it.
+	if !d.decodeTransport(meta.proto, meta.l4payload) {
+		if m.logger.Enabled(nblog.LevelTrace) {
+			m.logger.Trace3("dropping first fragment without full L4 header: src=%v dst=%v id=%v",
+				srcIP, dstIP, meta.key.id)
+		}
+		return true
+	}
+
+	return m.filterFirstFragment(d, meta, srcIP, dstIP, size)
+}
+
+// filterTrailingFragment applies a recorded first-fragment verdict to a
+// non-first fragment.
+func (m *Manager) filterTrailingFragment(meta fragmentMeta, srcIP, dstIP netip.Addr) bool {
+	switch m.fragments.verdict(meta.key, meta.offset) {
+	case fragmentAllow:
+		return false
+	case fragmentOverlap:
+		if m.logger.Enabled(nblog.LevelTrace) {
+			m.logger.Trace3("dropping overlapping fragment rewriting inspected header: src=%v dst=%v id=%v",
+				srcIP, dstIP, meta.key.id)
+		}
+		return true
+	default:
+		if m.logger.Enabled(nblog.LevelTrace) {
+			m.logger.Trace3("dropping fragment with no allowed first fragment: src=%v dst=%v id=%v",
+				srcIP, dstIP, meta.key.id)
+		}
+		return true
+	}
+}
+
+// filterFirstFragment runs the verdict part of the inbound pipeline on a first
+// fragment with its transport header decoded. It mirrors filterInboundDecoded
+// but skips DNAT (port rewriting on fragments is unsupported) and forwarder
+// injection (fragments are left to the stack to reassemble, not forwarded).
+// Allowed fragments have their verdict recorded so the datagram's trailing
+// fragments inherit it.
+func (m *Manager) filterFirstFragment(d *decoder, meta fragmentMeta, srcIP, dstIP netip.Addr, size int) bool {
+	if m.stateful && m.isValidTrackedConnection(d, srcIP, dstIP, size) {
+		m.recordFirstFragment(meta)
+		return false
+	}
+
+	if m.localipmanager.IsLocalIP(dstIP) {
+		ruleID, blocked := m.peerACLsBlock(srcIP, d, nil)
+		if blocked {
+			m.storeDropFlow("Dropping local first fragment (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
+				d, srcIP, dstIP, ruleID, size)
+			return true
+		}
+		m.trackInbound(d, srcIP, dstIP, ruleID, size)
+		m.recordFirstFragment(meta)
+		return false
+	}
+
+	if !m.routingEnabled.Load() {
+		if m.logger.Enabled(nblog.LevelTrace) {
+			m.logger.Trace2("Dropping routed fragment (routing disabled): src=%s dst=%s", srcIP, dstIP)
+		}
+		return true
+	}
+	if m.nativeRouter.Load() {
+		m.trackInbound(d, srcIP, dstIP, nil, size)
+		m.recordFirstFragment(meta)
+		return false
+	}
+
+	// TODO: pass fragments of routed packets to the forwarder; until then
+	// allowed routed fragments go to the native stack.
+	srcPort, dstPort := getPortsFromPacket(d)
+	ruleID, pass := m.routeACLsPass(srcIP, dstIP, d.decoded[1], srcPort, dstPort)
+	if !pass {
+		m.storeDropFlow("Dropping routed first fragment (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
+			d, srcIP, dstIP, ruleID, size)
+		return true
+	}
+
+	m.recordFirstFragment(meta)
+	return false
+}
+
+// recordFirstFragment caches an allowed first fragment's verdict for its
+// trailing fragments to inherit. Atomic fragments (no More Fragments bit) are
+// complete datagrams with no trailing fragments, so they are not cached and
+// cannot exhaust the verdict table.
+func (m *Manager) recordFirstFragment(meta fragmentMeta) {
+	if !meta.moreFragments {
+		return
+	}
+	m.fragments.recordAllowed(meta.key, meta.headerEndOctets)
+}
+
+// storeDropFlow logs and records a netflow drop event for an inbound packet
+// denied by the ACLs. msg is the trace format taking rule id, protocol, source
+// and destination.
+func (m *Manager) storeDropFlow(msg string, d *decoder, srcIP, dstIP netip.Addr, ruleID []byte, size int) {
+	pnum := getProtocolFromPacket(d)
+	srcPort, dstPort := getPortsFromPacket(d)
+
+	if m.logger.Enabled(nblog.LevelTrace) {
+		m.logger.Trace6(msg, ruleID, pnum, srcIP, srcPort, dstIP, dstPort)
+	}
+
+	m.flowLogger.StoreEvent(nftypes.EventFields{
+		FlowID:     uuid.New(),
+		Type:       nftypes.TypeDrop,
+		RuleID:     ruleID,
+		Direction:  nftypes.Ingress,
+		Protocol:   pnum,
+		SourceIP:   srcIP,
+		DestIP:     dstIP,
+		SourcePort: srcPort,
+		DestPort:   dstPort,
+		// TODO: icmp type/code
+		RxPackets: 1,
+		RxBytes:   uint64(size),
+	})
+}
+
 // handleLocalTraffic handles local traffic.
 // If it returns true, the packet should be dropped.
 func (m *Manager) handleLocalTraffic(d *decoder, srcIP, dstIP netip.Addr, packetData []byte, size int) bool {
 	ruleID, blocked := m.peerACLsBlock(srcIP, d, packetData)
 	if blocked {
-		pnum := getProtocolFromPacket(d)
-		srcPort, dstPort := getPortsFromPacket(d)
-
-		if m.logger.Enabled(nblog.LevelTrace) {
-			m.logger.Trace6("Dropping local packet (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
-				ruleID, pnum, srcIP, srcPort, dstIP, dstPort)
-		}
-
-		m.flowLogger.StoreEvent(nftypes.EventFields{
-			FlowID:     uuid.New(),
-			Type:       nftypes.TypeDrop,
-			RuleID:     ruleID,
-			Direction:  nftypes.Ingress,
-			Protocol:   pnum,
-			SourceIP:   srcIP,
-			DestIP:     dstIP,
-			SourcePort: srcPort,
-			DestPort:   dstPort,
-			// TODO: icmp type/code
-			RxPackets: 1,
-			RxBytes:   uint64(size),
-		})
+		m.storeDropFlow("Dropping local packet (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
+			d, srcIP, dstIP, ruleID, size)
 		return true
 	}
 
@@ -1168,27 +1405,8 @@ func (m *Manager) handleRoutedTraffic(d *decoder, srcIP, dstIP netip.Addr, packe
 
 	ruleID, pass := m.routeACLsPass(srcIP, dstIP, protoLayer, srcPort, dstPort)
 	if !pass {
-		proto := getProtocolFromPacket(d)
-
-		if m.logger.Enabled(nblog.LevelTrace) {
-			m.logger.Trace6("Dropping routed packet (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
-				ruleID, proto, srcIP, srcPort, dstIP, dstPort)
-		}
-
-		m.flowLogger.StoreEvent(nftypes.EventFields{
-			FlowID:     uuid.New(),
-			Type:       nftypes.TypeDrop,
-			RuleID:     ruleID,
-			Direction:  nftypes.Ingress,
-			Protocol:   proto,
-			SourceIP:   srcIP,
-			DestIP:     dstIP,
-			SourcePort: srcPort,
-			DestPort:   dstPort,
-			// TODO: icmp type/code
-			RxPackets: 1,
-			RxBytes:   uint64(size),
-		})
+		m.storeDropFlow("Dropping routed packet (ACL denied): rule_id=%s proto=%v src=%s:%d dst=%s:%d",
+			d, srcIP, dstIP, ruleID, size)
 		return true
 	}
 

client/firewall/uspfilter/forwarder/forwarder.go

@@ -5,7 +5,9 @@ import (
 	"fmt"
 	"net"
 	"net/netip"
+	"os"
 	"runtime"
+	"strconv"
 	"sync"
 	"time"
 
@@ -31,6 +33,11 @@ const (
 	defaultMaxInFlight   = 1024
 	iosReceiveWindow     = 16384
 	iosMaxInFlight       = 256
+
+	// envForceTCPRACK overrides the platform default for gVisor's RACK loss
+	// detection. Set to a truthy value to force RACK on, or a falsy value to
+	// force it off, on any platform.
+	envForceTCPRACK = "NB_FORCE_TCP_RACK"
 )
 
 type Forwarder struct {
@@ -152,6 +159,8 @@ func New(iface common.IFaceMapper, logger *nblog.Logger, flowLogger nftypes.Flow
 		maxInFlight = iosMaxInFlight
 	}
 
+	configureTCPRecovery(s)
+
 	tcpForwarder := tcp.NewForwarder(s, receiveWindow, maxInFlight, f.handleTCP)
 	s.SetTransportProtocolHandler(tcp.ProtocolNumber, tcpForwarder.HandlePacket)
 
@@ -466,3 +475,31 @@ func probeRawICMP(network, addr string, logger *nblog.Logger) bool {
 	logger.Debug1("forwarder: raw %s socket access available", network)
 	return true
 }
+
+// configureTCPRecovery disables gVisor's RACK loss detection on Windows, where
+// it interacts poorly with the host and collapses throughput on routed TCP
+// connections (gVisor issue #9778). Other platforms keep the default. The
+// EnvForceTCPRACK environment variable overrides the platform default.
+func configureTCPRecovery(s *stack.Stack) {
+	disableRACK := runtime.GOOS == "windows"
+
+	if val := os.Getenv(envForceTCPRACK); val != "" {
+		force, err := strconv.ParseBool(val)
+		if err != nil {
+			log.Warnf("parse %s: %v", envForceTCPRACK, err)
+		} else {
+			disableRACK = !force
+		}
+	}
+
+	if !disableRACK {
+		return
+	}
+
+	opt := tcpip.TCPRecovery(0)
+	if err := s.SetTransportProtocolOption(tcp.ProtocolNumber, &opt); err != nil {
+		log.Warnf("disable TCP RACK loss detection: %v", err)
+		return
+	}
+	log.Info("forwarder: TCP RACK loss detection disabled")
+}

client/firewall/uspfilter/fragment.go

@@ -0,0 +1,204 @@
+package uspfilter
+
+import (
+	"context"
+	"net/netip"
+	"os"
+	"strconv"
+	"sync"
+	"time"
+
+	nblog "github.com/netbirdio/netbird/client/firewall/uspfilter/log"
+)
+
+const (
+	// defaultFragmentTimeout bounds how long a first-fragment verdict is kept
+	// while the remaining fragments arrive. It mirrors the Linux IP reassembly
+	// timeout (net.ipv4.ipfrag_time).
+	defaultFragmentTimeout = 30 * time.Second
+	// fragmentCleanupInterval is how often expired verdicts are purged.
+	fragmentCleanupInterval = 10 * time.Second
+	// defaultMaxFragmentEntries caps the number of concurrently tracked
+	// fragmented datagrams. The table stays bounded because each datagram is a
+	// single small entry regardless of how many fragments it is split into, and
+	// the 13-bit IPv4 fragment-offset field limits any datagram to 64 KiB.
+	defaultMaxFragmentEntries = 16384
+
+	// EnvFragmentMaxEntries overrides defaultMaxFragmentEntries.
+	EnvFragmentMaxEntries = "NB_FRAGMENT_MAX_ENTRIES"
+)
+
+// fragmentVerdict is the decision for a trailing (headerless) fragment.
+type fragmentVerdict int
+
+const (
+	// fragmentDeny drops the fragment: no allowed first fragment is on record.
+	fragmentDeny fragmentVerdict = iota
+	// fragmentAllow passes the fragment: it belongs to an allowed datagram and
+	// does not overlap the already-inspected transport header.
+	fragmentAllow
+	// fragmentOverlap drops the fragment and poisons its datagram: it overlaps
+	// the transport header the ACL inspected (RFC 1858 §4, RFC 3128; RFC 5722
+	// requires discarding the whole datagram on overlap for IPv6).
+	fragmentOverlap
+)
+
+// fragmentKey identifies a fragmented datagram. It matches the RFC 791 / RFC
+// 8200 reassembly key: source, destination, protocol and identification. The id
+// is 32-bit to hold both the IPv4 (16-bit) and IPv6 (32-bit) identification.
+type fragmentKey struct {
+	srcIP netip.Addr
+	dstIP netip.Addr
+	id    uint32
+	proto uint8
+}
+
+// fragmentEntry records the verdict of an allowed first fragment.
+type fragmentEntry struct {
+	// headerEndOctets is the offset, in 8-byte units, at which the first
+	// fragment's payload ended. A trailing fragment starting before this
+	// overlaps bytes the ACL already inspected and is rejected.
+	headerEndOctets uint16
+	// recordedAt is when the first fragment was accepted. The verdict expires a
+	// fixed timeout later and is not refreshed, mirroring the kernel reassembly
+	// timer so a trailing-fragment flood can't keep a datagram alive.
+	recordedAt time.Time
+}
+
+// fragmentTracker records the ACL verdict of a datagram's first fragment so the
+// remaining fragments, which carry no L4 header, can inherit the decision
+// without reassembling the datagram. Only allowed first fragments are stored;
+// anything that cannot be tied to an allowed, non-overlapping first fragment is
+// dropped (fail closed).
+type fragmentTracker struct {
+	logger  *nblog.Logger
+	mutex   sync.Mutex
+	entries map[fragmentKey]fragmentEntry
+	timeout time.Duration
+	// maxEntries caps the table; atCapacity dedups the capacity warning until
+	// the table drains below the cap again.
+	maxEntries    int
+	atCapacity    bool
+	cleanupTicker *time.Ticker
+	cancel        context.CancelFunc
+}
+
+func newFragmentTracker(logger *nblog.Logger) *fragmentTracker {
+	ctx, cancel := context.WithCancel(context.Background())
+	t := &fragmentTracker{
+		logger:        logger,
+		entries:       make(map[fragmentKey]fragmentEntry),
+		timeout:       defaultFragmentTimeout,
+		maxEntries:    fragmentMaxEntries(logger),
+		cleanupTicker: time.NewTicker(fragmentCleanupInterval),
+		cancel:        cancel,
+	}
+	go t.cleanupRoutine(ctx)
+	return t
+}
+
+func fragmentMaxEntries(logger *nblog.Logger) int {
+	v := os.Getenv(EnvFragmentMaxEntries)
+	if v == "" {
+		return defaultMaxFragmentEntries
+	}
+	n, err := strconv.Atoi(v)
+	if err != nil || n <= 0 {
+		logger.Warn2("invalid %s=%q, using default", EnvFragmentMaxEntries, v)
+		return defaultMaxFragmentEntries
+	}
+	return n
+}
+
+// recordAllowed stores the verdict of an allowed first fragment. headerEndOctets
+// is the first fragment's payload length in 8-byte units. When the table is full
+// the record is dropped, which fails closed: the datagram's trailing fragments
+// will be denied.
+func (t *fragmentTracker) recordAllowed(key fragmentKey, headerEndOctets uint16) {
+	t.mutex.Lock()
+	defer t.mutex.Unlock()
+
+	if t.entries == nil {
+		return
+	}
+	if _, ok := t.entries[key]; !ok && len(t.entries) >= t.maxEntries {
+		if !t.atCapacity {
+			t.atCapacity = true
+			t.logger.Warn2("fragment verdict table at capacity (%d/%d): trailing fragments of new datagrams will be dropped",
+				len(t.entries), t.maxEntries)
+		}
+		return
+	}
+	t.entries[key] = fragmentEntry{
+		headerEndOctets: headerEndOctets,
+		recordedAt:      time.Now(),
+	}
+}
+
+// poison drops any recorded verdict for a datagram, so its later fragments are
+// denied until a new allowed first fragment is recorded. Called on every
+// offset-zero fragment to defeat offset-zero overlap rewrites (RFC 3128).
+func (t *fragmentTracker) poison(key fragmentKey) {
+	t.mutex.Lock()
+	defer t.mutex.Unlock()
+	delete(t.entries, key)
+}
+
+// verdict decides the fate of a trailing fragment at fragOffsetOctets (the IPv4
+// fragment offset, in 8-byte units). A fragment overlapping the inspected
+// header poisons the datagram: the entry is removed so all further fragments of
+// that datagram are denied too.
+func (t *fragmentTracker) verdict(key fragmentKey, fragOffsetOctets uint16) fragmentVerdict {
+	t.mutex.Lock()
+	defer t.mutex.Unlock()
+
+	entry, ok := t.entries[key]
+	if !ok {
+		return fragmentDeny
+	}
+	if time.Since(entry.recordedAt) > t.timeout {
+		delete(t.entries, key)
+		return fragmentDeny
+	}
+	if fragOffsetOctets < entry.headerEndOctets {
+		delete(t.entries, key)
+		return fragmentOverlap
+	}
+	return fragmentAllow
+}
+
+func (t *fragmentTracker) cleanupRoutine(ctx context.Context) {
+	defer t.cleanupTicker.Stop()
+	for {
+		select {
+		case <-t.cleanupTicker.C:
+			t.cleanup()
+		case <-ctx.Done():
+			return
+		}
+	}
+}
+
+func (t *fragmentTracker) cleanup() {
+	t.mutex.Lock()
+	defer t.mutex.Unlock()
+
+	for key, entry := range t.entries {
+		if time.Since(entry.recordedAt) > t.timeout {
+			delete(t.entries, key)
+		}
+	}
+
+	if len(t.entries) < t.maxEntries {
+		t.atCapacity = false
+	}
+}
+
+// Close stops the cleanup routine and releases resources.
+func (t *fragmentTracker) Close() {
+	t.cancel()
+
+	t.mutex.Lock()
+	t.entries = nil
+	t.mutex.Unlock()
+}

client/firewall/uspfilter/fragment_bench_test.go

@@ -0,0 +1,115 @@
+package uspfilter
+
+import (
+	"encoding/binary"
+	"testing"
+)
+
+// benchFilterInbound drives filterInbound over a fixed packet in a tight loop.
+// Packets are built once, outside the timed region, so the benchmark measures
+// only pipeline cost, which is what an attacker can amplify.
+func benchFilterInbound(b *testing.B, pkt []byte) {
+	b.Helper()
+	b.ReportAllocs()
+	b.SetBytes(int64(len(pkt)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		m := benchManager
+		m.filterInbound(pkt, len(pkt))
+	}
+}
+
+// benchManager is a package-level manager reused across fragment benchmarks so
+// setup cost stays out of the timed region.
+var benchManager *Manager
+
+func setupBenchManager(b *testing.B) *Manager {
+	b.Helper()
+	m := newFragmentTestManager(b)
+	allowUDP(b, m, 8080)
+	// Disable conntrack so the allowed-first-fragment path measures transport
+	// decode + ACL every iteration instead of matching the connection tracked
+	// on the first iteration.
+	m.stateful = false
+	benchManager = m
+	return m
+}
+
+// BenchmarkInbound_NormalPacket is the baseline: a full, non-fragmented UDP
+// packet that passes the ACL. Fragment paths should stay comparable to this.
+func BenchmarkInbound_NormalPacket(b *testing.B) {
+	setupBenchManager(b)
+	pkt := normalUDPPacket(b, 8080, 32)
+	benchFilterInbound(b, pkt)
+}
+
+// BenchmarkInbound_FirstFragmentAllowed measures the first-fragment path:
+// transport decode + ACL evaluation + verdict record.
+func BenchmarkInbound_FirstFragmentAllowed(b *testing.B) {
+	setupBenchManager(b)
+	pkt := firstFragmentUDP(b, 0x2000, 8080, 32)
+	benchFilterInbound(b, pkt)
+}
+
+// BenchmarkInbound_TrailingFragmentAllowed measures the common trailing-fragment
+// path: a single map lookup after the first fragment is on record.
+func BenchmarkInbound_TrailingFragmentAllowed(b *testing.B) {
+	m := setupBenchManager(b)
+	first := firstFragmentUDP(b, 0x3000, 8080, 32)
+	m.filterInbound(first, len(first))
+	pkt := trailingFragment(b, 0x3000, 5, false, 24)
+	benchFilterInbound(b, pkt)
+}
+
+// BenchmarkInbound_TrailingFragmentNoFirst is the primary DoS vector: an
+// attacker floods trailing fragments with no first fragment on record. Each is
+// a map miss and must be cheap.
+func BenchmarkInbound_TrailingFragmentNoFirst(b *testing.B) {
+	setupBenchManager(b)
+	pkt := trailingFragment(b, 0x4000, 185, false, 40)
+	benchFilterInbound(b, pkt)
+}
+
+// BenchmarkInbound_TinyFirstFragment measures the tiny-fragment drop path: a
+// first fragment too small to decode a transport header.
+func BenchmarkInbound_TinyFirstFragment(b *testing.B) {
+	setupBenchManager(b)
+	pkt := trailingFragment(b, 0x5000, 0, true, 4)
+	benchFilterInbound(b, pkt)
+}
+
+// BenchmarkInbound_TrailingFragmentDistinctIDs is the worst case for the
+// verdict table: an attacker varies the datagram id on every packet so no first
+// fragment ever matches. Verdict lookups always miss and nothing is recorded,
+// so the table cannot grow. Each iteration rewrites the id field in place.
+func BenchmarkInbound_TrailingFragmentDistinctIDs(b *testing.B) {
+	setupBenchManager(b)
+	pkt := trailingFragment(b, 0x6000, 185, false, 40)
+	m := benchManager
+
+	b.ReportAllocs()
+	b.SetBytes(int64(len(pkt)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		// IPv4 identification field is at bytes 4:6.
+		binary.BigEndian.PutUint16(pkt[4:6], uint16(i))
+		m.filterInbound(pkt, len(pkt))
+	}
+}
+
+// BenchmarkInbound_FirstFragmentDistinctIDs measures sustained first-fragment
+// pressure with distinct ids: transport decode + ACL + verdict insert until the
+// table caps, exercising the map growth and capacity guard.
+func BenchmarkInbound_FirstFragmentDistinctIDs(b *testing.B) {
+	setupBenchManager(b)
+	pkt := firstFragmentUDP(b, 0x7000, 8080, 32)
+	m := benchManager
+
+	b.ReportAllocs()
+	b.SetBytes(int64(len(pkt)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		binary.BigEndian.PutUint16(pkt[4:6], uint16(i))
+		m.filterInbound(pkt, len(pkt))
+	}
+}

client/firewall/uspfilter/fragment_test.go

@@ -0,0 +1,554 @@
+package uspfilter
+
+import (
+	"encoding/binary"
+	"net"
+	"net/netip"
+	"testing"
+	"time"
+
+	"github.com/google/gopacket"
+	"github.com/google/gopacket/layers"
+	"github.com/stretchr/testify/require"
+
+	fw "github.com/netbirdio/netbird/client/firewall/manager"
+	nbiface "github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/iface/device"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+const (
+	fragTestSrc   = "100.10.0.1"
+	fragTestDst   = "100.10.0.100"
+	fragTestSrcV6 = "fd00::1"
+	fragTestDstV6 = "fd00::100"
+)
+
+func newFragmentTestManager(tb testing.TB) *Manager {
+	tb.Helper()
+
+	ifaceMock := &IFaceMock{
+		SetFilterFunc: func(device.PacketFilter) error { return nil },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr(fragTestDst),
+				Network: netip.MustParsePrefix("100.10.0.0/16"),
+				IPv6:    netip.MustParseAddr(fragTestDstV6),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+
+	m, err := Create(ifaceMock, false, flowLogger, nbiface.DefaultMTU)
+	require.NoError(tb, err)
+	require.NoError(tb, m.UpdateLocalIPs())
+	tb.Cleanup(func() { require.NoError(tb, m.Close(nil)) })
+	return m
+}
+
+// firstFragmentUDPTo builds the first fragment of a fragmented UDP datagram to
+// the given destination: it carries the full UDP header plus payloadLen bytes
+// of data, with the More Fragments flag set and offset zero.
+func firstFragmentUDPTo(tb testing.TB, dst string, id uint16, dstPort uint16, payloadLen int) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv4{
+		Version:  4,
+		TTL:      64,
+		Id:       id,
+		Protocol: layers.IPProtocolUDP,
+		SrcIP:    net.ParseIP(fragTestSrc),
+		DstIP:    net.ParseIP(dst),
+		Flags:    layers.IPv4MoreFragments,
+	}
+	udp := &layers.UDP{SrcPort: 40000, DstPort: layers.UDPPort(dstPort)}
+	require.NoError(tb, udp.SetNetworkLayerForChecksum(ip))
+
+	buf := gopacket.NewSerializeBuffer()
+	opts := gopacket.SerializeOptions{ComputeChecksums: true, FixLengths: true}
+	require.NoError(tb, gopacket.SerializeLayers(buf, opts, ip, udp, gopacket.Payload(make([]byte, payloadLen))))
+	return buf.Bytes()
+}
+
+func firstFragmentUDP(tb testing.TB, id uint16, dstPort uint16, payloadLen int) []byte {
+	tb.Helper()
+	return firstFragmentUDPTo(tb, fragTestDst, id, dstPort, payloadLen)
+}
+
+// firstFragmentTCP builds the first fragment of a fragmented TCP datagram: the
+// full 20-byte TCP header plus 12 bytes of data, with the More Fragments flag
+// set and offset zero.
+func firstFragmentTCP(tb testing.TB, id uint16, dstPort uint16) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv4{
+		Version:  4,
+		TTL:      64,
+		Id:       id,
+		Protocol: layers.IPProtocolTCP,
+		SrcIP:    net.ParseIP(fragTestSrc),
+		DstIP:    net.ParseIP(fragTestDst),
+		Flags:    layers.IPv4MoreFragments,
+	}
+	tcp := &layers.TCP{SrcPort: 40000, DstPort: layers.TCPPort(dstPort), SYN: true, Window: 64240}
+	require.NoError(tb, tcp.SetNetworkLayerForChecksum(ip))
+
+	buf := gopacket.NewSerializeBuffer()
+	opts := gopacket.SerializeOptions{ComputeChecksums: true, FixLengths: true}
+	require.NoError(tb, gopacket.SerializeLayers(buf, opts, ip, tcp, gopacket.Payload(make([]byte, 12))))
+	return buf.Bytes()
+}
+
+// trailingFragmentTo builds a non-first fragment to the given destination: an
+// IPv4 header at the given fragment offset (in 8-byte units) carrying raw
+// payload and no L4 header.
+func trailingFragmentTo(tb testing.TB, dst string, proto layers.IPProtocol, id uint16, fragOffsetOctets uint16, moreFragments bool, payloadLen int) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv4{
+		Version:    4,
+		TTL:        64,
+		Id:         id,
+		Protocol:   proto,
+		SrcIP:      net.ParseIP(fragTestSrc),
+		DstIP:      net.ParseIP(dst),
+		FragOffset: fragOffsetOctets,
+	}
+	if moreFragments {
+		ip.Flags = layers.IPv4MoreFragments
+	}
+
+	buf := gopacket.NewSerializeBuffer()
+	opts := gopacket.SerializeOptions{FixLengths: true}
+	require.NoError(tb, gopacket.SerializeLayers(buf, opts, ip, gopacket.Payload(make([]byte, payloadLen))))
+	return buf.Bytes()
+}
+
+func trailingFragment(tb testing.TB, id uint16, fragOffsetOctets uint16, moreFragments bool, payloadLen int) []byte {
+	tb.Helper()
+	return trailingFragmentTo(tb, fragTestDst, layers.IPProtocolUDP, id, fragOffsetOctets, moreFragments, payloadLen)
+}
+
+// outboundUDPPacket builds a complete outbound UDP packet from the local
+// address, used to establish conntrack state for reply-direction tests.
+func outboundUDPPacket(tb testing.TB, srcPort, dstPort uint16) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv4{
+		Version:  4,
+		TTL:      64,
+		Id:       1,
+		Protocol: layers.IPProtocolUDP,
+		SrcIP:    net.ParseIP(fragTestDst),
+		DstIP:    net.ParseIP(fragTestSrc),
+	}
+	udp := &layers.UDP{SrcPort: layers.UDPPort(srcPort), DstPort: layers.UDPPort(dstPort)}
+	require.NoError(tb, udp.SetNetworkLayerForChecksum(ip))
+
+	buf := gopacket.NewSerializeBuffer()
+	opts := gopacket.SerializeOptions{ComputeChecksums: true, FixLengths: true}
+	require.NoError(tb, gopacket.SerializeLayers(buf, opts, ip, udp, gopacket.Payload(make([]byte, 16))))
+	return buf.Bytes()
+}
+
+// normalUDPPacket builds a complete, non-fragmented UDP packet for baseline
+// comparisons against the fragment paths.
+func normalUDPPacket(tb testing.TB, dstPort uint16, payloadLen int) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv4{
+		Version:  4,
+		TTL:      64,
+		Id:       1,
+		Protocol: layers.IPProtocolUDP,
+		SrcIP:    net.ParseIP(fragTestSrc),
+		DstIP:    net.ParseIP(fragTestDst),
+	}
+	udp := &layers.UDP{SrcPort: 40000, DstPort: layers.UDPPort(dstPort)}
+	require.NoError(tb, udp.SetNetworkLayerForChecksum(ip))
+
+	buf := gopacket.NewSerializeBuffer()
+	opts := gopacket.SerializeOptions{ComputeChecksums: true, FixLengths: true}
+	require.NoError(tb, gopacket.SerializeLayers(buf, opts, ip, udp, gopacket.Payload(make([]byte, payloadLen))))
+	return buf.Bytes()
+}
+
+func allowUDP(tb testing.TB, m *Manager, dstPort uint16) {
+	tb.Helper()
+	_, err := m.AddPeerFiltering(nil, net.ParseIP(fragTestSrc), fw.ProtocolUDP, nil,
+		&fw.Port{Values: []uint16{dstPort}}, fw.ActionAccept, "")
+	require.NoError(tb, err)
+}
+
+// TestFragment_TrailingWithoutFirstDropped is the core bypass repro: a trailing
+// fragment with no allowed first fragment on record must be dropped. Before the
+// fix, filterInbound returned false (allow) for any fragment.
+func TestFragment_TrailingWithoutFirstDropped(t *testing.T) {
+	m := newFragmentTestManager(t)
+
+	frag := trailingFragment(t, 0x1234, 185, false, 40)
+	require.True(t, m.filterInbound(frag, len(frag)),
+		"trailing fragment without an allowed first fragment must be dropped")
+}
+
+// TestFragment_AllowedFirstPassesTrailing verifies that once a first fragment
+// passes the ACL, its trailing fragments inherit the allow verdict.
+func TestFragment_AllowedFirstPassesTrailing(t *testing.T) {
+	m := newFragmentTestManager(t)
+	allowUDP(t, m, 8080)
+
+	// First fragment: UDP header (8) + 32 payload = 40 octets -> headerEnd = 5.
+	first := firstFragmentUDP(t, 0x2222, 8080, 32)
+	require.False(t, m.filterInbound(first, len(first)),
+		"allowed first fragment should pass and be recorded")
+
+	trailing := trailingFragment(t, 0x2222, 5, false, 24)
+	require.False(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment of an allowed datagram should pass")
+}
+
+// TestFragment_DeniedFirstDropsTrailing verifies that a first fragment blocked
+// by the ACL leaves no verdict, so its trailing fragments are dropped.
+func TestFragment_DeniedFirstDropsTrailing(t *testing.T) {
+	m := newFragmentTestManager(t)
+	// No accept rule: local traffic defaults to deny.
+
+	first := firstFragmentUDP(t, 0x3333, 9999, 32)
+	require.True(t, m.filterInbound(first, len(first)),
+		"first fragment to a blocked port should be dropped by the ACL")
+
+	trailing := trailingFragment(t, 0x3333, 5, false, 24)
+	require.True(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment of a denied datagram must be dropped")
+}
+
+// TestFragment_OverlappingHeaderDropped covers the RFC 1858 §4 / RFC 3128
+// overlapping-fragment rewrite: a trailing fragment starting inside the range
+// the ACL already inspected is dropped and poisons the datagram. TCP is used so
+// the overlap lands on real header bytes (the flags at byte 13).
+func TestFragment_OverlappingHeaderDropped(t *testing.T) {
+	m := newFragmentTestManager(t)
+	_, err := m.AddPeerFiltering(nil, net.ParseIP(fragTestSrc), fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{8080}}, fw.ActionAccept, "")
+	require.NoError(t, err)
+
+	// First fragment: TCP header (20) + 12 data = 32 bytes -> headerEnd = 4 octets.
+	first := firstFragmentTCP(t, 0x4444, 8080)
+	require.False(t, m.filterInbound(first, len(first)))
+
+	// Overlapping fragment at offset 1 (byte 8) falls inside the inspected TCP
+	// header, so it could rewrite the flags or port on reassembly.
+	overlap := trailingFragmentTo(t, fragTestDst, layers.IPProtocolTCP, 0x4444, 1, true, 32)
+	require.True(t, m.filterInbound(overlap, len(overlap)),
+		"fragment overlapping the inspected header must be dropped")
+
+	// The datagram is now poisoned: a later, non-overlapping fragment is also
+	// dropped because the verdict was removed.
+	later := trailingFragmentTo(t, fragTestDst, layers.IPProtocolTCP, 0x4444, 4, false, 24)
+	require.True(t, m.filterInbound(later, len(later)),
+		"fragments after an overlap must be dropped (datagram poisoned)")
+}
+
+// TestFragment_OffsetZeroOverlapPoisons covers the RFC 3128 offset-zero rewrite:
+// an allowed first fragment followed by a denied offset-zero fragment for the
+// same datagram must not leave the earlier allow verdict in place.
+func TestFragment_OffsetZeroOverlapPoisons(t *testing.T) {
+	m := newFragmentTestManager(t)
+	allowUDP(t, m, 8080)
+
+	allowed := firstFragmentUDP(t, 0x5A5A, 8080, 32)
+	require.False(t, m.filterInbound(allowed, len(allowed)),
+		"allowed first fragment should pass and be recorded")
+
+	// A second offset-zero fragment to a denied port supersedes the datagram's
+	// verdict; it is dropped and must not leave the allow in place.
+	denied := firstFragmentUDP(t, 0x5A5A, 9999, 32)
+	require.True(t, m.filterInbound(denied, len(denied)),
+		"denied offset-zero fragment must be dropped")
+
+	trailing := trailingFragment(t, 0x5A5A, 5, false, 24)
+	require.True(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment must be denied after the datagram was poisoned")
+}
+
+// TestFragment_TinyFirstDropped covers the tiny-fragment attack: a first
+// fragment too small to contain the full transport header can't be
+// ACL-evaluated and must be dropped.
+func TestFragment_TinyFirstDropped(t *testing.T) {
+	m := newFragmentTestManager(t)
+	allowUDP(t, m, 8080)
+
+	// IPv4 header + 4 raw bytes, MF set, offset 0: too small for the 8-byte UDP
+	// header, so it decodes to L3 only.
+	tiny := trailingFragment(t, 0x5555, 0, true, 4)
+	require.True(t, m.filterInbound(tiny, len(tiny)),
+		"tiny first fragment without a full L4 header must be dropped")
+}
+
+// TestFragment_TCPFirstFragment verifies the TCP arm of the transport decode: a
+// first fragment carrying the full 20-byte TCP header is ACL-evaluated and its
+// trailing fragments inherit the verdict.
+func TestFragment_TCPFirstFragment(t *testing.T) {
+	m := newFragmentTestManager(t)
+	_, err := m.AddPeerFiltering(nil, net.ParseIP(fragTestSrc), fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{8080}}, fw.ActionAccept, "")
+	require.NoError(t, err)
+
+	// TCP header (20) + 12 data = 32 bytes -> headerEnd = 4 octets.
+	first := firstFragmentTCP(t, 0x6666, 8080)
+	require.False(t, m.filterInbound(first, len(first)),
+		"allowed TCP first fragment should pass and be recorded")
+
+	trailing := trailingFragmentTo(t, fragTestDst, layers.IPProtocolTCP, 0x6666, 4, false, 24)
+	require.False(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment of an allowed TCP datagram should pass")
+}
+
+// TestFragment_TCPTinyFirstDropped verifies the TCP minimum header length: 12
+// bytes would satisfy a UDP header but falls short of the 20-byte TCP header.
+func TestFragment_TCPTinyFirstDropped(t *testing.T) {
+	m := newFragmentTestManager(t)
+	_, err := m.AddPeerFiltering(nil, net.ParseIP(fragTestSrc), fw.ProtocolTCP, nil,
+		&fw.Port{Values: []uint16{8080}}, fw.ActionAccept, "")
+	require.NoError(t, err)
+
+	tiny := trailingFragmentTo(t, fragTestDst, layers.IPProtocolTCP, 0x7777, 0, true, 12)
+	require.True(t, m.filterInbound(tiny, len(tiny)),
+		"first fragment shorter than the TCP header must be dropped")
+}
+
+// TestFragment_ConntrackAllowsFirstFragment verifies the conntrack branch: reply
+// fragments of an outbound-established UDP flow pass without any inbound rule.
+func TestFragment_ConntrackAllowsFirstFragment(t *testing.T) {
+	m := newFragmentTestManager(t)
+
+	out := outboundUDPPacket(t, 12345, 40000)
+	require.False(t, m.filterOutbound(out, len(out)))
+
+	first := firstFragmentUDP(t, 0x8888, 12345, 32)
+	require.False(t, m.filterInbound(first, len(first)),
+		"reply first fragment should pass via conntrack")
+
+	trailing := trailingFragment(t, 0x8888, 5, false, 24)
+	require.False(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment of a tracked flow should pass")
+}
+
+// TestFragment_RoutingDisabledDropsFragment verifies routed first fragments are
+// dropped when routing is disabled.
+func TestFragment_RoutingDisabledDropsFragment(t *testing.T) {
+	m := newFragmentTestManager(t)
+	m.routingEnabled.Store(false)
+
+	first := firstFragmentUDPTo(t, "198.51.100.10", 0x9999, 8080, 32)
+	require.True(t, m.filterInbound(first, len(first)),
+		"routed first fragment must be dropped when routing is disabled")
+}
+
+// TestFragment_RouteACL verifies the route-ACL branch: fragments to a non-local
+// destination follow the route rules, allowed datagrams pass their trailing
+// fragments and denied ones don't.
+func TestFragment_RouteACL(t *testing.T) {
+	m := newFragmentTestManager(t)
+	m.routingEnabled.Store(true)
+	m.nativeRouter.Store(false)
+
+	_, err := m.AddRouteFiltering(
+		[]byte("rt-1"),
+		[]netip.Prefix{netip.MustParsePrefix("100.10.0.0/16")},
+		fw.Network{Prefix: netip.MustParsePrefix("198.51.100.0/24")},
+		fw.ProtocolUDP,
+		nil,
+		&fw.Port{Values: []uint16{8080}},
+		fw.ActionAccept,
+	)
+	require.NoError(t, err)
+
+	first := firstFragmentUDPTo(t, "198.51.100.10", 0xAAAA, 8080, 32)
+	require.False(t, m.filterInbound(first, len(first)),
+		"route-ACL-allowed first fragment should pass")
+	trailing := trailingFragmentTo(t, "198.51.100.10", layers.IPProtocolUDP, 0xAAAA, 5, false, 24)
+	require.False(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment of an allowed routed datagram should pass")
+
+	denied := firstFragmentUDPTo(t, "198.51.100.10", 0xBBBB, 9999, 32)
+	require.True(t, m.filterInbound(denied, len(denied)),
+		"route-ACL-denied first fragment must be dropped")
+	deniedTrailing := trailingFragmentTo(t, "198.51.100.10", layers.IPProtocolUDP, 0xBBBB, 5, false, 24)
+	require.True(t, m.filterInbound(deniedTrailing, len(deniedTrailing)),
+		"trailing fragment of a denied routed datagram must be dropped")
+}
+
+// TestFragment_ExpiredVerdictDropsTrailing verifies a verdict older than the
+// tracker timeout no longer admits trailing fragments.
+func TestFragment_ExpiredVerdictDropsTrailing(t *testing.T) {
+	m := newFragmentTestManager(t)
+	allowUDP(t, m, 8080)
+
+	first := firstFragmentUDP(t, 0xCCCC, 8080, 32)
+	require.False(t, m.filterInbound(first, len(first)))
+
+	m.fragments.mutex.Lock()
+	for key, entry := range m.fragments.entries {
+		entry.recordedAt = time.Now().Add(-defaultFragmentTimeout - time.Second)
+		m.fragments.entries[key] = entry
+	}
+	m.fragments.mutex.Unlock()
+
+	trailing := trailingFragment(t, 0xCCCC, 5, false, 24)
+	require.True(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment after verdict expiry must be dropped")
+}
+
+// TestFragment_CapacityFailsClosed verifies the table cap: at capacity, new
+// datagram verdicts are not recorded (their trailing fragments are dropped)
+// while already-recorded datagrams keep working.
+func TestFragment_CapacityFailsClosed(t *testing.T) {
+	m := newFragmentTestManager(t)
+	allowUDP(t, m, 8080)
+
+	m.fragments.mutex.Lock()
+	m.fragments.maxEntries = 1
+	m.fragments.mutex.Unlock()
+
+	first1 := firstFragmentUDP(t, 0x0101, 8080, 32)
+	require.False(t, m.filterInbound(first1, len(first1)))
+
+	first2 := firstFragmentUDP(t, 0x0202, 8080, 32)
+	require.False(t, m.filterInbound(first2, len(first2)),
+		"first fragment itself still passes at capacity")
+
+	trailing2 := trailingFragment(t, 0x0202, 5, false, 24)
+	require.True(t, m.filterInbound(trailing2, len(trailing2)),
+		"trailing fragment of an unrecorded datagram must be dropped at capacity")
+
+	trailing1 := trailingFragment(t, 0x0101, 5, false, 24)
+	require.False(t, m.filterInbound(trailing1, len(trailing1)),
+		"already-recorded datagram should keep passing at capacity")
+}
+
+// v6FragmentHeader builds the 8-byte IPv6 fragment extension header for the
+// given inner protocol, offset (8-byte units), More Fragments bit and id.
+func v6FragmentHeader(proto layers.IPProtocol, offsetOctets uint16, moreFragments bool, id uint32) []byte {
+	offsetFlags := offsetOctets << 3
+	if moreFragments {
+		offsetFlags |= 1
+	}
+	hdr := make([]byte, 8)
+	hdr[0] = uint8(proto)
+	binary.BigEndian.PutUint16(hdr[2:4], offsetFlags)
+	binary.BigEndian.PutUint32(hdr[4:8], id)
+	return hdr
+}
+
+func v6UDPHeader(dstPort uint16, dataLen int) []byte {
+	hdr := make([]byte, 8)
+	binary.BigEndian.PutUint16(hdr[0:2], 40000)
+	binary.BigEndian.PutUint16(hdr[2:4], dstPort)
+	binary.BigEndian.PutUint16(hdr[4:6], uint16(8+dataLen))
+	return hdr
+}
+
+// firstFragmentUDPv6 builds the first fragment of a fragmented IPv6 UDP
+// datagram: fragment header (offset 0, More Fragments set) + full UDP header +
+// data.
+func firstFragmentUDPv6(tb testing.TB, id uint32, dstPort uint16, dataLen int) []byte {
+	tb.Helper()
+	return fragmentUDPv6(tb, id, dstPort, dataLen, true)
+}
+
+// fragmentUDPv6 builds an offset-zero IPv6 UDP fragment. With moreFragments
+// false it is an atomic fragment (a complete datagram, RFC 6946).
+func fragmentUDPv6(tb testing.TB, id uint32, dstPort uint16, dataLen int, moreFragments bool) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv6{
+		Version:    6,
+		NextHeader: layers.IPProtocolIPv6Fragment,
+		HopLimit:   64,
+		SrcIP:      net.ParseIP(fragTestSrcV6),
+		DstIP:      net.ParseIP(fragTestDstV6),
+	}
+	payload := append(v6FragmentHeader(layers.IPProtocolUDP, 0, moreFragments, id), v6UDPHeader(dstPort, dataLen)...)
+	payload = append(payload, make([]byte, dataLen)...)
+
+	buf := gopacket.NewSerializeBuffer()
+	require.NoError(tb, gopacket.SerializeLayers(buf, gopacket.SerializeOptions{FixLengths: true}, ip, gopacket.Payload(payload)))
+	return buf.Bytes()
+}
+
+// trailingFragmentV6 builds a non-first IPv6 fragment: fragment header at the
+// given offset carrying raw data and no transport header.
+func trailingFragmentV6(tb testing.TB, id uint32, offsetOctets uint16, moreFragments bool, dataLen int) []byte {
+	tb.Helper()
+
+	ip := &layers.IPv6{
+		Version:    6,
+		NextHeader: layers.IPProtocolIPv6Fragment,
+		HopLimit:   64,
+		SrcIP:      net.ParseIP(fragTestSrcV6),
+		DstIP:      net.ParseIP(fragTestDstV6),
+	}
+	payload := append(v6FragmentHeader(layers.IPProtocolUDP, offsetOctets, moreFragments, id), make([]byte, dataLen)...)
+
+	buf := gopacket.NewSerializeBuffer()
+	require.NoError(tb, gopacket.SerializeLayers(buf, gopacket.SerializeOptions{FixLengths: true}, ip, gopacket.Payload(payload)))
+	return buf.Bytes()
+}
+
+// TestFragmentV6_TrailingWithoutFirstDropped verifies the IPv6 bypass is closed:
+// a trailing fragment with no allowed first fragment is dropped.
+func TestFragmentV6_TrailingWithoutFirstDropped(t *testing.T) {
+	m := newFragmentTestManager(t)
+
+	frag := trailingFragmentV6(t, 0xAABBCCDD, 100, false, 40)
+	require.True(t, m.filterInbound(frag, len(frag)),
+		"IPv6 trailing fragment without an allowed first fragment must be dropped")
+}
+
+// TestFragmentV6_AllowedFirstPassesTrailing verifies IPv6 fragments are
+// evaluated like IPv4: an allowed first fragment lets its trailing fragments
+// through.
+func TestFragmentV6_AllowedFirstPassesTrailing(t *testing.T) {
+	m := newFragmentTestManager(t)
+	_, err := m.AddPeerFiltering(nil, net.ParseIP(fragTestSrcV6), fw.ProtocolUDP, nil,
+		&fw.Port{Values: []uint16{8080}}, fw.ActionAccept, "")
+	require.NoError(t, err)
+
+	// First fragment: UDP header (8) + 32 data = 40 octets -> headerEnd = 5.
+	first := firstFragmentUDPv6(t, 0xAABBCCDD, 8080, 32)
+	require.False(t, m.filterInbound(first, len(first)),
+		"allowed IPv6 first fragment should pass and be recorded")
+
+	trailing := trailingFragmentV6(t, 0xAABBCCDD, 5, false, 24)
+	require.False(t, m.filterInbound(trailing, len(trailing)),
+		"trailing fragment of an allowed IPv6 datagram should pass")
+}
+
+// TestFragmentV6_AtomicNotCached verifies an IPv6 atomic fragment (fragment
+// header with offset 0 and no More Fragments, a complete datagram per RFC 6946)
+// is evaluated but not recorded, so a flood of allowed atomic fragments can't
+// exhaust the verdict table.
+func TestFragmentV6_AtomicNotCached(t *testing.T) {
+	m := newFragmentTestManager(t)
+	_, err := m.AddPeerFiltering(nil, net.ParseIP(fragTestSrcV6), fw.ProtocolUDP, nil,
+		&fw.Port{Values: []uint16{8080}}, fw.ActionAccept, "")
+	require.NoError(t, err)
+
+	atomic := fragmentUDPv6(t, 0xA70301C, 8080, 16, false)
+	require.False(t, m.filterInbound(atomic, len(atomic)),
+		"allowed IPv6 atomic fragment should pass")
+
+	m.fragments.mutex.Lock()
+	n := len(m.fragments.entries)
+	m.fragments.mutex.Unlock()
+	require.Zero(t, n, "atomic fragment must not create a verdict entry")
+
+	// A genuine fragmented datagram (More Fragments set) is still recorded.
+	first := fragmentUDPv6(t, 0xBEEF, 8080, 32, true)
+	require.False(t, m.filterInbound(first, len(first)))
+	m.fragments.mutex.Lock()
+	n = len(m.fragments.entries)
+	m.fragments.mutex.Unlock()
+	require.Equal(t, 1, n, "genuine first fragment must record a verdict")
+}

client/iface/iface_test.go

@@ -464,6 +464,8 @@ func Test_RemovePeer(t *testing.T) {
 }
 
 func Test_ConnectPeers(t *testing.T) {
+	t.Setenv("NB_DISABLE_EBPF_WG_PROXY", "true")
+
 	peer1ifaceName := fmt.Sprintf("utun%d", WgIntNumber+400)
 	peer1wgIP := netip.MustParsePrefix("10.99.99.17/30")
 	peer1Key, _ := wgtypes.GeneratePrivateKey()
@@ -505,12 +507,8 @@ func Test_ConnectPeers(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	localIP, err := getLocalIP()
-	if err != nil {
-		t.Fatal(err)
-	}
-
-	peer1endpoint, err := net.ResolveUDPAddr("udp", fmt.Sprintf("%s:%d", localIP, peer1wgPort))
+	localIP1 := "127.0.0.1"
+	peer1endpoint, err := net.ResolveUDPAddr("udp", fmt.Sprintf("%s:%d", localIP1, peer1wgPort))
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -546,7 +544,8 @@ func Test_ConnectPeers(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	peer2endpoint, err := net.ResolveUDPAddr("udp", fmt.Sprintf("%s:%d", localIP, peer2wgPort))
+	localIP2 := "127.0.0.1"
+	peer2endpoint, err := net.ResolveUDPAddr("udp", fmt.Sprintf("%s:%d", localIP2, peer2wgPort))
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -569,17 +568,17 @@ func Test_ConnectPeers(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	// todo: investigate why in some tests execution we need 30s
+	// The peers use userspace WireGuard (stdnet transport). A tight busy-loop
+	// here starves the wireguard-go goroutines that process the handshake, so
+	// poll on a ticker instead and yield the CPU between checks. WireGuard also
+	// only retries a lost handshake initiation every REKEY_TIMEOUT (5s), which
+	// is why the overall wait can occasionally stretch to tens of seconds.
 	timeout := 30 * time.Second
 	timeoutChannel := time.After(timeout)
+	ticker := time.NewTicker(500 * time.Millisecond)
+	defer ticker.Stop()
 
 	for {
-		select {
-		case <-timeoutChannel:
-			t.Fatalf("waiting for peer handshake timeout after %s", timeout.String())
-		default:
-		}
-
 		peer, gpErr := getPeer(peer1ifaceName, peer2Key.PublicKey().String())
 		if gpErr != nil {
 			t.Fatal(gpErr)
@@ -588,6 +587,12 @@ func Test_ConnectPeers(t *testing.T) {
 			t.Log("peers successfully handshake")
 			break
 		}
+
+		select {
+		case <-timeoutChannel:
+			t.Fatalf("waiting for peer handshake timeout after %s", timeout.String())
+		case <-ticker.C:
+		}
 	}
 
 }
@@ -615,28 +620,3 @@ func getPeer(ifaceName, peerPubKey string) (wgtypes.Peer, error) {
 	}
 	return wgtypes.Peer{}, fmt.Errorf("peer not found")
 }
-
-func getLocalIP() (string, error) {
-	// Get all interfaces
-	addrs, err := net.InterfaceAddrs()
-	if err != nil {
-		return "", err
-	}
-
-	for _, addr := range addrs {
-		ipNet, ok := addr.(*net.IPNet)
-		if !ok {
-			continue
-		}
-		if ipNet.IP.IsLoopback() {
-			continue
-		}
-
-		if ipNet.IP.To4() == nil {
-			continue
-		}
-		return ipNet.IP.String(), nil
-	}
-
-	return "", fmt.Errorf("no local IP found")
-}

client/iface/netstack/env.go

@@ -3,14 +3,31 @@
 package netstack
 
 import (
-	"fmt"
+	"net"
 	"os"
 	"strconv"
 
 	log "github.com/sirupsen/logrus"
 )
 
-const EnvUseNetstackMode = "NB_USE_NETSTACK_MODE"
+const (
+	EnvUseNetstackMode = "NB_USE_NETSTACK_MODE"
+
+	// EnvSocks5ListenerPort overrides the port the SOCKS5 proxy listens on.
+	EnvSocks5ListenerPort = "NB_SOCKS5_LISTENER_PORT"
+
+	// EnvSocks5ListenerAddress overrides the host/IP the SOCKS5 proxy binds to.
+	// The proxy is a bridge for local host applications into the userspace
+	// WireGuard netstack, so it binds to loopback by default. Override this only
+	// when the proxy must be reachable from other hosts (e.g. a container
+	// gateway); doing so exposes an unauthenticated SOCKS5 proxy on that
+	// address.
+	EnvSocks5ListenerAddress = "NB_SOCKS5_LISTENER_ADDRESS"
+
+	// defaultSocks5Host is the loopback address the SOCKS5 proxy binds to unless
+	// overridden via EnvSocks5ListenerAddress.
+	defaultSocks5Host = "127.0.0.1"
+)
 
 // IsEnabled todo: move these function to cmd layer
 func IsEnabled() bool {
@@ -18,24 +35,40 @@ func IsEnabled() bool {
 }
 
 func ListenAddr() string {
-	sPort := os.Getenv("NB_SOCKS5_LISTENER_PORT")
+	return net.JoinHostPort(listenHost(), strconv.Itoa(listenPort()))
+}
+
+// listenHost returns the host/IP the SOCKS5 proxy binds to. It defaults to
+// loopback and only honors EnvSocks5ListenerAddress when it holds a valid IP.
+func listenHost() string {
+	addr := os.Getenv(EnvSocks5ListenerAddress)
+	if addr == "" {
+		return defaultSocks5Host
+	}
+	if net.ParseIP(addr) == nil {
+		log.Warnf("invalid socks5 listener address %q, falling back to default: %s", addr, defaultSocks5Host)
+		return defaultSocks5Host
+	}
+	return addr
+}
+
+// listenPort returns the port the SOCKS5 proxy binds to, defaulting to
+// DefaultSocks5Port when EnvSocks5ListenerPort is unset or invalid.
+func listenPort() int {
+	sPort := os.Getenv(EnvSocks5ListenerPort)
 	if sPort == "" {
-		return listenAddr(DefaultSocks5Port)
+		return DefaultSocks5Port
 	}
 
 	port, err := strconv.Atoi(sPort)
 	if err != nil {
 		log.Warnf("invalid socks5 listener port, unable to convert it to int, falling back to default: %d", DefaultSocks5Port)
-		return listenAddr(DefaultSocks5Port)
+		return DefaultSocks5Port
 	}
 	if port < 1 || port > 65535 {
 		log.Warnf("invalid socks5 listener port, it should be in the range 1-65535, falling back to default: %d", DefaultSocks5Port)
-		return listenAddr(DefaultSocks5Port)
+		return DefaultSocks5Port
 	}
 
-	return listenAddr(port)
-}
-
-func listenAddr(port int) string {
-	return fmt.Sprintf("0.0.0.0:%d", port)
+	return port
 }

client/iface/netstack/env_test.go

@@ -0,0 +1,63 @@
+//go:build !js
+
+package netstack
+
+import (
+	"net"
+	"strconv"
+	"testing"
+)
+
+func TestListenAddr_DefaultsToLoopback(t *testing.T) {
+	// No env overrides: must bind loopback, never all interfaces.
+	got := ListenAddr()
+	want := net.JoinHostPort("127.0.0.1", strconv.Itoa(DefaultSocks5Port))
+	if got != want {
+		t.Fatalf("ListenAddr() = %q, want %q", got, want)
+	}
+}
+
+func TestListenAddr_AddressOverride(t *testing.T) {
+	tests := []struct {
+		name string
+		env  string
+		want string
+	}{
+		{name: "valid override honored", env: "0.0.0.0", want: "0.0.0.0"},
+		{name: "valid specific ip honored", env: "10.0.0.5", want: "10.0.0.5"},
+		{name: "ipv6 loopback bracketed", env: "::1", want: "::1"},
+		{name: "invalid falls back to loopback", env: "not-an-ip", want: "127.0.0.1"},
+		{name: "empty falls back to loopback", env: "", want: "127.0.0.1"},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			t.Setenv(EnvSocks5ListenerAddress, tc.env)
+			want := net.JoinHostPort(tc.want, strconv.Itoa(DefaultSocks5Port))
+			if got := ListenAddr(); got != want {
+				t.Fatalf("ListenAddr() = %q, want %q", got, want)
+			}
+		})
+	}
+}
+
+func TestListenAddr_PortOverride(t *testing.T) {
+	tests := []struct {
+		name string
+		env  string
+		want int
+	}{
+		{name: "valid port honored", env: "1081", want: 1081},
+		{name: "non-numeric falls back", env: "abc", want: DefaultSocks5Port},
+		{name: "out of range falls back", env: "70000", want: DefaultSocks5Port},
+		{name: "zero falls back", env: "0", want: DefaultSocks5Port},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			t.Setenv(EnvSocks5ListenerPort, tc.env)
+			want := net.JoinHostPort("127.0.0.1", strconv.Itoa(tc.want))
+			if got := ListenAddr(); got != want {
+				t.Fatalf("ListenAddr() = %q, want %q", got, want)
+			}
+		})
+	}
+}

client/internal/auth/auth.go

@@ -351,6 +351,7 @@ func (a *Auth) setSystemInfoFlags(info *system.Info) {
 		a.config.BlockLANAccess,
 		a.config.BlockInbound,
 		a.config.DisableIPv6,
+		a.config.SyncMessageVersion,
 		a.config.EnableSSHRoot,
 		a.config.EnableSSHSFTP,
 		a.config.EnableSSHLocalPortForwarding,

client/internal/auth/device_flow.go

@@ -299,7 +299,7 @@ func (d *DeviceAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowIn
 				UseIDToken:   d.providerConfig.UseIDToken,
 			}
 
-			err = isValidAccessToken(tokenInfo.GetTokenToUse(), d.providerConfig.Audience)
+			err = validateTokenAudience(tokenInfo.GetTokenToUse(), d.providerConfig.Audience)
 			if err != nil {
 				return TokenInfo{}, fmt.Errorf("validate access token failed with error: %v", err)
 			}

client/internal/auth/pkce_flow.go

@@ -306,7 +306,7 @@ func (p *PKCEAuthorizationFlow) parseOAuthToken(token *oauth2.Token) (TokenInfo,
 		audience = p.providerConfig.ClientID
 	}
 
-	if err := isValidAccessToken(tokenInfo.GetTokenToUse(), audience); err != nil {
+	if err := validateTokenAudience(tokenInfo.GetTokenToUse(), audience); err != nil {
 		return TokenInfo{}, fmt.Errorf("authentication failed: invalid access token - %w", err)
 	}
 
@@ -320,6 +320,11 @@ func (p *PKCEAuthorizationFlow) parseOAuthToken(token *oauth2.Token) (TokenInfo,
 	return tokenInfo, nil
 }
 
+// parseEmailFromIDToken extracts the email (or name) claim from an ID token
+// without verifying its signature. The value is best-effort and used only as a
+// UX convenience (login hint prefill and display); it never drives an
+// authorization decision. The authoritative identity is established server-side
+// from the signature-verified token.
 func parseEmailFromIDToken(token string) (string, error) {
 	parts := strings.Split(token, ".")
 	if len(parts) < 2 {

client/internal/auth/sessionwatch/watcher.go

@@ -24,11 +24,7 @@ import (
 )
 
 const (
-	// Skew tolerates a small clock difference between the management
-	// server and this peer before treating a deadline as "in the past".
-	// Slightly above typical NTP drift; tight enough that the UI doesn't
-	// paint a stale expiry as if it were valid.
-	Skew = 30 * time.Second
+	maxPastHorizon = 30 * 24 * time.Hour
 
 	// maxDeadlineHorizon caps how far in the future an accepted deadline
 	// can sit. A timestamp beyond this is almost certainly a protocol
@@ -57,7 +53,7 @@ var (
 	ErrDeadlineTooFarFuture = errors.New("session deadline too far in the future")
 
 	// ErrDeadlineInPast is returned by Update when the supplied deadline
-	// is more than Skew in the past.
+	// is more than maxPastHorizon in the past.
 	ErrDeadlineInPast = errors.New("session deadline in the past")
 )
 
@@ -66,15 +62,14 @@ var (
 // for deadline change/clear, PublishEvent for the two warnings); tests pass
 // a fake recorder so the same surface is observable without an engine.
 //
-// The watcher is the single owner of the deadline propagated to the
-// recorder: every set, clear, sanity-check rejection and Close routes the
-// value through SetSessionExpiresAt, so the SubscribeStatus snapshot the UI
-// reads can never drift from the watcher's timer state. (SetSessionExpiresAt
-// fans out its own state-change notification, so no separate notify is
-// needed.) The recorder is server-scoped and outlives this engine-scoped
-// watcher — without the Close-time clear a teardown (Down, or the Down+Up of
-// a profile switch) would leave the next session showing the previous one's
-// stale "expires in" value.
+// While the watcher runs, it owns the deadline propagated to the recorder:
+// every set, clear and sanity-check rejection routes the value through
+// SetSessionExpiresAt, so the SubscribeStatus snapshot the UI reads can
+// never drift from the watcher's timer state. (SetSessionExpiresAt fans
+// out its own state-change notification, so no separate notify is needed.)
+// The recorder is server-scoped and outlives this engine-scoped watcher;
+// Close deliberately leaves the recorder value in place so transient engine
+// restarts don't blank it — the client run loop clears it on real teardown.
 //
 // PublishEvent's signature mirrors peer.Status.PublishEvent: the watcher
 // composes the metadata internally so the wire format (MetaSession*) is
@@ -135,10 +130,13 @@ func NewWithLeads(lead, final time.Duration, recorder StatusRecorder) *Watcher {
 // was disabled).
 //
 // Same-value updates are no-ops. A different non-zero value cancels any
-// pending timer, resets the "already fired" guard, and arms a new one.
+// pending timer, resets the "already fired" guards, and — when the
+// deadline lies in the future — arms fresh warning timers. A deadline
+// already in the past (within maxPastHorizon) is recorded as-is with no
+// timers: the session has expired and consumers render it that way.
 //
 // Returns one of the sentinel Err* values when the deadline fails the
-// sanity checks (pre-epoch, far future, or in the past beyond Skew).
+// sanity checks (pre-epoch, far future, or past beyond maxPastHorizon).
 // In every error case the watcher first clears its state so it stays
 // consistent with what the caller will push into its other sinks (e.g.
 // applySessionDeadline forces a zero deadline into the status recorder
@@ -163,7 +161,7 @@ func (w *Watcher) Update(deadline time.Time) error {
 	case deadline.After(now.Add(maxDeadlineHorizon)):
 		w.clearLocked()
 		return fmt.Errorf("%w: %v", ErrDeadlineTooFarFuture, deadline)
-	case deadline.Before(now.Add(-Skew)):
+	case deadline.Before(now.Add(-maxPastHorizon)):
 		w.clearLocked()
 		return fmt.Errorf("%w: %v (now=%v)", ErrDeadlineInPast, deadline, now)
 	}
@@ -183,7 +181,9 @@ func (w *Watcher) Update(deadline time.Time) error {
 	w.finalFiredAt = time.Time{}
 	w.dismissedAt = time.Time{}
 
-	w.armTimerLocked(deadline)
+	if deadline.After(now) {
+		w.armTimerLocked(deadline)
+	}
 	recorder := w.recorder
 	w.mu.Unlock()
 	if recorder != nil {
@@ -227,30 +227,25 @@ func (w *Watcher) Dismiss() {
 	log.Infof("auth session final-warning dismissed for deadline %s", w.current.Format(time.RFC3339))
 }
 
-// Close stops any pending timer and drops the deadline on the status
-// recorder. Update calls after Close are ignored. Clearing the recorder
-// here is what keeps a teardown (Down, or the Down+Up of a profile switch)
-// from leaving the next session showing this one's stale "expires in"
-// value — the recorder is server-scoped and outlives this engine-scoped
-// watcher, so nothing else drops the anchor on teardown.
+// Close stops any pending timer. Update calls after Close are ignored.
+// The recorder keeps its deadline: the watcher is engine-scoped and closes
+// on every engine restart (network change, sleep/wake, stream errors)
+// while the SSO deadline stays valid across those, so clearing here would
+// blank the UI's "expires in" row on every transient reconnect. The
+// client run loop clears the server-scoped recorder when it exits for
+// real (Down, profile switch, permanent login failure).
 func (w *Watcher) Close() {
 	w.mu.Lock()
+	defer w.mu.Unlock()
 	if w.closed {
-		w.mu.Unlock()
 		return
 	}
 	w.closed = true
 	w.stopTimerLocked()
-	hadDeadline := !w.current.IsZero()
 	w.current = time.Time{}
 	w.firedAt = time.Time{}
 	w.finalFiredAt = time.Time{}
 	w.dismissedAt = time.Time{}
-	recorder := w.recorder
-	w.mu.Unlock()
-	if recorder != nil && hadDeadline {
-		recorder.SetSessionExpiresAt(time.Time{})
-	}
 }
 
 // clearLocked drops the tracked deadline and notifies the recorder so

client/internal/auth/sessionwatch/watcher_test.go

@@ -224,11 +224,13 @@ func TestNewDeadlineCancelsPriorTimer(t *testing.T) {
 
 func TestRefreshAfterFireArmsNewWarning(t *testing.T) {
 	r := &fakeRecorder{}
-	lead := 30 * time.Millisecond
+	lead := 150 * time.Millisecond
 	w := newWatcher(lead, r)
 	defer w.Close()
 
-	first := time.Now().Add(50 * time.Millisecond)
+	// Warning fires ~20ms in; the deadline itself stays 150ms away so the
+	// replacement below lands well before it.
+	first := time.Now().Add(170 * time.Millisecond)
 	_ = w.Update(first)
 
 	// Wait for stateChange + warning of the first cycle.
@@ -306,7 +308,29 @@ func TestUpdateRejectsTooFarFuture(t *testing.T) {
 	}
 }
 
-func TestUpdateInPastClearsDeadline(t *testing.T) {
+func TestUpdateRecentPastRecordedAsExpired(t *testing.T) {
+	r := &fakeRecorder{}
+	w := newWatcher(50*time.Millisecond, r)
+	defer w.Close()
+
+	d := time.Now().Add(-1 * time.Hour)
+	if err := w.Update(d); err != nil {
+		t.Fatalf("recent-past Update should succeed, got %v", err)
+	}
+	if !w.Deadline().Equal(d) {
+		t.Fatalf("expected deadline to be recorded, got %v want %v", w.Deadline(), d)
+	}
+	if got := r.deadline(); !got.Equal(d) {
+		t.Fatalf("recorder deadline = %v, want %v", got, d)
+	}
+
+	time.Sleep(80 * time.Millisecond)
+	if n := countWhere(r.snapshot(), func(e event) bool { return e.kind == publish }); n != 0 {
+		t.Fatalf("no warning events may fire for an already-past deadline, got %+v", r.snapshot())
+	}
+}
+
+func TestUpdateAncientPastRejected(t *testing.T) {
 	r := &fakeRecorder{}
 	w := newWatcher(50*time.Millisecond, r)
 	defer w.Close()
@@ -318,12 +342,12 @@ func TestUpdateInPastClearsDeadline(t *testing.T) {
 	// Drain the stateChange from the seed.
 	waitForEvents(t, r, 1)
 
-	err := w.Update(time.Now().Add(-1 * time.Hour))
+	err := w.Update(time.Now().Add(-31 * 24 * time.Hour))
 	if !errors.Is(err, ErrDeadlineInPast) {
 		t.Fatalf("want ErrDeadlineInPast, got %v", err)
 	}
 	if !w.Deadline().IsZero() {
-		t.Fatalf("in-past update must clear the deadline, got %v", w.Deadline())
+		t.Fatalf("rejected ancient-past update must clear the deadline, got %v", w.Deadline())
 	}
 	events := waitForEvents(t, r, 2)
 	if events[1].kind != stateChange {
@@ -331,39 +355,25 @@ func TestUpdateInPastClearsDeadline(t *testing.T) {
 	}
 }
 
-func TestUpdateWithinSkewAccepted(t *testing.T) {
-	r := &fakeRecorder{}
-	w := newWatcher(50*time.Millisecond, r)
-	defer w.Close()
-
-	// 5 seconds in the past is within the 30s Skew tolerance — accept it.
-	d := time.Now().Add(-5 * time.Second)
-	if err := w.Update(d); err != nil {
-		t.Fatalf("within-skew Update should succeed, got %v", err)
-	}
-	if !w.Deadline().Equal(d) {
-		t.Fatalf("expected deadline to be applied, got %v want %v", w.Deadline(), d)
-	}
-}
-
 func TestCloseSilencesUpdates(t *testing.T) {
 	r := &fakeRecorder{}
 	w := newWatcher(50*time.Millisecond, r)
 	w.Close()
 
-	_ = w.Update(time.Now().Add(time.Hour))
-
-	time.Sleep(20 * time.Millisecond)
+	if err := w.Update(time.Now().Add(time.Hour)); err != nil {
+		t.Fatalf("Update after Close: want nil, got %v", err)
+	}
 	if got := r.snapshot(); len(got) != 0 {
 		t.Fatalf("expected no events after Close, got %+v", got)
 	}
 }
 
-// TestCloseClearsRecorderDeadline pins the profile-switch fix: a watcher
-// holding a live deadline must zero the recorder on Close so the next
-// engine's watcher (and the UI reading the shared server-scoped recorder)
-// doesn't start out showing the previous session's stale "expires in".
-func TestCloseClearsRecorderDeadline(t *testing.T) {
+// TestCloseKeepsRecorderDeadline pins the reconnect-flap fix: the watcher
+// closes on every engine restart (network change, sleep/wake) while the
+// SSO deadline stays valid across those, so Close must leave the
+// server-scoped recorder's value in place. The client run loop clears the
+// recorder when it exits for real.
+func TestCloseKeepsRecorderDeadline(t *testing.T) {
 	r := &fakeRecorder{}
 	w := newWatcher(time.Hour, r)
 
@@ -377,8 +387,8 @@ func TestCloseClearsRecorderDeadline(t *testing.T) {
 
 	w.Close()
 
-	if got := r.deadline(); !got.IsZero() {
-		t.Fatalf("recorder deadline after Close = %v, want zero", got)
+	if got := r.deadline(); !got.Equal(d) {
+		t.Fatalf("recorder deadline after Close = %v, want %v", got, d)
 	}
 }
 

client/internal/auth/util.go

@@ -20,14 +20,26 @@ func randomBytesInHex(count int) (string, error) {
 	return hex.EncodeToString(buf), nil
 }
 
-// isValidAccessToken is a simple validation of the access token
-func isValidAccessToken(token string, audience string) error {
+// validateTokenAudience checks that the token is a well-formed JWT whose
+// audience claim matches the expected audience.
+//
+// It does NOT verify the token's cryptographic signature and therefore must not
+// be treated as an authenticity check. The token is obtained by the client
+// directly from the IdP token endpoint over TLS, and its signature is verified
+// server-side by the management server against the IdP's JWKS
+// (see shared/auth/jwt/validator.go). This function is only a client-side
+// sanity check that the returned token targets the expected audience.
+func validateTokenAudience(token string, audience string) error {
 	if token == "" {
 		return fmt.Errorf("token received is empty")
 	}
 
-	encodedClaims := strings.Split(token, ".")[1]
-	claimsString, err := base64.RawURLEncoding.DecodeString(encodedClaims)
+	parts := strings.Split(token, ".")
+	if len(parts) != 3 {
+		return fmt.Errorf("token is not a well-formed JWT")
+	}
+
+	claimsString, err := base64.RawURLEncoding.DecodeString(parts[1])
 	if err != nil {
 		return err
 	}

client/internal/auth/util_test.go

@@ -0,0 +1,108 @@
+package auth
+
+import (
+	"encoding/base64"
+	"encoding/json"
+	"testing"
+)
+
+// makeJWT builds an unsigned JWT-shaped string (header.payload.signature) with
+// the given claims payload. The signature part is arbitrary because
+// validateTokenAudience intentionally does not verify it.
+func makeJWT(t *testing.T, claims map[string]interface{}) string {
+	t.Helper()
+	header := base64.RawURLEncoding.EncodeToString([]byte(`{"alg":"RS256","typ":"JWT"}`))
+	payloadBytes, err := json.Marshal(claims)
+	if err != nil {
+		t.Fatalf("marshal claims: %v", err)
+	}
+	payload := base64.RawURLEncoding.EncodeToString(payloadBytes)
+	return header + "." + payload + ".unverified-signature"
+}
+
+func TestValidateTokenAudience(t *testing.T) {
+	tests := []struct {
+		name     string
+		token    string
+		audience string
+		wantErr  bool
+	}{
+		{
+			name:     "empty token",
+			token:    "",
+			audience: "netbird",
+			wantErr:  true,
+		},
+		{
+			name:     "not a JWT - no dots",
+			token:    "notajwt",
+			audience: "netbird",
+			wantErr:  true,
+		},
+		{
+			name:     "not a JWT - two parts only",
+			token:    "header.payload",
+			audience: "netbird",
+			wantErr:  true,
+		},
+		{
+			name:     "matching string audience",
+			token:    makeJWT(t, map[string]interface{}{"aud": "netbird"}),
+			audience: "netbird",
+			wantErr:  false,
+		},
+		{
+			name:     "mismatching string audience",
+			token:    makeJWT(t, map[string]interface{}{"aud": "other"}),
+			audience: "netbird",
+			wantErr:  true,
+		},
+		{
+			name:     "matching audience in array",
+			token:    makeJWT(t, map[string]interface{}{"aud": []interface{}{"other", "netbird"}}),
+			audience: "netbird",
+			wantErr:  false,
+		},
+		{
+			name:     "mismatching audience array",
+			token:    makeJWT(t, map[string]interface{}{"aud": []interface{}{"a", "b"}}),
+			audience: "netbird",
+			wantErr:  true,
+		},
+		{
+			name:     "missing audience claim",
+			token:    makeJWT(t, map[string]interface{}{"sub": "user"}),
+			audience: "netbird",
+			wantErr:  true,
+		},
+		{
+			name:     "invalid base64 payload",
+			token:    "header.!!!not-base64!!!.sig",
+			audience: "netbird",
+			wantErr:  true,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			err := validateTokenAudience(tc.token, tc.audience)
+			if tc.wantErr && err == nil {
+				t.Fatalf("expected error, got nil")
+			}
+			if !tc.wantErr && err != nil {
+				t.Fatalf("expected no error, got %v", err)
+			}
+		})
+	}
+}
+
+// TestValidateTokenAudienceNoPanic guards the regression where a non-empty
+// token without the JWT dot structure caused an index-out-of-range panic.
+func TestValidateTokenAudienceNoPanic(t *testing.T) {
+	inputs := []string{"a", ".", "a.", "aaaa", "no-dots-here"}
+	for _, in := range inputs {
+		if err := validateTokenAudience(in, "netbird"); err == nil {
+			t.Fatalf("expected error for malformed token %q, got nil", in)
+		}
+	}
+}

client/internal/conn_mgr.go

@@ -34,6 +34,8 @@ const (
 // - Handling connection establishment based on peer signaling
 //
 // The implementation is not thread-safe; it is protected by engine.syncMsgMux.
+// The only exception is ActivatePeer, which is safe for concurrent use so the
+// DNS warm-up path can call it without contending on the engine mutex.
 type ConnMgr struct {
 	peerStore        *peerstore.Store
 	statusRecorder   *peer.Status
@@ -42,12 +44,26 @@ type ConnMgr struct {
 	rosenpassEnabled bool
 
 	lazyConnMgr *manager.Manager
+	// lazyConnMgrMu guards the lazyConnMgr pointer for readers outside the
+	// engine loop (ActivatePeer). Writers hold it in addition to
+	// engine.syncMsgMux; all other reads stay under engine.syncMsgMux only.
+	lazyConnMgrMu sync.RWMutex
+
+	// reconcileRoutedIPs re-applies a peer's routed allowed IPs after its lazy wake endpoint is
+	// (re)armed (Mode A at arm time). Injected by the engine; nil disables the reconcile.
+	reconcileRoutedIPs func(peerKey string) error
 
 	wg            sync.WaitGroup
 	lazyCtx       context.Context
 	lazyCtxCancel context.CancelFunc
 }
 
+// SetRoutedIPsReconciler injects the callback used to re-apply a peer's routed allowed IPs when
+// its lazy wake endpoint is (re)armed. Must be called before the lazy manager starts.
+func (e *ConnMgr) SetRoutedIPsReconciler(fn func(peerKey string) error) {
+	e.reconcileRoutedIPs = fn
+}
+
 func NewConnMgr(engineConfig *EngineConfig, statusRecorder *peer.Status, peerStore *peerstore.Store, iface lazyconn.WGIface) *ConnMgr {
 	e := &ConnMgr{
 		peerStore:        peerStore,
@@ -238,12 +254,20 @@ func (e *ConnMgr) RemovePeerConn(peerKey string) {
 	conn.Log.Infof("removed peer from lazy conn manager")
 }
 
+// ActivatePeer wakes an idle lazy connection. Unlike the rest of ConnMgr it is
+// safe for concurrent use: the lazy manager pointer is read under lazyConnMgrMu
+// and the manager itself is internally synchronized, so callers outside the
+// engine loop (DNS warm-up) do not need engine.syncMsgMux.
 func (e *ConnMgr) ActivatePeer(ctx context.Context, conn *peer.Conn) {
-	if !e.isStartedWithLazyMgr() {
+	e.lazyConnMgrMu.RLock()
+	lazyConnMgr := e.lazyConnMgr
+	started := lazyConnMgr != nil && e.lazyCtxCancel != nil
+	e.lazyConnMgrMu.RUnlock()
+	if !started {
 		return
 	}
 
-	if found := e.lazyConnMgr.ActivatePeer(conn.GetKey()); found {
+	if found := lazyConnMgr.ActivatePeer(conn.GetKey()); found {
 		if err := conn.Open(ctx); err != nil {
 			conn.Log.Errorf("failed to open connection: %v", err)
 		}
@@ -268,16 +292,22 @@ func (e *ConnMgr) Close() {
 
 	e.lazyCtxCancel()
 	e.wg.Wait()
+
+	e.lazyConnMgrMu.Lock()
 	e.lazyConnMgr = nil
+	e.lazyConnMgrMu.Unlock()
 }
 
 func (e *ConnMgr) initLazyManager(engineCtx context.Context) {
 	cfg := manager.Config{
 		InactivityThreshold: inactivityThresholdEnv(),
+		ReconcileAllowedIPs: e.reconcileRoutedIPs,
 	}
-	e.lazyConnMgr = manager.NewManager(cfg, engineCtx, e.peerStore, e.iface)
 
+	e.lazyConnMgrMu.Lock()
+	e.lazyConnMgr = manager.NewManager(cfg, engineCtx, e.peerStore, e.iface)
 	e.lazyCtx, e.lazyCtxCancel = context.WithCancel(engineCtx)
+	e.lazyConnMgrMu.Unlock()
 
 	e.wg.Add(1)
 	go func() {
@@ -316,7 +346,10 @@ func (e *ConnMgr) closeManager(ctx context.Context) {
 
 	e.lazyCtxCancel()
 	e.wg.Wait()
+
+	e.lazyConnMgrMu.Lock()
 	e.lazyConnMgr = nil
+	e.lazyConnMgrMu.Unlock()
 
 	for _, peerID := range e.peerStore.PeersPubKey() {
 		e.peerStore.PeerConnOpen(ctx, peerID)

client/internal/conn_mgr_test.go

@@ -1,10 +1,21 @@
 package internal
 
 import (
+	"context"
+	"net"
+	"net/netip"
 	"os"
+	"sync"
 	"testing"
+	"time"
 
+	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
 	"github.com/netbirdio/netbird/client/internal/lazyconn"
+	"github.com/netbirdio/netbird/client/internal/peer"
+	"github.com/netbirdio/netbird/client/internal/peerstore"
+	"github.com/netbirdio/netbird/monotime"
 )
 
 func TestResolveLazyForce(t *testing.T) {
@@ -38,3 +49,58 @@ func TestResolveLazyForce(t *testing.T) {
 		})
 	}
 }
+
+type mockLazyWGIface struct{}
+
+func (mockLazyWGIface) RemovePeer(string) error { return nil }
+func (mockLazyWGIface) UpdatePeer(string, []netip.Prefix, time.Duration, *net.UDPAddr, *wgtypes.Key) error {
+	return nil
+}
+func (mockLazyWGIface) IsUserspaceBind() bool                    { return false }
+func (mockLazyWGIface) Address() wgaddr.Address                  { return wgaddr.Address{} }
+func (mockLazyWGIface) LastActivities() map[string]monotime.Time { return nil }
+func (mockLazyWGIface) MTU() uint16                              { return 1280 }
+
+// TestConnMgr_ActivatePeerConcurrentWithLifecycle exercises ActivatePeer from
+// non-engine goroutines (the DNS warm-up path) racing the manager lifecycle,
+// which stays on the engine loop. Run with -race: it fails if ActivatePeer
+// still requires engine.syncMsgMux for safety.
+func TestConnMgr_ActivatePeerConcurrentWithLifecycle(t *testing.T) {
+	t.Setenv(lazyconn.EnvLazyConn, "on")
+
+	status := peer.NewRecorder("https://mgm")
+	store := peerstore.NewConnStore()
+	connMgr := NewConnMgr(&EngineConfig{}, status, store, mockLazyWGIface{})
+
+	conn := newTestPeerConn(t, "peerA")
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	connMgr.Start(ctx)
+
+	done := make(chan struct{})
+	var wg sync.WaitGroup
+	for range 4 {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			for {
+				select {
+				case <-done:
+					return
+				default:
+					connMgr.ActivatePeer(ctx, conn)
+				}
+			}
+		}()
+	}
+
+	// Let the activators spin against the started manager, then tear it down
+	// underneath them and let them spin against the stopped manager.
+	time.Sleep(100 * time.Millisecond)
+	connMgr.Close()
+	time.Sleep(50 * time.Millisecond)
+
+	close(done)
+	wg.Wait()
+}

client/internal/connect.go

@@ -257,7 +257,10 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 		log.Errorf("failed to clean up temporary installer file: %v", err)
 	}
 
-	defer c.statusRecorder.ClientStop()
+	defer func() {
+		c.statusRecorder.SetSessionExpiresAt(time.Time{})
+		c.statusRecorder.ClientStop()
+	}()
 	operation := func() error {
 		// if context cancelled we not start new backoff cycle
 		if c.ctx.Err() != nil {
@@ -618,6 +621,7 @@ func createEngineConfig(key wgtypes.Key, config *profilemanager.Config, peerConf
 		BlockLANAccess:      config.BlockLANAccess,
 		BlockInbound:        config.BlockInbound,
 		DisableIPv6:         config.DisableIPv6,
+		SyncMessageVersion:  config.SyncMessageVersion,
 
 		LazyConnection: lazyconn.ParseState(config.LazyConnection),
 
@@ -693,6 +697,7 @@ func loginToManagement(ctx context.Context, client mgm.Client, pubSSHKey []byte,
 		config.BlockLANAccess,
 		config.BlockInbound,
 		config.DisableIPv6,
+		config.SyncMessageVersion,
 		config.EnableSSHRoot,
 		config.EnableSSHSFTP,
 		config.EnableSSHLocalPortForwarding,

client/internal/debug/debug.go

@@ -676,6 +676,7 @@ 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("SyncMessageVersion: %v\n", g.internalConfig.SyncMessageVersion))
 
 	if g.internalConfig.DisableNotifications != nil {
 		configContent.WriteString(fmt.Sprintf("DisableNotifications: %v\n", *g.internalConfig.DisableNotifications))

client/internal/debug/debug_test.go

@@ -887,6 +887,8 @@ func TestAddConfig_AllFieldsCovered(t *testing.T) {
 		ClientCertKeyPath:             "/tmp/key",
 		LazyConnection:                "on",
 		MTU:                           1280,
+		DisableIPv6:                   true,
+		SyncMessageVersion:            func(v int) *int { return &v }(1),
 	}
 
 	for _, anonymize := range []bool{false, true} {

client/internal/dns/handler_chain.go

@@ -3,6 +3,7 @@ package dns
 import (
 	"context"
 	"fmt"
+	"maps"
 	"math"
 	"net"
 	"slices"
@@ -31,6 +32,26 @@ type SubdomainMatcher interface {
 	MatchSubdomains() bool
 }
 
+// responseMeta holds the annotations handlers attach to a request to explain the
+// response the chain ends up writing. It survives a deferral, so an answer that
+// did not come from the handler that owns the name still says who stepped aside
+// and why.
+type responseMeta map[resutil.MetaKey]string
+
+// format renders the annotations for the response log line. The order is stable
+// so the same event reads the same way every time; a map's own order is not.
+func (m responseMeta) format() string {
+	if len(m) == 0 {
+		return ""
+	}
+
+	var b strings.Builder
+	for _, k := range slices.Sorted(maps.Keys(m)) {
+		b.WriteString(" " + string(k) + "=" + m[k])
+	}
+	return b.String()
+}
+
 type HandlerEntry struct {
 	Handler         dns.Handler
 	Priority        int
@@ -52,8 +73,19 @@ type ResponseWriterChain struct {
 	origPattern    string
 	requestID      string
 	shouldContinue bool
-	response       *dns.Msg
-	meta           map[string]string
+	// softNegative suppresses a poisoning negative verdict for this request. A
+	// handler that owns the name but cannot answer this query type sets it
+	// before deferring, and it stays set for every handler that runs after.
+	softNegative bool
+	// clientHasEdns records whether the original query advertised EDNS0, taken
+	// before any handler ran: handlers add EDNS0 to the query they forward
+	// upstream, so the message itself no longer answers the question later.
+	clientHasEdns bool
+	response      *dns.Msg
+	// meta is handed to the next handler when this one defers, so the same map
+	// outlives the writer that created it. Handlers must only set metadata from
+	// within ServeDNS, never from a goroutine that outlives the call.
+	meta responseMeta
 }
 
 // RequestID returns the request ID for tracing
@@ -62,26 +94,64 @@ func (w *ResponseWriterChain) RequestID() string {
 }
 
 // SetMeta sets a metadata key-value pair for logging
-func (w *ResponseWriterChain) SetMeta(key, value string) {
+func (w *ResponseWriterChain) SetMeta(key resutil.MetaKey, value string) {
 	if w.meta == nil {
-		w.meta = make(map[string]string)
+		w.meta = make(responseMeta)
 	}
 	w.meta[key] = value
 }
 
+// RequestSoftNegative marks the request so a downstream NXDOMAIN is turned into
+// NODATA before it reaches the client. Set by a handler that defers a query for
+// a name it owns but a type it cannot resolve.
+func (w *ResponseWriterChain) RequestSoftNegative() {
+	w.softNegative = true
+}
+
 func (w *ResponseWriterChain) WriteMsg(m *dns.Msg) error {
 	// Check if this is a continue signal (NXDOMAIN with Zero bit set)
 	if m.Rcode == dns.RcodeNameError && m.MsgHdr.Zero {
 		w.shouldContinue = true
 		return nil
 	}
+	if w.softNegative && m.Rcode == dns.RcodeNameError {
+		m = softenNegative(m, w.clientHasEdns)
+		w.SetMeta(resutil.MetaKeySoftened, "nxdomain->nodata")
+	}
 	w.response = m
 	if m.MsgHdr.Truncated {
-		w.SetMeta("truncated", "true")
+		w.SetMeta(resutil.MetaKeyTruncated, "true")
 	}
 	return w.ResponseWriter.WriteMsg(m)
 }
 
+// softenNegative downgrades an NXDOMAIN to NODATA for a request a handler that
+// owns the name deferred. NXDOMAIN is cached for the name and every type below
+// it, so a resolver that has never heard of a routed name would take out the
+// record types the route does serve; NODATA is cached for this name and type
+// only. The authority section goes with it: the negative TTL of a zone we just
+// overruled does not apply, and RFC 2308 keeps a negative answer that carries no
+// SOA out of caches altogether, so the rewrite cannot outlive the route.
+//
+// The rewrite is ours, not the answering resolver's, so an EDNS0 client is told
+// as much: the reply we hand back travels a path no capture on this host sees,
+// and an empty answer is otherwise indistinguishable from a real one. Returns a
+// copy so the answering handler keeps whatever it may hold on to.
+func softenNegative(m *dns.Msg, clientHasEdns bool) *dns.Msg {
+	out := m.Copy()
+	out.Rcode = dns.RcodeSuccess
+	out.Ns = nil
+
+	if clientHasEdns {
+		resutil.AttachEDE(out, resutil.EDENetbirdSoftenedNegative,
+			"netbird: name is served locally, NXDOMAIN from the fallthrough resolver suppressed")
+	} else {
+		resutil.StripOPT(out)
+	}
+
+	return out
+}
+
 func NewHandlerChain() *HandlerChain {
 	return &HandlerChain{
 		handlers: make([]HandlerEntry, 0),
@@ -223,6 +293,19 @@ func (c *HandlerChain) dispatch(w dns.ResponseWriter, r *dns.Msg, maxPriority in
 	handlers := slices.Clone(c.handlers)
 	c.mu.RUnlock()
 
+	// Carried across deferrals: once a handler that owns the name defers, no
+	// handler after it may answer with a verdict that poisons the name. The
+	// metadata of a handler that stepped aside is carried too, so the response
+	// log line explains an answer that did not come from the handler that owns
+	// the name.
+	var softNegative bool
+	var carried responseMeta
+
+	// Taken before any handler runs: handlers advertise EDNS0 on the query they
+	// forward upstream, so afterwards the message no longer tells us whether the
+	// client did.
+	clientHasEdns := r.IsEdns0() != nil
+
 	// Try handlers in priority order
 	for _, entry := range handlers {
 		if entry.Priority > maxPriority {
@@ -245,11 +328,16 @@ func (c *HandlerChain) dispatch(w dns.ResponseWriter, r *dns.Msg, maxPriority in
 			ResponseWriter: w,
 			origPattern:    entry.OrigPattern,
 			requestID:      requestID,
+			softNegative:   softNegative,
+			clientHasEdns:  clientHasEdns,
+			meta:           carried,
 		}
 		entry.Handler.ServeDNS(chainWriter, r)
 
 		// If handler wants to continue, try next handler
 		if chainWriter.shouldContinue {
+			softNegative = softNegative || chainWriter.softNegative
+			carried = chainWriter.meta
 			if entry.Priority != PriorityMgmtCache {
 				logger.Tracef("handler requested continue for domain=%s", qname)
 			}
@@ -265,30 +353,59 @@ func (c *HandlerChain) dispatch(w dns.ResponseWriter, r *dns.Msg, maxPriority in
 		qname, dns.TypeToString[question.Qtype], dns.ClassToString[question.Qclass])
 	resp := &dns.Msg{}
 	resp.SetRcode(r, dns.RcodeRefused)
+	// A handler that owns the name deferred and nothing below it could answer
+	// (a client with no primary nameserver group). The name exists as far as
+	// this client is concerned, since the route serves its addresses, so REFUSED
+	// would contradict the route: a stub that takes it as "not served here" and
+	// retries another resolver can come back with an NXDOMAIN that takes the
+	// whole name down. Answer "no records of this type" instead, without an SOA,
+	// so RFC 2308 keeps it out of negative caches, and tell an EDNS0 client the
+	// empty answer is ours rather than a resolver's.
+	if softNegative {
+		resp.Rcode = dns.RcodeSuccess
+		if clientHasEdns {
+			resutil.AttachEDE(resp, resutil.EDENetbirdSoftenedNegative,
+				"netbird: name is served locally, no fallthrough resolver for this query type")
+		}
+		// logResponse never runs on this path, so the carried metadata is
+		// appended here or the reason for the deferral is lost in exactly the
+		// case that is hardest to diagnose.
+		logger.Tracef("no handler below the deferring one for domain=%s type=%s, answering NODATA%s",
+			qname, dns.TypeToString[question.Qtype], carried.format())
+	}
 	if err := w.WriteMsg(resp); err != nil {
 		logger.Errorf("failed to write DNS response: %v", err)
 	}
 }
 
 func (c *HandlerChain) logResponse(logger *log.Entry, cw *ResponseWriterChain, qname string, startTime time.Time) {
+	// Runs for every query, and the arguments below are not free: Len() packs
+	// the message to measure it, and formatting the answers and the metadata
+	// allocates. None of it is worth doing when the line is discarded.
+	if !log.IsLevelEnabled(log.TraceLevel) {
+		return
+	}
+
 	if cw.response == nil {
 		return
 	}
 
-	var meta string
-	for k, v := range cw.meta {
-		meta += " " + k + "=" + v
-	}
-
 	logger.Tracef("response: domain=%s rcode=%s answers=%s size=%dB%s took=%s",
 		qname, dns.RcodeToString[cw.response.Rcode], resutil.FormatAnswers(cw.response.Answer),
-		cw.response.Len(), meta, time.Since(startTime))
+		cw.response.Len(), cw.meta.format(), time.Since(startTime))
 }
 
 // ResolveInternal runs an in-process DNS query against the chain, skipping any
 // handler with priority > maxPriority. Used by internal callers (e.g. the mgmt
-// cache refresher) that must bypass themselves to avoid loops. Honors ctx
-// cancellation; on ctx.Done the dispatch goroutine is left to drain on its own
+// cache refresher) that must bypass themselves to avoid loops.
+//
+// "Nothing answered" is read off RcodeRefused, which a request soft-negatived by
+// a deferring handler never carries: it ends in an empty NOERROR instead, and
+// would look resolved. No caller can reach that today, since every handler that
+// defers sits above the maxPriority any caller passes. Lowering one below it
+// means this check needs the soft-negative case too.
+//
+// Honors ctx cancellation; on ctx.Done the dispatch goroutine is left to drain on its own
 // (bounded by the invoked handler's internal timeout).
 func (c *HandlerChain) ResolveInternal(ctx context.Context, r *dns.Msg, maxPriority int) (*dns.Msg, error) {
 	if len(r.Question) == 0 {

client/internal/dns/handler_chain_test.go

@@ -3,15 +3,19 @@ package dns_test
 import (
 	"context"
 	"net"
+	"strings"
 	"testing"
 	"time"
 
 	"github.com/miekg/dns"
+	log "github.com/sirupsen/logrus"
+	logtest "github.com/sirupsen/logrus/hooks/test"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/mock"
 	"github.com/stretchr/testify/require"
 
 	nbdns "github.com/netbirdio/netbird/client/internal/dns"
+	"github.com/netbirdio/netbird/client/internal/dns/resutil"
 	"github.com/netbirdio/netbird/client/internal/dns/test"
 )
 
@@ -1238,6 +1242,276 @@ func TestHandlerChain_ResolveInternal_HonorsContextTimeout(t *testing.T) {
 	assert.Less(t, elapsed, 500*time.Millisecond, "ResolveInternal must return shortly after ctx deadline")
 }
 
+// requestSoftNegative asks the chain to soften a negative verdict produced by
+// the handlers that run after the caller defers, reporting whether the chain
+// supports the signal. Written as a type assertion so these tests compile
+// against a chain that does not support it yet.
+func requestSoftNegative(w dns.ResponseWriter) bool {
+	sn, ok := w.(interface{ RequestSoftNegative() })
+	if ok {
+		sn.RequestSoftNegative()
+	}
+	return ok
+}
+
+// deferringHandler defers to the next handler in the chain, optionally asking
+// for the negative verdict of whatever answers instead to be softened. This is
+// what a DNS route handler does for a record type its routing peer cannot
+// resolve.
+type deferringHandler struct {
+	softNegative bool
+	called       bool
+	supported    bool
+}
+
+func (h *deferringHandler) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
+	h.called = true
+	if h.softNegative {
+		h.supported = requestSoftNegative(w)
+		// The real handler records why it stepped aside, for the response log
+		// line of whichever handler answers instead.
+		resutil.SetMeta(w, resutil.MetaKeyDeferredBy, "test handler")
+	}
+	resp := new(dns.Msg)
+	resp.SetRcode(r, dns.RcodeNameError)
+	resp.MsgHdr.Zero = true
+	_ = w.WriteMsg(resp)
+}
+
+// nxdomainHandler answers an authoritative NXDOMAIN with an SOA in the
+// authority section, the way a public resolver answers for a name that only
+// exists inside the routed network.
+type nxdomainHandler struct{}
+
+func (h *nxdomainHandler) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
+	resp := new(dns.Msg)
+	resp.SetRcode(r, dns.RcodeNameError)
+	resp.Ns = []dns.RR{&dns.SOA{
+		Hdr:     dns.RR_Header{Name: "example.com.", Rrtype: dns.TypeSOA, Class: dns.ClassINET, Ttl: 3600},
+		Ns:      "ns1.example.com.",
+		Mbox:    "hostmaster.example.com.",
+		Minttl:  3600,
+		Expire:  604800,
+		Refresh: 7200,
+		Retry:   3600,
+	}}
+	_ = w.WriteMsg(resp)
+}
+
+// TestHandlerChain_SoftNegative_DowngradesDownstreamNXDOMAIN is the whole point
+// of the soft-negative signal: a route handler may only defer a query to the
+// public chain if the answer cannot poison the routed name. NXDOMAIN is cached
+// for the name and every type under it (RFC 2308, RFC 8020), so it has to be
+// rewritten to NODATA, which is cached per name and type only.
+func TestHandlerChain_SoftNegative_DowngradesDownstreamNXDOMAIN(t *testing.T) {
+	chain := nbdns.NewHandlerChain()
+
+	route := &deferringHandler{softNegative: true}
+	chain.AddHandler("*.example.com.", route, nbdns.PriorityDNSRoute)
+	chain.AddHandler(".", &nxdomainHandler{}, nbdns.PriorityDefault)
+
+	r := new(dns.Msg)
+	r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+
+	mw := &test.MockResponseWriter{}
+	chain.ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+	resp := mw.GetLastResponse()
+	require.NotNil(t, resp, "a response must reach the client")
+	assert.True(t, route.called, "the route handler must run first")
+	require.True(t, route.supported, "the chain writer must accept a soft-negative request")
+	assert.Equal(t, dns.RcodeSuccess, resp.Rcode, "NXDOMAIN must be softened to NODATA")
+	assert.Empty(t, resp.Answer, "a softened negative carries no answer")
+	assert.Empty(t, resp.Ns, "the downstream zone's SOA must not set the negative TTL for a name we overrode")
+}
+
+// responseLineFor returns the chain's response log line for one query. Raising
+// the level to trace also unmutes whatever else is logging in this package, so
+// the line has to be picked by the name it was asked about rather than by being
+// the last one seen.
+func responseLineFor(hook *logtest.Hook, qname string) string {
+	for _, e := range hook.AllEntries() {
+		if strings.HasPrefix(e.Message, "response:") && strings.Contains(e.Message, qname) {
+			return e.Message
+		}
+	}
+	return ""
+}
+
+// TestHandlerChain_SoftNegative_IsVisibleToTheClient covers observability of the
+// rewrite. The reply we hand the application travels over loopback, which the
+// bundle capture does not see, so a softened verdict has to say so on the wire:
+// without it an empty answer is indistinguishable from a real "no such record"
+// in a dig output or a capture taken next to the application.
+func TestHandlerChain_SoftNegative_IsVisibleToTheClient(t *testing.T) {
+	// One hook for the whole test: logtest installs it on the standard logger
+	// and logrus has no way to take it off again, so a hook per subtest would
+	// leave several behind buffering every later line in the package.
+	hook := logtest.NewGlobal()
+	t.Cleanup(hook.Reset)
+
+	newChain := func() *nbdns.HandlerChain {
+		chain := nbdns.NewHandlerChain()
+		chain.AddHandler("*.example.com.", &deferringHandler{softNegative: true}, nbdns.PriorityDNSRoute)
+		chain.AddHandler(".", &nxdomainHandler{}, nbdns.PriorityDefault)
+		return chain
+	}
+
+	t.Run("EDNS0 client gets an extended error", func(t *testing.T) {
+		r := new(dns.Msg)
+		r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+		r.SetEdns0(dns.DefaultMsgSize, false)
+
+		mw := &test.MockResponseWriter{}
+		newChain().ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+		resp := mw.GetLastResponse()
+		require.NotNil(t, resp)
+		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
+
+		ede, ok := resutil.ExtractEDE(resp)
+		require.True(t, ok, "a softened verdict must carry an extended DNS error")
+		assert.Equal(t, resutil.EDENetbirdSoftenedNegative, ede.InfoCode)
+		assert.Contains(t, ede.ExtraText, "netbird", "the text must name us as the source of the rewrite")
+	})
+
+	t.Run("plain client gets no OPT", func(t *testing.T) {
+		r := new(dns.Msg)
+		r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+
+		mw := &test.MockResponseWriter{}
+		newChain().ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+		resp := mw.GetLastResponse()
+		require.NotNil(t, resp)
+		assert.Nil(t, resp.IsEdns0(), "RFC 6891 forbids an OPT toward a client that did not advertise EDNS0")
+	})
+
+	// The response line is what support reads out of a debug bundle, and it now
+	// carries several annotations at once. Built from a map, their order would
+	// differ on every query, so the same event never looks the same twice.
+	t.Run("log fields keep a stable order", func(t *testing.T) {
+		const qname = "_mongodb._tcp.stable.example.com."
+
+		prev := log.GetLevel()
+		log.SetLevel(log.TraceLevel)
+		t.Cleanup(func() { log.SetLevel(prev) })
+
+		lineFor := func() string {
+			hook.Reset()
+
+			r := new(dns.Msg)
+			r.SetQuestion(qname, dns.TypeSRV)
+
+			mw := &test.MockResponseWriter{}
+			newChain().ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+			line := responseLineFor(hook, qname)
+			// The duration differs per query and is not what we compare.
+			line, _, _ = strings.Cut(line, " took=")
+			return line
+		}
+
+		first := lineFor()
+		require.NotEmpty(t, first, "the chain must log the response it wrote")
+		require.Contains(t, first, "deferred_by=", "the line must carry more than one annotation to be worth ordering")
+		require.Contains(t, first, "softened=")
+
+		for range 20 {
+			assert.Equal(t, first, lineFor(), "the same event must produce the same line")
+		}
+	})
+
+	t.Run("logged with the reason it was deferred", func(t *testing.T) {
+		const qname = "_mongodb._tcp.reason.example.com."
+
+		hook.Reset()
+
+		prev := log.GetLevel()
+		log.SetLevel(log.TraceLevel)
+		t.Cleanup(func() { log.SetLevel(prev) })
+
+		r := new(dns.Msg)
+		r.SetQuestion(qname, dns.TypeSRV)
+
+		mw := &test.MockResponseWriter{}
+		newChain().ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+		response := responseLineFor(hook, qname)
+		require.NotEmpty(t, response, "the chain must log the response it wrote")
+		assert.Contains(t, response, "softened=", "the log must show the verdict was rewritten")
+		assert.Contains(t, response, "deferred_by=", "the log must name the handler that deferred")
+	})
+}
+
+// TestHandlerChain_SoftNegative_NoHandlerBelow covers a client with no primary
+// nameserver group: the deferred query reaches the end of the chain unanswered.
+// REFUSED would say the name is not served here while the route serves its
+// addresses, and a stub that acts on that by asking elsewhere can bring back an
+// NXDOMAIN for the whole name. The answer must be an empty, uncacheable NODATA.
+func TestHandlerChain_SoftNegative_NoHandlerBelow(t *testing.T) {
+	chain := nbdns.NewHandlerChain()
+
+	route := &deferringHandler{softNegative: true}
+	chain.AddHandler("*.example.com.", route, nbdns.PriorityDNSRoute)
+
+	r := new(dns.Msg)
+	r.SetQuestion("db.example.com.", dns.TypeHTTPS)
+
+	mw := &test.MockResponseWriter{}
+	chain.ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+	resp := mw.GetLastResponse()
+	require.NotNil(t, resp, "a response must reach the client")
+	assert.Equal(t, dns.RcodeSuccess, resp.Rcode, "an unanswered soft-negative query must be NODATA, not REFUSED")
+	assert.Empty(t, resp.Answer)
+	assert.Empty(t, resp.Ns,
+		"no SOA, so RFC 2308 keeps the empty answer out of negative caches and it cannot outlive the route")
+}
+
+// TestHandlerChain_SoftNegative_KeepsRealAnswers guards the other direction:
+// softening applies to negative verdicts only. A real answer from a downstream
+// handler must reach the client untouched.
+func TestHandlerChain_SoftNegative_KeepsRealAnswers(t *testing.T) {
+	chain := nbdns.NewHandlerChain()
+
+	route := &deferringHandler{softNegative: true}
+	chain.AddHandler("*.example.com.", route, nbdns.PriorityDNSRoute)
+	chain.AddHandler(".", &answeringHandler{name: "public", ip: "203.0.113.10"}, nbdns.PriorityDefault)
+
+	r := new(dns.Msg)
+	r.SetQuestion("db.example.com.", dns.TypeA)
+
+	mw := &test.MockResponseWriter{}
+	chain.ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+	resp := mw.GetLastResponse()
+	require.NotNil(t, resp, "a response must reach the client")
+	assert.Equal(t, dns.RcodeSuccess, resp.Rcode)
+	require.Len(t, resp.Answer, 1, "the downstream answer must pass through")
+}
+
+// TestHandlerChain_NXDOMAINPreservedWithoutSoftNegative makes sure the
+// softening is opt-in: an ordinary chain continuation still yields NXDOMAIN, so
+// genuine non-existence keeps being reported.
+func TestHandlerChain_NXDOMAINPreservedWithoutSoftNegative(t *testing.T) {
+	chain := nbdns.NewHandlerChain()
+
+	route := &deferringHandler{}
+	chain.AddHandler("*.example.com.", route, nbdns.PriorityDNSRoute)
+	chain.AddHandler(".", &nxdomainHandler{}, nbdns.PriorityDefault)
+
+	r := new(dns.Msg)
+	r.SetQuestion("nope.example.com.", dns.TypeA)
+
+	mw := &test.MockResponseWriter{}
+	chain.ServeDNS(&nbdns.ResponseWriterChain{ResponseWriter: mw}, r)
+
+	resp := mw.GetLastResponse()
+	require.NotNil(t, resp, "a response must reach the client")
+	assert.Equal(t, dns.RcodeNameError, resp.Rcode, "without the signal a real NXDOMAIN must survive")
+}
+
 func TestHandlerChain_HasRootHandlerAtOrBelow(t *testing.T) {
 	chain := nbdns.NewHandlerChain()
 	h := &answeringHandler{name: "h", ip: "10.0.0.1"}

client/internal/dns/local/local.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"net"
 	"net/netip"
+	"os"
 	"slices"
 	"strings"
 	"sync"
@@ -36,7 +37,43 @@ type resolver interface {
 // record is left alone (it points at something outside our mesh, e.g.
 // a non-peer upstream).
 type PeerConnectivity interface {
-	IsConnectedByIP(ip string) (known, connected bool)
+	IsConnectedByIP(ip netip.Addr) (known, connected bool)
+}
+
+// PeerActivator wakes lazy-connection peers on demand. The local resolver calls
+// it with the tunnel IPs an answer points at, so a peer that is idle (lazily
+// disconnected) starts connecting at DNS-resolution time rather than racing the
+// client's first request packet. nil disables warm-up.
+type PeerActivator interface {
+	// ActivatePeersByIP triggers wake-up for the peer(s) owning addrs and blocks
+	// until one is connected or ctx (a short per-query budget) expires. It is a
+	// fast no-op for unknown or already-connected addresses.
+	ActivatePeersByIP(ctx context.Context, addrs []netip.Addr)
+}
+
+const (
+	defaultLazyWarmupTimeout = 2 * time.Second
+	envLazyWarmupTimeout     = "NB_DNS_LAZY_WARMUP_TIMEOUT"
+)
+
+// lazyWarmupTimeoutFromEnv returns the per-query budget for waking a
+// lazy-connection peer a DNS answer points at. Tunable via
+// NB_DNS_LAZY_WARMUP_TIMEOUT (a Go duration). Parsed once at construction time.
+func lazyWarmupTimeoutFromEnv() time.Duration {
+	v := os.Getenv(envLazyWarmupTimeout)
+	if v == "" {
+		return defaultLazyWarmupTimeout
+	}
+	d, err := time.ParseDuration(v)
+	if err != nil {
+		log.Warnf("invalid %s value %q, using default %s: %v", envLazyWarmupTimeout, v, defaultLazyWarmupTimeout, err)
+		return defaultLazyWarmupTimeout
+	}
+	if d <= 0 {
+		log.Warnf("non-positive %s value %q, using default %s", envLazyWarmupTimeout, v, defaultLazyWarmupTimeout)
+		return defaultLazyWarmupTimeout
+	}
+	return d
 }
 
 type Resolver struct {
@@ -51,6 +88,12 @@ type Resolver struct {
 	// filter and preserves the legacy "return whatever is registered"
 	// behaviour for callers that never wire a status source.
 	peerConn PeerConnectivity
+	// peerActivator, when non-nil, is called at resolution time to warm the
+	// lazy connection to the peer(s) an answer points at. nil disables warm-up.
+	peerActivator PeerActivator
+	// warmupTimeout is the per-query budget for the lazy-connection warm-up
+	// wait, resolved from the environment once at construction time.
+	warmupTimeout time.Duration
 
 	ctx    context.Context
 	cancel context.CancelFunc
@@ -59,11 +102,12 @@ type Resolver struct {
 func NewResolver() *Resolver {
 	ctx, cancel := context.WithCancel(context.Background())
 	return &Resolver{
-		records: make(map[dns.Question][]dns.RR),
-		domains: make(map[domain.Domain]struct{}),
-		zones:   make(map[domain.Domain]bool),
-		ctx:     ctx,
-		cancel:  cancel,
+		records:       make(map[dns.Question][]dns.RR),
+		domains:       make(map[domain.Domain]struct{}),
+		zones:         make(map[domain.Domain]bool),
+		warmupTimeout: lazyWarmupTimeoutFromEnv(),
+		ctx:           ctx,
+		cancel:        cancel,
 	}
 }
 
@@ -76,6 +120,14 @@ func (d *Resolver) SetPeerConnectivity(p PeerConnectivity) {
 	d.peerConn = p
 }
 
+// SetPeerActivator wires the DNS-time lazy-connection warm-up. Pass nil to
+// disable. Safe to call multiple times; the latest value wins.
+func (d *Resolver) SetPeerActivator(a PeerActivator) {
+	d.mu.Lock()
+	defer d.mu.Unlock()
+	d.peerActivator = a
+}
+
 func (d *Resolver) MatchSubdomains() bool {
 	return true
 }
@@ -122,6 +174,9 @@ func (d *Resolver) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 	replyMessage.RecursionAvailable = true
 
 	result := d.lookupRecords(logger, question)
+	// Warm before filtering: activation flips a lazily-idle target to connected,
+	// which then lets it survive the disconnected-peer filter below.
+	d.warmLazyPeers(question, result.records)
 	result.records = d.filterDisconnectedPeerAnswers(logger, question, result.records)
 	replyMessage.Authoritative = !result.hasExternalData
 	replyMessage.Answer = result.records
@@ -495,8 +550,8 @@ func (d *Resolver) filterDisconnectedPeerAnswers(logger *log.Entry, question dns
 	kept := make([]dns.RR, 0, len(records))
 	var dropped int
 	for _, rr := range records {
-		ip := extractRecordIP(rr)
-		if ip == "" {
+		ip, ok := extractRecordAddr(rr)
+		if !ok {
 			kept = append(kept, rr)
 			continue
 		}
@@ -518,22 +573,57 @@ func (d *Resolver) filterDisconnectedPeerAnswers(logger *log.Entry, question dns
 	return kept
 }
 
-// extractRecordIP returns the dotted-decimal / colon-hex IP carried by
-// an A or AAAA record, or "" for any other record type.
-func extractRecordIP(rr dns.RR) string {
+// warmLazyPeers triggers lazy-connection wake-up for the peers a resolved
+// answer points at and waits briefly for one to connect, so the caller's first
+// request doesn't race the connection establishment. Warm-up is scoped to
+// match-only (non-authoritative) zones — the synthesized private-service zones
+// and user-created zones whose records point at specific peers. The account's
+// peer zone is authoritative, so plain peer-name lookups never trigger warm-up;
+// otherwise resolving any peer's name would wake its idle connection, defeating
+// laziness mesh-wide. No-op when no activator is wired (lazy connections
+// disabled) or the answer carries no peer IPs.
+func (d *Resolver) warmLazyPeers(question dns.Question, records []dns.RR) {
+	if len(records) < 2 {
+		return
+	}
+	d.mu.RLock()
+	activator := d.peerActivator
+	var nonAuth, found bool
+	if activator != nil {
+		nonAuth, found = d.findZone(question.Name)
+	}
+	d.mu.RUnlock()
+	if activator == nil || !found || !nonAuth {
+		return
+	}
+
+	var addrs []netip.Addr
+	for _, rr := range records {
+		if addr, ok := extractRecordAddr(rr); ok {
+			addrs = append(addrs, addr)
+		}
+	}
+	if len(addrs) == 0 {
+		return
+	}
+
+	ctx, cancel := context.WithTimeout(d.ctx, d.warmupTimeout)
+	defer cancel()
+	activator.ActivatePeersByIP(ctx, addrs)
+}
+
+// extractRecordAddr returns the IP address carried by an A or AAAA record.
+// ok is false for any other record type or a record with no address.
+func extractRecordAddr(rr dns.RR) (netip.Addr, bool) {
 	switch r := rr.(type) {
 	case *dns.A:
-		if r.A == nil {
-			return ""
-		}
-		return r.A.String()
+		addr, ok := netip.AddrFromSlice(r.A)
+		return addr.Unmap(), ok
 	case *dns.AAAA:
-		if r.AAAA == nil {
-			return ""
-		}
-		return r.AAAA.String()
+		addr, ok := netip.AddrFromSlice(r.AAAA)
+		return addr.Unmap(), ok
 	}
-	return ""
+	return netip.Addr{}, false
 }
 
 // Update replaces all zones and their records

client/internal/dns/local/local_test.go

@@ -37,8 +37,8 @@ type mockPeerConnectivity struct {
 	byIP map[string]struct{ known, connected bool }
 }
 
-func (m mockPeerConnectivity) IsConnectedByIP(ip string) (known, connected bool) {
-	v, ok := m.byIP[ip]
+func (m mockPeerConnectivity) IsConnectedByIP(ip netip.Addr) (known, connected bool) {
+	v, ok := m.byIP[ip.String()]
 	if !ok {
 		return false, false
 	}

client/internal/dns/local/warmup_test.go

@@ -0,0 +1,204 @@
+package local
+
+import (
+	"context"
+	"net"
+	"net/netip"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/miekg/dns"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/netbirdio/netbird/client/internal/dns/test"
+	nbdns "github.com/netbirdio/netbird/dns"
+)
+
+// recordingActivator records the addresses it was asked to warm and returns
+// immediately, so ServeDNS is not blocked by the test.
+type recordingActivator struct {
+	mu     sync.Mutex
+	called bool
+	addrs  []netip.Addr
+}
+
+func (r *recordingActivator) ActivatePeersByIP(_ context.Context, addrs []netip.Addr) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	r.called = true
+	r.addrs = append(r.addrs, addrs...)
+}
+
+func serveA(t *testing.T, resolver *Resolver, name string) *dns.Msg {
+	t.Helper()
+	var resp *dns.Msg
+	w := &test.MockResponseWriter{WriteMsgFunc: func(m *dns.Msg) error { resp = m; return nil }}
+	resolver.ServeDNS(w, new(dns.Msg).SetQuestion(name, dns.TypeA))
+	return resp
+}
+
+// serviceZone registers rec in a match-only (non-authoritative) zone, the shape
+// the synthesized private-service zones arrive in.
+func serviceZone(t *testing.T, resolver *Resolver, zone string, records ...nbdns.SimpleRecord) {
+	t.Helper()
+	resolver.Update([]nbdns.CustomZone{{
+		Domain:           zone,
+		Records:          records,
+		NonAuthoritative: true,
+	}})
+}
+
+func TestLocalResolver_WarmsLazyPeerOnResolve(t *testing.T) {
+	// Warm-up fires only for multi-record answers (the HA / round-robin shape of
+	// the synthesized private-service zones), so register two peer targets.
+	const name = "svc.proxy.netbird.cloud."
+	recs := []nbdns.SimpleRecord{
+		{Name: name, Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.7"},
+		{Name: name, Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.8"},
+	}
+	resolver := NewResolver()
+	serviceZone(t, resolver, "proxy.netbird.cloud", recs...)
+
+	act := &recordingActivator{}
+	resolver.SetPeerActivator(act)
+
+	resp := serveA(t, resolver, name)
+	require.NotNil(t, resp, "resolver must answer")
+	require.NotEmpty(t, resp.Answer, "answer must carry the A records")
+
+	act.mu.Lock()
+	defer act.mu.Unlock()
+	assert.True(t, act.called, "activator must be invoked for a multi-record service-zone answer")
+	assert.Contains(t, act.addrs, netip.MustParseAddr("100.64.0.7"), "activator must receive the first peer IP")
+	assert.Contains(t, act.addrs, netip.MustParseAddr("100.64.0.8"), "activator must receive the second peer IP")
+}
+
+func TestLocalResolver_NoWarmupForSingleRecord(t *testing.T) {
+	// A single-record answer does not trigger warm-up; the resolver only warms
+	// multi-record answers.
+	rec := nbdns.SimpleRecord{Name: "svc.proxy.netbird.cloud.", Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.7"}
+	resolver := NewResolver()
+	serviceZone(t, resolver, "proxy.netbird.cloud", rec)
+
+	act := &recordingActivator{}
+	resolver.SetPeerActivator(act)
+
+	resp := serveA(t, resolver, rec.Name)
+	require.NotNil(t, resp, "resolver must answer")
+	require.NotEmpty(t, resp.Answer, "answer must carry the A record")
+
+	act.mu.Lock()
+	defer act.mu.Unlock()
+	assert.False(t, act.called, "activator must not be invoked for a single-record answer")
+}
+
+func TestLocalResolver_NoActivatorNoWarmup(t *testing.T) {
+	// With no activator wired the resolver behaves exactly as before.
+	rec := nbdns.SimpleRecord{Name: "svc.proxy.netbird.cloud.", Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.7"}
+	resolver := NewResolver()
+	serviceZone(t, resolver, "proxy.netbird.cloud", rec)
+
+	resp := serveA(t, resolver, rec.Name)
+	require.NotNil(t, resp, "resolver must still answer without an activator")
+	require.NotEmpty(t, resp.Answer, "answer must carry the A record")
+}
+
+func TestLocalResolver_NoWarmupForMissingRecord(t *testing.T) {
+	// A query that resolves to nothing must not invoke the activator (no IPs).
+	resolver := NewResolver()
+	serviceZone(t, resolver, "proxy.netbird.cloud",
+		nbdns.SimpleRecord{Name: "svc.proxy.netbird.cloud.", Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.7"})
+
+	act := &recordingActivator{}
+	resolver.SetPeerActivator(act)
+
+	serveA(t, resolver, "absent.proxy.netbird.cloud.")
+
+	act.mu.Lock()
+	defer act.mu.Unlock()
+	assert.False(t, act.called, "activator must not be invoked when there is no answer")
+}
+
+func TestLocalResolver_NoWarmupInAuthoritativeZone(t *testing.T) {
+	// The account's peer zone is authoritative; resolving a peer's name there
+	// must not wake its lazy connection — warm-up is scoped to match-only
+	// (non-authoritative) zones such as the synthesized private-service zones.
+	// Use a multi-record answer so the authoritative-zone scoping is the only
+	// reason warm-up is skipped, not the single-record guard.
+	const name = "peer.netbird.cloud."
+	recs := []nbdns.SimpleRecord{
+		{Name: name, Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.9"},
+		{Name: name, Type: 1, Class: nbdns.DefaultClass, TTL: 300, RData: "100.64.0.10"},
+	}
+	resolver := NewResolver()
+	resolver.Update([]nbdns.CustomZone{{
+		Domain:  "netbird.cloud",
+		Records: recs,
+	}})
+
+	act := &recordingActivator{}
+	resolver.SetPeerActivator(act)
+
+	resp := serveA(t, resolver, name)
+	require.NotNil(t, resp, "resolver must answer")
+	require.NotEmpty(t, resp.Answer, "answer must carry the A records")
+
+	act.mu.Lock()
+	defer act.mu.Unlock()
+	assert.False(t, act.called, "activator must not be invoked for authoritative-zone answers")
+}
+
+func TestLazyWarmupTimeoutFromEnv(t *testing.T) {
+	tests := []struct {
+		name   string
+		value  string
+		envSet bool
+		want   time.Duration
+	}{
+		{name: "unset uses default", want: defaultLazyWarmupTimeout},
+		{name: "valid overrides", value: "5s", envSet: true, want: 5 * time.Second},
+		{name: "invalid falls back", value: "not-a-duration", envSet: true, want: defaultLazyWarmupTimeout},
+		{name: "negative falls back", value: "-1s", envSet: true, want: defaultLazyWarmupTimeout},
+		{name: "zero falls back", value: "0s", envSet: true, want: defaultLazyWarmupTimeout},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if tt.envSet {
+				t.Setenv(envLazyWarmupTimeout, tt.value)
+			}
+			assert.Equal(t, tt.want, lazyWarmupTimeoutFromEnv())
+			assert.Equal(t, tt.want, NewResolver().warmupTimeout, "constructor must resolve the timeout once")
+		})
+	}
+}
+
+func TestExtractRecordAddr(t *testing.T) {
+	t.Run("A record yields unmapped v4", func(t *testing.T) {
+		// net.ParseIP returns the 16-byte v4-in-v6 form, the same shape
+		// miekg/dns stores after parsing an A record; the extracted address
+		// must compare equal to a plain v4 netip.Addr.
+		addr, ok := extractRecordAddr(&dns.A{A: net.ParseIP("100.64.0.7")})
+		require.True(t, ok)
+		assert.True(t, addr.Is4())
+		assert.Equal(t, netip.MustParseAddr("100.64.0.7"), addr)
+	})
+
+	t.Run("AAAA record yields v6", func(t *testing.T) {
+		addr, ok := extractRecordAddr(&dns.AAAA{AAAA: net.ParseIP("fd00::1")})
+		require.True(t, ok)
+		assert.Equal(t, netip.MustParseAddr("fd00::1"), addr)
+	})
+
+	t.Run("A record without address", func(t *testing.T) {
+		_, ok := extractRecordAddr(&dns.A{})
+		assert.False(t, ok)
+	})
+
+	t.Run("non-address record", func(t *testing.T) {
+		_, ok := extractRecordAddr(&dns.CNAME{Target: "target.netbird.cloud."})
+		assert.False(t, ok)
+	})
+}

client/internal/dns/mock_server.go

@@ -8,6 +8,7 @@ import (
 	"github.com/miekg/dns"
 
 	dnsconfig "github.com/netbirdio/netbird/client/internal/dns/config"
+	"github.com/netbirdio/netbird/client/internal/dns/local"
 	nbdns "github.com/netbirdio/netbird/dns"
 	"github.com/netbirdio/netbird/route"
 	"github.com/netbirdio/netbird/shared/management/domain"
@@ -92,6 +93,11 @@ func (m *MockServer) SetFirewall(Firewall) {
 	// Mock implementation - no-op
 }
 
+// SetPeerActivator mock implementation of SetPeerActivator from Server interface
+func (m *MockServer) SetPeerActivator(local.PeerActivator) {
+	// Mock implementation - no-op
+}
+
 // BeginBatch mock implementation of BeginBatch from Server interface
 func (m *MockServer) BeginBatch() {
 	// Mock implementation - no-op

client/internal/dns/resutil/resolve.go

@@ -19,6 +19,34 @@ import (
 // uses when a resolved host has no addresses of the requested family.
 const errNoSuitableAddress = "no suitable address found"
 
+// Extended DNS Error info codes NetBird emits so a client can see why an answer
+// looks the way it does without reading this peer's logs. They live in the RFC
+// 8914 Private Use range (49152-65535) and are registered here, in one place,
+// because nothing else guarantees two NetBird components pick distinct codes.
+const (
+	// EDENetbirdUpstreamTimeout: a DNS forwarder's upstream did not answer.
+	EDENetbirdUpstreamTimeout uint16 = 49152
+	// EDENetbirdUpstreamFailure: a DNS forwarder's upstream failed.
+	EDENetbirdUpstreamFailure uint16 = 49153
+	// EDENetbirdSoftenedNegative: the empty answer is ours, not the answering
+	// resolver's. A handler that owns the name deferred this query type, and the
+	// negative verdict that came back was downgraded so it cannot poison the
+	// name.
+	EDENetbirdSoftenedNegative uint16 = 49154
+)
+
+// AttachEDE adds an Extended DNS Error (RFC 8914) option to a message, creating
+// the OPT pseudo-record if it has none. Callers must only use it toward a client
+// that advertised EDNS0: per RFC 6891 an OPT must not appear otherwise.
+func AttachEDE(msg *dns.Msg, code uint16, text string) {
+	opt := msg.IsEdns0()
+	if opt == nil {
+		msg.SetEdns0(dns.DefaultMsgSize, false)
+		opt = msg.IsEdns0()
+	}
+	opt.Option = append(opt.Option, &dns.EDNS0_EDE{InfoCode: code, ExtraText: text})
+}
+
 // GenerateRequestID creates a random 8-character hex string for request tracing.
 func GenerateRequestID() string {
 	bytes := make([]byte, 4)
@@ -78,10 +106,38 @@ type resolver interface {
 	LookupNetIP(ctx context.Context, network, host string) ([]netip.Addr, error)
 }
 
+// MetaKey names an annotation a handler attaches to a request to explain the
+// response the chain ends up writing. The set is closed and rendered onto a
+// single log line, so an unrecognized key is a typo inventing a field rather
+// than a new fact.
+type MetaKey string
+
+const (
+	// MetaKeyProtocol: the transport a query arrived on.
+	MetaKeyProtocol MetaKey = "protocol"
+	// MetaKeyUpstream: the upstream that answered.
+	MetaKeyUpstream MetaKey = "upstream"
+	// MetaKeyUpstreamProtocol: the transport used toward that upstream.
+	MetaKeyUpstreamProtocol MetaKey = "upstream_protocol"
+	// MetaKeyPeer: the routing peer whose forwarder answered.
+	MetaKeyPeer MetaKey = "peer"
+	// MetaKeyEDE: an Extended DNS Error carried by the answer.
+	MetaKeyEDE MetaKey = "ede"
+	// MetaKeyTruncated: the answer did not fit and was truncated.
+	MetaKeyTruncated MetaKey = "truncated"
+	// MetaKeySoftened: a negative verdict was rewritten so it cannot poison a
+	// name served locally.
+	MetaKeySoftened MetaKey = "softened"
+	// MetaKeyDeferredBy: the handler that owned the name and stepped aside.
+	MetaKeyDeferredBy MetaKey = "deferred_by"
+	// MetaKeyDeferredReason: why it stepped aside.
+	MetaKeyDeferredReason MetaKey = "deferred_reason"
+)
+
 // chainedWriter is implemented by ResponseWriters that carry request metadata
 type chainedWriter interface {
 	RequestID() string
-	SetMeta(key, value string)
+	SetMeta(key MetaKey, value string)
 }
 
 // GetRequestID extracts a request ID from the ResponseWriter if available,
@@ -96,12 +152,30 @@ func GetRequestID(w dns.ResponseWriter) string {
 }
 
 // SetMeta sets metadata on the ResponseWriter if it supports it.
-func SetMeta(w dns.ResponseWriter, key, value string) {
+func SetMeta(w dns.ResponseWriter, key MetaKey, value string) {
 	if cw, ok := w.(chainedWriter); ok {
 		cw.SetMeta(key, value)
 	}
 }
 
+// softNegativeRequester is implemented by chain writers that can soften the
+// negative verdict of the handlers a deferring handler falls through to.
+type softNegativeRequester interface {
+	RequestSoftNegative()
+}
+
+// RequestSoftNegative asks the handler chain to downgrade an NXDOMAIN from the
+// handlers that run after the caller defers. A handler that owns a name but
+// cannot answer one query type for it needs this: the resolvers it falls
+// through to cannot prove the name absent, and an NXDOMAIN from them is cached
+// for the name and every type under it (RFC 2308, RFC 8020), taking the
+// addresses the handler does serve down with it.
+func RequestSoftNegative(w dns.ResponseWriter) {
+	if sn, ok := w.(softNegativeRequester); ok {
+		sn.RequestSoftNegative()
+	}
+}
+
 // LookupResult contains the result of an external DNS lookup
 type LookupResult struct {
 	IPs   []netip.Addr
@@ -199,11 +273,27 @@ type RecordResolver interface {
 	LookupAddr(ctx context.Context, addr string) ([]string, error)
 }
 
+// SupportedRecordQtype reports whether LookupRecords can resolve qtype. The
+// set is bounded by the net.Resolver API, which exposes no way to query an
+// arbitrary record type, and going around it with a raw exchange would mean
+// picking nameservers ourselves instead of resolving the way the host does.
+//
+// Both ends read this: a DNS forwarder answers these types for the domains it
+// routes, and a client uses it to tell which types are worth forwarding to a
+// peer at all.
+func SupportedRecordQtype(qtype uint16) bool {
+	switch qtype {
+	case dns.TypeMX, dns.TypeTXT, dns.TypeNS, dns.TypeSRV, dns.TypeCNAME, dns.TypePTR:
+		return true
+	default:
+		return false
+	}
+}
+
 // LookupRecords resolves a non-address DNS record type through the host
-// resolver and returns the resource records and the DNS rcode. Types the host
-// resolver cannot answer (anything not covered by the net.Resolver Lookup*
-// methods) yield NODATA so that a routed name is never poisoned with NXDOMAIN
-// for an unsupported type.
+// resolver and returns the resource records and the DNS rcode. Types outside
+// SupportedRecordQtype yield NODATA so that a routed name is never poisoned
+// with NXDOMAIN for a type we cannot look up.
 func LookupRecords(ctx context.Context, r RecordResolver, name string, qtype uint16, ttl uint32) ([]dns.RR, int) {
 	fqdn := dns.Fqdn(name)
 

client/internal/dns/server.go

@@ -82,6 +82,7 @@ type Server interface {
 	PopulateManagementDomain(mgmtURL *url.URL) error
 	SetRouteSources(selected, active func() route.HAMap)
 	SetFirewall(Firewall)
+	SetPeerActivator(local.PeerActivator)
 }
 
 type nsGroupsByDomain struct {
@@ -491,6 +492,13 @@ func (s *DefaultServer) SetFirewall(fw Firewall) {
 	}
 }
 
+// SetPeerActivator wires the DNS-time lazy-connection warm-up on the local
+// resolver. Injected after the connection manager exists (it does not at
+// DNS-server construction time). Pass nil to disable.
+func (s *DefaultServer) SetPeerActivator(a local.PeerActivator) {
+	s.localResolver.SetPeerActivator(a)
+}
+
 // Stop stops the server
 func (s *DefaultServer) Stop() {
 	s.ctxCancel()
@@ -1435,11 +1443,11 @@ type localPeerConnectivity struct {
 
 // IsConnectedByIP looks the IP up in the peerstore and surfaces both
 // the known and connected bits. Used by Resolver.filterDisconnectedPeerAnswers.
-func (l localPeerConnectivity) IsConnectedByIP(ip string) (known, connected bool) {
+func (l localPeerConnectivity) IsConnectedByIP(ip netip.Addr) (known, connected bool) {
 	if l.status == nil {
 		return false, false
 	}
-	state, ok := l.status.PeerStateByIP(ip)
+	state, ok := l.status.PeerStateByIP(ip.String())
 	if !ok {
 		return false, false
 	}

client/internal/dns/service_listener.go

@@ -292,18 +292,16 @@ func (s *serviceViaListener) generateFreePort() (uint16, error) {
 		return customPort, nil
 	}
 
-	udpAddr := net.UDPAddrFromAddrPort(netip.MustParseAddrPort("0.0.0.0:0"))
-	probeListener, err := net.ListenUDP("udp", udpAddr)
+	probeListener, err := net.ListenUDP("udp4", &net.UDPAddr{})
 	if err != nil {
 		log.Debugf("failed to bind random port for DNS: %s", err)
 		return 0, err
 	}
 
-	addrPort := netip.MustParseAddrPort(probeListener.LocalAddr().String()) // might panic if address is incorrect
-	err = probeListener.Close()
-	if err != nil {
+	port := uint16(probeListener.LocalAddr().(*net.UDPAddr).Port)
+	if err = probeListener.Close(); err != nil {
 		log.Debugf("failed to free up DNS port: %s", err)
 		return 0, err
 	}
-	return addrPort.Port(), nil
+	return port, nil
 }

client/internal/dns/upstream.go

@@ -295,7 +295,7 @@ func (u *upstreamResolverBase) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 	if addr := w.RemoteAddr(); addr != nil {
 		network := addr.Network()
 		ctx = contextWithDNSProtocol(ctx, network)
-		resutil.SetMeta(w, "protocol", network)
+		resutil.SetMeta(w, resutil.MetaKeyProtocol, network)
 	}
 
 	ok, failures := u.tryUpstreamServers(ctx, w, r, logger)
@@ -331,7 +331,7 @@ func (u *upstreamResolverBase) tryOnlyRace(ctx context.Context, w dns.ResponseWr
 		return false, res.failures
 	}
 	if res.ede != "" {
-		resutil.SetMeta(w, "ede", res.ede)
+		resutil.SetMeta(w, resutil.MetaKeyEDE, res.ede)
 	}
 	u.writeSuccessResponse(w, res.msg, res.upstream, r.Question[0].Name, res.protocol, logger)
 	return true, res.failures
@@ -361,7 +361,7 @@ func (u *upstreamResolverBase) raceAll(ctx context.Context, w dns.ResponseWriter
 			failures = append(failures, res.failures...)
 			if res.msg != nil {
 				if res.ede != "" {
-					resutil.SetMeta(w, "ede", res.ede)
+					resutil.SetMeta(w, resutil.MetaKeyEDE, res.ede)
 				}
 				u.writeSuccessResponse(w, res.msg, res.upstream, r.Question[0].Name, res.protocol, logger)
 				return true, failures
@@ -550,9 +550,9 @@ func (u *upstreamResolverBase) debugUpstreamTimeout(upstream netip.AddrPort) str
 }
 
 func (u *upstreamResolverBase) writeSuccessResponse(w dns.ResponseWriter, rm *dns.Msg, upstream netip.AddrPort, domain string, proto string, logger *log.Entry) {
-	resutil.SetMeta(w, "upstream", upstream.String())
+	resutil.SetMeta(w, resutil.MetaKeyUpstream, upstream.String())
 	if proto != "" {
-		resutil.SetMeta(w, "upstream_protocol", proto)
+		resutil.SetMeta(w, resutil.MetaKeyUpstreamProtocol, proto)
 	}
 
 	// Clear Zero bit from external responses to prevent upstream servers from

client/internal/dns_peer_activator.go

@@ -0,0 +1,76 @@
+package internal
+
+import (
+	"context"
+	"net/netip"
+	"time"
+
+	"github.com/netbirdio/netbird/client/internal/peer"
+	"github.com/netbirdio/netbird/client/internal/peerstore"
+)
+
+const dnsActivationPollInterval = 50 * time.Millisecond
+
+// dnsPeerActivator wakes lazy-connection peers from the DNS resolution path. It
+// implements dns/local.PeerActivator. DNS queries run on their own goroutines,
+// so it only touches state that is safe for concurrent use — ConnMgr.ActivatePeer,
+// peerstore.Store and peer.Status — and never takes the engine's syncMsgMux,
+// keeping DNS resolution from contending with network-map processing.
+type dnsPeerActivator struct {
+	connMgr   *ConnMgr
+	peerStore *peerstore.Store
+	status    *peer.Status
+	// ctx is the engine's long-lived context. The connection dial is tied to it
+	// (not the per-query DNS wait budget) so a handshake that outlasts the wait
+	// still completes in the background rather than being cancelled at the deadline.
+	ctx context.Context
+}
+
+// ActivatePeersByIP triggers wake-up for the peer(s) owning addrs and waits
+// until one is connected or ctx (the per-query DNS wait budget) expires.
+// Activation itself is tied to the engine's long-lived context so the dial
+// survives a wait that times out. Unknown or already-connected addresses are
+// skipped, so the steady-state (warm) path adds no latency.
+func (a *dnsPeerActivator) ActivatePeersByIP(ctx context.Context, addrs []netip.Addr) {
+	if a == nil || a.connMgr == nil {
+		return
+	}
+
+	var pending []string
+	for _, addr := range addrs {
+		ip := addr.String()
+		st, ok := a.status.PeerStateByIP(ip)
+		if !ok || st.ConnStatus == peer.StatusConnected {
+			continue
+		}
+		conn, ok := a.peerStore.PeerConn(st.PubKey)
+		if !ok {
+			continue
+		}
+		a.connMgr.ActivatePeer(a.ctx, conn)
+		pending = append(pending, ip)
+	}
+
+	if len(pending) == 0 {
+		return
+	}
+	a.waitConnected(ctx, pending)
+}
+
+// waitConnected blocks until any of ips reports a connected peer or ctx expires.
+func (a *dnsPeerActivator) waitConnected(ctx context.Context, ips []string) {
+	ticker := time.NewTicker(dnsActivationPollInterval)
+	defer ticker.Stop()
+	for {
+		for _, ip := range ips {
+			if st, ok := a.status.PeerStateByIP(ip); ok && st.ConnStatus == peer.StatusConnected {
+				return
+			}
+		}
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+		}
+	}
+}

client/internal/dns_peer_activator_test.go

@@ -0,0 +1,129 @@
+package internal
+
+import (
+	"context"
+	"net/netip"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+
+	"github.com/netbirdio/netbird/client/internal/peer"
+	"github.com/netbirdio/netbird/client/internal/peerstore"
+)
+
+func newTestPeerConn(t *testing.T, key string) *peer.Conn {
+	t.Helper()
+	conn, err := peer.NewConn(peer.ConnConfig{
+		Key:      key,
+		LocalKey: "local",
+		WgConfig: peer.WgConfig{
+			AllowedIps: []netip.Prefix{netip.MustParsePrefix("100.64.0.1/32")},
+		},
+	}, peer.ServiceDependencies{})
+	require.NoError(t, err)
+	return conn
+}
+
+func newTestDNSPeerActivator(t *testing.T) (*dnsPeerActivator, *peer.Status, *peerstore.Store) {
+	t.Helper()
+	status := peer.NewRecorder("https://mgm")
+	store := peerstore.NewConnStore()
+	// ConnMgr without Start: the lazy manager is nil, so ActivatePeer is a
+	// no-op — these tests exercise the activator's skip/wait logic.
+	connMgr := NewConnMgr(&EngineConfig{}, status, store, nil)
+	return &dnsPeerActivator{
+		connMgr:   connMgr,
+		peerStore: store,
+		status:    status,
+		ctx:       context.Background(),
+	}, status, store
+}
+
+func TestDNSPeerActivator_NilSafe(t *testing.T) {
+	var a *dnsPeerActivator
+	a.ActivatePeersByIP(context.Background(), []netip.Addr{netip.MustParseAddr("100.64.0.1")})
+}
+
+// TestDNSPeerActivator_SkipsUnknownAndConnectedPeers verifies the steady-state
+// (warm) path adds no latency: already-connected and unknown addresses never
+// enter the wait loop.
+func TestDNSPeerActivator_SkipsUnknownAndConnectedPeers(t *testing.T) {
+	a, status, store := newTestDNSPeerActivator(t)
+
+	require.NoError(t, status.AddPeer("peerA", "a.netbird.cloud", "100.64.0.1", "fd00::1"))
+	require.NoError(t, status.UpdatePeerState(peer.State{PubKey: "peerA", ConnStatus: peer.StatusConnected}))
+	store.AddPeerConn("peerA", newTestPeerConn(t, "peerA"))
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	start := time.Now()
+	a.ActivatePeersByIP(ctx, []netip.Addr{
+		netip.MustParseAddr("100.64.0.1"),  // known, connected -> skipped
+		netip.MustParseAddr("fd00::1"),     // known via IPv6, connected -> skipped
+		netip.MustParseAddr("100.64.0.99"), // unknown -> skipped
+	})
+	require.Less(t, time.Since(start), time.Second, "no pending peer must mean no wait")
+}
+
+// TestDNSPeerActivator_WaitsForPendingPeerToConnect verifies the wait loop
+// returns as soon as a pending peer reports connected, well before the
+// per-query budget expires.
+func TestDNSPeerActivator_WaitsForPendingPeerToConnect(t *testing.T) {
+	a, status, store := newTestDNSPeerActivator(t)
+
+	require.NoError(t, status.AddPeer("peerA", "a.netbird.cloud", "100.64.0.1", ""))
+	store.AddPeerConn("peerA", newTestPeerConn(t, "peerA"))
+
+	go func() {
+		time.Sleep(150 * time.Millisecond)
+		_ = status.UpdatePeerState(peer.State{PubKey: "peerA", ConnStatus: peer.StatusConnected})
+	}()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	defer cancel()
+
+	start := time.Now()
+	a.ActivatePeersByIP(ctx, []netip.Addr{netip.MustParseAddr("100.64.0.1")})
+	elapsed := time.Since(start)
+
+	require.GreaterOrEqual(t, elapsed, 100*time.Millisecond, "must wait for the pending peer")
+	require.Less(t, elapsed, 5*time.Second, "must return on connect, not at the deadline")
+}
+
+// TestDNSPeerActivator_ReturnsAtBudgetWhenPeerStaysIdle verifies a peer that
+// never connects releases the DNS response at the per-query budget instead of
+// blocking it indefinitely.
+func TestDNSPeerActivator_ReturnsAtBudgetWhenPeerStaysIdle(t *testing.T) {
+	a, status, store := newTestDNSPeerActivator(t)
+
+	require.NoError(t, status.AddPeer("peerA", "a.netbird.cloud", "100.64.0.1", ""))
+	store.AddPeerConn("peerA", newTestPeerConn(t, "peerA"))
+
+	ctx, cancel := context.WithTimeout(context.Background(), 300*time.Millisecond)
+	defer cancel()
+
+	start := time.Now()
+	a.ActivatePeersByIP(ctx, []netip.Addr{netip.MustParseAddr("100.64.0.1")})
+	elapsed := time.Since(start)
+
+	require.GreaterOrEqual(t, elapsed, 250*time.Millisecond, "must wait out the budget for a pending peer")
+	require.Less(t, elapsed, 5*time.Second, "must not block past the budget")
+}
+
+// TestDNSPeerActivator_NoWaitWithoutPeerConn verifies a known-but-idle peer
+// with no connection object in the store is not waited on: there is nothing to
+// activate, so waiting could only ever time out.
+func TestDNSPeerActivator_NoWaitWithoutPeerConn(t *testing.T) {
+	a, status, _ := newTestDNSPeerActivator(t)
+
+	require.NoError(t, status.AddPeer("peerA", "a.netbird.cloud", "100.64.0.1", ""))
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	start := time.Now()
+	a.ActivatePeersByIP(ctx, []netip.Addr{netip.MustParseAddr("100.64.0.1")})
+	require.Less(t, time.Since(start), time.Second, "peer without a conn must not be waited on")
+}

client/internal/dnsfwd/forwarder.go

@@ -26,15 +26,6 @@ import (
 const errResolveFailed = "failed to resolve query for domain=%s: %v"
 const upstreamTimeout = 15 * time.Second
 
-// EDE info codes the forwarder emits on upstream failures so the querying
-// client can see the reason without inspecting this peer's logs. They live in
-// the RFC 8914 Private Use range (49152-65535); the Go resolver never exposes a
-// real upstream EDE here, so these cannot collide with a genuine code.
-const (
-	edeNetbirdUpstreamTimeout uint16 = 49152
-	edeNetbirdUpstreamFailure uint16 = 49153
-)
-
 type resolver interface {
 	LookupNetIP(ctx context.Context, network, host string) ([]netip.Addr, error)
 	LookupMX(ctx context.Context, name string) ([]*net.MX, error)
@@ -216,6 +207,9 @@ func (f *DNSForwarder) handleDNSQuery(logger *log.Entry, w dns.ResponseWriter, q
 		qname, dns.TypeToString[question.Qtype], dns.ClassToString[question.Qclass])
 
 	resp := query.SetReply(query)
+	// Every answer here comes from a recursive lookup on this peer. SetReply
+	// leaves RA unset, which reads to a client as a server that cannot recurse.
+	resp.RecursionAvailable = true
 
 	mostSpecificResId, matchingEntries := f.getMatchingEntries(strings.TrimSuffix(qname, "."))
 	if mostSpecificResId == "" {
@@ -229,20 +223,22 @@ func (f *DNSForwarder) handleDNSQuery(logger *log.Entry, w dns.ResponseWriter, q
 
 	reqHasEdns := query.IsEdns0() != nil
 
-	switch question.Qtype {
-	case dns.TypeA, dns.TypeAAAA:
+	switch {
+	case question.Qtype == dns.TypeA || question.Qtype == dns.TypeAAAA:
 		f.handleAddressQuery(ctx, logger, w, resp, mostSpecificResId, matchingEntries, reqHasEdns, startTime)
-	case dns.TypeMX, dns.TypeTXT, dns.TypeNS, dns.TypeSRV, dns.TypeCNAME, dns.TypePTR:
+	case resutil.SupportedRecordQtype(question.Qtype):
 		f.handleRecordQuery(ctx, logger, w, resp, startTime)
 	default:
 		// The domain is routed here, so any other type is answered NODATA
 		// (NOERROR, empty answer) rather than falling back to a resolver that
 		// would poison the name with NXDOMAIN. The Extended DNS Error lets a
 		// client tell this capability-driven NODATA apart from an
-		// authoritative one. The OPT pseudo-record must not appear unless the
-		// query advertised EDNS0.
+		// authoritative one; a current client knows the type is unsupported and
+		// resolves it through its own handler chain instead of asking at all.
+		// The OPT pseudo-record must not appear unless the query advertised
+		// EDNS0.
 		if reqHasEdns {
-			attachEDE(resp, dns.ExtendedErrorCodeNotSupported, "netbird forwarder: unsupported query type")
+			resutil.AttachEDE(resp, dns.ExtendedErrorCodeNotSupported, "netbird forwarder: unsupported query type")
 		}
 		f.writeResponse(logger, w, resp, qname, startTime)
 	}
@@ -441,7 +437,7 @@ func (f *DNSForwarder) handleDNSError(
 	}
 
 	if reqHasEdns {
-		attachEDE(resp, edeCodeFor(dnsErr), edeText(dnsErr))
+		resutil.AttachEDE(resp, edeCodeFor(dnsErr), edeText(dnsErr))
 	}
 
 	f.writeResponse(logger, w, resp, domain, startTime)
@@ -488,9 +484,9 @@ func (f *DNSForwarder) getMatchingEntries(domain string) (route.ResID, []*Forwar
 // edeCodeFor maps an upstream lookup error to the NetBird EDE info code.
 func edeCodeFor(dnsErr *net.DNSError) uint16 {
 	if dnsErr != nil && dnsErr.IsTimeout {
-		return edeNetbirdUpstreamTimeout
+		return resutil.EDENetbirdUpstreamTimeout
 	}
-	return edeNetbirdUpstreamFailure
+	return resutil.EDENetbirdUpstreamFailure
 }
 
 // edeText builds the EDE extra-text describing the class of upstream failure.
@@ -503,14 +499,3 @@ func edeText(dnsErr *net.DNSError) string {
 	}
 	return "netbird forwarder: upstream failure"
 }
-
-// attachEDE adds an Extended DNS Error (RFC 8914) option to the response,
-// creating the OPT pseudo-record if the response does not already carry one.
-func attachEDE(resp *dns.Msg, code uint16, text string) {
-	opt := resp.IsEdns0()
-	if opt == nil {
-		resp.SetEdns0(dns.DefaultMsgSize, false)
-		opt = resp.IsEdns0()
-	}
-	opt.Option = append(opt.Option, &dns.EDNS0_EDE{InfoCode: code, ExtraText: text})
-}

client/internal/dnsfwd/forwarder_test.go

@@ -649,6 +649,39 @@ func TestDNSForwarder_ResponseCodes(t *testing.T) {
 	}
 }
 
+// TestDNSForwarder_RecursionAvailable covers the RA bit. Every answer this
+// forwarder produces comes from a recursive lookup on the routing peer, but
+// dns.Msg.SetReply leaves RA unset, so clients report "recursion not available"
+// and some stub resolvers treat the server as unable to serve the query.
+func TestDNSForwarder_RecursionAvailable(t *testing.T) {
+	t.Run("record answer", func(t *testing.T) {
+		mockResolver := &MockResolver{}
+		forwarder := newRecordTestForwarder(t, mockResolver, "example.com")
+
+		mockResolver.On("LookupMX", mock.Anything, "example.com.").
+			Return([]*net.MX{{Host: "mail.example.com.", Pref: 10}}, nil).Once()
+
+		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeMX)
+		assert.True(t, resp.RecursionAvailable, "the forwarder resolves recursively")
+	})
+
+	t.Run("unsupported type NODATA", func(t *testing.T) {
+		forwarder := newRecordTestForwarder(t, &MockResolver{}, "example.com")
+
+		resp := runRecordQuery(t, forwarder, "example.com", dns.TypeCAA)
+		require.Equal(t, dns.RcodeSuccess, resp.Rcode)
+		assert.True(t, resp.RecursionAvailable, "RA describes the server, not the query type")
+	})
+
+	t.Run("unauthorized domain", func(t *testing.T) {
+		forwarder := newRecordTestForwarder(t, &MockResolver{}, "example.com")
+
+		resp := runRecordQuery(t, forwarder, "other.com", dns.TypeMX)
+		require.Equal(t, dns.RcodeRefused, resp.Rcode)
+		assert.True(t, resp.RecursionAvailable, "RA describes the server, not the verdict")
+	})
+}
+
 func hasEDE(m *dns.Msg, code uint16) bool {
 	opt := m.IsEdns0()
 	if opt == nil {
@@ -859,7 +892,7 @@ func TestDNSForwarder_UpstreamFailureEDE(t *testing.T) {
 			lookupErr:   &net.DNSError{Err: "i/o timeout", Server: "10.0.0.53:53", IsTimeout: true},
 			reqEdns:     true,
 			wantEDE:     true,
-			wantCode:    edeNetbirdUpstreamTimeout,
+			wantCode:    resutil.EDENetbirdUpstreamTimeout,
 			wantTextHas: "netbird forwarder: upstream timeout",
 		},
 		{
@@ -867,7 +900,7 @@ func TestDNSForwarder_UpstreamFailureEDE(t *testing.T) {
 			lookupErr:   &net.DNSError{Err: "server misbehaving", Server: "10.0.0.53:53"},
 			reqEdns:     true,
 			wantEDE:     true,
-			wantCode:    edeNetbirdUpstreamFailure,
+			wantCode:    resutil.EDENetbirdUpstreamFailure,
 			wantTextHas: "netbird forwarder: upstream failure",
 		},
 		{

client/internal/ebpf/ebpf/src/dns_fwd.c

@@ -52,11 +52,14 @@ int xdp_dns_fwd(struct iphdr  *ip, struct udphdr *udp) {
 
     if (udp->dest == GENERAL_DNS_PORT && ip->daddr == dns_ip) {
         udp->dest = dns_port;
+        // Clear the now-stale checksum; zero means "not computed" for IPv4.
+        udp->check = 0;
         return XDP_PASS;
     }
 
     if (udp->source == dns_port && ip->saddr == dns_ip) {
         udp->source = GENERAL_DNS_PORT;
+        udp->check = 0;
         return XDP_PASS;
     }
 

client/internal/ebpf/ebpf/src/wg_proxy.c

@@ -50,5 +50,11 @@ int xdp_wg_proxy(struct iphdr  *ip, struct udphdr *udp) {
     __be16 new_dst_port = htons(proxy_port);
     udp->dest = new_dst_port;
     udp->source = new_src_port;
+
+    // The ports are covered by the UDP checksum. This is an IPv4 loopback hop
+    // and the payload is already integrity-protected, so clear the checksum (a
+    // zero UDP checksum means "not computed" for IPv4) rather than leave a
+    // stale value the kernel would drop as UDP_CSUM.
+    udp->check = 0;
 	return XDP_PASS;
 }

client/internal/engine.go

@@ -64,7 +64,10 @@ import (
 	"github.com/netbirdio/netbird/route"
 	mgm "github.com/netbirdio/netbird/shared/management/client"
 	"github.com/netbirdio/netbird/shared/management/domain"
+	sharedgrpc "github.com/netbirdio/netbird/shared/management/grpc"
+	nbnetworkmap "github.com/netbirdio/netbird/shared/management/networkmap"
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
+	types "github.com/netbirdio/netbird/shared/management/types"
 	"github.com/netbirdio/netbird/shared/netiputil"
 	auth "github.com/netbirdio/netbird/shared/relay/auth/hmac"
 	relayClient "github.com/netbirdio/netbird/shared/relay/client"
@@ -147,6 +150,7 @@ type EngineConfig struct {
 	BlockLANAccess      bool
 	BlockInbound        bool
 	DisableIPv6         bool
+	SyncMessageVersion  *int
 
 	// LazyConnection is the MDM-sourced lazy-connection override; StateUnset defers to
 	// the env var and management feature flag.
@@ -220,6 +224,13 @@ type Engine struct {
 	// networkSerial is the latest CurrentSerial (state ID) of the network sent by the Management service
 	networkSerial uint64
 
+	// latestComponents is the most-recent NetworkMapComponents decoded from
+	// a NetworkMapEnvelope (capability=3 peers only). Held alongside the
+	// NetworkMap that Calculate() produced from it so future incremental
+	// updates have a base to apply changes against. nil for legacy-format
+	// peers. Guarded by syncMsgMux.
+	latestComponents *types.NetworkMapComponents
+
 	networkMonitor *networkmonitor.NetworkMonitor
 
 	sshServer sshServer
@@ -652,8 +663,24 @@ func (e *Engine) Start(netbirdConfig *mgmProto.NetbirdConfig, mgmtURL *url.URL)
 	iceCfg := e.createICEConfig()
 
 	e.connMgr = NewConnMgr(e.config, e.statusRecorder, e.peerStore, wgIface)
+	e.connMgr.SetRoutedIPsReconciler(func(peerKey string) error {
+		if e.routeManager == nil {
+			return nil
+		}
+		return e.routeManager.ReconcilePeerAllowedIPs(peerKey)
+	})
 	e.connMgr.Start(e.ctx)
 
+	// Wire DNS-time lazy-connection warm-up now that the connection manager
+	// exists (it does not at DNS-server construction time). A DNS answer that
+	// points at an idle peer then wakes it before the client's first request.
+	e.dnsServer.SetPeerActivator(&dnsPeerActivator{
+		connMgr:   e.connMgr,
+		peerStore: e.peerStore,
+		status:    e.statusRecorder,
+		ctx:       e.ctx,
+	})
+
 	e.srWatcher = guard.NewSRWatcher(e.signal, e.relayManager, e.mobileDep.IFaceDiscover, iceCfg)
 	e.srWatcher.Start(peer.IsForceRelayed())
 
@@ -963,8 +990,12 @@ func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
 
 	e.ApplySessionDeadline(update.GetSessionExpiresAt())
 
-	if update.NetworkMap != nil && update.NetworkMap.PeerConfig != nil {
-		e.handleAutoUpdateVersion(update.NetworkMap.PeerConfig.AutoUpdate)
+	// Envelope sync responses carry PeerConfig at the top level; legacy
+	// NetworkMap syncs carry it under NetworkMap.PeerConfig.
+	if pc := update.GetPeerConfig(); pc != nil {
+		e.handleAutoUpdateVersion(pc.GetAutoUpdate())
+	} else if nm := update.GetNetworkMap(); nm != nil && nm.GetPeerConfig() != nil {
+		e.handleAutoUpdateVersion(nm.GetPeerConfig().GetAutoUpdate())
 	}
 
 	done := e.phase("netbird_config")
@@ -974,12 +1005,47 @@ func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
 		return err
 	}
 
+	// Decode the network map from either the components envelope or the
+	// legacy proto.NetworkMap before the posture-check gating below, so the
+	// "is there a network map" decision covers both wire shapes.
+	var (
+		nm         *mgmProto.NetworkMap
+		components *types.NetworkMapComponents
+	)
+	if version := update.GetVersion(); version == int32(sharedgrpc.ComponentNetworkMap) {
+		// Components-format peer: decode the envelope back to typed
+		// components, run Calculate() locally, and convert to the wire
+		// NetworkMap shape the rest of the engine consumes. Components are
+		// retained so future incremental updates can apply deltas instead
+		// of doing a full reconstruction.
+		envelope := update.GetNetworkMapEnvelope()
+		if envelope == nil {
+			return fmt.Errorf("received a SyncReponse indicating use of components network map, but components are missing")
+		}
+
+		localKey := e.config.WgPrivateKey.PublicKey().String()
+		dnsName := ""
+		if pc := update.GetPeerConfig(); pc != nil {
+			// PeerConfig.Fqdn = "<dns_label>.<dns_domain>" — extract the
+			// shared domain by stripping the peer's own label prefix. Falls
+			// back to empty if the FQDN doesn't have the expected shape.
+			dnsName = extractDNSDomainFromFQDN(pc.GetFqdn())
+		}
+		result, err := nbnetworkmap.EnvelopeToNetworkMap(e.ctx, envelope, localKey, dnsName)
+		if err != nil {
+			return fmt.Errorf("decode network map envelope: %w", err)
+		}
+		nm = result.NetworkMap
+		components = result.Components
+	} else {
+		nm = update.GetNetworkMap()
+	}
+
 	// Posture checks are bound to the network map presence:
 	//   NetworkMap != nil, checks present -> apply the received checks
 	//   NetworkMap != nil, checks nil     -> posture checks were removed, clear them
 	//   NetworkMap == nil                 -> config-only update (e.g. relay token rotation),
 	//                                        leave the previously applied checks untouched
-	nm := update.GetNetworkMap()
 	if nm == nil {
 		return nil
 	}
@@ -992,6 +1058,14 @@ func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
 	}
 
 	done = e.phase("persist")
+	// Only retain the components view when the server sent the envelope
+	// path. A legacy proto.NetworkMap means components == nil; writing it
+	// here would clobber a previously-cached snapshot, breaking the
+	// incremental-delta base on a future envelope sync.
+	if components != nil {
+		e.latestComponents = components
+	}
+
 	e.persistSyncResponse(update)
 	done()
 
@@ -1005,6 +1079,19 @@ func (e *Engine) handleSync(update *mgmProto.SyncResponse) error {
 	return nil
 }
 
+// extractDNSDomainFromFQDN returns the trailing dotted domain part of the
+// receiving peer's FQDN — the same value the management server fills as
+// dnsName when it builds the legacy NetworkMap. "peer42.netbird.cloud" →
+// "netbird.cloud". An empty string is returned for unrecognized formats.
+func extractDNSDomainFromFQDN(fqdn string) string {
+	for i := 0; i < len(fqdn); i++ {
+		if fqdn[i] == '.' && i+1 < len(fqdn) {
+			return fqdn[i+1:]
+		}
+	}
+	return ""
+}
+
 // updateNetbirdConfig applies the management-provided NetBird configuration:
 // STUN/TURN and relay servers, flow logging and DNS settings. A nil config is a no-op,
 // which is the case for sync updates carrying only a network map.
@@ -1164,6 +1251,7 @@ func (e *Engine) applyInfoFlags(info *system.Info) {
 		e.config.BlockLANAccess,
 		e.config.BlockInbound,
 		e.config.DisableIPv6,
+		e.config.SyncMessageVersion,
 		e.config.EnableSSHRoot,
 		e.config.EnableSSHSFTP,
 		e.config.EnableSSHLocalPortForwarding,
@@ -2032,6 +2120,7 @@ func (e *Engine) readInitialSettings() ([]*route.Route, *nbdns.Config, bool, err
 		e.config.BlockLANAccess,
 		e.config.BlockInbound,
 		e.config.DisableIPv6,
+		e.config.SyncMessageVersion,
 		e.config.EnableSSHRoot,
 		e.config.EnableSSHSFTP,
 		e.config.EnableSSHLocalPortForwarding,

client/internal/engine_session_deadline_test.go

@@ -75,4 +75,14 @@ func TestApplySessionDeadline_ThreeState(t *testing.T) {
 		require.True(t, e.statusRecorder.GetSessionExpiresAt().IsZero(),
 			"invalid timestamp must clear the deadline")
 	})
+
+	t.Run("recently expired timestamp stays visible as expired", func(t *testing.T) {
+		e := newEngine()
+		expired := time.Now().Add(-5 * time.Minute).UTC().Truncate(time.Second)
+
+		e.ApplySessionDeadline(timestamppb.New(expired))
+
+		require.True(t, e.statusRecorder.GetSessionExpiresAt().Equal(expired),
+			"recently-expired deadline must stay on the recorder so consumers render it as expired")
+	})
 }

client/internal/lazyconn/manager/manager.go

@@ -29,6 +29,11 @@ type managedPeer struct {
 
 type Config struct {
 	InactivityThreshold *time.Duration
+	// ReconcileAllowedIPs re-applies a peer's routed allowed IPs after its wake endpoint is
+	// armed. The activity listener creates the wake peer with the overlay /32 only; without the
+	// routed prefixes WireGuard would not steer subnet-bound traffic to the wake endpoint, so an
+	// idle routing peer could never be woken by that traffic. Optional; nil disables the reconcile.
+	ReconcileAllowedIPs func(peerKey string) error
 }
 
 // Manager manages lazy connections
@@ -56,6 +61,9 @@ type Manager struct {
 	peerToHAGroups map[string][]route.HAUniqueID // peer ID -> HA groups they belong to
 	haGroupToPeers map[route.HAUniqueID][]string // HA group -> peer IDs in the group
 	routesMu       sync.RWMutex
+
+	// reconcileAllowedIPs re-applies a peer's routed allowed IPs after its wake endpoint is armed.
+	reconcileAllowedIPs func(peerKey string) error
 }
 
 // NewManager creates a new lazy connection manager
@@ -73,6 +81,7 @@ func NewManager(config Config, engineCtx context.Context, peerStore *peerstore.S
 		activityManager:      activity.NewManager(wgIface),
 		peerToHAGroups:       make(map[string][]route.HAUniqueID),
 		haGroupToPeers:       make(map[route.HAUniqueID][]string),
+		reconcileAllowedIPs:  config.ReconcileAllowedIPs,
 	}
 
 	if wgIface.IsUserspaceBind() {
@@ -201,7 +210,7 @@ func (m *Manager) AddPeer(peerCfg lazyconn.PeerConfig) (bool, error) {
 		return false, nil
 	}
 
-	if err := m.activityManager.MonitorPeerActivity(peerCfg); err != nil {
+	if err := m.armActivityListener(peerCfg); err != nil {
 		return false, err
 	}
 
@@ -288,7 +297,7 @@ func (m *Manager) DeactivatePeer(peerID peerid.ConnID) {
 
 	m.inactivityManager.RemovePeer(mp.peerCfg.PublicKey)
 
-	if err := m.activityManager.MonitorPeerActivity(*mp.peerCfg); err != nil {
+	if err := m.armActivityListener(*mp.peerCfg); err != nil {
 		mp.peerCfg.Log.Errorf("failed to create activity monitor: %v", err)
 		return
 	}
@@ -465,6 +474,31 @@ func (m *Manager) close() {
 }
 
 // shouldDeferIdleForHA checks if peer should stay connected due to HA group requirements
+// armRoutedAllowedIPs re-applies the peer's routed allowed IPs onto its freshly armed wake
+// endpoint. The activity listener creates the wake peer with the overlay /32 only, so without
+// this the routed prefixes would be missing and traffic to a routed subnet could not wake the
+// idle routing peer. It is a no-op when no reconciler is configured.
+// armActivityListener (re)arms the peer's wake endpoint via the activity manager and then
+// re-applies its routed allowed IPs, so traffic to a routed subnet can wake an idle routing
+// peer. The routed prefixes must be re-applied after the wake endpoint exists because the
+// listener creates it with the overlay /32 only.
+func (m *Manager) armActivityListener(peerCfg lazyconn.PeerConfig) error {
+	if err := m.activityManager.MonitorPeerActivity(peerCfg); err != nil {
+		return err
+	}
+	m.armRoutedAllowedIPs(&peerCfg)
+	return nil
+}
+
+func (m *Manager) armRoutedAllowedIPs(peerCfg *lazyconn.PeerConfig) {
+	if m.reconcileAllowedIPs == nil {
+		return
+	}
+	if err := m.reconcileAllowedIPs(peerCfg.PublicKey); err != nil {
+		peerCfg.Log.Errorf("failed to reconcile routed allowed IPs on wake endpoint: %v", err)
+	}
+}
+
 func (m *Manager) shouldDeferIdleForHA(inactivePeers map[string]struct{}, peerID string) bool {
 	m.routesMu.RLock()
 	defer m.routesMu.RUnlock()
@@ -577,7 +611,7 @@ func (m *Manager) onPeerInactivityTimedOut(peerIDs map[string]struct{}) {
 
 		mp.peerCfg.Log.Infof("start activity monitor")
 
-		if err := m.activityManager.MonitorPeerActivity(*mp.peerCfg); err != nil {
+		if err := m.armActivityListener(*mp.peerCfg); err != nil {
 			mp.peerCfg.Log.Errorf("failed to create activity monitor: %v", err)
 			continue
 		}

client/internal/peer/status.go

@@ -813,19 +813,14 @@ func (d *Status) SetSessionExpiresAt(deadline time.Time) {
 }
 
 // GetSessionExpiresAt returns the most recently recorded SSO session deadline,
-// or the zero value when no deadline is tracked. A deadline that has already
-// slipped into the past reports as "none": once the session has expired it is
-// no longer a meaningful countdown, and the sessionwatch.Watcher does not
-// arm a timer at the deadline itself to clear it (only the two pre-expiry
-// warnings). Without this guard the UI would keep painting a stale
-// "expires in …" against a moment that has passed until the next login,
-// extend, or teardown rewrote the value.
+// or the zero value when no deadline is tracked. A deadline in the past is
+// returned as-is: it means the session has expired, and consumers (tray row,
+// CLI status) render it as "expired" rather than hiding it — masking it as
+// "none" would blank the UI at the exact moment it should say the session
+// ended.
 func (d *Status) GetSessionExpiresAt() time.Time {
 	d.mux.Lock()
 	defer d.mux.Unlock()
-	if !d.sessionExpiresAt.IsZero() && d.sessionExpiresAt.Before(time.Now()) {
-		return time.Time{}
-	}
 	return d.sessionExpiresAt
 }
 

client/internal/profilemanager/config.go

@@ -96,6 +96,7 @@ type ConfigInput struct {
 	BlockLANAccess      *bool
 	BlockInbound        *bool
 	DisableIPv6         *bool
+	SyncMessageVersion  *int
 
 	DisableNotifications *bool
 
@@ -137,6 +138,7 @@ type Config struct {
 	BlockLANAccess      bool
 	BlockInbound        bool
 	DisableIPv6         bool
+	SyncMessageVersion  *int
 
 	DisableNotifications *bool
 
@@ -587,6 +589,12 @@ func (config *Config) apply(input ConfigInput) (updated bool, err error) {
 		updated = true
 	}
 
+	if input.SyncMessageVersion != nil && *input.SyncMessageVersion != *config.SyncMessageVersion {
+		log.Infof("setting SyncMessageVersion to %v", *input.SyncMessageVersion)
+		*config.SyncMessageVersion = *input.SyncMessageVersion
+		updated = true
+	}
+
 	if input.DisableNotifications != nil && (config.DisableNotifications == nil || *input.DisableNotifications != *config.DisableNotifications) {
 		if *input.DisableNotifications {
 			log.Infof("disabling notifications")

client/internal/routemanager/dnsinterceptor/handler.go

@@ -22,7 +22,6 @@ import (
 	nbdns "github.com/netbirdio/netbird/client/internal/dns"
 	"github.com/netbirdio/netbird/client/internal/dns/resutil"
 	"github.com/netbirdio/netbird/client/internal/peer"
-	"github.com/netbirdio/netbird/client/internal/peerstore"
 	"github.com/netbirdio/netbird/client/internal/routemanager/common"
 	"github.com/netbirdio/netbird/client/internal/routemanager/fakeip"
 	iface "github.com/netbirdio/netbird/client/internal/routemanager/iface"
@@ -40,6 +39,61 @@ type internalDNATer interface {
 	AddInternalDNATMapping(netip.Addr, netip.Addr) error
 }
 
+// peerAllowedIPs reports the tunnel address a peer is reachable on.
+type peerAllowedIPs interface {
+	AllowedIP(pubKey string) (netip.Addr, bool)
+}
+
+// disposition says where a query for a given record type has to be answered.
+type disposition int
+
+const (
+	// dispositionPeer: the routing peer owns the answer. Address records are
+	// what a DNS route exists for, and their answer programs the routes,
+	// allowed IPs and firewall sets, so they are never resolved anywhere else.
+	dispositionPeer disposition = iota
+	// dispositionPeerFirst: the peer's forwarder resolves the type, and the
+	// records may only exist inside the routed network, so it has to be asked.
+	// A peer running a client that predates that support cannot answer, which
+	// only its reply reveals.
+	dispositionPeerFirst
+	// dispositionChain: no forwarder resolves the type, so asking would cost a
+	// tunnel round trip for a reply that says nothing. Public records for the
+	// name are still better than none, so the query goes to the rest of the
+	// chain.
+	dispositionChain
+)
+
+func dispositionFor(qtype uint16) disposition {
+	switch {
+	case qtype == dns.TypeA || qtype == dns.TypeAAAA:
+		return dispositionPeer
+	case resutil.SupportedRecordQtype(qtype):
+		return dispositionPeerFirst
+	default:
+		return dispositionChain
+	}
+}
+
+// peerCannotAnswer reports whether a forwarder reply means the peer is unable to
+// resolve this query type, as opposed to having resolved it and found nothing.
+// A forwarder older than the one that learned to resolve non-address types
+// answers NOTIMP to all of them, before it even looks at the domain; FORMERR
+// covers one that cannot parse the EDNS0 we add to the query.
+//
+// REFUSED is deliberately not here. It says the peer does not hold the name,
+// which happens while the client and the peer disagree about the route set, and
+// falling through then would hand the name of an internal-only domain to a
+// public resolver. No forwarder version reports a missing record type that way.
+func peerCannotAnswer(rcode int) bool {
+	switch rcode {
+	case dns.RcodeNotImplemented, dns.RcodeFormatError:
+		return true
+	default:
+		return false
+	}
+}
+
 type DnsInterceptor struct {
 	mu                   sync.RWMutex
 	route                *route.Route
@@ -50,7 +104,7 @@ type DnsInterceptor struct {
 	currentPeerKey       string
 	interceptedDomains   domainMap
 	wgInterface          iface.WGIface
-	peerStore            *peerstore.Store
+	peerStore            peerAllowedIPs
 	firewall             firewall.Manager
 	fakeIPManager        *fakeip.Manager
 	forwarderPort        *atomic.Uint32
@@ -226,11 +280,14 @@ func (d *DnsInterceptor) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 		return
 	}
 
-	// All query types for an intercepted domain are forwarded to the peer's
-	// DNS forwarder, which owns the name. Falling through to the system
-	// resolver would let it answer NXDOMAIN for a name it isn't authoritative
-	// for, poisoning the whole name (including the A/AAAA records the route
-	// does serve). The forwarder answers NODATA for types it cannot resolve.
+	qtype := r.Question[0].Qtype
+	dispose := dispositionFor(qtype)
+
+	if dispose == dispositionChain {
+		d.deferToChain(w, r, logger, "unsupported-qtype")
+		return
+	}
+
 	d.mu.RLock()
 	peerKey := d.currentPeerKey
 	d.mu.RUnlock()
@@ -246,35 +303,32 @@ func (d *DnsInterceptor) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 		return
 	}
 
-	if r.Extra == nil {
-		r.MsgHdr.AuthenticatedData = true
-	}
-
-	// Advertise EDNS0 to the forwarder so it may return an Extended DNS Error
-	// describing why a lookup failed. The OPT is stripped from the reply when
-	// the original client did not request EDNS0.
-	hadEdns := r.IsEdns0() != nil
-	if !hadEdns {
-		r.SetEdns0(dns.DefaultMsgSize, false)
-	}
+	query, hadEdns := peerQuery(r)
 
 	upstream := net.JoinHostPort(upstreamIP.String(), strconv.FormatUint(uint64(d.forwarderPort.Load()), 10))
 	ctx, cancel := context.WithTimeout(context.Background(), dnsTimeout)
 	defer cancel()
 
-	reply := d.queryUpstreamDNS(ctx, w, r, upstream, upstreamIP, peerKey, logger)
+	reply := d.queryUpstreamDNS(ctx, w, query, upstream, upstreamIP, peerKey, logger)
 	if reply == nil {
 		return
 	}
 
+	// The peer owns the name but its forwarder cannot resolve this type. No
+	// capability is announced anywhere, so the round trip above is the probe.
+	if dispose == dispositionPeerFirst && peerCannotAnswer(reply.Rcode) {
+		d.deferToChain(w, r, logger, "peer-rcode-"+dns.RcodeToString[reply.Rcode])
+		return
+	}
+
 	if ede, ok := resutil.ExtractEDE(reply); ok {
-		resutil.SetMeta(w, "ede", fmt.Sprintf("%d %s", ede.InfoCode, ede.ExtraText))
+		resutil.SetMeta(w, resutil.MetaKeyEDE, fmt.Sprintf("%d %s", ede.InfoCode, ede.ExtraText))
 	}
 	if !hadEdns {
 		resutil.StripOPT(reply)
 	}
 
-	resutil.SetMeta(w, "peer", peerKey)
+	resutil.SetMeta(w, resutil.MetaKeyPeer, peerKey)
 
 	reply.Id = r.Id
 	if err := d.writeMsg(w, reply, logger); err != nil {
@@ -282,6 +336,30 @@ func (d *DnsInterceptor) ServeDNS(w dns.ResponseWriter, r *dns.Msg) {
 	}
 }
 
+// deferToChain hands the query to the next handler in the chain and asks the
+// chain to soften the negative verdict of whatever answers instead. The
+// resolvers below cannot prove a name inside the routed network absent, so their
+// NXDOMAIN must not reach the client: it would be cached for the name and every
+// type under it, taking the addresses this route does serve with it.
+func (d *DnsInterceptor) deferToChain(w dns.ResponseWriter, r *dns.Msg, logger *log.Entry, reason string) {
+	logger.Tracef("continuing to next handler for domain=%s type=%s reason=%s",
+		r.Question[0].Name, dns.TypeToString[r.Question[0].Qtype], reason)
+
+	resutil.RequestSoftNegative(w)
+	// Carried to the chain's response log line: without it the answer looks like
+	// it came from the fallthrough resolver on its own.
+	resutil.SetMeta(w, resutil.MetaKeyDeferredBy, "dns-route")
+	resutil.SetMeta(w, resutil.MetaKeyDeferredReason, reason)
+
+	resp := new(dns.Msg)
+	resp.SetRcode(r, dns.RcodeNameError)
+	// Set Zero bit to signal handler chain to continue
+	resp.MsgHdr.Zero = true
+	if err := w.WriteMsg(resp); err != nil {
+		logger.Errorf("failed writing DNS continue response: %v", err)
+	}
+}
+
 func (d *DnsInterceptor) writeDNSError(w dns.ResponseWriter, r *dns.Msg, logger *log.Entry, reason string) {
 	logger.Warnf("failed to query upstream for domain=%s: %s", r.Question[0].Name, reason)
 
@@ -621,3 +699,26 @@ func (d *DnsInterceptor) debugPeerTimeout(peerIP netip.Addr, peerKey string) str
 
 	return fmt.Sprintf(" (peer %s)", nbdns.FormatPeerStatus(&peerState))
 }
+
+// peerQuery builds the query sent to a peer's DNS forwarder and reports whether
+// the client itself advertised EDNS0. EDNS0 is added so the forwarder can return
+// an Extended DNS Error describing an upstream failure, and the OPT is stripped
+// from the reply again when the client did not ask for it. The client's message
+// is left untouched: it may still be handed to the next handler in the chain,
+// and neither the OPT nor the AD bit is ours to put on the wire on its behalf.
+func peerQuery(r *dns.Msg) (query *dns.Msg, hadEdns bool) {
+	query = r.Copy()
+	hadEdns = query.IsEdns0() != nil
+
+	// AD tells the forwarder we understand authenticated data. Only set when the
+	// client sent no additional section of its own, so we never overrule what it
+	// asked for.
+	if len(query.Extra) == 0 {
+		query.MsgHdr.AuthenticatedData = true
+	}
+	if !hadEdns {
+		query.SetEdns0(dns.DefaultMsgSize, false)
+	}
+
+	return query, hadEdns
+}

client/internal/routemanager/dnsinterceptor/handler_test.go

@@ -0,0 +1,399 @@
+package dnsinterceptor
+
+import (
+	"net"
+	"net/netip"
+	"sync"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/miekg/dns"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.zx2c4.com/wireguard/tun/netstack"
+
+	"github.com/netbirdio/netbird/client/iface/device"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+	"github.com/netbirdio/netbird/client/internal/dns/resutil"
+	"github.com/netbirdio/netbird/client/internal/dns/test"
+	"github.com/netbirdio/netbird/client/internal/peer"
+	"github.com/netbirdio/netbird/route"
+)
+
+// softNegativeWriter records what the handler told the chain: whether to soften
+// a negative verdict from the handlers it defers to, and the metadata that ends
+// up on the chain's response log line.
+type softNegativeWriter struct {
+	test.MockResponseWriter
+	softNegative bool
+	meta         map[resutil.MetaKey]string
+}
+
+func (w *softNegativeWriter) RequestSoftNegative() { w.softNegative = true }
+
+func (w *softNegativeWriter) RequestID() string { return "test" }
+
+func (w *softNegativeWriter) SetMeta(key resutil.MetaKey, value string) {
+	if w.meta == nil {
+		w.meta = make(map[resutil.MetaKey]string)
+	}
+	w.meta[key] = value
+}
+
+// TestServeDNS_QtypeTheForwarderCannotResolve covers the record types no DNS
+// forwarder can answer, because the host resolver exposes no API for them.
+// Asking the peer only burns a tunnel round trip, so the query goes straight to
+// the rest of the chain. No peer key is configured, proving no round trip is
+// attempted.
+func TestServeDNS_QtypeTheForwarderCannotResolve(t *testing.T) {
+	qtypes := []uint16{
+		dns.TypeHTTPS,
+		dns.TypeSVCB,
+		dns.TypeCAA,
+		dns.TypeNAPTR,
+		dns.TypeTLSA,
+		dns.TypeSOA,
+	}
+
+	for _, qtype := range qtypes {
+		t.Run(dns.TypeToString[qtype], func(t *testing.T) {
+			d := &DnsInterceptor{}
+			w := &softNegativeWriter{}
+
+			r := new(dns.Msg)
+			r.SetQuestion("db.example.com.", qtype)
+
+			d.ServeDNS(w, r)
+
+			resp := w.GetLastResponse()
+			require.NotNil(t, resp, "a response must be written")
+			assert.Equal(t, dns.RcodeNameError, resp.Rcode, "chain continuation is signalled as NXDOMAIN")
+			assert.True(t, resp.MsgHdr.Zero, "Zero bit must be set so the chain continues")
+			assert.True(t, w.softNegative,
+				"a downstream NXDOMAIN must be softened, or the fallthrough poisons the routed name")
+			assert.Equal(t, "dns-route", w.meta[resutil.MetaKeyDeferredBy],
+				"the chain's response log line must name us as the handler that stepped aside")
+			assert.Equal(t, "unsupported-qtype", w.meta[resutil.MetaKeyDeferredReason],
+				"the reason must survive to the log line, or the fallthrough is invisible")
+		})
+	}
+}
+
+// TestServeDNS_AddressQtypeNeverFallsThrough locks in the opposite case: A and
+// AAAA are what the route exists for, and the peer is authoritative for them.
+// A missing peer is an error the client must see, never a fallthrough to a
+// resolver that knows nothing about the routed name.
+func TestServeDNS_AddressQtypeNeverFallsThrough(t *testing.T) {
+	for _, qtype := range []uint16{dns.TypeA, dns.TypeAAAA} {
+		t.Run(dns.TypeToString[qtype], func(t *testing.T) {
+			d := &DnsInterceptor{}
+			w := &softNegativeWriter{}
+
+			r := new(dns.Msg)
+			r.SetQuestion("db.example.com.", qtype)
+
+			d.ServeDNS(w, r)
+
+			resp := w.GetLastResponse()
+			require.NotNil(t, resp, "a response must be written")
+			assert.Equal(t, dns.RcodeServerFailure, resp.Rcode, "an unusable route must fail, not fall through")
+			assert.False(t, resp.MsgHdr.Zero, "the chain must not continue for address queries")
+			assert.False(t, w.softNegative, "no fallthrough means nothing to soften")
+		})
+	}
+}
+
+// TestDispositionFor pins the query-type policy, and ties the types we ask a
+// peer about to the types a forwarder can actually resolve, so the two ends
+// cannot drift apart.
+func TestDispositionFor(t *testing.T) {
+	tests := []struct {
+		qtype uint16
+		want  disposition
+	}{
+		{dns.TypeA, dispositionPeer},
+		{dns.TypeAAAA, dispositionPeer},
+		{dns.TypeMX, dispositionPeerFirst},
+		{dns.TypeTXT, dispositionPeerFirst},
+		{dns.TypeNS, dispositionPeerFirst},
+		{dns.TypeSRV, dispositionPeerFirst},
+		{dns.TypeCNAME, dispositionPeerFirst},
+		{dns.TypePTR, dispositionPeerFirst},
+		{dns.TypeHTTPS, dispositionChain},
+		{dns.TypeSVCB, dispositionChain},
+		{dns.TypeCAA, dispositionChain},
+		{dns.TypeNAPTR, dispositionChain},
+		{dns.TypeTLSA, dispositionChain},
+		{dns.TypeDS, dispositionChain},
+		{dns.TypeDNSKEY, dispositionChain},
+		{dns.TypeSOA, dispositionChain},
+		{dns.TypeANY, dispositionChain},
+	}
+
+	for _, tt := range tests {
+		t.Run(dns.TypeToString[tt.qtype], func(t *testing.T) {
+			assert.Equal(t, tt.want, dispositionFor(tt.qtype), "disposition for %s", dns.TypeToString[tt.qtype])
+
+			// Address types are resolved by the forwarder's address path, not by
+			// its record path, so they are deliberately outside
+			// SupportedRecordQtype and their disposition does not depend on it.
+			if tt.want == dispositionPeer {
+				assert.False(t, resutil.SupportedRecordQtype(tt.qtype),
+					"address types take the forwarder's address path")
+				return
+			}
+
+			assert.Equal(t, tt.want == dispositionPeerFirst, resutil.SupportedRecordQtype(tt.qtype),
+				"only types a forwarder resolves may be sent to a peer")
+		})
+	}
+}
+
+// TestPeerCannotAnswer separates a peer that cannot resolve a type from one that
+// resolved it and found nothing. Only the former may be retried elsewhere:
+// treating NODATA or NXDOMAIN as a capability failure would send the client to a
+// public resolver behind the peer's back and prefer a public record over the
+// private one the route exists to reach.
+func TestPeerCannotAnswer(t *testing.T) {
+	cannot := []int{dns.RcodeNotImplemented, dns.RcodeFormatError}
+	for _, rcode := range cannot {
+		assert.True(t, peerCannotAnswer(rcode), "%s means the peer cannot answer", dns.RcodeToString[rcode])
+	}
+
+	// REFUSED belongs here, not above: it says the peer does not hold the name,
+	// so retrying elsewhere would leak the name of an internal-only domain to a
+	// public resolver. No forwarder version reports a missing record type this
+	// way, so nothing is lost by keeping it out.
+	can := []int{
+		dns.RcodeSuccess,
+		dns.RcodeNameError,
+		dns.RcodeServerFailure,
+		dns.RcodeNotAuth,
+		dns.RcodeRefused,
+	}
+	for _, rcode := range can {
+		assert.False(t, peerCannotAnswer(rcode), "%s is the peer's answer, not a capability failure", dns.RcodeToString[rcode])
+	}
+}
+
+// fakeWGIface is the minimum an interceptor needs to reach a peer's DNS
+// forwarder over a plain socket. A nil netstack keeps the exchange on the host
+// stack so the test can point it at a loopback listener.
+type fakeWGIface struct{}
+
+func (fakeWGIface) AddAllowedIP(string, netip.Prefix) error    { return nil }
+func (fakeWGIface) RemoveAllowedIP(string, netip.Prefix) error { return nil }
+func (fakeWGIface) Name() string                               { return "wt0" }
+func (fakeWGIface) Address() wgaddr.Address                    { return wgaddr.Address{} }
+func (fakeWGIface) ToInterface() *net.Interface                { return nil }
+func (fakeWGIface) IsUserspaceBind() bool                      { return false }
+func (fakeWGIface) GetFilter() device.PacketFilter             { return nil }
+func (fakeWGIface) GetDevice() *device.FilteredDevice          { return nil }
+func (fakeWGIface) GetNet() *netstack.Net                      { return nil }
+
+// fakePeerIPs resolves every peer key to the loopback address the test's
+// forwarder stub listens on.
+type fakePeerIPs struct {
+	addr netip.Addr
+}
+
+func (p fakePeerIPs) AllowedIP(string) (netip.Addr, bool) { return p.addr, p.addr.IsValid() }
+
+// forwarderStub stands in for the DNS forwarder of a routing peer, recording
+// what it was asked and answering with a canned reply.
+type forwarderStub struct {
+	mu       sync.Mutex
+	queries  []*dns.Msg
+	reply    func(q *dns.Msg) *dns.Msg
+	port     uint16
+	shutdown func()
+}
+
+func (s *forwarderStub) received() []*dns.Msg {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return append([]*dns.Msg(nil), s.queries...)
+}
+
+func newForwarderStub(t *testing.T, reply func(q *dns.Msg) *dns.Msg) *forwarderStub {
+	t.Helper()
+
+	conn, err := net.ListenUDP("udp", net.UDPAddrFromAddrPort(netip.MustParseAddrPort("127.0.0.1:0")))
+	require.NoError(t, err, "listen for the forwarder stub")
+
+	stub := &forwarderStub{reply: reply}
+	stub.port = uint16(conn.LocalAddr().(*net.UDPAddr).Port)
+
+	mux := dns.NewServeMux()
+	mux.HandleFunc(".", func(w dns.ResponseWriter, q *dns.Msg) {
+		stub.mu.Lock()
+		stub.queries = append(stub.queries, q.Copy())
+		stub.mu.Unlock()
+		_ = w.WriteMsg(stub.reply(q))
+	})
+
+	srv := &dns.Server{PacketConn: conn, Handler: mux}
+	started := make(chan struct{})
+	srv.NotifyStartedFunc = func() { close(started) }
+	go func() {
+		_ = srv.ActivateAndServe()
+	}()
+	select {
+	case <-started:
+	case <-time.After(5 * time.Second):
+		t.Fatal("forwarder stub did not start")
+	}
+
+	stub.shutdown = func() { _ = srv.Shutdown() }
+	t.Cleanup(stub.shutdown)
+
+	return stub
+}
+
+func newTestInterceptor(t *testing.T, stub *forwarderStub) *DnsInterceptor {
+	t.Helper()
+
+	port := new(atomic.Uint32)
+	port.Store(uint32(stub.port))
+
+	return &DnsInterceptor{
+		route:              &route.Route{Domains: nil},
+		statusRecorder:     peer.NewRecorder("https://mgm"),
+		currentPeerKey:     "peer-key",
+		interceptedDomains: make(domainMap),
+		wgInterface:        fakeWGIface{},
+		peerStore:          fakePeerIPs{addr: netip.MustParseAddr("127.0.0.1")},
+		forwarderPort:      port,
+	}
+}
+
+// TestServeDNS_PeerCannotResolveQtype is the reported regression: a routing peer
+// running a client older than the one that taught the forwarder about non-address
+// record types answers NOTIMP for every SRV query, and the interceptor used to
+// hand that straight to the application, breaking SRV-based service discovery.
+// The client cannot know the peer's capability up front, so it must try the peer
+// and fall back to the chain when the peer says it cannot answer.
+func TestServeDNS_PeerCannotResolveQtype(t *testing.T) {
+	for _, rcode := range []int{dns.RcodeNotImplemented, dns.RcodeFormatError} {
+		t.Run(dns.RcodeToString[rcode], func(t *testing.T) {
+			stub := newForwarderStub(t, func(q *dns.Msg) *dns.Msg {
+				resp := new(dns.Msg)
+				resp.SetRcode(q, rcode)
+				return resp
+			})
+			d := newTestInterceptor(t, stub)
+			w := &softNegativeWriter{}
+
+			r := new(dns.Msg)
+			r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+
+			d.ServeDNS(w, r)
+
+			require.Len(t, stub.received(), 1, "the peer must be asked before giving up on it")
+
+			resp := w.GetLastResponse()
+			require.NotNil(t, resp, "a response must be written")
+			assert.Equal(t, dns.RcodeNameError, resp.Rcode, "chain continuation is signalled as NXDOMAIN")
+			assert.True(t, resp.MsgHdr.Zero, "Zero bit must be set so the chain continues")
+			assert.True(t, w.softNegative, "the fallthrough must not be allowed to poison the routed name")
+			assert.Equal(t, "dns-route", w.meta[resutil.MetaKeyDeferredBy])
+			assert.Equal(t, "peer-rcode-"+dns.RcodeToString[rcode], w.meta[resutil.MetaKeyDeferredReason],
+				"the peer's verdict must be visible in the log, it is the only trace of the probe")
+		})
+	}
+}
+
+// TestServeDNS_PeerVerdictPassesThrough covers the replies that are the peer's
+// answer rather than a statement about its capability. The peer is authoritative
+// for a name routed to it, so its verdict reaches the client as it is: retrying
+// these elsewhere would send the name of an internal-only domain to a public
+// resolver and prefer whatever that resolver says over the route.
+func TestServeDNS_PeerVerdictPassesThrough(t *testing.T) {
+	// REFUSED means the peer does not consider the name routed to it, which
+	// happens while the client and the peer disagree about the route set. It is
+	// never how a peer reports a record type it cannot resolve: a forwarder too
+	// old for non-address types answers NOTIMP before it looks at the domain.
+	for _, rcode := range []int{dns.RcodeRefused, dns.RcodeNameError, dns.RcodeSuccess} {
+		t.Run(dns.RcodeToString[rcode], func(t *testing.T) {
+			stub := newForwarderStub(t, func(q *dns.Msg) *dns.Msg {
+				resp := new(dns.Msg)
+				resp.SetRcode(q, rcode)
+				return resp
+			})
+			d := newTestInterceptor(t, stub)
+			w := &softNegativeWriter{}
+
+			r := new(dns.Msg)
+			r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+
+			d.ServeDNS(w, r)
+
+			require.Len(t, stub.received(), 1, "the peer must be asked exactly once")
+
+			resp := w.GetLastResponse()
+			require.NotNil(t, resp, "a response must be written")
+			assert.Equal(t, rcode, resp.Rcode, "the peer's verdict must reach the client unchanged")
+			assert.False(t, resp.MsgHdr.Zero, "the chain must not continue past an answered query")
+			assert.False(t, w.softNegative, "nothing was deferred, so nothing may be softened")
+		})
+	}
+}
+
+// TestServeDNS_PeerResolvesQtype guards the common case: a peer that can answer
+// the type stays authoritative, and its records reach the client unchanged.
+func TestServeDNS_PeerResolvesQtype(t *testing.T) {
+	stub := newForwarderStub(t, func(q *dns.Msg) *dns.Msg {
+		resp := new(dns.Msg)
+		resp.SetReply(q)
+		resp.Answer = []dns.RR{&dns.SRV{
+			Hdr:    dns.RR_Header{Name: q.Question[0].Name, Rrtype: dns.TypeSRV, Class: dns.ClassINET, Ttl: 60},
+			Target: "shard-00.db.example.com.",
+			Port:   27017,
+		}}
+		return resp
+	})
+	d := newTestInterceptor(t, stub)
+	w := &softNegativeWriter{}
+
+	r := new(dns.Msg)
+	r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+
+	d.ServeDNS(w, r)
+
+	resp := w.GetLastResponse()
+	require.NotNil(t, resp, "a response must be written")
+	assert.Equal(t, dns.RcodeSuccess, resp.Rcode)
+	require.Len(t, resp.Answer, 1, "the peer's records must pass through")
+	assert.False(t, resp.MsgHdr.Zero, "an answered query must not continue the chain")
+	assert.False(t, w.softNegative)
+}
+
+// TestServeDNS_DeferredQueryIsPristine covers what the fallthrough hands to the
+// next handler. The interceptor advertises EDNS0 to the peer so the forwarder can
+// return an Extended DNS Error, and sets AD, but neither belongs to the client's
+// query: on the deferred path they would travel to the public resolver and the
+// reply could come back carrying an OPT the client never advertised, which
+// RFC 6891 forbids us from passing on.
+func TestServeDNS_DeferredQueryIsPristine(t *testing.T) {
+	stub := newForwarderStub(t, func(q *dns.Msg) *dns.Msg {
+		resp := new(dns.Msg)
+		resp.SetRcode(q, dns.RcodeNotImplemented)
+		return resp
+	})
+	d := newTestInterceptor(t, stub)
+	w := &softNegativeWriter{}
+
+	r := new(dns.Msg)
+	r.SetQuestion("_mongodb._tcp.db.example.com.", dns.TypeSRV)
+
+	d.ServeDNS(w, r)
+
+	sent := stub.received()
+	require.Len(t, sent, 1)
+	assert.NotNil(t, sent[0].IsEdns0(), "the peer must be asked with EDNS0 so it can return an EDE")
+
+	assert.Nil(t, r.IsEdns0(), "the deferred query must not carry an OPT the client never sent")
+	assert.False(t, r.AuthenticatedData, "the deferred query must not carry an AD bit the client never set")
+	assert.Empty(t, r.Extra, "the deferred query must reach the next handler as the client sent it")
+}

client/internal/routemanager/dnsinterceptor/handler_windows_test.go

@@ -0,0 +1,7 @@
+//go:build windows
+
+package dnsinterceptor
+
+// GetInterfaceGUIDString completes iface.WGIface on Windows, which requires the
+// interface GUID for DNS registration.
+func (fakeWGIface) GetInterfaceGUIDString() (string, error) { return "", nil }

client/internal/routemanager/dynamic/route.go

@@ -185,7 +185,7 @@ func (r *Route) startResolver(ctx context.Context) {
 }
 
 func (r *Route) update(ctx context.Context) error {
-	resolved, err := r.resolveDomains()
+	resolved, err := r.resolveDomains(ctx)
 	if err != nil {
 		if len(resolved) == 0 {
 			return fmt.Errorf("resolve domains: %w", err)
@@ -199,9 +199,9 @@ func (r *Route) update(ctx context.Context) error {
 	return nil
 }
 
-func (r *Route) resolveDomains() (domainMap, error) {
+func (r *Route) resolveDomains(ctx context.Context) (domainMap, error) {
 	results := make(chan resolveResult)
-	go r.resolve(results)
+	go r.resolve(ctx, results)
 
 	resolved := domainMap{}
 	var merr *multierror.Error
@@ -217,7 +217,7 @@ func (r *Route) resolveDomains() (domainMap, error) {
 	return resolved, nberrors.FormatErrorOrNil(merr)
 }
 
-func (r *Route) resolve(results chan resolveResult) {
+func (r *Route) resolve(ctx context.Context, results chan resolveResult) {
 	var wg sync.WaitGroup
 
 	for _, d := range r.route.Domains {
@@ -225,10 +225,10 @@ func (r *Route) resolve(results chan resolveResult) {
 		go func(domain domain.Domain) {
 			defer wg.Done()
 
-			ips, err := r.getIPsFromResolver(domain)
+			ips, err := r.getIPsFromResolver(ctx, domain)
 			if err != nil {
 				log.Tracef("Failed to resolve domain %s with private resolver: %v", domain.SafeString(), err)
-				ips, err = net.LookupIP(domain.PunycodeString())
+				ips, err = lookupHostIPs(ctx, domain)
 				if err != nil {
 					results <- resolveResult{domain: domain, err: fmt.Errorf("resolve d %s: %w", domain.SafeString(), err)}
 					return
@@ -364,6 +364,20 @@ func determinePrefixChanges(oldPrefixes, newPrefixes []netip.Prefix) (toAdd, toR
 	return
 }
 
+// lookupHostIPs resolves d via the system resolver, honoring ctx cancellation.
+func lookupHostIPs(ctx context.Context, d domain.Domain) ([]net.IP, error) {
+	addrs, err := net.DefaultResolver.LookupIPAddr(ctx, d.PunycodeString())
+	if err != nil {
+		return nil, err
+	}
+
+	ips := make([]net.IP, 0, len(addrs))
+	for _, addr := range addrs {
+		ips = append(ips, addr.IP)
+	}
+	return ips, nil
+}
+
 func combinePrefixes(oldPrefixes, removedPrefixes, addedPrefixes []netip.Prefix) []netip.Prefix {
 	prefixSet := make(map[netip.Prefix]struct{})
 	for _, prefix := range oldPrefixes {

client/internal/routemanager/dynamic/route_generic.go

@@ -3,11 +3,12 @@
 package dynamic
 
 import (
+	"context"
 	"net"
 
 	"github.com/netbirdio/netbird/shared/management/domain"
 )
 
-func (r *Route) getIPsFromResolver(domain domain.Domain) ([]net.IP, error) {
-	return net.LookupIP(domain.PunycodeString())
+func (r *Route) getIPsFromResolver(ctx context.Context, domain domain.Domain) ([]net.IP, error) {
+	return lookupHostIPs(ctx, domain)
 }

client/internal/routemanager/dynamic/route_ios.go

@@ -3,6 +3,7 @@
 package dynamic
 
 import (
+	"context"
 	"fmt"
 	"net"
 	"time"
@@ -16,7 +17,7 @@ import (
 
 const dialTimeout = 10 * time.Second
 
-func (r *Route) getIPsFromResolver(domain domain.Domain) ([]net.IP, error) {
+func (r *Route) getIPsFromResolver(ctx context.Context, domain domain.Domain) ([]net.IP, error) {
 	privateClient, err := nbdns.GetClientPrivate(r.wgInterface, r.resolverAddr.Addr(), dialTimeout)
 	if err != nil {
 		return nil, fmt.Errorf("error while creating private client: %s", err)
@@ -32,7 +33,7 @@ func (r *Route) getIPsFromResolver(domain domain.Domain) ([]net.IP, error) {
 		msg := new(dns.Msg)
 		msg.SetQuestion(fqdn, qtype)
 
-		response, _, err := nbdns.ExchangeWithFallback(nil, privateClient, msg, r.resolverAddr.String())
+		response, _, err := nbdns.ExchangeWithFallback(ctx, privateClient, msg, r.resolverAddr.String())
 		if err != nil {
 			if queryErr == nil {
 				queryErr = fmt.Errorf("DNS query for %s (type %d) after %s: %w", domain.SafeString(), qtype, time.Since(startTime), err)

client/internal/routemanager/manager.go

@@ -61,6 +61,7 @@ type Manager interface {
 	InitialRouteRange() []string
 	SetFirewall(firewall.Manager) error
 	SetDNSForwarderPort(port uint16)
+	ReconcilePeerAllowedIPs(peerKey string) error
 	Stop(stateManager *statemanager.Manager)
 }
 
@@ -232,6 +233,30 @@ func (m *DefaultManager) setupRefCounters(useNoop bool) {
 	)
 }
 
+// ReconcilePeerAllowedIPs re-applies every routed allowed IP currently tracked for the peer
+// onto the WireGuard device. The allowed-IP refcounter only calls its AddFunc (which pushes to
+// the device) on a prefix's 0->1 transition, so a peer whose device entry was rebuilt without a
+// matching refcounter change — e.g. a lazy connection cycling through idle->wake, which recreates
+// the WireGuard peer with the overlay /32 only — ends up missing routed prefixes the refcounter
+// still considers installed, and nothing retries. Calling this when the peer's WireGuard entry is
+// (re)created restores convergence. It is add-only and idempotent: AddAllowedIP is update-only, so
+// prefixes are re-added to an existing peer and an absent peer is left untouched.
+func (m *DefaultManager) ReconcilePeerAllowedIPs(peerKey string) error {
+	if m.allowedIPsRefCounter == nil {
+		return nil
+	}
+
+	return m.allowedIPsRefCounter.ReapplyMatching(
+		func(out string) bool { return out == peerKey },
+		func(prefix netip.Prefix) error {
+			if err := m.wgInterface.AddAllowedIP(peerKey, prefix); err != nil {
+				return fmt.Errorf("add allowed IP %s for peer %s: %w", prefix, peerKey, err)
+			}
+			return nil
+		},
+	)
+}
+
 // Init sets up the routing
 func (m *DefaultManager) Init() error {
 	m.routeSelector = m.initSelector()

client/internal/routemanager/mock.go

@@ -112,6 +112,11 @@ func (m *MockManager) SetFirewall(firewall.Manager) error {
 func (m *MockManager) SetDNSForwarderPort(port uint16) {
 }
 
+// ReconcilePeerAllowedIPs mock implementation of ReconcilePeerAllowedIPs from Manager interface
+func (m *MockManager) ReconcilePeerAllowedIPs(peerKey string) error {
+	return nil
+}
+
 // Stop mock implementation of Stop from Manager interface
 func (m *MockManager) Stop(stateManager *statemanager.Manager) {
 	if m.StopFunc != nil {

client/internal/routemanager/reconcile_test.go

@@ -0,0 +1,90 @@
+//go:build !windows
+
+package routemanager
+
+import (
+	"net"
+	"net/netip"
+	"sync"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.zx2c4.com/wireguard/tun/netstack"
+
+	"github.com/netbirdio/netbird/client/iface/device"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+	"github.com/netbirdio/netbird/client/internal/routemanager/refcounter"
+)
+
+// reconcileWGMock is a minimal iface.WGIface that only records AddAllowedIP calls; every other
+// method is an inert stub because ReconcilePeerAllowedIPs exercises none of them.
+type reconcileWGMock struct {
+	mu   sync.Mutex
+	adds map[string][]netip.Prefix
+}
+
+func (m *reconcileWGMock) AddAllowedIP(peerKey string, allowedIP netip.Prefix) error {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	if m.adds == nil {
+		m.adds = map[string][]netip.Prefix{}
+	}
+	m.adds[peerKey] = append(m.adds[peerKey], allowedIP)
+	return nil
+}
+
+func (m *reconcileWGMock) added(peerKey string) []netip.Prefix {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.adds[peerKey]
+}
+
+func (m *reconcileWGMock) RemoveAllowedIP(string, netip.Prefix) error { return nil }
+func (m *reconcileWGMock) Name() string                               { return "utun-test" }
+func (m *reconcileWGMock) Address() wgaddr.Address                    { return wgaddr.Address{} }
+func (m *reconcileWGMock) ToInterface() *net.Interface                { return nil }
+func (m *reconcileWGMock) IsUserspaceBind() bool                      { return false }
+func (m *reconcileWGMock) GetFilter() device.PacketFilter             { return nil }
+func (m *reconcileWGMock) GetDevice() *device.FilteredDevice          { return nil }
+func (m *reconcileWGMock) GetNet() *netstack.Net                      { return nil }
+
+// TestReconcilePeerAllowedIPs verifies the declarative reconcile re-applies every routed prefix
+// tracked for the peer (self-heal, independent of refcount level) and stays scoped to that peer.
+func TestReconcilePeerAllowedIPs(t *testing.T) {
+	wg := &reconcileWGMock{}
+	m := &DefaultManager{wgInterface: wg}
+	m.allowedIPsRefCounter = refcounter.New[netip.Prefix, string, string](
+		func(_ netip.Prefix, peerKey string) (string, error) { return peerKey, nil },
+		func(netip.Prefix, string) error { return nil },
+	)
+
+	peerA1 := netip.MustParsePrefix("10.0.0.0/24")
+	peerA2 := netip.MustParsePrefix("10.1.0.0/24")
+	peerB1 := netip.MustParsePrefix("10.2.0.0/24")
+
+	for prefix, peer := range map[netip.Prefix]string{peerA1: "peerA", peerA2: "peerA", peerB1: "peerB"} {
+		_, err := m.allowedIPsRefCounter.Increment(prefix, peer)
+		require.NoError(t, err)
+	}
+	// Extra reference: reconcile must still re-apply the prefix even though its refcount never
+	// hit 0 again (the exact case the plain incremental path skips).
+	_, err := m.allowedIPsRefCounter.Increment(peerA1, "peerA")
+	require.NoError(t, err)
+
+	require.NoError(t, m.ReconcilePeerAllowedIPs("peerA"))
+
+	assert.ElementsMatch(t, []netip.Prefix{peerA1, peerA2}, wg.added("peerA"),
+		"reconcile must re-apply all routed prefixes of the peer")
+	assert.Empty(t, wg.added("peerB"), "reconcile must not touch another peer's prefixes")
+}
+
+// TestReconcilePeerAllowedIPsNoCounter verifies reconcile is a safe no-op before the refcounter is
+// set up.
+func TestReconcilePeerAllowedIPsNoCounter(t *testing.T) {
+	wg := &reconcileWGMock{}
+	m := &DefaultManager{wgInterface: wg}
+
+	require.NoError(t, m.ReconcilePeerAllowedIPs("peerA"))
+	assert.Empty(t, wg.added("peerA"))
+}

client/internal/routemanager/refcounter/refcounter.go

@@ -94,6 +94,26 @@ func (rm *Counter[Key, I, O]) Get(key Key) (Ref[O], bool) {
 	return ref, ok
 }
 
+// ReapplyMatching calls apply for every key whose stored Out satisfies pred, holding the
+// counter lock for the whole pass. Running apply under the lock keeps it atomic with respect
+// to Increment/Decrement: a prefix dropped to zero is removed from the map (and had its
+// RemoveFunc called) before this pass observes it, so a stale key can never be re-applied.
+// pred and apply are invoked under the lock, so they must not call back into the counter.
+func (rm *Counter[Key, I, O]) ReapplyMatching(pred func(out O) bool, apply func(key Key) error) error {
+	rm.mu.Lock()
+	defer rm.mu.Unlock()
+
+	var merr *multierror.Error
+	for key, ref := range rm.refCountMap {
+		if pred(ref.Out) {
+			if err := apply(key); err != nil {
+				merr = multierror.Append(merr, err)
+			}
+		}
+	}
+	return nberrors.FormatErrorOrNil(merr)
+}
+
 // Increment increments the reference count for the given key.
 // If this is the first reference to the key, the AddFunc is called.
 func (rm *Counter[Key, I, O]) Increment(key Key, in I) (Ref[O], error) {

client/internal/routemanager/refcounter/refcounter_test.go

@@ -0,0 +1,47 @@
+package refcounter
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// TestReapplyMatching verifies ReapplyMatching invokes apply for exactly the keys whose stored
+// Out satisfies the predicate (no duplicates for multiply-referenced keys) — the primitive
+// ReconcilePeerAllowedIPs relies on to re-apply a single peer's routed prefixes.
+func TestReapplyMatching(t *testing.T) {
+	rc := New[netip.Prefix, string, string](
+		func(_ netip.Prefix, peerKey string) (string, error) { return peerKey, nil },
+		func(netip.Prefix, string) error { return nil },
+	)
+
+	peerA1 := netip.MustParsePrefix("10.0.0.0/24")
+	peerA2 := netip.MustParsePrefix("10.1.0.0/24")
+	peerB1 := netip.MustParsePrefix("10.2.0.0/24")
+
+	for prefix, peer := range map[netip.Prefix]string{peerA1: "peerA", peerA2: "peerA", peerB1: "peerB"} {
+		_, err := rc.Increment(prefix, peer)
+		require.NoError(t, err)
+	}
+	// a second reference must not make the key applied twice
+	_, err := rc.Increment(peerA1, "peerA")
+	require.NoError(t, err)
+
+	var applied []netip.Prefix
+	err = rc.ReapplyMatching(
+		func(out string) bool { return out == "peerA" },
+		func(key netip.Prefix) error { applied = append(applied, key); return nil },
+	)
+	require.NoError(t, err)
+	assert.ElementsMatch(t, []netip.Prefix{peerA1, peerA2}, applied)
+
+	var none []netip.Prefix
+	err = rc.ReapplyMatching(
+		func(out string) bool { return out == "missing" },
+		func(key netip.Prefix) error { none = append(none, key); return nil },
+	)
+	require.NoError(t, err)
+	assert.Empty(t, none)
+}

client/internal/statemanager/manager.go

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

client/ios/NetBirdSDK/client.go

@@ -233,6 +233,9 @@ func (c *Client) DebugBundle(anonymize bool) (string, error) {
 		deps.SyncResponse = resp
 
 		if e := cc.Engine(); e != nil {
+			deps.RefreshStatus = func() {
+				e.RunHealthProbes(context.Background(), true)
+			}
 			if cm := e.GetClientMetrics(); cm != nil {
 				deps.ClientMetrics = cm
 			}

client/ios/NetBirdSDK/login.go

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

client/server/server.go

@@ -1081,7 +1081,10 @@ func (s *Server) Down(ctx context.Context, _ *proto.DownRequest) (*proto.DownRes
 
 	if err := s.cleanupConnection(); err != nil {
 		s.mutex.Unlock()
-		// todo review to update the status in case any type of error
+		if errors.Is(err, ErrServiceNotUp) {
+			log.Debugf("Down called while service not up: %v", err)
+			return nil, err
+		}
 		log.Errorf("failed to shut down properly: %v", err)
 		return nil, err
 	}
@@ -1154,7 +1157,7 @@ func (s *Server) cleanupConnection() error {
 	// making the run loop the sole owner of engine shutdown.
 	if engine != nil {
 		if err := engine.Stop(); err != nil {
-			return err
+			log.Errorf("failed to stop engine during cleanup: %v", err)
 		}
 	}
 

client/ssh/config/manager.go

@@ -14,6 +14,7 @@ import (
 	log "github.com/sirupsen/logrus"
 
 	nbssh "github.com/netbirdio/netbird/client/ssh"
+	"github.com/netbirdio/netbird/shared/management/domain"
 )
 
 const (
@@ -218,11 +219,20 @@ func (m *Manager) buildHostPatterns(peer PeerSSHInfo) []string {
 	if peer.IPv6.IsValid() {
 		hostPatterns = append(hostPatterns, peer.IPv6.String())
 	}
-	if peer.FQDN != "" {
+	// Peer FQDNs and hostnames originate from remote peers, so they must be
+	// validated as plain DNS names before being embedded in the ssh_config
+	// "Match host" pattern list. This prevents injection of arbitrary
+	// ssh_config directives via embedded quotes, whitespace, newlines, the
+	// comma pattern separator, or the "*"/"?" pattern metacharacters.
+	if domain.IsValidDomainNoWildcard(peer.FQDN) {
 		hostPatterns = append(hostPatterns, peer.FQDN)
+	} else if peer.FQDN != "" {
+		log.Warnf("skipping peer FQDN with invalid characters in SSH config: %q", peer.FQDN)
 	}
-	if peer.Hostname != "" && peer.Hostname != peer.FQDN {
+	if peer.Hostname != peer.FQDN && domain.IsValidDomainNoWildcard(peer.Hostname) {
 		hostPatterns = append(hostPatterns, peer.Hostname)
+	} else if peer.Hostname != "" && peer.Hostname != peer.FQDN {
+		log.Warnf("skipping peer hostname with invalid characters in SSH config: %q", peer.Hostname)
 	}
 	return hostPatterns
 }

client/ssh/config/manager_test.go

@@ -148,6 +148,45 @@ func TestManager_MatchHostFormat(t *testing.T) {
 		"should use Match host with comma-separated patterns")
 }
 
+func TestManager_HostPatternInjection(t *testing.T) {
+	tempDir, err := os.MkdirTemp("", "netbird-ssh-config-test")
+	require.NoError(t, err)
+	defer func() { assert.NoError(t, os.RemoveAll(tempDir)) }()
+
+	manager := &Manager{
+		sshConfigDir:  filepath.Join(tempDir, "ssh_config.d"),
+		sshConfigFile: "99-netbird.conf",
+	}
+
+	// A malicious peer FQDN/hostname attempts to break out of the Match host
+	// directive and inject arbitrary ssh_config (a ProxyCommand executing a
+	// command). It must be rejected, not written to the config.
+	peers := []PeerSSHInfo{
+		{
+			Hostname: "evil\"\n    ProxyCommand touch /tmp/pwned\nHost x",
+			IP:       netip.MustParseAddr("100.125.1.1"),
+			FQDN:     "evil\"\n    ProxyCommand touch /tmp/pwned\nHost x.nb.internal",
+		},
+		{Hostname: "peer2", IP: netip.MustParseAddr("100.125.1.2"), FQDN: "peer2.nb.internal"},
+	}
+
+	err = manager.SetupSSHClientConfig(peers)
+	require.NoError(t, err)
+
+	configPath := filepath.Join(manager.sshConfigDir, manager.sshConfigFile)
+	content, err := os.ReadFile(configPath)
+	require.NoError(t, err)
+	configStr := string(content)
+
+	assert.NotContains(t, configStr, "ProxyCommand touch /tmp/pwned",
+		"injected directive must not appear in generated config")
+	assert.NotContains(t, configStr, "evil",
+		"malicious pattern must be dropped entirely")
+	// The valid peer must still be present, on a single Match host line.
+	assert.Contains(t, configStr, "Match host \"100.125.1.1,100.125.1.2,peer2.nb.internal,peer2\"",
+		"valid peers must survive, injected patterns dropped")
+}
+
 func TestManager_ForcedSSHConfig(t *testing.T) {
 	// Set force environment variable
 	t.Setenv(EnvForceSSHConfig, "true")

client/ssh/server/getent_unix.go

@@ -69,7 +69,8 @@ func parseGetentPasswd(output string) (*user.User, string, error) {
 
 // validateGetentInput checks that the input is safe to pass to getent or id.
 // Allows POSIX usernames, numeric UIDs, and common NSS extensions
-// (@ for Kerberos, $ for Samba, + for NIS compat).
+// (@ for Kerberos, $ for Samba, + for NIS compat). A leading hyphen is
+// rejected so the input can never be parsed as a command-line flag.
 func validateGetentInput(input string) bool {
 	maxLen := 32
 	if runtime.GOOS == "linux" {
@@ -80,6 +81,10 @@ func validateGetentInput(input string) bool {
 		return false
 	}
 
+	if input[0] == '-' {
+		return false
+	}
+
 	for _, r := range input {
 		if isAllowedGetentChar(r) {
 			continue

client/ssh/server/getent_unix_test.go

@@ -157,6 +157,9 @@ func TestValidateGetentInput(t *testing.T) {
 		{"numeric UID", "1001", true},
 		{"dots and underscores", "alice.bob_test", true},
 		{"hyphen", "alice-bob", true},
+		{"leading hyphen rejected", "-i", false},
+		{"leading double hyphen rejected", "--no-idn", false},
+		{"lone hyphen rejected", "-", false},
 		{"kerberos principal", "user@REALM", true},
 		{"samba machine account", "MACHINE$", true},
 		{"NIS compat", "+user", true},

client/system/info.go

@@ -79,13 +79,15 @@ type Info struct {
 	EnableSSHLocalPortForwarding  bool
 	EnableSSHRemotePortForwarding bool
 	DisableSSHAuth                bool
+
+	SyncMessageVersion *int
 }
 
 func (i *Info) SetFlags(
 	rosenpassEnabled, rosenpassPermissive bool,
 	serverSSHAllowed *bool,
 	disableClientRoutes, disableServerRoutes,
-	disableDNS, disableFirewall, blockLANAccess, blockInbound, disableIPv6 bool,
+	disableDNS, disableFirewall, blockLANAccess, blockInbound, disableIPv6 bool, syncMessageVersion *int,
 	enableSSHRoot, enableSSHSFTP, enableSSHLocalPortForwarding, enableSSHRemotePortForwarding *bool,
 	disableSSHAuth *bool,
 ) {
@@ -103,6 +105,8 @@ func (i *Info) SetFlags(
 	i.BlockInbound = blockInbound
 	i.DisableIPv6 = disableIPv6
 
+	i.SyncMessageVersion = syncMessageVersion
+
 	if enableSSHRoot != nil {
 		i.EnableSSHRoot = *enableSSHRoot
 	}

client/ui/autostart_default.go

@@ -51,7 +51,7 @@ func autostartDisabledByMDM(policy *mdm.Policy) bool {
 // netbirdFootprintExists reports whether the machine already carries NetBird
 // daemon config or state, meaning this is not a genuinely fresh install. It is
 // the update-safety gate for the autostart default: upgrading users always
-// have a footprint, so an update can never trigger a login-item write.
+// have a footprint, so an update can never trigger a autostart entry write.
 func netbirdFootprintExists() bool {
 	candidates := []string{
 		profilemanager.DefaultConfigPath,
@@ -69,9 +69,23 @@ func netbirdFootprintExists() bool {
 // applyAutostartDefault runs the one-time launch-on-login default for genuinely
 // fresh installs. The autostartInitialized marker is persisted before any
 // enable attempt so a crash mid-flow degrades to "never enabled" instead of
-// retrying login-item writes on every launch. A user's later disable in
+// retrying autostart entry writes on every launch. A user's later disable in
 // Settings is never overridden: the marker guarantees at-most-once, ever.
 func applyAutostartDefault(ctx context.Context, autostart *services.Autostart, prefs *preferences.Store, prefsFileExisted bool) {
+	mdmDisabled := autostartDisabledByMDM(mdm.LoadPolicy())
+
+	if mdmDisabled {
+		if enabled, err := autostart.IsEnabled(ctx); err != nil {
+			log.Warnf("MDM disableAutostart: read autostart state: %v", err)
+		} else if enabled {
+			if err := autostart.SetEnabled(ctx, false); err != nil {
+				log.Warnf("MDM disableAutostart: force off failed: %v", err)
+			} else {
+				log.Info("MDM disableAutostart enforced: autostart turned off")
+			}
+		}
+	}
+
 	priorFootprint := netbirdFootprintExists() || prefsFileExisted
 
 	if prefs.Get().AutostartInitialized {
@@ -84,7 +98,7 @@ func applyAutostartDefault(ctx context.Context, autostart *services.Autostart, p
 
 	state := autostartDefaultState{
 		supported:    autostart.Supported(ctx),
-		mdmDisabled:  autostartDisabledByMDM(mdm.LoadPolicy()),
+		mdmDisabled:  mdmDisabled,
 		priorInstall: priorFootprint,
 	}
 	enable, reason := shouldEnableAutostartDefault(state)

client/ui/frontend/src/components/empty-state/DaemonOutdatedOverlay.tsx

@@ -1,10 +1,13 @@
+import { useEffect, useState } from "react";
 import { useTranslation } from "react-i18next";
 import { AlertTriangleIcon, DownloadIcon } from "lucide-react";
 import { Browser } from "@wailsio/runtime";
+import { Version } from "@bindings/services";
 import { Button } from "@/components/buttons/Button";
 import { useStatus } from "@/contexts/StatusContext.tsx";
 
 const RELEASES_URL = "https://github.com/netbirdio/netbird/releases/latest";
+const RC_RELEASES_URL = "https://pkgs.netbird.io/releases/rc";
 
 function openUrl(url: string) {
     Browser.OpenURL(url).catch(() => globalThis.open(url, "_blank"));
@@ -12,7 +15,26 @@ function openUrl(url: string) {
 
 export const DaemonOutdatedOverlay = () => {
     const { t } = useTranslation();
-    const { isDaemonOutdated } = useStatus();
+    const { status, isDaemonOutdated } = useStatus();
+
+    const [guiVersion, setGuiVersion] = useState<string>("-");
+    const clientVersion = status?.daemonVersion ?? "—";
+
+    const isRc = /-rc/i.test(guiVersion) || /-rc/i.test(clientVersion);
+    const downloadUrl = isRc ? RC_RELEASES_URL : RELEASES_URL;
+
+    useEffect(() => {
+        if (!isDaemonOutdated) return;
+        let cancelled = false;
+        Version.GUI()
+            .then((v) => {
+                if (!cancelled) setGuiVersion(v);
+            })
+            .catch((err) => console.error("[DaemonOutdatedOverlay] GUI version error", err));
+        return () => {
+            cancelled = true;
+        };
+    }, [isDaemonOutdated]);
 
     if (!isDaemonOutdated) return null;
 
@@ -38,10 +60,37 @@ export const DaemonOutdatedOverlay = () => {
                     <p className={"text-sm text-nb-gray-300"}>{t("daemon.outdated.description")}</p>
                 </div>
 
+                <div className={"flex flex-col items-center gap-0.5 text-center"}>
+                    <p className={"text-sm font-semibold text-nb-gray-100"}>
+                        {clientVersion === "development" ? (
+                            <span>
+                                {t("settings.about.clientName")}{" "}
+                                <span className={"font-mono text-yellow-400"}>
+                                    {t("settings.about.development")}
+                                </span>
+                            </span>
+                        ) : (
+                            t("settings.about.client", { version: clientVersion })
+                        )}
+                    </p>
+                    <p className={"text-sm font-medium text-nb-gray-250"}>
+                        {guiVersion === "development" ? (
+                            <span>
+                                {t("settings.about.guiName")}{" "}
+                                <span className={"font-mono text-yellow-400"}>
+                                    {t("settings.about.development")}
+                                </span>
+                            </span>
+                        ) : (
+                            t("settings.about.gui", { version: guiVersion })
+                        )}
+                    </p>
+                </div>
+
                 <div className={"wails-no-draggable"}>
-                    <Button variant={"primary"} size={"xs"} onClick={() => openUrl(RELEASES_URL)}>
+                    <Button variant={"primary"} size={"xs"} onClick={() => openUrl(downloadUrl)}>
                         <DownloadIcon size={14} />
-                        {t("update.card.getInstaller")}
+                        {t("daemon.outdated.download")}
                     </Button>
                 </div>
             </div>

client/ui/frontend/src/contexts/ProfileContext.tsx

@@ -28,6 +28,7 @@ type ProfileContextValue = {
     loaded: boolean;
     refresh: () => Promise<void>;
     switchProfile: (id: string) => Promise<void>;
+    switchProfileNoConnect: (id: string) => Promise<void>;
     addProfile: (name: string) => Promise<string>;
     removeProfile: (id: string) => Promise<void>;
     renameProfile: (id: string, newName: string) => Promise<void>;
@@ -112,6 +113,16 @@ export const ProfileProvider = ({ children }: { children: ReactNode }) => {
         [username, refresh],
     );
 
+    // Manage-profiles variant: switches without connecting, so the user can
+    // still adjust the management URL before bringing the connection up.
+    const switchProfileNoConnect = useCallback(
+        async (id: string) => {
+            await ProfileSwitcher.SwitchActiveNoConnect({ profileName: id, username });
+            await refresh();
+        },
+        [username, refresh],
+    );
+
     // addProfile creates a profile by display name and returns the
     // daemon-generated ID, so the caller can immediately address it by ID.
     const addProfile = useCallback(
@@ -158,6 +169,7 @@ export const ProfileProvider = ({ children }: { children: ReactNode }) => {
             loaded,
             refresh,
             switchProfile,
+            switchProfileNoConnect,
             addProfile,
             removeProfile,
             renameProfile,
@@ -171,6 +183,7 @@ export const ProfileProvider = ({ children }: { children: ReactNode }) => {
             loaded,
             refresh,
             switchProfile,
+            switchProfileNoConnect,
             addProfile,
             removeProfile,
             renameProfile,

client/ui/frontend/src/modules/profiles/ProfilesTab.tsx

@@ -45,7 +45,7 @@ export function ProfilesTab() {
         activeProfileId,
         loaded,
         username,
-        switchProfile,
+        switchProfileNoConnect,
         addProfile,
         removeProfile,
         renameProfile,
@@ -100,7 +100,7 @@ export function ProfilesTab() {
             confirmLabel: t("profile.switch.confirm"),
         });
         if (!ok) return;
-        await guarded(i18next.t("profile.error.switchTitle"), () => switchProfile(id));
+        await guarded(i18next.t("profile.error.switchTitle"), () => switchProfileNoConnect(id));
     };
 
     const handleDeregister = async (id: string, name: string) => {
@@ -129,14 +129,13 @@ export function ProfilesTab() {
         await guarded(i18next.t("profile.error.createTitle"), async () => {
             const id = await addProfile(name);
             // SetConfig is keyed by the new profile's ID, so it writes the
-            // not-yet-active profile. Write before switching so any reconnect
-            // targets the right deployment.
+            // not-yet-active profile before the switch makes it current.
             if (!isNetbirdCloud(managementUrl)) {
                 await SettingsSvc.SetConfig(
                     new SetConfigParams({ profileName: id, username, managementUrl }),
                 );
             }
-            await switchProfile(id);
+            await switchProfileNoConnect(id);
         });
     };
 

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

@@ -73,6 +73,13 @@ export default function SessionExpirationDialog() {
 
         let offCancel: (() => void) | undefined;
 
+        // Return the dialog to its interactive state and dismiss the browser popup
+        const resetDialog = () => {
+            offCancel?.();
+            WindowManager.CloseBrowserLogin().catch(console.error);
+            setBusy(false);
+        };
+
         try {
             const start = await Session.RequestExtend({ hint: "" });
             const uri = start.verificationUriComplete || start.verificationUri;
@@ -105,25 +112,22 @@ export default function SessionExpirationDialog() {
             if (outcome.kind === "cancel") {
                 waitPromise.cancel?.();
                 waitPromise.catch(() => {});
+                resetDialog();
                 return;
             }
 
             // Another surface owns this flow; keep the dialog open to retry.
             if (outcome.result.preempted) {
+                resetDialog();
                 return;
             }
-
-            // Close before the popup so the restore can't flash this window back.
-            WindowManager.CloseSessionExpiration().catch(console.error);
+            WindowManager.CloseRenewFlow().catch(console.error);
         } catch (e) {
+            resetDialog();
             await errorDialog({
                 Title: t("sessionExpiration.extendFailedTitle"),
                 Message: formatErrorMessage(e),
             });
-        } finally {
-            offCancel?.();
-            WindowManager.CloseBrowserLogin().catch(console.error);
-            setBusy(false);
         }
     }, [busy, t]);
 
@@ -139,12 +143,11 @@ export default function SessionExpirationDialog() {
             });
             WindowManager.CloseSessionExpiration().catch(console.error);
         } catch (e) {
+            setBusy(false);
             await errorDialog({
                 Title: t("sessionExpiration.logoutFailedTitle"),
                 Message: formatErrorMessage(e),
             });
-        } finally {
-            setBusy(false);
         }
     }, [busy, t]);
 

client/ui/frontend/src/modules/welcome/WelcomeStepTray.tsx

@@ -22,6 +22,9 @@ type WelcomeStepTrayProps = {
 export function WelcomeStepTray({ onContinue }: Readonly<WelcomeStepTrayProps>) {
     const { t } = useTranslation();
     const trayScreenshot = trayScreenshotForOS();
+    // macOS has no tray — the icon sits in the menu bar, so the copy says so.
+    const titleKey = isMacOS() ? "welcome.titleMac" : "welcome.title";
+    const descriptionKey = isMacOS() ? "welcome.descriptionMac" : "welcome.description";
 
     return (
         <>
@@ -36,9 +39,9 @@ export function WelcomeStepTray({ onContinue }: Readonly<WelcomeStepTrayProps>)
 
             <div className={"flex w-full flex-col gap-1"}>
                 <DialogHeading id={"nb-welcome-title"} align={"left"}>
-                    {t("welcome.title")}
+                    {t(titleKey)}
                 </DialogHeading>
-                <DialogDescription align={"left"}>{t("welcome.description")}</DialogDescription>
+                <DialogDescription align={"left"}>{t(descriptionKey)}</DialogDescription>
             </div>
 
             <DialogActions>

client/ui/i18n/locales/_index.json

@@ -8,6 +8,7 @@
         {"code": "fr", "displayName": "Français", "englishName": "French"},
         {"code": "it", "displayName": "Italiano", "englishName": "Italian"},
         {"code": "pt", "displayName": "Português", "englishName": "Portuguese"},
-        {"code": "zh-CN", "displayName": "简体中文", "englishName": "Simplified Chinese"}
+        {"code": "zh-CN", "displayName": "简体中文", "englishName": "Simplified Chinese"},
+        {"code": "ja", "displayName": "日本語", "englishName": "Japanese"}
     ]
 }

client/ui/i18n/locales/de/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Suchen Sie NetBird in der Taskleiste"
     },
+    "welcome.titleMac": {
+        "message": "Suchen Sie NetBird in der Menüleiste"
+    },
     "welcome.description": {
         "message": "NetBird läuft in Ihrer Taskleiste. Klicken Sie auf das Symbol, um sich zu verbinden, Profile zu wechseln oder die Einstellungen zu öffnen."
     },
+    "welcome.descriptionMac": {
+        "message": "NetBird läuft in Ihrer Menüleiste. Klicken Sie auf das Symbol, um sich zu verbinden, Profile zu wechseln oder die Einstellungen zu öffnen."
+    },
     "welcome.continue": {
         "message": "Weiter"
     },
@@ -1293,10 +1299,13 @@
         "message": "Dokumentation"
     },
     "daemon.outdated.title": {
-        "message": "NetBird-Dienst ist veraltet"
+        "message": "NetBird Client ist veraltet"
     },
     "daemon.outdated.description": {
-        "message": "Aktualisieren Sie den NetBird-Dienst, um diese App zu verwenden."
+        "message": "Die neue GUI ist nicht mit Ihrem älteren Client kompatibel. Aktualisieren Sie Ihren Client, um die neue Anwendung zu verwenden."
+    },
+    "daemon.outdated.download": {
+        "message": "Neueste Version herunterladen"
     },
     "error.jwt_clock_skew": {
         "message": "Anmeldung fehlgeschlagen: Die Uhr dieses Geräts ist nicht mit dem Server synchron. Bitte synchronisieren Sie die Systemuhr und versuchen Sie es erneut."

client/ui/i18n/locales/en/common.json

@@ -1377,11 +1377,19 @@
     },
     "welcome.title": {
         "message": "Look for NetBird in your tray",
-        "description": "Heading on the first onboarding step, pointing the user to the tray icon. 'tray' = system tray / menu bar."
+        "description": "Heading on the first onboarding step, pointing the user to the tray icon. Shown on Windows and Linux; macOS uses welcome.titleMac."
+    },
+    "welcome.titleMac": {
+        "message": "Look for NetBird in your menu bar",
+        "description": "Heading on the first onboarding step on macOS, pointing the user to the menu bar icon. Use your language's Apple term for the macOS menu bar."
     },
     "welcome.description": {
         "message": "NetBird lives in your tray. Click the icon to connect, switch profiles, or open settings.",
-        "description": "Body of the first onboarding step explaining the tray icon."
+        "description": "Body of the first onboarding step explaining the tray icon. Shown on Windows and Linux; macOS uses welcome.descriptionMac."
+    },
+    "welcome.descriptionMac": {
+        "message": "NetBird lives in your menu bar. Click the icon to connect, switch profiles, or open settings.",
+        "description": "Body of the first onboarding step on macOS explaining the menu bar icon. Use your language's Apple term for the macOS menu bar."
     },
     "welcome.continue": {
         "message": "Continue",
@@ -1724,12 +1732,16 @@
         "description": "Documentation link on the daemon-unavailable overlay."
     },
     "daemon.outdated.title": {
-        "message": "NetBird Service Is Outdated",
-        "description": "Title of the overlay shown when the NetBird background service is too old to drive this UI."
+        "message": "NetBird Client Is Outdated",
+        "description": "Title of the overlay shown when the NetBird client (daemon) is too old to drive this UI."
     },
     "daemon.outdated.description": {
-        "message": "Update the NetBird service to use this app.",
-        "description": "Body of the daemon-outdated overlay telling the user to upgrade the service."
+        "message": "The new GUI isn't compatible with the older NetBird client. Update your client to use the new application.",
+        "description": "Body of the daemon-outdated overlay explaining that the GUI is newer than the client and the client must be updated."
+    },
+    "daemon.outdated.download": {
+        "message": "Download Latest",
+        "description": "Button on the daemon-outdated overlay that opens the download page for the latest release."
     },
     "error.jwt_clock_skew": {
         "message": "Sign-in failed: this device's clock is out of sync with the server. Please sync your system clock and try again.",

client/ui/i18n/locales/es/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Busque NetBird en su bandeja del sistema"
     },
+    "welcome.titleMac": {
+        "message": "Busque NetBird en su barra de menús"
+    },
     "welcome.description": {
         "message": "NetBird reside en su bandeja del sistema. Haga clic en el icono para conectarse, cambiar de perfil o abrir la configuración."
     },
+    "welcome.descriptionMac": {
+        "message": "NetBird reside en su barra de menús. Haga clic en el icono para conectarse, cambiar de perfil o abrir la configuración."
+    },
     "welcome.continue": {
         "message": "Continuar"
     },
@@ -1293,10 +1299,13 @@
         "message": "Documentación"
     },
     "daemon.outdated.title": {
-        "message": "El servicio de NetBird está desactualizado"
+        "message": "NetBird Client está desactualizado"
     },
     "daemon.outdated.description": {
-        "message": "Actualice el servicio de NetBird para usar esta aplicación."
+        "message": "La nueva GUI no es compatible con su cliente anterior. Actualice su cliente para usar la nueva aplicación."
+    },
+    "daemon.outdated.download": {
+        "message": "Descargar la última versión"
     },
     "error.jwt_clock_skew": {
         "message": "Error al iniciar sesión: el reloj de este dispositivo no está sincronizado con el servidor. Sincronice el reloj del sistema e inténtelo de nuevo."

client/ui/i18n/locales/fr/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Cherchez NetBird dans votre barre d’état système"
     },
+    "welcome.titleMac": {
+        "message": "Cherchez NetBird dans votre barre des menus"
+    },
     "welcome.description": {
         "message": "NetBird se trouve dans votre barre d’état système. Cliquez sur l’icône pour vous connecter, changer de profil ou ouvrir les paramètres."
     },
+    "welcome.descriptionMac": {
+        "message": "NetBird se trouve dans votre barre des menus. Cliquez sur l’icône pour vous connecter, changer de profil ou ouvrir les paramètres."
+    },
     "welcome.continue": {
         "message": "Continuer"
     },
@@ -1293,10 +1299,13 @@
         "message": "Documentation"
     },
     "daemon.outdated.title": {
-        "message": "Le service NetBird est obsolète"
+        "message": "Le Client NetBird est obsolète"
     },
     "daemon.outdated.description": {
-        "message": "Mettez à jour le service NetBird pour utiliser cette application."
+        "message": "La nouvelle GUI n'est pas compatible avec votre ancien client. Mettez à jour votre client pour utiliser la nouvelle application."
+    },
+    "daemon.outdated.download": {
+        "message": "Télécharger la dernière version"
     },
     "error.jwt_clock_skew": {
         "message": "Échec de la connexion : l’horloge de cet appareil n’est pas synchronisée avec le serveur. Veuillez synchroniser l’horloge de votre système et réessayer."

client/ui/i18n/locales/hu/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Keresse a NetBirdöt a tálcán"
     },
+    "welcome.titleMac": {
+        "message": "Keresse a NetBirdöt a menüsorban"
+    },
     "welcome.description": {
         "message": "A NetBird a tálcán fut. Kattintson az ikonra a csatlakozáshoz, profilváltáshoz vagy a beállítások megnyitásához."
     },
+    "welcome.descriptionMac": {
+        "message": "A NetBird a menüsorban fut. Kattintson az ikonra a csatlakozáshoz, profilváltáshoz vagy a beállítások megnyitásához."
+    },
     "welcome.continue": {
         "message": "Folytatás"
     },
@@ -1293,10 +1299,13 @@
         "message": "Dokumentáció"
     },
     "daemon.outdated.title": {
-        "message": "A NetBird szolgáltatás elavult"
+        "message": "A NetBird Kliens elavult"
     },
     "daemon.outdated.description": {
-        "message": "Frissítsd a NetBird szolgáltatást az alkalmazás használatához."
+        "message": "Az új GUI nem kompatibilis a régebbi klienseddel. Frissítsd a klienst az új alkalmazás használatához."
+    },
+    "daemon.outdated.download": {
+        "message": "Legújabb letöltése"
     },
     "error.jwt_clock_skew": {
         "message": "A bejelentkezés sikertelen: az eszköz órája eltér a szerverétől. Kérjük, szinkronizálja a rendszer óráját, majd próbálja újra."

client/ui/i18n/locales/it/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Cerchi NetBird nella tray"
     },
+    "welcome.titleMac": {
+        "message": "Cerchi NetBird nella barra dei menu"
+    },
     "welcome.description": {
         "message": "NetBird risiede nella tray. Clicchi sull'icona per connettersi, cambiare profilo o aprire le impostazioni."
     },
+    "welcome.descriptionMac": {
+        "message": "NetBird risiede nella barra dei menu. Clicchi sull'icona per connettersi, cambiare profilo o aprire le impostazioni."
+    },
     "welcome.continue": {
         "message": "Continua"
     },
@@ -1293,10 +1299,13 @@
         "message": "Documentazione"
     },
     "daemon.outdated.title": {
-        "message": "Il servizio NetBird è obsoleto"
+        "message": "NetBird Client è obsoleto"
     },
     "daemon.outdated.description": {
-        "message": "Aggiorna il servizio NetBird per usare questa app."
+        "message": "La nuova GUI non è compatibile con il tuo client precedente. Aggiorna il client per usare la nuova applicazione."
+    },
+    "daemon.outdated.download": {
+        "message": "Scarica l'ultima versione"
     },
     "error.jwt_clock_skew": {
         "message": "Accesso non riuscito: l'orologio di questo dispositivo non è sincronizzato con il server. Sincronizzi l'orologio di sistema e riprovi."

client/ui/i18n/locales/pt/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Procure o NetBird na sua bandeja"
     },
+    "welcome.titleMac": {
+        "message": "Procure o NetBird na sua barra de menus"
+    },
     "welcome.description": {
         "message": "O NetBird fica na sua bandeja. Clique no ícone para conectar, alternar perfis ou abrir as configurações."
     },
+    "welcome.descriptionMac": {
+        "message": "O NetBird fica na sua barra de menus. Clique no ícone para conectar, alternar perfis ou abrir as configurações."
+    },
     "welcome.continue": {
         "message": "Continuar"
     },
@@ -1293,10 +1299,13 @@
         "message": "Documentação"
     },
     "daemon.outdated.title": {
-        "message": "O serviço NetBird está desatualizado"
+        "message": "O NetBird Client está desatualizado"
     },
     "daemon.outdated.description": {
-        "message": "Atualize o serviço NetBird para usar este aplicativo."
+        "message": "A nova GUI não é compatível com o seu cliente mais antigo. Atualize o seu cliente para usar o novo aplicativo."
+    },
+    "daemon.outdated.download": {
+        "message": "Baixar a versão mais recente"
     },
     "error.jwt_clock_skew": {
         "message": "Falha no login: o relógio deste dispositivo está fora de sincronia com o servidor. Sincronize o relógio do sistema e tente novamente."

client/ui/i18n/locales/ru/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "Найдите NetBird в системном трее"
     },
+    "welcome.titleMac": {
+        "message": "Найдите NetBird в строке меню"
+    },
     "welcome.description": {
         "message": "NetBird находится в системном трее. Нажмите на значок, чтобы подключиться, переключить профиль или открыть настройки."
     },
+    "welcome.descriptionMac": {
+        "message": "NetBird находится в строке меню. Нажмите на значок, чтобы подключиться, переключить профиль или открыть настройки."
+    },
     "welcome.continue": {
         "message": "Продолжить"
     },
@@ -1293,10 +1299,13 @@
         "message": "Документация"
     },
     "daemon.outdated.title": {
-        "message": "Служба NetBird устарела"
+        "message": "Клиент NetBird устарел"
     },
     "daemon.outdated.description": {
-        "message": "Обновите службу NetBird, чтобы использовать это приложение."
+        "message": "Новый GUI несовместим с вашим более старым клиентом. Обновите клиент, чтобы использовать новое приложение."
+    },
+    "daemon.outdated.download": {
+        "message": "Скачать последнюю версию"
     },
     "error.jwt_clock_skew": {
         "message": "Не удалось войти: часы этого устройства рассинхронизированы с сервером. Синхронизируйте системные часы и повторите попытку."

client/ui/i18n/locales/zh-CN/common.json

@@ -1034,9 +1034,15 @@
     "welcome.title": {
         "message": "在托盘中查找 NetBird"
     },
+    "welcome.titleMac": {
+        "message": "在菜单栏中查找 NetBird"
+    },
     "welcome.description": {
         "message": "NetBird 驻留在您的托盘中。点击图标即可连接、切换配置文件或打开设置。"
     },
+    "welcome.descriptionMac": {
+        "message": "NetBird 驻留在您的菜单栏中。点击图标即可连接、切换配置文件或打开设置。"
+    },
     "welcome.continue": {
         "message": "继续"
     },
@@ -1293,10 +1299,13 @@
         "message": "文档"
     },
     "daemon.outdated.title": {
-        "message": "NetBird 服务版本过旧"
+        "message": "NetBird 客户端版本过旧"
     },
     "daemon.outdated.description": {
-        "message": "请更新 NetBird 服务以使用此应用。"
+        "message": "新版 GUI 与您较旧的客户端不兼容。请更新客户端以使用新应用。"
+    },
+    "daemon.outdated.download": {
+        "message": "下载最新版本"
     },
     "error.jwt_clock_skew": {
         "message": "登录失败：此设备的时钟与服务器不同步。请同步您的系统时钟后重试。"

client/ui/services/autostart.go

@@ -10,7 +10,7 @@ import (
 	"github.com/wailsapp/wails/v3/pkg/application"
 )
 
-// Autostart facade over Wails' AutostartManager. The OS login-item registration
+// Autostart facade over Wails' AutostartManager. The OS autostart entry registration
 // is the single source of truth; nothing is mirrored to preferences.
 type Autostart struct {
 	mgr *application.AutostartManager

client/ui/services/profileswitcher.go

@@ -12,13 +12,15 @@ import (
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 )
 
-// ProfileSwitcher holds the reconnect policy shared by the tray and React
-// frontend so both flip profiles identically. The policy keys off prevStatus
-// from DaemonFeed.Get at SwitchActive entry:
+// ProfileSwitcher holds the switch policy shared by the tray and React
+// frontend so both flip profiles identically. SwitchActive (plain selection:
+// header dropdown, tray submenu) always connects after the switch;
+// SwitchActiveNoConnect (manage-profiles screen) never does, so the user can
+// still adjust the management URL before connecting. prevStatus from
+// DaemonFeed.Get at entry only decides the teardown:
 //
-//	Connected/Connecting → Switch + Down + Up; optimistic Connecting paint.
-//	NeedsLogin/LoginFailed/SessionExpired → Switch + Down; clear stale error for re-login.
-//	Idle → Switch only.
+//	Connected/Connecting/NeedsLogin/LoginFailed/SessionExpired → Down first.
+//	Idle → no Down.
 type ProfileSwitcher struct {
 	profiles   *Profiles
 	connection *Connection
@@ -29,29 +31,40 @@ func NewProfileSwitcher(profiles *Profiles, connection *Connection, feed *Daemon
 	return &ProfileSwitcher{profiles: profiles, connection: connection, feed: feed}
 }
 
-// SwitchActive switches to the named profile applying the reconnect policy.
+// SwitchActive switches to the named profile and always connects afterwards.
 func (s *ProfileSwitcher) SwitchActive(ctx context.Context, p ProfileRef) error {
+	return s.switchActive(ctx, p, true)
+}
+
+// SwitchActiveNoConnect switches to the named profile without connecting,
+// tearing down any existing connection first.
+func (s *ProfileSwitcher) SwitchActiveNoConnect(ctx context.Context, p ProfileRef) error {
+	return s.switchActive(ctx, p, false)
+}
+
+func (s *ProfileSwitcher) switchActive(ctx context.Context, p ProfileRef, connect bool) error {
 	prevStatus := ""
-	if st, err := s.feed.Get(ctx); err == nil {
-		prevStatus = st.Status
-	} else {
-		log.Warnf("profileswitcher: get status: %v", err)
+	if s.feed != nil {
+		if st, err := s.feed.Get(ctx); err == nil {
+			prevStatus = st.Status
+		} else {
+			log.Warnf("profileswitcher: get status: %v", err)
+		}
 	}
 
-	wasActive := strings.EqualFold(prevStatus, StatusConnected) ||
-		strings.EqualFold(prevStatus, StatusConnecting)
-	needsDown := wasActive ||
+	needsDown := strings.EqualFold(prevStatus, StatusConnected) ||
+		strings.EqualFold(prevStatus, StatusConnecting) ||
 		strings.EqualFold(prevStatus, StatusNeedsLogin) ||
 		strings.EqualFold(prevStatus, StatusLoginFailed) ||
 		strings.EqualFold(prevStatus, StatusSessionExpired)
 
-	log.Infof("profileswitcher: switch profile=%q prevStatus=%q wasActive=%v needsDown=%v",
-		p.ProfileName, prevStatus, wasActive, needsDown)
+	log.Infof("profileswitcher: switch profile=%q prevStatus=%q connect=%v needsDown=%v",
+		p.ProfileName, prevStatus, connect, needsDown)
 
-	// Optimistic Connecting paint only when wasActive: those prevStatuses emit
-	// stale Connected + transient Idle pushes during Down that must be
-	// suppressed until Up resumes the stream (see DaemonFeed suppression table).
-	if wasActive {
+	// Optimistic Connecting paint plus stale-push suppression during Down (see
+	// DaemonFeed suppression table); also arms the login-watch that pops
+	// browser-login when the new profile turns out to need SSO.
+	if connect && s.feed != nil {
 		s.feed.BeginProfileSwitch()
 	}
 
@@ -76,9 +89,9 @@ func (s *ProfileSwitcher) SwitchActive(ctx context.Context, p ProfileRef) error
 		}
 	}
 
-	if wasActive {
+	if connect {
 		if err := s.connection.Up(ctx, UpParams(p)); err != nil {
-			return fmt.Errorf("reconnect %q: %w", p.ProfileName, err)
+			return fmt.Errorf("connect %q: %w", p.ProfileName, err)
 		}
 	}
 

client/ui/services/windowmanager.go

@@ -185,37 +185,38 @@ func (s *WindowManager) OpenBrowserLogin(uri string) {
 			startURL = "/#/dialog/browser-login?uri=" + url.QueryEscape(uri)
 		}
 		s.hideOtherWindowsLocked("browser-login")
-		// Prefer the main window's screen (multi-monitor); falls back to OS-default centering.
-		var screen *application.Screen
-		if s.mainWindow != nil {
-			if sc, err := s.mainWindow.GetScreen(); err == nil {
-				screen = sc
-			}
-		}
 		opts := DialogWindowOptions("browser-login", s.title("window.title.signIn"), startURL, s.linuxIcon)
 		// Not always-on-top: it would obscure the browser tab the user logs in through.
 		opts.AlwaysOnTop = false
 		opts.InitialPosition = application.WindowCentered
-		opts.Screen = screen
+		// Open on the active (where users cursor is) display, like the session-expiration dialog.
+		opts.Screen = s.getScreenBasedOnCursorPosition()
 		s.browserLogin = s.app.Window.NewWithOptions(opts)
 		bl := s.browserLogin
-		// Red-X close means cancel: emit the event so startLogin() tears down the SSO wait.
 		bl.OnWindowEvent(events.Common.WindowClosing, func(_ *application.WindowEvent) {
-			s.app.Event.Emit(EventBrowserLoginCancel)
 			s.mu.Lock()
-			s.browserLogin = nil
-			s.restoreHiddenWindowsLocked()
+			// Only a live user red-X still has this registered; programmatic closers
+			// nil s.browserLogin first and clean up themselves. Guarding here stops a
+			// stale close event from wiping a replacement popup's state.
+			userClosed := s.browserLogin == bl
+			if userClosed {
+				s.browserLogin = nil
+				s.restoreHiddenWindowsLocked()
+			}
 			s.mu.Unlock()
+			if userClosed {
+				s.app.Event.Emit(EventBrowserLoginCancel)
+			}
 		})
-		s.centerWhenReady(s.browserLogin)
+		s.centerOnCursorScreen(s.browserLogin)
 		return
 	}
 	if uri != "" {
 		s.browserLogin.SetURL("/#/dialog/browser-login?uri=" + url.QueryEscape(uri))
 	}
+	s.centerOnCursorScreen(s.browserLogin)
 	s.browserLogin.Show()
 	s.browserLogin.Focus()
-	s.centerWhenReady(s.browserLogin)
 }
 
 // BrowserLoginWindow returns the live SSO popup, or nil. While non-nil it is the
@@ -238,6 +239,15 @@ func (s *WindowManager) CloseBrowserLogin() {
 	s.mu.Lock()
 	w := s.browserLogin
 	s.browserLogin = nil
+	// The WindowClosing hook no-ops on a programmatic close, so restore here —
+	// but only if a popup was actually open. The frontend calls this even when no
+	// popup was ever shown (e.g. resetDialog() after an early RequestExtend failure,
+	// or connection.ts's catch path), and hiddenForLogin is shared with
+	// OpenInstallProgress, so an unconditional restore could re-show windows a
+	// still-running install-progress is hiding.
+	if w != nil {
+		s.restoreHiddenWindowsLocked()
+	}
 	s.mu.Unlock()
 	if w != nil {
 		w.Close()
@@ -279,6 +289,35 @@ func (s *WindowManager) CloseSessionExpiration() {
 	}
 }
 
+// CloseRenewFlow tears down the SSO session-renewal UI in a single call: it
+// closes the browser-login popup and the session-expiration window together.
+func (s *WindowManager) CloseRenewFlow() {
+	s.mu.Lock()
+	bl := s.browserLogin
+	se := s.sessionExpiration
+	s.browserLogin = nil
+	s.sessionExpiration = nil
+	if se != nil {
+		kept := s.hiddenForLogin[:0]
+		for _, w := range s.hiddenForLogin {
+			if w != se {
+				kept = append(kept, w)
+			}
+		}
+		s.hiddenForLogin = kept
+	}
+	s.restoreHiddenWindowsLocked()
+	s.mu.Unlock()
+
+	// Close after unlock so the re-entrant handlers can take s.mu.
+	if bl != nil {
+		bl.Close()
+	}
+	if se != nil {
+		se.Close()
+	}
+}
+
 // OpenInstallProgress shows the install-progress window and hides the rest for the duration
 // (restored on close). It owns its own result polling since the daemon restarts mid-install.
 func (s *WindowManager) OpenInstallProgress(version string) {

client/ui/tray.go

@@ -30,6 +30,8 @@ const (
 
 	statusError = "Error"
 
+	quitDownTimeout = 5 * time.Second
+
 	urlGitHubRepo     = "https://github.com/netbirdio/netbird"
 	urlGitHubReleases = "https://github.com/netbirdio/netbird/releases/latest"
 	urlDocs           = "https://docs.netbird.io"
@@ -315,8 +317,7 @@ func (t *Tray) relayoutMenu() {
 		if sessionDeadline.IsZero() {
 			t.sessionExpiresItem.SetHidden(true)
 		} else {
-			remaining := t.formatSessionRemaining(time.Until(sessionDeadline))
-			t.sessionExpiresItem.SetLabel(t.loc.T("tray.session.expiresIn", "remaining", remaining))
+			t.sessionExpiresItem.SetLabel(t.sessionRowLabel(sessionDeadline))
 			t.sessionExpiresItem.SetHidden(false)
 		}
 	}
@@ -446,11 +447,28 @@ func (t *Tray) buildMenu() *application.Menu {
 	menu.AddSeparator()
 	menu.Add(t.loc.T("tray.menu.quit")).
 		SetAccelerator("CmdOrCtrl+Q").
-		OnClick(func(*application.Context) { t.app.Quit() })
+		OnClick(func(*application.Context) { t.handleQuit() })
 
 	return menu
 }
 
+func (t *Tray) handleQuit() {
+	t.profileMu.Lock()
+	if t.switchCancel != nil {
+		t.switchCancel()
+		t.switchCancel = nil
+	}
+	t.profileMu.Unlock()
+	t.svc.DaemonFeed.CancelProfileSwitch()
+
+	ctx, cancel := context.WithTimeout(context.Background(), quitDownTimeout)
+	defer cancel()
+	if err := t.svc.Connection.Down(ctx); err != nil {
+		log.Errorf("disconnect on quit: %v", err)
+	}
+	t.app.Quit()
+}
+
 // handleConnect receives the clicked item from the buildMenu closure —
 // t.upItem is menuMu-guarded and must not be read here.
 func (t *Tray) handleConnect(upItem *application.MenuItem) {

client/ui/tray_session.go

@@ -64,11 +64,42 @@ func (t *Tray) applySessionExpiry(deadline *time.Time, connected bool) bool {
 	return changed
 }
 
-// runSessionExpiryTicker recomputes the "Expires in …" row label every 30s. Runs until process exit.
+// runSessionExpiryTicker recomputes the "Expires in …" row label until process exit.
+// The interval scales with the remaining time: coarse when the deadline is far off,
+// down to 10s in the final two minutes so the label doesn't lag the ceiling-rounded
+// countdown near expiry. The cached deadline is re-read every iteration, so an extend
+// or reconnect that moves it is picked up on the next tick.
 func (t *Tray) runSessionExpiryTicker() {
-	tk := time.NewTicker(30 * time.Second)
-	for range tk.C {
+	tm := time.NewTimer(sessionRefreshInterval(t.sessionRemaining()))
+	defer tm.Stop()
+	for range tm.C {
 		t.refreshSessionExpiresLabel()
+		tm.Reset(sessionRefreshInterval(t.sessionRemaining()))
+	}
+}
+
+// sessionRemaining returns the time left on the cached SSO deadline, or 0 when unknown.
+func (t *Tray) sessionRemaining() time.Duration {
+	t.sessionMu.Lock()
+	deadline := t.sessionExpiresAt
+	t.sessionMu.Unlock()
+	if deadline.IsZero() {
+		return 0
+	}
+	return time.Until(deadline)
+}
+
+// sessionRefreshInterval picks how long to wait before the next label recompute.
+func sessionRefreshInterval(remaining time.Duration) time.Duration {
+	switch {
+	case remaining <= 0:
+		return 30 * time.Second
+	case remaining <= 2*time.Minute:
+		return 10 * time.Second
+	case remaining <= time.Hour:
+		return 30 * time.Second
+	default:
+		return time.Minute
 	}
 }
 
@@ -87,30 +118,39 @@ func (t *Tray) refreshSessionExpiresLabel() {
 	if deadline.IsZero() {
 		return
 	}
-	remaining := t.formatSessionRemaining(time.Until(deadline))
-	item.SetLabel(t.loc.T("tray.session.expiresIn", "remaining", remaining))
+	item.SetLabel(t.sessionRowLabel(deadline))
+}
+
+func (t *Tray) sessionRowLabel(deadline time.Time) string {
+	remaining := time.Until(deadline)
+	if remaining <= 0 {
+		return t.loc.T("tray.status.sessionExpired")
+	}
+	return t.loc.T("tray.session.expiresIn", "remaining", t.formatSessionRemaining(remaining))
 }
 
 // formatSessionRemaining renders d as a localised long-form string picking the largest non-zero unit.
+// Each unit is rounded up so the label never claims less time than actually remains, matching the
+// upper-bound sense of the sub-minute "less than a minute" fragment.
 // Singular/plural keys are split per language for proper translation.
 func (t *Tray) formatSessionRemaining(d time.Duration) string {
 	switch {
 	case d < time.Minute:
 		return t.loc.T("tray.session.unit.lessThanMinute")
-	case d < time.Hour:
-		m := int(d / time.Minute)
+	case d <= 59*time.Minute:
+		m := ceilDiv(d, time.Minute)
 		if m == 1 {
 			return t.loc.T("tray.session.unit.minute")
 		}
 		return t.loc.T("tray.session.unit.minutes", "count", strconv.Itoa(m))
-	case d < 24*time.Hour:
-		h := int((d + 30*time.Minute) / time.Hour)
+	case d <= 23*time.Hour:
+		h := ceilDiv(d, time.Hour)
 		if h == 1 {
 			return t.loc.T("tray.session.unit.hour")
 		}
 		return t.loc.T("tray.session.unit.hours", "count", strconv.Itoa(h))
 	default:
-		days := int((d + 12*time.Hour) / (24 * time.Hour))
+		days := ceilDiv(d, 24*time.Hour)
 		if days == 1 {
 			return t.loc.T("tray.session.unit.day")
 		}
@@ -118,6 +158,11 @@ func (t *Tray) formatSessionRemaining(d time.Duration) string {
 	}
 }
 
+// ceilDiv divides d by unit rounding up, assuming d > 0.
+func ceilDiv(d, unit time.Duration) int {
+	return int((d + unit - time.Nanosecond) / unit)
+}
+
 // registerSessionWarningCategory wires the OS notification category and response handler for the expiry warning.
 // Errors are swallowed since the worst case is a plain notification without buttons.
 func (t *Tray) registerSessionWarningCategory() {
@@ -252,11 +297,9 @@ func (t *Tray) openSessionExpiration() {
 }
 
 // openSessionExtendFlow opens the SessionExpiration window seeded with the cached deadline's remaining time,
-// for the "Expires in …" tray row. No-ops when the deadline is unknown or elapsed.
+// for the "Expires in …" tray row. Once the deadline has elapsed the row reads "Session expired" and the
+// click routes to the login flow instead. No-op when the deadline is unknown.
 func (t *Tray) openSessionExtendFlow() {
-	if t.svc.WindowManager == nil {
-		return
-	}
 	t.sessionMu.Lock()
 	deadline := t.sessionExpiresAt
 	t.sessionMu.Unlock()
@@ -265,6 +308,14 @@ 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()
+		}
+		return
+	}
+	if t.svc.WindowManager == nil {
 		return
 	}
 	t.svc.WindowManager.OpenSessionExpiration(seconds)

combined/cmd/admin.go

@@ -0,0 +1,151 @@
+package cmd
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/dexidp/dex/storage"
+	log "github.com/sirupsen/logrus"
+	"github.com/spf13/cobra"
+
+	"github.com/netbirdio/netbird/formatter/hook"
+	admincmd "github.com/netbirdio/netbird/management/cmd/admin"
+	tokencmd "github.com/netbirdio/netbird/management/cmd/token"
+	nbconfig "github.com/netbirdio/netbird/management/internals/server/config"
+	"github.com/netbirdio/netbird/management/server/activity"
+	activitystore "github.com/netbirdio/netbird/management/server/activity/store"
+	"github.com/netbirdio/netbird/management/server/store"
+	"github.com/netbirdio/netbird/management/server/types"
+	"github.com/netbirdio/netbird/util"
+)
+
+// newAdminCommands creates the admin command tree with combined-specific resource openers.
+func newAdminCommands() *cobra.Command {
+	return admincmd.NewCommands(admincmd.Openers{
+		Resources: withAdminResources,
+		Store:     withAdminStoreOnly,
+		IDP:       withAdminIDPOnly,
+	})
+}
+
+func newLegacyTokenCommand() *cobra.Command {
+	cmd := tokencmd.NewCommands(tokencmd.StoreOpener(withAdminStoreOnly))
+	cmd.Deprecated = "use 'admin token' instead"
+	return cmd
+}
+
+// withAdminResources loads the combined YAML config, initializes stores, and calls fn.
+func withAdminResources(cmd *cobra.Command, fn func(ctx context.Context, resources admincmd.Resources) error) error {
+	return withAdminConfig(cmd, func(ctx context.Context, cfg *CombinedConfig) error {
+		mgmtConfig, err := adminManagementConfig(cfg)
+		if err != nil {
+			return err
+		}
+
+		managementStore, err := openAdminStore(ctx, cfg)
+		if err != nil {
+			return err
+		}
+		defer admincmd.CloseStore(ctx, managementStore)
+
+		idpStorage, idpStorageFile, err := admincmd.OpenIDPStorage(mgmtConfig)
+		if err != nil {
+			return err
+		}
+		defer admincmd.CloseIDPStorage(idpStorage)
+
+		eventStore, esErr := openAdminEventStore(ctx, cfg, mgmtConfig)
+		if esErr != nil {
+			_, _ = fmt.Fprintf(cmd.ErrOrStderr(), "Warning: audit events will not be recorded: %v\n", esErr)
+		}
+		if eventStore != nil {
+			defer func() {
+				if err := eventStore.Close(ctx); err != nil {
+					log.Debugf("close activity event store: %v", err)
+				}
+			}()
+		}
+
+		return fn(ctx, admincmd.Resources{Store: managementStore, IDPStorage: idpStorage, IDPStorageFile: idpStorageFile, EventStore: eventStore})
+	})
+}
+
+// withAdminStoreOnly opens only the management store for admin subcommands that do not
+// need embedded IdP storage.
+func withAdminStoreOnly(cmd *cobra.Command, fn func(ctx context.Context, s store.Store) error) error {
+	return withAdminConfig(cmd, func(ctx context.Context, cfg *CombinedConfig) error {
+		managementStore, err := openAdminStore(ctx, cfg)
+		if err != nil {
+			return err
+		}
+		defer admincmd.CloseStore(ctx, managementStore)
+
+		return fn(ctx, managementStore)
+	})
+}
+
+func withAdminIDPOnly(cmd *cobra.Command, fn func(ctx context.Context, idpStorage storage.Storage, storageFile string) error) error {
+	return withAdminConfig(cmd, func(ctx context.Context, cfg *CombinedConfig) error {
+		mgmtConfig, err := adminManagementConfig(cfg)
+		if err != nil {
+			return err
+		}
+		idpStorage, idpStorageFile, err := admincmd.OpenIDPStorage(mgmtConfig)
+		if err != nil {
+			return err
+		}
+		defer admincmd.CloseIDPStorage(idpStorage)
+
+		return fn(ctx, idpStorage, idpStorageFile)
+	})
+}
+
+func withAdminConfig(cmd *cobra.Command, fn func(ctx context.Context, cfg *CombinedConfig) error) error {
+	if err := util.InitLog("error", "console"); err != nil {
+		return fmt.Errorf("init log: %w", err)
+	}
+
+	ctx := context.WithValue(cmd.Context(), hook.ExecutionContextKey, hook.SystemSource) //nolint:staticcheck
+
+	cfg, err := LoadConfig(configPath)
+	if err != nil {
+		return fmt.Errorf("load config: %w", err)
+	}
+	cfg.ApplyAdminDefaults()
+	applyServerStoreEnv(cfg.Server.Store)
+
+	return fn(ctx, cfg)
+}
+
+func adminManagementConfig(cfg *CombinedConfig) (*nbconfig.Config, error) {
+	mgmtConfig, err := cfg.ToManagementConfig()
+	if err != nil {
+		return nil, fmt.Errorf("create management config: %w", err)
+	}
+	return mgmtConfig, nil
+}
+
+func openAdminStore(ctx context.Context, cfg *CombinedConfig) (store.Store, error) {
+	managementStore, err := store.NewStore(ctx, types.Engine(cfg.Management.Store.Engine), cfg.Management.DataDir, nil, true)
+	if err != nil {
+		return nil, fmt.Errorf("create store: %w", err)
+	}
+	return managementStore, nil
+}
+
+func openAdminEventStore(ctx context.Context, cfg *CombinedConfig, config *nbconfig.Config) (activity.Store, error) {
+	if config.DataStoreEncryptionKey == "" {
+		return nil, fmt.Errorf("data store encryption key is not configured")
+	}
+	if err := applyActivityStoreEnv(cfg.Server.ActivityStore); err != nil {
+		return nil, fmt.Errorf("configure activity event store: %w", err)
+	}
+	eventStore, err := activitystore.NewSqlStore(ctx, config.Datadir, config.DataStoreEncryptionKey)
+	if err != nil {
+		return nil, fmt.Errorf("open activity event store: %w", err)
+	}
+	if eventStore == nil {
+		return nil, fmt.Errorf("open activity event store: returned nil store")
+	}
+	return eventStore, nil
+}

combined/cmd/admin_config_test.go

@@ -0,0 +1,47 @@
+package cmd
+
+import (
+	"context"
+	"os"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+
+	nbconfig "github.com/netbirdio/netbird/management/internals/server/config"
+)
+
+func TestApplyAdminDefaultsCopiesServerStoreWithoutExposedAddress(t *testing.T) {
+	cfg := DefaultConfig()
+	cfg.Server.ExposedAddress = ""
+	cfg.Server.DataDir = "/srv/netbird"
+	cfg.Server.Store = StoreConfig{
+		Engine: "postgres",
+		DSN:    "postgres://user:pass@example.com/netbird",
+	}
+
+	cfg.ApplyAdminDefaults()
+
+	require.Equal(t, "/srv/netbird", cfg.Management.DataDir)
+	require.Equal(t, "postgres", cfg.Management.Store.Engine)
+	require.Equal(t, cfg.Server.Store.DSN, cfg.Management.Store.DSN)
+}
+
+func TestOpenAdminEventStoreMissingEncryptionKeyReturnsNilInterface(t *testing.T) {
+	eventStore, err := openAdminEventStore(context.Background(), &CombinedConfig{}, &nbconfig.Config{})
+	require.Error(t, err)
+	require.Contains(t, err.Error(), "encryption key")
+	require.Nil(t, eventStore)
+}
+
+func TestApplyServerStoreEnv(t *testing.T) {
+	t.Setenv("NB_STORE_ENGINE_POSTGRES_DSN", "")
+	t.Setenv("NB_STORE_ENGINE_MYSQL_DSN", "")
+	t.Setenv("NB_STORE_ENGINE_SQLITE_FILE", "")
+
+	applyServerStoreEnv(StoreConfig{Engine: "postgres", DSN: "postgres-dsn", File: "store.db"})
+	require.Equal(t, "postgres-dsn", os.Getenv("NB_STORE_ENGINE_POSTGRES_DSN"))
+	require.Equal(t, "store.db", os.Getenv("NB_STORE_ENGINE_SQLITE_FILE"))
+
+	applyServerStoreEnv(StoreConfig{Engine: "mysql", DSN: "mysql-dsn"})
+	require.Equal(t, "mysql-dsn", os.Getenv("NB_STORE_ENGINE_MYSQL_DSN"))
+}

combined/cmd/config.go

@@ -6,8 +6,7 @@ import (
 	"net"
 	"net/netip"
 	"os"
-	"path"
-	"path/filepath"
+	filePath "path/filepath"
 	"strings"
 	"time"
 
@@ -74,6 +73,9 @@ type ServerConfig struct {
 	ActivityStore           StoreConfig        `yaml:"activityStore"`
 	AuthStore               StoreConfig        `yaml:"authStore"`
 	ReverseProxy            ReverseProxyConfig `yaml:"reverseProxy"`
+
+	SupportedSyncMessageVersions           *int           `yaml:"supportedSyncMessageVersions,omitempty"`
+	PerAccountSupportedSyncMessageVersions map[string]int `yaml:"perAccountSupportedSyncMessageVersions,omitempty"`
 }
 
 // TLSConfig contains TLS/HTTPS settings
@@ -145,6 +147,7 @@ type AuthConfig struct {
 	CLIRedirectURIs                 []string          `yaml:"cliRedirectURIs"`
 	Owner                           *AuthOwnerConfig  `yaml:"owner,omitempty"`
 	DashboardPostLogoutRedirectURIs []string          `yaml:"dashboardPostLogoutRedirectURIs"`
+	GrantTypes                      []string          `yaml:"grantTypes"`
 }
 
 // AuthStorageConfig contains auth storage settings
@@ -299,6 +302,19 @@ func (c *CombinedConfig) ApplySimplifiedDefaults() {
 	c.autoConfigureClientSettings(exposedProto, exposedHost, exposedHostPort, hasExternalStuns, hasExternalRelay, hasExternalSignal)
 }
 
+// ApplyAdminDefaults applies the management settings needed by admin commands even
+// when the full server config is invalid and ApplySimplifiedDefaults cannot run.
+func (c *CombinedConfig) ApplyAdminDefaults() {
+	if c.Management.DataDir == "" || c.Management.DataDir == "/var/lib/netbird/" {
+		c.Management.DataDir = c.Server.DataDir
+	}
+	if c.Management.Store.Engine == "" || c.Management.Store.Engine == "sqlite" {
+		if c.Server.Store.Engine != "" || c.Server.Store.File != "" || c.Server.Store.DSN != "" {
+			c.Management.Store = c.Server.Store
+		}
+	}
+}
+
 // applyRelayDefaults configures the relay service if no external relay is configured.
 func (c *CombinedConfig) applyRelayDefaults(exposedProto, exposedHostPort string, hasExternalRelay, hasExternalStuns bool) {
 	if hasExternalRelay {
@@ -576,11 +592,11 @@ func (c *CombinedConfig) buildEmbeddedIdPConfig(mgmt ManagementConfig) (*idp.Emb
 			return nil, fmt.Errorf("authStore.dsn is required when authStore.engine is postgres")
 		}
 	} else {
-		authStorageFile = path.Join(mgmt.DataDir, "idp.db")
+		authStorageFile = filePath.Join(mgmt.DataDir, "idp.db")
 		if c.Server.AuthStore.File != "" {
 			authStorageFile = c.Server.AuthStore.File
-			if !filepath.IsAbs(authStorageFile) {
-				authStorageFile = filepath.Join(mgmt.DataDir, authStorageFile)
+			if !filePath.IsAbs(authStorageFile) {
+				authStorageFile = filePath.Join(mgmt.DataDir, authStorageFile)
 			}
 		}
 	}
@@ -604,6 +620,7 @@ func (c *CombinedConfig) buildEmbeddedIdPConfig(mgmt ManagementConfig) (*idp.Emb
 		DashboardRedirectURIs:           mgmt.Auth.DashboardRedirectURIs,
 		CLIRedirectURIs:                 mgmt.Auth.CLIRedirectURIs,
 		DashboardPostLogoutRedirectURIs: mgmt.Auth.DashboardPostLogoutRedirectURIs,
+		GrantTypes:                      mgmt.Auth.GrantTypes,
 	}
 
 	if mgmt.Auth.Owner != nil && mgmt.Auth.Owner.Email != "" {
@@ -694,16 +711,18 @@ func (c *CombinedConfig) ToManagementConfig() (*nbconfig.Config, error) {
 	httpConfig.AuthCallbackURL = callbackURL + types.ProxyCallbackEndpointFull
 
 	return &nbconfig.Config{
-		Stuns:                  stuns,
-		Relay:                  relayConfig,
-		Signal:                 signalConfig,
-		Datadir:                mgmt.DataDir,
-		DataStoreEncryptionKey: mgmt.Store.EncryptionKey,
-		HttpConfig:             httpConfig,
-		StoreConfig:            storeConfig,
-		ReverseProxy:           reverseProxy,
-		DisableDefaultPolicy:   mgmt.DisableDefaultPolicy,
-		EmbeddedIdP:            embeddedIdP,
+		Stuns:                              stuns,
+		Relay:                              relayConfig,
+		Signal:                             signalConfig,
+		Datadir:                            mgmt.DataDir,
+		DataStoreEncryptionKey:             mgmt.Store.EncryptionKey,
+		HttpConfig:                         httpConfig,
+		StoreConfig:                        storeConfig,
+		ReverseProxy:                       reverseProxy,
+		DisableDefaultPolicy:               mgmt.DisableDefaultPolicy,
+		EmbeddedIdP:                        embeddedIdP,
+		HighestSupportedSyncMessageVersion: c.Server.SupportedSyncMessageVersions,
+		PerAccountHighestSupportedSyncMessageVersion: c.Server.PerAccountSupportedSyncMessageVersions,
 	}, nil
 }
 
@@ -727,7 +746,7 @@ func ApplyEmbeddedIdPConfig(ctx context.Context, cfg *nbconfig.Config, mgmtPort
 		cfg.EmbeddedIdP.Storage.Type = "sqlite3"
 	}
 	if cfg.EmbeddedIdP.Storage.Config.File == "" && cfg.Datadir != "" {
-		cfg.EmbeddedIdP.Storage.Config.File = path.Join(cfg.Datadir, "idp.db")
+		cfg.EmbeddedIdP.Storage.Config.File = filePath.Join(cfg.Datadir, "idp.db")
 	}
 
 	issuer := cfg.EmbeddedIdP.Issuer

combined/cmd/root.go

@@ -31,6 +31,7 @@ import (
 	relayServer "github.com/netbirdio/netbird/relay/server"
 	"github.com/netbirdio/netbird/relay/server/listener"
 	"github.com/netbirdio/netbird/relay/server/listener/ws"
+	syncgrpc "github.com/netbirdio/netbird/shared/management/grpc"
 	sharedMetrics "github.com/netbirdio/netbird/shared/metrics"
 	"github.com/netbirdio/netbird/shared/relay/auth"
 	"github.com/netbirdio/netbird/shared/signal/proto"
@@ -64,7 +65,8 @@ func init() {
 	rootCmd.PersistentFlags().StringVarP(&configPath, "config", "c", "", "path to YAML configuration file (required)")
 	_ = rootCmd.MarkPersistentFlagRequired("config")
 
-	rootCmd.AddCommand(newTokenCommands())
+	rootCmd.AddCommand(newAdminCommands())
+	rootCmd.AddCommand(newLegacyTokenCommand())
 }
 
 func RootCmd() *cobra.Command {
@@ -122,6 +124,37 @@ func execute(cmd *cobra.Command, _ []string) error {
 }
 
 // initializeConfig loads and validates the configuration, then initializes logging.
+func applyServerStoreEnv(storeConfig StoreConfig) {
+	if dsn := storeConfig.DSN; dsn != "" {
+		switch strings.ToLower(storeConfig.Engine) {
+		case "postgres":
+			os.Setenv("NB_STORE_ENGINE_POSTGRES_DSN", dsn)
+		case "mysql":
+			os.Setenv("NB_STORE_ENGINE_MYSQL_DSN", dsn)
+		}
+	}
+	if file := storeConfig.File; file != "" {
+		os.Setenv("NB_STORE_ENGINE_SQLITE_FILE", file)
+	}
+}
+
+func applyActivityStoreEnv(storeConfig StoreConfig) error {
+	if engine := storeConfig.Engine; engine != "" {
+		engineLower := strings.ToLower(engine)
+		if engineLower == "postgres" && storeConfig.DSN == "" {
+			return fmt.Errorf("activityStore.dsn is required when activityStore.engine is postgres")
+		}
+		os.Setenv("NB_ACTIVITY_EVENT_STORE_ENGINE", engineLower)
+		if dsn := storeConfig.DSN; dsn != "" {
+			os.Setenv("NB_ACTIVITY_EVENT_POSTGRES_DSN", dsn)
+		}
+	}
+	if file := storeConfig.File; file != "" {
+		os.Setenv("NB_ACTIVITY_EVENT_SQLITE_FILE", file)
+	}
+	return nil
+}
+
 func initializeConfig() error {
 	var err error
 	config, err = LoadConfig(configPath)
@@ -137,30 +170,10 @@ func initializeConfig() error {
 		return fmt.Errorf("failed to initialize log: %w", err)
 	}
 
-	if dsn := config.Server.Store.DSN; dsn != "" {
-		switch strings.ToLower(config.Server.Store.Engine) {
-		case "postgres":
-			os.Setenv("NB_STORE_ENGINE_POSTGRES_DSN", dsn)
-		case "mysql":
-			os.Setenv("NB_STORE_ENGINE_MYSQL_DSN", dsn)
-		}
-	}
-	if file := config.Server.Store.File; file != "" {
-		os.Setenv("NB_STORE_ENGINE_SQLITE_FILE", file)
-	}
+	applyServerStoreEnv(config.Server.Store)
 
-	if engine := config.Server.ActivityStore.Engine; engine != "" {
-		engineLower := strings.ToLower(engine)
-		if engineLower == "postgres" && config.Server.ActivityStore.DSN == "" {
-			return fmt.Errorf("activityStore.dsn is required when activityStore.engine is postgres")
-		}
-		os.Setenv("NB_ACTIVITY_EVENT_STORE_ENGINE", engineLower)
-		if dsn := config.Server.ActivityStore.DSN; dsn != "" {
-			os.Setenv("NB_ACTIVITY_EVENT_POSTGRES_DSN", dsn)
-		}
-	}
-	if file := config.Server.ActivityStore.File; file != "" {
-		os.Setenv("NB_ACTIVITY_EVENT_SQLITE_FILE", file)
+	if err := applyActivityStoreEnv(config.Server.ActivityStore); err != nil {
+		return err
 	}
 
 	log.Infof("Starting combined NetBird server")
@@ -226,7 +239,7 @@ func (s *serverInstances) createRelayServer(cfg *CombinedConfig, tlsSupport bool
 	}
 
 	hashedSecret := sha256.Sum256([]byte(cfg.Relay.AuthSecret))
-	authenticator := auth.NewTimedHMACValidator(hashedSecret[:], 24*time.Hour)
+	authenticator := auth.NewTimedHMACValidator(hashedSecret[:])
 
 	relayCfg := relayServer.Config{
 		Meter:          s.metricsServer.Meter,
@@ -505,6 +518,16 @@ func createManagementServer(cfg *CombinedConfig, mgmtConfig *nbconfig.Config) (m
 	}
 	mgmtPort, _ := strconv.Atoi(portStr)
 
+	if err := syncgrpc.ValidateSyncMessageVersion(mgmtConfig.HighestSupportedSyncMessageVersion); err != nil {
+		return nil, err
+	}
+
+	for accountId, version := range mgmtConfig.PerAccountHighestSupportedSyncMessageVersion {
+		if err := syncgrpc.ValidateSyncMessageVersion(&version); err != nil {
+			return nil, fmt.Errorf("unrecognized sync message version in perAccountSupportedSyncMessageVersions for account %s %w", accountId, err)
+		}
+	}
+
 	mgmtSrv := newServer(
 		&mgmtServer.Config{
 			NbConfig:                mgmtConfig,

combined/cmd/token.go

@@ -1,63 +0,0 @@
-package cmd
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"strings"
-
-	log "github.com/sirupsen/logrus"
-	"github.com/spf13/cobra"
-
-	"github.com/netbirdio/netbird/formatter/hook"
-	tokencmd "github.com/netbirdio/netbird/management/cmd/token"
-	"github.com/netbirdio/netbird/management/server/store"
-	"github.com/netbirdio/netbird/management/server/types"
-	"github.com/netbirdio/netbird/util"
-)
-
-// newTokenCommands creates the token command tree with combined-specific store opener.
-func newTokenCommands() *cobra.Command {
-	return tokencmd.NewCommands(withTokenStore)
-}
-
-// withTokenStore loads the combined YAML config, initializes the store, and calls fn.
-func withTokenStore(cmd *cobra.Command, fn func(ctx context.Context, s store.Store) error) error {
-	if err := util.InitLog("error", "console"); err != nil {
-		return fmt.Errorf("init log: %w", err)
-	}
-
-	ctx := context.WithValue(cmd.Context(), hook.ExecutionContextKey, hook.SystemSource) //nolint:staticcheck
-
-	cfg, err := LoadConfig(configPath)
-	if err != nil {
-		return fmt.Errorf("load config: %w", err)
-	}
-
-	if dsn := cfg.Server.Store.DSN; dsn != "" {
-		switch strings.ToLower(cfg.Server.Store.Engine) {
-		case "postgres":
-			os.Setenv("NB_STORE_ENGINE_POSTGRES_DSN", dsn)
-		case "mysql":
-			os.Setenv("NB_STORE_ENGINE_MYSQL_DSN", dsn)
-		}
-	}
-	if file := cfg.Server.Store.File; file != "" {
-		os.Setenv("NB_STORE_ENGINE_SQLITE_FILE", file)
-	}
-
-	datadir := cfg.Management.DataDir
-	engine := types.Engine(cfg.Management.Store.Engine)
-
-	s, err := store.NewStore(ctx, engine, datadir, nil, true)
-	if err != nil {
-		return fmt.Errorf("create store: %w", err)
-	}
-	defer func() {
-		if err := s.Close(ctx); err != nil {
-			log.Debugf("close store: %v", err)
-		}
-	}()
-
-	return fn(ctx, s)
-}

dns/nameserver.go

@@ -53,6 +53,7 @@ type NameServerGroup struct {
 	ID string `gorm:"primaryKey"`
 	// AccountID is a reference to Account that this object belongs
 	AccountID string `gorm:"index"`
+	PublicID  string `json:"-"`
 	// Name group name
 	Name string
 	// Description group description

docs/io.netbird.client.plist

@@ -66,6 +66,9 @@
     <key>disableAutoConnect</key>
     <false/>
 
+    <key>disableAutostart</key>
+    <false/>
+
     <key>disableClientRoutes</key>
     <false/>
 

docs/netbird-macos.mobileconfig

@@ -103,6 +103,8 @@
                                 <!--
                                 <key>disableAutoConnect</key>
                                 <false/>
+                                <key>disableAutostart</key>
+                                <false/>
                                 <key>disableClientRoutes</key>
                                 <false/>
                                 <key>disableServerRoutes</key>

docs/netbird-macos.sh

@@ -58,6 +58,7 @@ preSharedKey="$NULL"                       # secret; redacted in log
 allowServerSSH='true'
 blockInbound="$NULL"
 disableAutoConnect="$NULL"
+disableAutostart="$NULL"
 disableClientRoutes="$NULL"
 disableServerRoutes="$NULL"
 disableMetricsCollection="$NULL"
@@ -155,6 +156,7 @@ main() {
   is_set "$allowServerSSH"            && emit_bool    allowServerSSH            "$allowServerSSH"
   is_set "$blockInbound"              && emit_bool    blockInbound              "$blockInbound"
   is_set "$disableAutoConnect"        && emit_bool    disableAutoConnect        "$disableAutoConnect"
+  is_set "$disableAutostart"          && emit_bool    disableAutostart          "$disableAutostart"
   is_set "$disableClientRoutes"       && emit_bool    disableClientRoutes       "$disableClientRoutes"
   is_set "$disableServerRoutes"       && emit_bool    disableServerRoutes       "$disableServerRoutes"
   is_set "$disableMetricsCollection"  && emit_bool    disableMetricsCollection  "$disableMetricsCollection"

docs/netbird.adml

@@ -24,6 +24,9 @@
       <string id="DisableAutoConnect_Name">Disable auto-connect</string>
       <string id="DisableAutoConnect_Help">When enabled, the NetBird tunnel does not auto-connect at daemon startup. Equivalent to --disable-auto-connect.</string>
 
+      <string id="DisableAutostart_Name">Disable autostart</string>
+      <string id="DisableAutostart_Help">When enabled, the NetBird GUI is prevented from registering itself as an OS autostart entry on fresh installs, and any existing OS autostart entry registration is removed on the next GUI launch (Windows Registry Run key, macOS Login Item, Linux .desktop). Once the admin lifts the policy, the setting stays off until the user re-enables it in Settings.</string>
+
       <string id="DisableClientRoutes_Name">Disable client routes</string>
       <string id="DisableClientRoutes_Help">When enabled, this client will not consume routes advertised by routing peers. Equivalent to --disable-client-routes.</string>