[client] Distinguish between NXDOMAIN and NODATA in the dns forwarder (#4321)

client/internal/dns/file_parser_unix_test.go

@@ -176,4 +176,3 @@ nameserver 192.168.0.1
 		t.Errorf("unexpected resolv.conf content: %v", cfg)
 	}
 }
-

client/internal/dns/host_windows.go

@@ -64,9 +64,10 @@ const (
 )
 
 type registryConfigurator struct {
-	guid       string
+	guid           string
-	routingAll bool
+	routingAll     bool
-	gpo        bool
+	gpo            bool
+	nrptEntryCount int
 }
 
 func newHostManager(wgInterface WGIface) (*registryConfigurator, error) {
@@ -177,7 +178,11 @@ func (r *registryConfigurator) applyDNSConfig(config HostDNSConfig, stateManager
 		log.Infof("removed %s as main DNS forwarder for this peer", config.ServerIP)
 	}
 
-	if err := stateManager.UpdateState(&ShutdownState{Guid: r.guid, GPO: r.gpo}); err != nil {
+	if err := stateManager.UpdateState(&ShutdownState{
+		Guid:           r.guid,
+		GPO:            r.gpo,
+		NRPTEntryCount: r.nrptEntryCount,
+	}); err != nil {
 		log.Errorf("failed to update shutdown state: %s", err)
 	}
 
@@ -193,13 +198,24 @@ func (r *registryConfigurator) applyDNSConfig(config HostDNSConfig, stateManager
 	}
 
 	if len(matchDomains) != 0 {
-		if err := r.addDNSMatchPolicy(matchDomains, config.ServerIP); err != nil {
+		count, err := r.addDNSMatchPolicy(matchDomains, config.ServerIP)
+		if err != nil {
 			return fmt.Errorf("add dns match policy: %w", err)
 		}
+		r.nrptEntryCount = count
 	} else {
 		if err := r.removeDNSMatchPolicies(); err != nil {
 			return fmt.Errorf("remove dns match policies: %w", err)
 		}
+		r.nrptEntryCount = 0
+	}
+
+	if err := stateManager.UpdateState(&ShutdownState{
+		Guid:           r.guid,
+		GPO:            r.gpo,
+		NRPTEntryCount: r.nrptEntryCount,
+	}); err != nil {
+		log.Errorf("failed to update shutdown state: %s", err)
 	}
 
 	if err := r.updateSearchDomains(searchDomains); err != nil {
@@ -220,28 +236,34 @@ func (r *registryConfigurator) addDNSSetupForAll(ip netip.Addr) error {
 	return nil
 }
 
-func (r *registryConfigurator) addDNSMatchPolicy(domains []string, ip netip.Addr) error {
+func (r *registryConfigurator) addDNSMatchPolicy(domains []string, ip netip.Addr) (int, error) {
 	// if the gpo key is present, we need to put our DNS settings there, otherwise our config might be ignored
 	// see https://learn.microsoft.com/en-us/openspecs/windows_protocols/ms-gpnrpt/8cc31cb9-20cb-4140-9e85-3e08703b4745
-	if r.gpo {
+	for i, domain := range domains {
-		if err := r.configureDNSPolicy(gpoDnsPolicyConfigMatchPath, domains, ip); err != nil {
+		policyPath := fmt.Sprintf("%s-%d", dnsPolicyConfigMatchPath, i)
-			return fmt.Errorf("configure GPO DNS policy: %w", err)
+		if r.gpo {
+			policyPath = fmt.Sprintf("%s-%d", gpoDnsPolicyConfigMatchPath, i)
 		}
 
+		singleDomain := []string{domain}
+
+		if err := r.configureDNSPolicy(policyPath, singleDomain, ip); err != nil {
+			return i, fmt.Errorf("configure DNS policy for domain %s: %w", domain, err)
+		}
+
+		log.Debugf("added NRPT entry for domain: %s", domain)
+	}
+
+	if r.gpo {
 		if err := refreshGroupPolicy(); err != nil {
 			log.Warnf("failed to refresh group policy: %v", err)
 		}
-	} else {
-		if err := r.configureDNSPolicy(dnsPolicyConfigMatchPath, domains, ip); err != nil {
-			return fmt.Errorf("configure local DNS policy: %w", err)
-		}
 	}
 
-	log.Infof("added %d match domains. Domain list: %s", len(domains), domains)
+	log.Infof("added %d separate NRPT entries. Domain list: %s", len(domains), domains)
-	return nil
+	return len(domains), nil
 }
 
-// configureDNSPolicy handles the actual configuration of a DNS policy at the specified path
 func (r *registryConfigurator) configureDNSPolicy(policyPath string, domains []string, ip netip.Addr) error {
 	if err := removeRegistryKeyFromDNSPolicyConfig(policyPath); err != nil {
 		return fmt.Errorf("remove existing dns policy: %w", err)
@@ -374,12 +396,25 @@ func (r *registryConfigurator) restoreHostDNS() error {
 
 func (r *registryConfigurator) removeDNSMatchPolicies() error {
 	var merr *multierror.Error
+
+	// Try to remove the base entries (for backward compatibility)
 	if err := removeRegistryKeyFromDNSPolicyConfig(dnsPolicyConfigMatchPath); err != nil {
-		merr = multierror.Append(merr, fmt.Errorf("remove local registry key: %w", err))
+		merr = multierror.Append(merr, fmt.Errorf("remove local base entry: %w", err))
+	}
+	if err := removeRegistryKeyFromDNSPolicyConfig(gpoDnsPolicyConfigMatchPath); err != nil {
+		merr = multierror.Append(merr, fmt.Errorf("remove GPO base entry: %w", err))
 	}
 
-	if err := removeRegistryKeyFromDNSPolicyConfig(gpoDnsPolicyConfigMatchPath); err != nil {
+	for i := 0; i < r.nrptEntryCount; i++ {
-		merr = multierror.Append(merr, fmt.Errorf("remove GPO registry key: %w", err))
+		localPath := fmt.Sprintf("%s-%d", dnsPolicyConfigMatchPath, i)
+		gpoPath := fmt.Sprintf("%s-%d", gpoDnsPolicyConfigMatchPath, i)
+
+		if err := removeRegistryKeyFromDNSPolicyConfig(localPath); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove local entry %d: %w", i, err))
+		}
+		if err := removeRegistryKeyFromDNSPolicyConfig(gpoPath); err != nil {
+			merr = multierror.Append(merr, fmt.Errorf("remove GPO entry %d: %w", i, err))
+		}
 	}
 
 	if err := refreshGroupPolicy(); err != nil {

client/internal/dns/server.go

@@ -695,6 +695,12 @@ func (s *DefaultServer) createHandlersForDomainGroup(domainGroup nsGroupsByDomai
 					ns.IP.String(), ns.NSType.String(), nbdns.UDPNameServerType.String())
 				continue
 			}
+
+			if ns.IP == s.service.RuntimeIP() {
+				log.Warnf("skipping nameserver %s as it matches our DNS server IP, preventing potential loop", ns.IP)
+				continue
+			}
+
 			handler.upstreamServers = append(handler.upstreamServers, ns.AddrPort())
 		}
 

client/internal/dns/server_test.go

@@ -2056,3 +2056,124 @@ func TestLocalResolverPriorityConstants(t *testing.T) {
 	assert.Equal(t, PriorityLocal, localMuxUpdates[0].priority, "Local handler should use PriorityLocal")
 	assert.Equal(t, "local.example.com", localMuxUpdates[0].domain)
 }
+
+func TestDNSLoopPrevention(t *testing.T) {
+	wgInterface := &mocWGIface{}
+	service := NewServiceViaMemory(wgInterface)
+	dnsServerIP := service.RuntimeIP()
+
+	server := &DefaultServer{
+		ctx:           context.Background(),
+		wgInterface:   wgInterface,
+		service:       service,
+		localResolver: local.NewResolver(),
+		handlerChain:  NewHandlerChain(),
+		hostManager:   &noopHostConfigurator{},
+		dnsMuxMap:     make(registeredHandlerMap),
+	}
+
+	tests := []struct {
+		name              string
+		nsGroups          []*nbdns.NameServerGroup
+		expectedHandlers  int
+		expectedServers   []netip.Addr
+		shouldFilterOwnIP bool
+	}{
+		{
+			name: "FilterOwnDNSServerIP",
+			nsGroups: []*nbdns.NameServerGroup{
+				{
+					Primary: true,
+					NameServers: []nbdns.NameServer{
+						{IP: netip.MustParseAddr("8.8.8.8"), NSType: nbdns.UDPNameServerType, Port: 53},
+						{IP: dnsServerIP, NSType: nbdns.UDPNameServerType, Port: 53},
+						{IP: netip.MustParseAddr("1.1.1.1"), NSType: nbdns.UDPNameServerType, Port: 53},
+					},
+					Domains: []string{},
+				},
+			},
+			expectedHandlers:  1,
+			expectedServers:   []netip.Addr{netip.MustParseAddr("8.8.8.8"), netip.MustParseAddr("1.1.1.1")},
+			shouldFilterOwnIP: true,
+		},
+		{
+			name: "AllServersFiltered",
+			nsGroups: []*nbdns.NameServerGroup{
+				{
+					Primary: false,
+					NameServers: []nbdns.NameServer{
+						{IP: dnsServerIP, NSType: nbdns.UDPNameServerType, Port: 53},
+					},
+					Domains: []string{"example.com"},
+				},
+			},
+			expectedHandlers:  0,
+			expectedServers:   []netip.Addr{},
+			shouldFilterOwnIP: true,
+		},
+		{
+			name: "MixedServersWithOwnIP",
+			nsGroups: []*nbdns.NameServerGroup{
+				{
+					Primary: false,
+					NameServers: []nbdns.NameServer{
+						{IP: netip.MustParseAddr("8.8.8.8"), NSType: nbdns.UDPNameServerType, Port: 53},
+						{IP: dnsServerIP, NSType: nbdns.UDPNameServerType, Port: 53},
+						{IP: netip.MustParseAddr("1.1.1.1"), NSType: nbdns.UDPNameServerType, Port: 53},
+						{IP: dnsServerIP, NSType: nbdns.UDPNameServerType, Port: 53}, // duplicate
+					},
+					Domains: []string{"test.com"},
+				},
+			},
+			expectedHandlers:  1,
+			expectedServers:   []netip.Addr{netip.MustParseAddr("8.8.8.8"), netip.MustParseAddr("1.1.1.1")},
+			shouldFilterOwnIP: true,
+		},
+		{
+			name: "NoOwnIPInList",
+			nsGroups: []*nbdns.NameServerGroup{
+				{
+					Primary: true,
+					NameServers: []nbdns.NameServer{
+						{IP: netip.MustParseAddr("8.8.8.8"), NSType: nbdns.UDPNameServerType, Port: 53},
+						{IP: netip.MustParseAddr("1.1.1.1"), NSType: nbdns.UDPNameServerType, Port: 53},
+					},
+					Domains: []string{},
+				},
+			},
+			expectedHandlers:  1,
+			expectedServers:   []netip.Addr{netip.MustParseAddr("8.8.8.8"), netip.MustParseAddr("1.1.1.1")},
+			shouldFilterOwnIP: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			muxUpdates, err := server.buildUpstreamHandlerUpdate(tt.nsGroups)
+			assert.NoError(t, err)
+			assert.Len(t, muxUpdates, tt.expectedHandlers)
+
+			if tt.expectedHandlers > 0 {
+				handler := muxUpdates[0].handler.(*upstreamResolver)
+				assert.Len(t, handler.upstreamServers, len(tt.expectedServers))
+
+				if tt.shouldFilterOwnIP {
+					for _, upstream := range handler.upstreamServers {
+						assert.NotEqual(t, dnsServerIP, upstream.Addr())
+					}
+				}
+
+				for _, expected := range tt.expectedServers {
+					found := false
+					for _, upstream := range handler.upstreamServers {
+						if upstream.Addr() == expected {
+							found = true
+							break
+						}
+					}
+					assert.True(t, found, "Expected server %s not found", expected)
+				}
+			}
+		})
+	}
+}

client/internal/dns/unclean_shutdown_windows.go

@@ -5,8 +5,9 @@ import (
 )
 
 type ShutdownState struct {
-	Guid string
+	Guid           string
-	GPO  bool
+	GPO            bool
+	NRPTEntryCount int
 }
 
 func (s *ShutdownState) Name() string {
@@ -15,8 +16,9 @@ func (s *ShutdownState) Name() string {
 
 func (s *ShutdownState) Cleanup() error {
 	manager := &registryConfigurator{
-		guid: s.Guid,
+		guid:           s.Guid,
-		gpo:  s.GPO,
+		gpo:            s.GPO,
+		nrptEntryCount: s.NRPTEntryCount,
 	}
 
 	if err := manager.restoreUncleanShutdownDNS(); err != nil {

client/internal/dnsfwd/forwarder.go

@@ -165,7 +165,7 @@ func (f *DNSForwarder) handleDNSQuery(w dns.ResponseWriter, query *dns.Msg) *dns
 	defer cancel()
 	ips, err := f.resolver.LookupNetIP(ctx, network, domain)
 	if err != nil {
-		f.handleDNSError(w, query, resp, domain, err)
+		f.handleDNSError(ctx, w, question, resp, domain, err)
 		return nil
 	}
 
@@ -244,20 +244,57 @@ func (f *DNSForwarder) updateFirewall(matchingEntries []*ForwarderEntry, prefixe
 	}
 }
 
+// setResponseCodeForNotFound determines and sets the appropriate response code when IsNotFound is true
+// It distinguishes between NXDOMAIN (domain doesn't exist) and NODATA (domain exists but no records of requested type)
+//
+// LIMITATION: This function only checks A and AAAA record types to determine domain existence.
+// If a domain has only other record types (MX, TXT, CNAME, etc.) but no A/AAAA records,
+// it may incorrectly return NXDOMAIN instead of NODATA. This is acceptable since the forwarder
+// only handles A/AAAA queries and returns NOTIMP for other types.
+func (f *DNSForwarder) setResponseCodeForNotFound(ctx context.Context, resp *dns.Msg, domain string, originalQtype uint16) {
+	// Try querying for a different record type to see if the domain exists
+	// If the original query was for AAAA, try A. If it was for A, try AAAA.
+	// This helps distinguish between NXDOMAIN and NODATA.
+	var alternativeNetwork string
+	switch originalQtype {
+	case dns.TypeAAAA:
+		alternativeNetwork = "ip4"
+	case dns.TypeA:
+		alternativeNetwork = "ip6"
+	default:
+		resp.Rcode = dns.RcodeNameError
+		return
+	}
+
+	if _, err := f.resolver.LookupNetIP(ctx, alternativeNetwork, domain); err != nil {
+		var dnsErr *net.DNSError
+		if errors.As(err, &dnsErr) && dnsErr.IsNotFound {
+			// Alternative query also returned not found - domain truly doesn't exist
+			resp.Rcode = dns.RcodeNameError
+			return
+		}
+		// Some other error (timeout, server failure, etc.) - can't determine, assume domain exists
+		resp.Rcode = dns.RcodeSuccess
+		return
+	}
+
+	// Alternative query succeeded - domain exists but has no records of this type
+	resp.Rcode = dns.RcodeSuccess
+}
+
 // handleDNSError processes DNS lookup errors and sends an appropriate error response
-func (f *DNSForwarder) handleDNSError(w dns.ResponseWriter, query, resp *dns.Msg, domain string, err error) {
+func (f *DNSForwarder) handleDNSError(ctx context.Context, w dns.ResponseWriter, question dns.Question, resp *dns.Msg, domain string, err error) {
 	var dnsErr *net.DNSError
 
 	switch {
 	case errors.As(err, &dnsErr):
 		resp.Rcode = dns.RcodeServerFailure
 		if dnsErr.IsNotFound {
-			// Pass through NXDOMAIN
+			f.setResponseCodeForNotFound(ctx, resp, domain, question.Qtype)
-			resp.Rcode = dns.RcodeNameError
 		}
 
 		if dnsErr.Server != "" {
-			log.Warnf("failed to resolve query for type=%s domain=%s server=%s: %v", dns.TypeToString[query.Question[0].Qtype], domain, dnsErr.Server, err)
+			log.Warnf("failed to resolve query for type=%s domain=%s server=%s: %v", dns.TypeToString[question.Qtype], domain, dnsErr.Server, err)
 		} else {
 			log.Warnf(errResolveFailed, domain, err)
 		}

client/internal/dnsfwd/forwarder_test.go

@@ -3,6 +3,7 @@ package dnsfwd
 import (
 	"context"
 	"fmt"
+	"net"
 	"net/netip"
 	"strings"
 	"testing"
@@ -16,8 +17,8 @@ import (
 	firewall "github.com/netbirdio/netbird/client/firewall/manager"
 	"github.com/netbirdio/netbird/client/internal/dns/test"
 	"github.com/netbirdio/netbird/client/internal/peer"
-	"github.com/netbirdio/netbird/shared/management/domain"
 	"github.com/netbirdio/netbird/route"
+	"github.com/netbirdio/netbird/shared/management/domain"
 )
 
 func Test_getMatchingEntries(t *testing.T) {
@@ -708,6 +709,131 @@ func TestDNSForwarder_MultipleOverlappingPatterns(t *testing.T) {
 	assert.Len(t, matches, 3, "Should match 3 patterns")
 }
 
+// TestDNSForwarder_NodataVsNxdomain tests that the forwarder correctly distinguishes
+// between NXDOMAIN (domain doesn't exist) and NODATA (domain exists but no records of that type)
+func TestDNSForwarder_NodataVsNxdomain(t *testing.T) {
+	mockFirewall := &MockFirewall{}
+	mockResolver := &MockResolver{}
+
+	forwarder := NewDNSForwarder("127.0.0.1:0", 300, mockFirewall, &peer.Status{})
+	forwarder.resolver = mockResolver
+
+	d, err := domain.FromString("example.com")
+	require.NoError(t, err)
+
+	set := firewall.NewDomainSet([]domain.Domain{d})
+	entries := []*ForwarderEntry{{Domain: d, ResID: "test-res", Set: set}}
+	forwarder.UpdateDomains(entries)
+
+	tests := []struct {
+		name           string
+		queryType      uint16
+		setupMocks     func()
+		expectedCode   int
+		expectNoAnswer bool // true if we expect NOERROR with empty answer (NODATA case)
+		description    string
+	}{
+		{
+			name:      "domain exists but no AAAA records (NODATA)",
+			queryType: dns.TypeAAAA,
+			setupMocks: func() {
+				// First query for AAAA returns not found
+				mockResolver.On("LookupNetIP", mock.Anything, "ip6", "example.com.").
+					Return([]netip.Addr{}, &net.DNSError{IsNotFound: true, Name: "example.com"}).Once()
+				// Check query for A records succeeds (domain exists)
+				mockResolver.On("LookupNetIP", mock.Anything, "ip4", "example.com.").
+					Return([]netip.Addr{netip.MustParseAddr("1.2.3.4")}, nil).Once()
+			},
+			expectedCode:   dns.RcodeSuccess,
+			expectNoAnswer: true,
+			description:    "Should return NOERROR when domain exists but has no records of requested type",
+		},
+		{
+			name:      "domain exists but no A records (NODATA)",
+			queryType: dns.TypeA,
+			setupMocks: func() {
+				// First query for A returns not found
+				mockResolver.On("LookupNetIP", mock.Anything, "ip4", "example.com.").
+					Return([]netip.Addr{}, &net.DNSError{IsNotFound: true, Name: "example.com"}).Once()
+				// Check query for AAAA records succeeds (domain exists)
+				mockResolver.On("LookupNetIP", mock.Anything, "ip6", "example.com.").
+					Return([]netip.Addr{netip.MustParseAddr("2001:db8::1")}, nil).Once()
+			},
+			expectedCode:   dns.RcodeSuccess,
+			expectNoAnswer: true,
+			description:    "Should return NOERROR when domain exists but has no A records",
+		},
+		{
+			name:      "domain doesn't exist (NXDOMAIN)",
+			queryType: dns.TypeA,
+			setupMocks: func() {
+				// First query for A returns not found
+				mockResolver.On("LookupNetIP", mock.Anything, "ip4", "example.com.").
+					Return([]netip.Addr{}, &net.DNSError{IsNotFound: true, Name: "example.com"}).Once()
+				// Check query for AAAA also returns not found (domain doesn't exist)
+				mockResolver.On("LookupNetIP", mock.Anything, "ip6", "example.com.").
+					Return([]netip.Addr{}, &net.DNSError{IsNotFound: true, Name: "example.com"}).Once()
+			},
+			expectedCode:   dns.RcodeNameError,
+			expectNoAnswer: true,
+			description:    "Should return NXDOMAIN when domain doesn't exist at all",
+		},
+		{
+			name:      "domain exists with records (normal success)",
+			queryType: dns.TypeA,
+			setupMocks: func() {
+				mockResolver.On("LookupNetIP", mock.Anything, "ip4", "example.com.").
+					Return([]netip.Addr{netip.MustParseAddr("1.2.3.4")}, nil).Once()
+				// Expect firewall update for successful resolution
+				expectedPrefix := netip.PrefixFrom(netip.MustParseAddr("1.2.3.4"), 32)
+				mockFirewall.On("UpdateSet", set, []netip.Prefix{expectedPrefix}).Return(nil).Once()
+			},
+			expectedCode:   dns.RcodeSuccess,
+			expectNoAnswer: false,
+			description:    "Should return NOERROR with answer when records exist",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			// Reset mock expectations
+			mockResolver.ExpectedCalls = nil
+			mockResolver.Calls = nil
+			mockFirewall.ExpectedCalls = nil
+			mockFirewall.Calls = nil
+
+			tt.setupMocks()
+
+			query := &dns.Msg{}
+			query.SetQuestion(dns.Fqdn("example.com"), tt.queryType)
+
+			var writtenResp *dns.Msg
+			mockWriter := &test.MockResponseWriter{
+				WriteMsgFunc: func(m *dns.Msg) error {
+					writtenResp = m
+					return nil
+				},
+			}
+
+			resp := forwarder.handleDNSQuery(mockWriter, query)
+
+			// If a response was returned, it means it should be written (happens in wrapper functions)
+			if resp != nil && writtenResp == nil {
+				writtenResp = resp
+			}
+
+			require.NotNil(t, writtenResp, "Expected response to be written")
+			assert.Equal(t, tt.expectedCode, writtenResp.Rcode, tt.description)
+
+			if tt.expectNoAnswer {
+				assert.Empty(t, writtenResp.Answer, "Response should have no answer records")
+			}
+
+			mockResolver.AssertExpectations(t)
+		})
+	}
+}
+
 func TestDNSForwarder_EmptyQuery(t *testing.T) {
 	// Test handling of malformed query with no questions
 	forwarder := NewDNSForwarder("127.0.0.1:0", 300, nil, &peer.Status{})

relay/cmd/pprof.go

@@ -0,0 +1,33 @@
+//go:build pprof
+// +build pprof
+
+package cmd
+
+import (
+	"net/http"
+	_ "net/http/pprof"
+	"os"
+
+	log "github.com/sirupsen/logrus"
+)
+
+func init() {
+	addr := pprofAddr()
+	go pprof(addr)
+}
+
+func pprofAddr() string {
+	listenAddr := os.Getenv("NB_PPROF_ADDR")
+	if listenAddr == "" {
+		return "localhost:6969"
+	}
+
+	return listenAddr
+}
+
+func pprof(listenAddr string) {
+	log.Infof("listening pprof on: %s\n", listenAddr)
+	if err := http.ListenAndServe(listenAddr, nil); err != nil {
+		log.Fatalf("Failed to start pprof: %v", err)
+	}
+}