update synctrace

.github/ISSUE_TEMPLATE/config.yml

@@ -17,5 +17,5 @@ contact_links:
     about: 'The documentation is the best place to start if you are new to Nebula.'
 
   - name: 💁 Support/Chat
-    url: https://join.slack.com/t/nebulaoss/shared_invite/zt-39pk4xopc-CUKlGcb5Z39dQ0cK1v7ehA
+    url: https://join.slack.com/t/nebulaoss/shared_invite/zt-2xqe6e7vn-k_KGi8s13nsr7cvHVvHvuQ
     about: 'For faster support, join us on Slack for assistance!'

.github/workflows/gofmt.yml

@@ -16,9 +16,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version: '1.24'
         check-latest: true
 
     - name: Install goimports

.github/workflows/release.yml

@@ -12,9 +12,9 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - uses: actions/setup-go@v6
+      - uses: actions/setup-go@v5
         with:
-          go-version: '1.25'
+          go-version: '1.24'
           check-latest: true
 
       - name: Build
@@ -35,9 +35,9 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - uses: actions/setup-go@v6
+      - uses: actions/setup-go@v5
         with:
-          go-version: '1.25'
+          go-version: '1.24'
           check-latest: true
 
       - name: Build
@@ -68,9 +68,9 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - uses: actions/setup-go@v6
+      - uses: actions/setup-go@v5
         with:
-          go-version: '1.25'
+          go-version: '1.24'
           check-latest: true
 
       - name: Import certificates

.github/workflows/smoke-extra.yml

@@ -22,9 +22,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version-file: 'go.mod'
         check-latest: true
 
     - name: add hashicorp source

.github/workflows/smoke.yml

@@ -20,9 +20,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version: '1.24'
         check-latest: true
 
     - name: build

.github/workflows/test.yml

@@ -20,9 +20,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version: '1.24'
         check-latest: true
 
     - name: Build
@@ -32,9 +32,9 @@ jobs:
       run: make vet
 
     - name: golangci-lint
-      uses: golangci/golangci-lint-action@v8
+      uses: golangci/golangci-lint-action@v7
       with:
-        version: v2.5
+        version: v2.0
 
     - name: Test
       run: make test
@@ -58,9 +58,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version: '1.24'
         check-latest: true
 
     - name: Build
@@ -79,9 +79,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version: '1.22'
         check-latest: true
 
     - name: Build
@@ -100,9 +100,9 @@ jobs:
 
     - uses: actions/checkout@v4
 
-    - uses: actions/setup-go@v6
+    - uses: actions/setup-go@v5
       with:
-        go-version: '1.25'
+        go-version: '1.24'
         check-latest: true
 
     - name: Build nebula
@@ -115,9 +115,9 @@ jobs:
       run: make vet
 
     - name: golangci-lint
-      uses: golangci/golangci-lint-action@v8
+      uses: golangci/golangci-lint-action@v7
       with:
-        version: v2.5
+        version: v2.0
 
     - name: Test
       run: make test

Makefile

@@ -61,7 +61,7 @@ ALL = $(ALL_LINUX) \
 	windows-arm64
 
 e2e:
-	$(TEST_ENV) go test -tags=e2e_testing -count=1 $(TEST_FLAGS) ./e2e
+	$(TEST_ENV) go test -tags=synctrace,e2e_testing -count=1 $(TEST_FLAGS) ./e2e
 
 e2ev: TEST_FLAGS += -v
 e2ev: e2e
@@ -215,6 +215,7 @@ ifeq ($(words $(MAKECMDGOALS)),1)
 	@$(MAKE) service ${.DEFAULT_GOAL} --no-print-directory
 endif
 
+bin-docker: BUILD_ARGS = -tags=synctrace
 bin-docker: bin build/linux-amd64/nebula build/linux-amd64/nebula-cert
 
 smoke-docker: bin-docker

README.md

@@ -12,7 +12,7 @@ Further documentation can be found [here](https://nebula.defined.net/docs/).
 
 You can read more about Nebula [here](https://medium.com/p/884110a5579).
 
-You can also join the NebulaOSS Slack group [here](https://join.slack.com/t/nebulaoss/shared_invite/zt-39pk4xopc-CUKlGcb5Z39dQ0cK1v7ehA).
+You can also join the NebulaOSS Slack group [here](https://join.slack.com/t/nebulaoss/shared_invite/zt-2xqe6e7vn-k_KGi8s13nsr7cvHVvHvuQ).
 
 ## Supported Platforms
 

calculated_remote.go

@@ -84,11 +84,16 @@ func NewCalculatedRemotesFromConfig(c *config.C, k string) (*bart.Table[[]*calcu
 
 	calculatedRemotes := new(bart.Table[[]*calculatedRemote])
 
-	rawMap, ok := value.(map[string]any)
+	rawMap, ok := value.(map[any]any)
 	if !ok {
 		return nil, fmt.Errorf("config `%s` has invalid type: %T", k, value)
 	}
-	for rawCIDR, rawValue := range rawMap {
+	for rawKey, rawValue := range rawMap {
+		rawCIDR, ok := rawKey.(string)
+		if !ok {
+			return nil, fmt.Errorf("config `%s` has invalid key (type %T): %v", k, rawKey, rawKey)
+		}
+
 		cidr, err := netip.ParsePrefix(rawCIDR)
 		if err != nil {
 			return nil, fmt.Errorf("config `%s` has invalid CIDR: %s", k, rawCIDR)
@@ -124,7 +129,7 @@ func newCalculatedRemotesListFromConfig(cidr netip.Prefix, raw any) ([]*calculat
 }
 
 func newCalculatedRemotesEntryFromConfig(cidr netip.Prefix, raw any) (*calculatedRemote, error) {
-	rawMap, ok := raw.(map[string]any)
+	rawMap, ok := raw.(map[any]any)
 	if !ok {
 		return nil, fmt.Errorf("invalid type: %T", raw)
 	}

cert/cert.go

@@ -135,7 +135,8 @@ func Recombine(v Version, rawCertBytes, publicKey []byte, curve Curve) (Certific
 	case Version2:
 		c, err = unmarshalCertificateV2(rawCertBytes, publicKey, curve)
 	default:
-		return nil, ErrUnknownVersion
+		//TODO: CERT-V2 make a static var
+		return nil, fmt.Errorf("unknown certificate version %d", v)
 	}
 
 	if err != nil {

cert/cert_v1.go

@@ -110,10 +110,8 @@ func (c *certificateV1) CheckSignature(key []byte) bool {
 	case Curve_CURVE25519:
 		return ed25519.Verify(key, b, c.signature)
 	case Curve_P256:
-		pubKey, err := ecdsa.ParseUncompressedPublicKey(elliptic.P256(), key)
+		x, y := elliptic.Unmarshal(elliptic.P256(), key)
-		if err != nil {
+		pubKey := &ecdsa.PublicKey{Curve: elliptic.P256(), X: x, Y: y}
-			return false
-		}
 		hashed := sha256.Sum256(b)
 		return ecdsa.VerifyASN1(pubKey, hashed[:], c.signature)
 	default:

cert/cert_v2.go

@@ -149,10 +149,8 @@ func (c *certificateV2) CheckSignature(key []byte) bool {
 	case Curve_CURVE25519:
 		return ed25519.Verify(key, b, c.signature)
 	case Curve_P256:
-		pubKey, err := ecdsa.ParseUncompressedPublicKey(elliptic.P256(), key)
+		x, y := elliptic.Unmarshal(elliptic.P256(), key)
-		if err != nil {
+		pubKey := &ecdsa.PublicKey{Curve: elliptic.P256(), X: x, Y: y}
-			return false
-		}
 		hashed := sha256.Sum256(b)
 		return ecdsa.VerifyASN1(pubKey, hashed[:], c.signature)
 	default:

cert/errors.go

@@ -20,7 +20,6 @@ var (
 	ErrPublicPrivateKeyMismatch   = errors.New("public key and private key are not a pair")
 	ErrPrivateKeyEncrypted        = errors.New("private key must be decrypted")
 	ErrCaNotFound                 = errors.New("could not find ca for the certificate")
-	ErrUnknownVersion             = errors.New("certificate version unrecognized")
 
 	ErrInvalidPEMBlock                   = errors.New("input did not contain a valid PEM encoded block")
 	ErrInvalidPEMCertificateBanner       = errors.New("bytes did not contain a proper certificate banner")

cert/sign.go

@@ -7,6 +7,7 @@ import (
 	"crypto/rand"
 	"crypto/sha256"
 	"fmt"
+	"math/big"
 	"net/netip"
 	"time"
 )
@@ -54,10 +55,15 @@ func (t *TBSCertificate) Sign(signer Certificate, curve Curve, key []byte) (Cert
 		}
 		return t.SignWith(signer, curve, sp)
 	case Curve_P256:
-		pk, err := ecdsa.ParseRawPrivateKey(elliptic.P256(), key)
+		pk := &ecdsa.PrivateKey{
-		if err != nil {
+			PublicKey: ecdsa.PublicKey{
-			return nil, err
+				Curve: elliptic.P256(),
+			},
+			// ref: https://github.com/golang/go/blob/go1.19/src/crypto/x509/sec1.go#L95
+			D: new(big.Int).SetBytes(key),
 		}
+		// ref: https://github.com/golang/go/blob/go1.19/src/crypto/x509/sec1.go#L119
+		pk.X, pk.Y = pk.Curve.ScalarBaseMult(key)
 		sp := func(certBytes []byte) ([]byte, error) {
 			// We need to hash first for ECDSA
 			// - https://pkg.go.dev/crypto/ecdsa#SignASN1

cert_test/cert.go

@@ -114,33 +114,6 @@ func NewTestCert(v cert.Version, curve cert.Curve, ca cert.Certificate, key []by
 	return c, pub, cert.MarshalPrivateKeyToPEM(curve, priv), pem
 }
 
-func NewTestCertDifferentVersion(c cert.Certificate, v cert.Version, ca cert.Certificate, key []byte) (cert.Certificate, []byte) {
-	nc := &cert.TBSCertificate{
-		Version:        v,
-		Curve:          c.Curve(),
-		Name:           c.Name(),
-		Networks:       c.Networks(),
-		UnsafeNetworks: c.UnsafeNetworks(),
-		Groups:         c.Groups(),
-		NotBefore:      time.Unix(c.NotBefore().Unix(), 0),
-		NotAfter:       time.Unix(c.NotAfter().Unix(), 0),
-		PublicKey:      c.PublicKey(),
-		IsCA:           false,
-	}
-
-	c, err := nc.Sign(ca, ca.Curve(), key)
-	if err != nil {
-		panic(err)
-	}
-
-	pem, err := c.MarshalPEM()
-	if err != nil {
-		panic(err)
-	}
-
-	return c, pem
-}
-
 func X25519Keypair() ([]byte, []byte) {
 	privkey := make([]byte, 32)
 	if _, err := io.ReadFull(rand.Reader, privkey); err != nil {

config/config.go

@@ -11,12 +11,12 @@ import (
 	"sort"
 	"strconv"
 	"strings"
-	"sync"
 	"syscall"
 	"time"
 
 	"dario.cat/mergo"
 	"github.com/sirupsen/logrus"
+	"github.com/wadey/synctrace"
 	"gopkg.in/yaml.v3"
 )
 
@@ -27,13 +27,14 @@ type C struct {
 	oldSettings map[string]any
 	callbacks   []func(*C)
 	l           *logrus.Logger
-	reloadLock  sync.Mutex
+	reloadLock  synctrace.Mutex
 }
 
 func NewC(l *logrus.Logger) *C {
 	return &C{
-		Settings: make(map[string]any),
+		Settings:   make(map[string]any),
-		l:        l,
+		l:          l,
+		reloadLock: synctrace.NewMutex("config-reload"),
 	}
 }
 

connection_manager.go

@@ -4,17 +4,14 @@ import (
 	"bytes"
 	"context"
 	"encoding/binary"
-	"fmt"
 	"net/netip"
-	"sync"
-	"sync/atomic"
 	"time"
 
 	"github.com/rcrowley/go-metrics"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
-	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/header"
+	"github.com/wadey/synctrace"
 )
 
 type trafficDecision int
@@ -30,124 +27,130 @@ const (
 )
 
 type connectionManager struct {
+	in     map[uint32]struct{}
+	inLock synctrace.RWMutex
+
+	out     map[uint32]struct{}
+	outLock synctrace.RWMutex
+
 	// relayUsed holds which relay localIndexs are in use
 	relayUsed     map[uint32]struct{}
-	relayUsedLock *sync.RWMutex
+	relayUsedLock synctrace.RWMutex
 
-	hostMap      *HostMap
+	hostMap                 *HostMap
-	trafficTimer *LockingTimerWheel[uint32]
+	trafficTimer            *LockingTimerWheel[uint32]
-	intf         *Interface
+	intf                    *Interface
-	punchy       *Punchy
+	pendingDeletion         map[uint32]struct{}
-
+	punchy                  *Punchy
-	// Configuration settings
 	checkInterval           time.Duration
 	pendingDeletionInterval time.Duration
-	inactivityTimeout       atomic.Int64
+	metricsTxPunchy         metrics.Counter
-	dropInactive            atomic.Bool
-
-	metricsTxPunchy metrics.Counter
 
 	l *logrus.Logger
 }
 
-func newConnectionManagerFromConfig(l *logrus.Logger, c *config.C, hm *HostMap, p *Punchy) *connectionManager {
+func newConnectionManager(ctx context.Context, l *logrus.Logger, intf *Interface, checkInterval, pendingDeletionInterval time.Duration, punchy *Punchy) *connectionManager {
-	cm := &connectionManager{
+	var max time.Duration
-		hostMap:         hm,
+	if checkInterval < pendingDeletionInterval {
-		l:               l,
+		max = pendingDeletionInterval
-		punchy:          p,
+	} else {
-		relayUsed:       make(map[uint32]struct{}),
+		max = checkInterval
-		relayUsedLock:   &sync.RWMutex{},
-		metricsTxPunchy: metrics.GetOrRegisterCounter("messages.tx.punchy", nil),
 	}
 
-	cm.reload(c, true)
+	nc := &connectionManager{
-	c.RegisterReloadCallback(func(c *config.C) {
+		hostMap:                 intf.hostMap,
-		cm.reload(c, false)
+		in:                      make(map[uint32]struct{}),
-	})
+		inLock:                  synctrace.NewRWMutex("connection-manager-in"),
-
+		out:                     make(map[uint32]struct{}),
-	return cm
+		outLock:                 synctrace.NewRWMutex("connection-manager-out"),
-}
+		relayUsed:               make(map[uint32]struct{}),
-
+		relayUsedLock:           synctrace.NewRWMutex("connection-manager-relay-used"),
-func (cm *connectionManager) reload(c *config.C, initial bool) {
+		trafficTimer:            NewLockingTimerWheel[uint32]("traffic-timer", time.Millisecond*500, max),
-	if initial {
+		intf:                    intf,
-		cm.checkInterval = time.Duration(c.GetInt("timers.connection_alive_interval", 5)) * time.Second
+		pendingDeletion:         make(map[uint32]struct{}),
-		cm.pendingDeletionInterval = time.Duration(c.GetInt("timers.pending_deletion_interval", 10)) * time.Second
+		checkInterval:           checkInterval,
-
+		pendingDeletionInterval: pendingDeletionInterval,
-		// We want at least a minimum resolution of 500ms per tick so that we can hit these intervals
+		punchy:                  punchy,
-		// pretty close to their configured duration.
+		metricsTxPunchy:         metrics.GetOrRegisterCounter("messages.tx.punchy", nil),
-		// The inactivity duration is checked each time a hostinfo ticks through so we don't need the wheel to contain it.
+		l:                       l,
-		minDuration := min(time.Millisecond*500, cm.checkInterval, cm.pendingDeletionInterval)
-		maxDuration := max(cm.checkInterval, cm.pendingDeletionInterval)
-		cm.trafficTimer = NewLockingTimerWheel[uint32](minDuration, maxDuration)
 	}
 
-	if initial || c.HasChanged("tunnels.inactivity_timeout") {
+	nc.Start(ctx)
-		old := cm.getInactivityTimeout()
+	return nc
-		cm.inactivityTimeout.Store((int64)(c.GetDuration("tunnels.inactivity_timeout", 10*time.Minute)))
-		if !initial {
-			cm.l.WithField("oldDuration", old).
-				WithField("newDuration", cm.getInactivityTimeout()).
-				Info("Inactivity timeout has changed")
-		}
-	}
-
-	if initial || c.HasChanged("tunnels.drop_inactive") {
-		old := cm.dropInactive.Load()
-		cm.dropInactive.Store(c.GetBool("tunnels.drop_inactive", false))
-		if !initial {
-			cm.l.WithField("oldBool", old).
-				WithField("newBool", cm.dropInactive.Load()).
-				Info("Drop inactive setting has changed")
-		}
-	}
 }
 
-func (cm *connectionManager) getInactivityTimeout() time.Duration {
+func (n *connectionManager) In(localIndex uint32) {
-	return (time.Duration)(cm.inactivityTimeout.Load())
+	n.inLock.RLock()
-}
-
-func (cm *connectionManager) In(h *HostInfo) {
-	h.in.Store(true)
-}
-
-func (cm *connectionManager) Out(h *HostInfo) {
-	h.out.Store(true)
-}
-
-func (cm *connectionManager) RelayUsed(localIndex uint32) {
-	cm.relayUsedLock.RLock()
 	// If this already exists, return
-	if _, ok := cm.relayUsed[localIndex]; ok {
+	if _, ok := n.in[localIndex]; ok {
-		cm.relayUsedLock.RUnlock()
+		n.inLock.RUnlock()
 		return
 	}
-	cm.relayUsedLock.RUnlock()
+	n.inLock.RUnlock()
-	cm.relayUsedLock.Lock()
+	n.inLock.Lock()
-	cm.relayUsed[localIndex] = struct{}{}
+	n.in[localIndex] = struct{}{}
-	cm.relayUsedLock.Unlock()
+	n.inLock.Unlock()
+}
+
+func (n *connectionManager) Out(localIndex uint32) {
+	n.outLock.RLock()
+	// If this already exists, return
+	if _, ok := n.out[localIndex]; ok {
+		n.outLock.RUnlock()
+		return
+	}
+	n.outLock.RUnlock()
+	n.outLock.Lock()
+	n.out[localIndex] = struct{}{}
+	n.outLock.Unlock()
+}
+
+func (n *connectionManager) RelayUsed(localIndex uint32) {
+	n.relayUsedLock.RLock()
+	// If this already exists, return
+	if _, ok := n.relayUsed[localIndex]; ok {
+		n.relayUsedLock.RUnlock()
+		return
+	}
+	n.relayUsedLock.RUnlock()
+	n.relayUsedLock.Lock()
+	n.relayUsed[localIndex] = struct{}{}
+	n.relayUsedLock.Unlock()
 }
 
 // getAndResetTrafficCheck returns if there was any inbound or outbound traffic within the last tick and
 // resets the state for this local index
-func (cm *connectionManager) getAndResetTrafficCheck(h *HostInfo, now time.Time) (bool, bool) {
+func (n *connectionManager) getAndResetTrafficCheck(localIndex uint32) (bool, bool) {
-	in := h.in.Swap(false)
+	n.inLock.Lock()
-	out := h.out.Swap(false)
+	n.outLock.Lock()
-	if in || out {
+	_, in := n.in[localIndex]
-		h.lastUsed = now
+	_, out := n.out[localIndex]
-	}
+	delete(n.in, localIndex)
+	delete(n.out, localIndex)
+	n.inLock.Unlock()
+	n.outLock.Unlock()
 	return in, out
 }
 
-// AddTrafficWatch must be called for every new HostInfo.
+func (n *connectionManager) AddTrafficWatch(localIndex uint32) {
-// We will continue to monitor the HostInfo until the tunnel is dropped.
+	// Use a write lock directly because it should be incredibly rare that we are ever already tracking this index
-func (cm *connectionManager) AddTrafficWatch(h *HostInfo) {
+	n.outLock.Lock()
-	if h.out.Swap(true) == false {
+	if _, ok := n.out[localIndex]; ok {
-		cm.trafficTimer.Add(h.localIndexId, cm.checkInterval)
+		n.outLock.Unlock()
+		return
 	}
+	n.out[localIndex] = struct{}{}
+	n.trafficTimer.Add(localIndex, n.checkInterval)
+	n.outLock.Unlock()
 }
 
-func (cm *connectionManager) Start(ctx context.Context) {
+func (n *connectionManager) Start(ctx context.Context) {
-	clockSource := time.NewTicker(cm.trafficTimer.t.tickDuration)
+	go n.Run(ctx)
+}
+
+func (n *connectionManager) Run(ctx context.Context) {
+	//TODO: this tick should be based on the min wheel tick? Check firewall
+	clockSource := time.NewTicker(500 * time.Millisecond)
 	defer clockSource.Stop()
 
 	p := []byte("")
@@ -160,61 +163,61 @@ func (cm *connectionManager) Start(ctx context.Context) {
 			return
 
 		case now := <-clockSource.C:
-			cm.trafficTimer.Advance(now)
+			n.trafficTimer.Advance(now)
 			for {
-				localIndex, has := cm.trafficTimer.Purge()
+				localIndex, has := n.trafficTimer.Purge()
 				if !has {
 					break
 				}
 
-				cm.doTrafficCheck(localIndex, p, nb, out, now)
+				n.doTrafficCheck(localIndex, p, nb, out, now)
 			}
 		}
 	}
 }
 
-func (cm *connectionManager) doTrafficCheck(localIndex uint32, p, nb, out []byte, now time.Time) {
+func (n *connectionManager) doTrafficCheck(localIndex uint32, p, nb, out []byte, now time.Time) {
-	decision, hostinfo, primary := cm.makeTrafficDecision(localIndex, now)
+	decision, hostinfo, primary := n.makeTrafficDecision(localIndex, now)
 
 	switch decision {
 	case deleteTunnel:
-		if cm.hostMap.DeleteHostInfo(hostinfo) {
+		if n.hostMap.DeleteHostInfo(hostinfo) {
 			// Only clearing the lighthouse cache if this is the last hostinfo for this vpn ip in the hostmap
-			cm.intf.lightHouse.DeleteVpnAddrs(hostinfo.vpnAddrs)
+			n.intf.lightHouse.DeleteVpnAddrs(hostinfo.vpnAddrs)
 		}
 
 	case closeTunnel:
-		cm.intf.sendCloseTunnel(hostinfo)
+		n.intf.sendCloseTunnel(hostinfo)
-		cm.intf.closeTunnel(hostinfo)
+		n.intf.closeTunnel(hostinfo)
 
 	case swapPrimary:
-		cm.swapPrimary(hostinfo, primary)
+		n.swapPrimary(hostinfo, primary)
 
 	case migrateRelays:
-		cm.migrateRelayUsed(hostinfo, primary)
+		n.migrateRelayUsed(hostinfo, primary)
 
 	case tryRehandshake:
-		cm.tryRehandshake(hostinfo)
+		n.tryRehandshake(hostinfo)
 
 	case sendTestPacket:
-		cm.intf.SendMessageToHostInfo(header.Test, header.TestRequest, hostinfo, p, nb, out)
+		n.intf.SendMessageToHostInfo(header.Test, header.TestRequest, hostinfo, p, nb, out)
 	}
 
-	cm.resetRelayTrafficCheck(hostinfo)
+	n.resetRelayTrafficCheck(hostinfo)
 }
 
-func (cm *connectionManager) resetRelayTrafficCheck(hostinfo *HostInfo) {
+func (n *connectionManager) resetRelayTrafficCheck(hostinfo *HostInfo) {
 	if hostinfo != nil {
-		cm.relayUsedLock.Lock()
+		n.relayUsedLock.Lock()
-		defer cm.relayUsedLock.Unlock()
+		defer n.relayUsedLock.Unlock()
 		// No need to migrate any relays, delete usage info now.
 		for _, idx := range hostinfo.relayState.CopyRelayForIdxs() {
-			delete(cm.relayUsed, idx)
+			delete(n.relayUsed, idx)
 		}
 	}
 }
 
-func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo) {
+func (n *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo) {
 	relayFor := oldhostinfo.relayState.CopyAllRelayFor()
 
 	for _, r := range relayFor {
@@ -224,51 +227,46 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 		var relayFrom netip.Addr
 		var relayTo netip.Addr
 		switch {
-		case ok:
+		case ok && existing.State == Established:
-			switch existing.State {
+			// This relay already exists in newhostinfo, then do nothing.
-			case Established, PeerRequested, Disestablished:
+			continue
-				// This relay already exists in newhostinfo, then do nothing.
+		case ok && existing.State == Requested:
-				continue
+			// The relay exists in a Requested state; re-send the request
-			case Requested:
+			index = existing.LocalIndex
-				// The relay exists in a Requested state; re-send the request
+			switch r.Type {
-				index = existing.LocalIndex
+			case TerminalType:
-				switch r.Type {
+				relayFrom = n.intf.myVpnAddrs[0]
-				case TerminalType:
+				relayTo = existing.PeerAddr
-					relayFrom = cm.intf.myVpnAddrs[0]
+			case ForwardingType:
-					relayTo = existing.PeerAddr
+				relayFrom = existing.PeerAddr
-				case ForwardingType:
+				relayTo = newhostinfo.vpnAddrs[0]
-					relayFrom = existing.PeerAddr
+			default:
-					relayTo = newhostinfo.vpnAddrs[0]
+				// should never happen
-				default:
-					// should never happen
-					panic(fmt.Sprintf("Migrating unknown relay type: %v", r.Type))
-				}
 			}
 		case !ok:
-			cm.relayUsedLock.RLock()
+			n.relayUsedLock.RLock()
-			if _, relayUsed := cm.relayUsed[r.LocalIndex]; !relayUsed {
+			if _, relayUsed := n.relayUsed[r.LocalIndex]; !relayUsed {
 				// The relay hasn't been used; don't migrate it.
-				cm.relayUsedLock.RUnlock()
+				n.relayUsedLock.RUnlock()
 				continue
 			}
-			cm.relayUsedLock.RUnlock()
+			n.relayUsedLock.RUnlock()
 			// The relay doesn't exist at all; create some relay state and send the request.
 			var err error
-			index, err = AddRelay(cm.l, newhostinfo, cm.hostMap, r.PeerAddr, nil, r.Type, Requested)
+			index, err = AddRelay(n.l, newhostinfo, n.hostMap, r.PeerAddr, nil, r.Type, Requested)
 			if err != nil {
-				cm.l.WithError(err).Error("failed to migrate relay to new hostinfo")
+				n.l.WithError(err).Error("failed to migrate relay to new hostinfo")
 				continue
 			}
 			switch r.Type {
 			case TerminalType:
-				relayFrom = cm.intf.myVpnAddrs[0]
+				relayFrom = n.intf.myVpnAddrs[0]
 				relayTo = r.PeerAddr
 			case ForwardingType:
 				relayFrom = r.PeerAddr
 				relayTo = newhostinfo.vpnAddrs[0]
 			default:
 				// should never happen
-				panic(fmt.Sprintf("Migrating unknown relay type: %v", r.Type))
 			}
 		}
 
@@ -281,12 +279,12 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 		switch newhostinfo.GetCert().Certificate.Version() {
 		case cert.Version1:
 			if !relayFrom.Is4() {
-				cm.l.Error("can not migrate v1 relay with a v6 network because the relay is not running a current nebula version")
+				n.l.Error("can not migrate v1 relay with a v6 network because the relay is not running a current nebula version")
 				continue
 			}
 
 			if !relayTo.Is4() {
-				cm.l.Error("can not migrate v1 relay with a v6 remote network because the relay is not running a current nebula version")
+				n.l.Error("can not migrate v1 relay with a v6 remote network because the relay is not running a current nebula version")
 				continue
 			}
 
@@ -298,16 +296,16 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 			req.RelayFromAddr = netAddrToProtoAddr(relayFrom)
 			req.RelayToAddr = netAddrToProtoAddr(relayTo)
 		default:
-			newhostinfo.logger(cm.l).Error("Unknown certificate version found while attempting to migrate relay")
+			newhostinfo.logger(n.l).Error("Unknown certificate version found while attempting to migrate relay")
 			continue
 		}
 
 		msg, err := req.Marshal()
 		if err != nil {
-			cm.l.WithError(err).Error("failed to marshal Control message to migrate relay")
+			n.l.WithError(err).Error("failed to marshal Control message to migrate relay")
 		} else {
-			cm.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
+			n.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
-			cm.l.WithFields(logrus.Fields{
+			n.l.WithFields(logrus.Fields{
 				"relayFrom":           req.RelayFromAddr,
 				"relayTo":             req.RelayToAddr,
 				"initiatorRelayIndex": req.InitiatorRelayIndex,
@@ -318,44 +316,46 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 	}
 }
 
-func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time) (trafficDecision, *HostInfo, *HostInfo) {
+func (n *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time) (trafficDecision, *HostInfo, *HostInfo) {
-	// Read lock the main hostmap to order decisions based on tunnels being the primary tunnel
+	n.hostMap.RLock()
-	cm.hostMap.RLock()
+	defer n.hostMap.RUnlock()
-	defer cm.hostMap.RUnlock()
 
-	hostinfo := cm.hostMap.Indexes[localIndex]
+	hostinfo := n.hostMap.Indexes[localIndex]
 	if hostinfo == nil {
-		cm.l.WithField("localIndex", localIndex).Debugln("Not found in hostmap")
+		n.l.WithField("localIndex", localIndex).Debugf("Not found in hostmap")
+		delete(n.pendingDeletion, localIndex)
 		return doNothing, nil, nil
 	}
 
-	if cm.isInvalidCertificate(now, hostinfo) {
+	if n.isInvalidCertificate(now, hostinfo) {
+		delete(n.pendingDeletion, hostinfo.localIndexId)
 		return closeTunnel, hostinfo, nil
 	}
 
-	primary := cm.hostMap.Hosts[hostinfo.vpnAddrs[0]]
+	primary := n.hostMap.Hosts[hostinfo.vpnAddrs[0]]
 	mainHostInfo := true
 	if primary != nil && primary != hostinfo {
 		mainHostInfo = false
 	}
 
 	// Check for traffic on this hostinfo
-	inTraffic, outTraffic := cm.getAndResetTrafficCheck(hostinfo, now)
+	inTraffic, outTraffic := n.getAndResetTrafficCheck(localIndex)
 
 	// A hostinfo is determined alive if there is incoming traffic
 	if inTraffic {
 		decision := doNothing
-		if cm.l.Level >= logrus.DebugLevel {
+		if n.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(cm.l).
+			hostinfo.logger(n.l).
 				WithField("tunnelCheck", m{"state": "alive", "method": "passive"}).
 				Debug("Tunnel status")
 		}
-		hostinfo.pendingDeletion.Store(false)
+		delete(n.pendingDeletion, hostinfo.localIndexId)
 
 		if mainHostInfo {
 			decision = tryRehandshake
+
 		} else {
-			if cm.shouldSwapPrimary(hostinfo) {
+			if n.shouldSwapPrimary(hostinfo, primary) {
 				decision = swapPrimary
 			} else {
 				// migrate the relays to the primary, if in use.
@@ -363,55 +363,46 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 			}
 		}
 
-		cm.trafficTimer.Add(hostinfo.localIndexId, cm.checkInterval)
+		n.trafficTimer.Add(hostinfo.localIndexId, n.checkInterval)
 
 		if !outTraffic {
 			// Send a punch packet to keep the NAT state alive
-			cm.sendPunch(hostinfo)
+			n.sendPunch(hostinfo)
 		}
 
 		return decision, hostinfo, primary
 	}
 
-	if hostinfo.pendingDeletion.Load() {
+	if _, ok := n.pendingDeletion[hostinfo.localIndexId]; ok {
 		// We have already sent a test packet and nothing was returned, this hostinfo is dead
-		hostinfo.logger(cm.l).
+		hostinfo.logger(n.l).
 			WithField("tunnelCheck", m{"state": "dead", "method": "active"}).
 			Info("Tunnel status")
 
+		delete(n.pendingDeletion, hostinfo.localIndexId)
 		return deleteTunnel, hostinfo, nil
 	}
 
 	decision := doNothing
 	if hostinfo != nil && hostinfo.ConnectionState != nil && mainHostInfo {
 		if !outTraffic {
-			inactiveFor, isInactive := cm.isInactive(hostinfo, now)
-			if isInactive {
-				// Tunnel is inactive, tear it down
-				hostinfo.logger(cm.l).
-					WithField("inactiveDuration", inactiveFor).
-					WithField("primary", mainHostInfo).
-					Info("Dropping tunnel due to inactivity")
-
-				return closeTunnel, hostinfo, primary
-			}
-
 			// If we aren't sending or receiving traffic then its an unused tunnel and we don't to test the tunnel.
 			// Just maintain NAT state if configured to do so.
-			cm.sendPunch(hostinfo)
+			n.sendPunch(hostinfo)
-			cm.trafficTimer.Add(hostinfo.localIndexId, cm.checkInterval)
+			n.trafficTimer.Add(hostinfo.localIndexId, n.checkInterval)
 			return doNothing, nil, nil
+
 		}
 
-		if cm.punchy.GetTargetEverything() {
+		if n.punchy.GetTargetEverything() {
 			// This is similar to the old punchy behavior with a slight optimization.
 			// We aren't receiving traffic but we are sending it, punch on all known
 			// ips in case we need to re-prime NAT state
-			cm.sendPunch(hostinfo)
+			n.sendPunch(hostinfo)
 		}
 
-		if cm.l.Level >= logrus.DebugLevel {
+		if n.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(cm.l).
+			hostinfo.logger(n.l).
 				WithField("tunnelCheck", m{"state": "testing", "method": "active"}).
 				Debug("Tunnel status")
 		}
@@ -420,33 +411,17 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 		decision = sendTestPacket
 
 	} else {
-		if cm.l.Level >= logrus.DebugLevel {
+		if n.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(cm.l).Debugf("Hostinfo sadness")
+			hostinfo.logger(n.l).Debugf("Hostinfo sadness")
 		}
 	}
 
-	hostinfo.pendingDeletion.Store(true)
+	n.pendingDeletion[hostinfo.localIndexId] = struct{}{}
-	cm.trafficTimer.Add(hostinfo.localIndexId, cm.pendingDeletionInterval)
+	n.trafficTimer.Add(hostinfo.localIndexId, n.pendingDeletionInterval)
 	return decision, hostinfo, nil
 }
 
-func (cm *connectionManager) isInactive(hostinfo *HostInfo, now time.Time) (time.Duration, bool) {
+func (n *connectionManager) shouldSwapPrimary(current, primary *HostInfo) bool {
-	if cm.dropInactive.Load() == false {
-		// We aren't configured to drop inactive tunnels
-		return 0, false
-	}
-
-	inactiveDuration := now.Sub(hostinfo.lastUsed)
-	if inactiveDuration < cm.getInactivityTimeout() {
-		// It's not considered inactive
-		return inactiveDuration, false
-	}
-
-	// The tunnel is inactive
-	return inactiveDuration, true
-}
-
-func (cm *connectionManager) shouldSwapPrimary(current *HostInfo) bool {
 	// The primary tunnel is the most recent handshake to complete locally and should work entirely fine.
 	// If we are here then we have multiple tunnels for a host pair and neither side believes the same tunnel is primary.
 	// Let's sort this out.
@@ -454,127 +429,83 @@ func (cm *connectionManager) shouldSwapPrimary(current *HostInfo) bool {
 	// Only one side should swap because if both swap then we may never resolve to a single tunnel.
 	// vpn addr is static across all tunnels for this host pair so lets
 	// use that to determine if we should consider swapping.
-	if current.vpnAddrs[0].Compare(cm.intf.myVpnAddrs[0]) < 0 {
+	if current.vpnAddrs[0].Compare(n.intf.myVpnAddrs[0]) < 0 {
 		// Their primary vpn addr is less than mine. Do not swap.
 		return false
 	}
 
-	crt := cm.intf.pki.getCertState().getCertificate(current.ConnectionState.myCert.Version())
+	crt := n.intf.pki.getCertState().getCertificate(current.ConnectionState.myCert.Version())
-	if crt == nil {
-		//my cert was reloaded away. We should definitely swap from this tunnel
-		return true
-	}
 	// If this tunnel is using the latest certificate then we should swap it to primary for a bit and see if things
 	// settle down.
 	return bytes.Equal(current.ConnectionState.myCert.Signature(), crt.Signature())
 }
 
-func (cm *connectionManager) swapPrimary(current, primary *HostInfo) {
+func (n *connectionManager) swapPrimary(current, primary *HostInfo) {
-	cm.hostMap.Lock()
+	n.hostMap.Lock()
 	// Make sure the primary is still the same after the write lock. This avoids a race with a rehandshake.
-	if cm.hostMap.Hosts[current.vpnAddrs[0]] == primary {
+	if n.hostMap.Hosts[current.vpnAddrs[0]] == primary {
-		cm.hostMap.unlockedMakePrimary(current)
+		n.hostMap.unlockedMakePrimary(current)
 	}
-	cm.hostMap.Unlock()
+	n.hostMap.Unlock()
 }
 
-// isInvalidCertificate decides if we should destroy a tunnel.
+// isInvalidCertificate will check if we should destroy a tunnel if pki.disconnect_invalid is true and
-// returns true if pki.disconnect_invalid is true and the certificate is no longer valid.
+// the certificate is no longer valid. Block listed certificates will skip the pki.disconnect_invalid
-// Blocklisted certificates will skip the pki.disconnect_invalid check and return true.
+// check and return true.
-func (cm *connectionManager) isInvalidCertificate(now time.Time, hostinfo *HostInfo) bool {
+func (n *connectionManager) isInvalidCertificate(now time.Time, hostinfo *HostInfo) bool {
 	remoteCert := hostinfo.GetCert()
 	if remoteCert == nil {
-		return false //don't tear down tunnels for handshakes in progress
-	}
-
-	caPool := cm.intf.pki.GetCAPool()
-	err := caPool.VerifyCachedCertificate(now, remoteCert)
-	if err == nil {
-		return false //cert is still valid! yay!
-	} else if err == cert.ErrBlockListed { //avoiding errors.Is for speed
-		// Block listed certificates should always be disconnected
-		hostinfo.logger(cm.l).WithError(err).
-			WithField("fingerprint", remoteCert.Fingerprint).
-			Info("Remote certificate is blocked, tearing down the tunnel")
-		return true
-	} else if cm.intf.disconnectInvalid.Load() {
-		hostinfo.logger(cm.l).WithError(err).
-			WithField("fingerprint", remoteCert.Fingerprint).
-			Info("Remote certificate is no longer valid, tearing down the tunnel")
-		return true
-	} else {
-		//if we reach here, the cert is no longer valid, but we're configured to keep tunnels from now-invalid certs open
 		return false
 	}
+
+	caPool := n.intf.pki.GetCAPool()
+	err := caPool.VerifyCachedCertificate(now, remoteCert)
+	if err == nil {
+		return false
+	}
+
+	if !n.intf.disconnectInvalid.Load() && err != cert.ErrBlockListed {
+		// Block listed certificates should always be disconnected
+		return false
+	}
+
+	hostinfo.logger(n.l).WithError(err).
+		WithField("fingerprint", remoteCert.Fingerprint).
+		Info("Remote certificate is no longer valid, tearing down the tunnel")
+
+	return true
 }
 
-func (cm *connectionManager) sendPunch(hostinfo *HostInfo) {
+func (n *connectionManager) sendPunch(hostinfo *HostInfo) {
-	if !cm.punchy.GetPunch() {
+	if !n.punchy.GetPunch() {
 		// Punching is disabled
 		return
 	}
 
-	if cm.intf.lightHouse.IsAnyLighthouseAddr(hostinfo.vpnAddrs) {
+	if n.punchy.GetTargetEverything() {
-		// Do not punch to lighthouses, we assume our lighthouse update interval is good enough.
+		hostinfo.remotes.ForEach(n.hostMap.GetPreferredRanges(), func(addr netip.AddrPort, preferred bool) {
-		// In the event the update interval is not sufficient to maintain NAT state then a publicly available lighthouse
+			n.metricsTxPunchy.Inc(1)
-		// would lose the ability to notify us and punchy.respond would become unreliable.
+			n.intf.outside.WriteTo([]byte{1}, addr)
-		return
-	}
-
-	if cm.punchy.GetTargetEverything() {
-		hostinfo.remotes.ForEach(cm.hostMap.GetPreferredRanges(), func(addr netip.AddrPort, preferred bool) {
-			cm.metricsTxPunchy.Inc(1)
-			cm.intf.outside.WriteTo([]byte{1}, addr)
 		})
 
 	} else if hostinfo.remote.IsValid() {
-		cm.metricsTxPunchy.Inc(1)
+		n.metricsTxPunchy.Inc(1)
-		cm.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
+		n.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
 	}
 }
 
-func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
+func (n *connectionManager) tryRehandshake(hostinfo *HostInfo) {
-	cs := cm.intf.pki.getCertState()
+	cs := n.intf.pki.getCertState()
 	curCrt := hostinfo.ConnectionState.myCert
-	curCrtVersion := curCrt.Version()
+	myCrt := cs.getCertificate(curCrt.Version())
-	myCrt := cs.getCertificate(curCrtVersion)
+	if curCrt.Version() >= cs.initiatingVersion && bytes.Equal(curCrt.Signature(), myCrt.Signature()) == true {
-	if myCrt == nil {
+		// The current tunnel is using the latest certificate and version, no need to rehandshake.
-		cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
-			WithField("version", curCrtVersion).
-			WithField("reason", "local certificate removed").
-			Info("Re-handshaking with remote")
-		cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
 		return
 	}
-	peerCrt := hostinfo.ConnectionState.peerCert
-	if peerCrt != nil && curCrtVersion < peerCrt.Certificate.Version() {
-		// if our certificate version is less than theirs, and we have a matching version available, rehandshake?
-		if cs.getCertificate(peerCrt.Certificate.Version()) != nil {
-			cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
-				WithField("version", curCrtVersion).
-				WithField("peerVersion", peerCrt.Certificate.Version()).
-				WithField("reason", "local certificate version lower than peer, attempting to correct").
-				Info("Re-handshaking with remote")
-			cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], func(hh *HandshakeHostInfo) {
-				hh.initiatingVersionOverride = peerCrt.Certificate.Version()
-			})
-			return
-		}
-	}
-	if !bytes.Equal(curCrt.Signature(), myCrt.Signature()) {
-		cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
-			WithField("reason", "local certificate is not current").
-			Info("Re-handshaking with remote")
 
-		cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
+	n.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
-		return
+		WithField("reason", "local certificate is not current").
-	}
+		Info("Re-handshaking with remote")
-	if curCrtVersion < cs.initiatingVersion {
-		cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
-			WithField("reason", "current cert version < pki.initiatingVersion").
-			Info("Re-handshaking with remote")
 
-		cm.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
+	n.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
-		return
-	}
 }

connection_manager_test.go

@@ -1,6 +1,7 @@
 package nebula
 
 import (
+	"context"
 	"crypto/ed25519"
 	"crypto/rand"
 	"net/netip"
@@ -22,7 +23,7 @@ func newTestLighthouse() *LightHouse {
 		addrMap:   map[netip.Addr]*RemoteList{},
 		queryChan: make(chan netip.Addr, 10),
 	}
-	lighthouses := []netip.Addr{}
+	lighthouses := map[netip.Addr]struct{}{}
 	staticList := map[netip.Addr]struct{}{}
 
 	lh.lighthouses.Store(&lighthouses)
@@ -63,10 +64,10 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	conf := config.NewC(l)
+	ctx, cancel := context.WithCancel(context.Background())
-	punchy := NewPunchyFromConfig(l, conf)
+	defer cancel()
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
-	nc.intf = ifce
+	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
@@ -84,33 +85,32 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
 
 	// We saw traffic out to vpnIp
-	nc.Out(hostinfo)
+	nc.Out(hostinfo.localIndexId)
-	nc.In(hostinfo)
+	nc.In(hostinfo.localIndexId)
-	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
-	assert.True(t, hostinfo.out.Load())
+	assert.Contains(t, nc.out, hostinfo.localIndexId)
-	assert.True(t, hostinfo.in.Load())
 
 	// Do a traffic check tick, should not be pending deletion but should not have any in/out packets recorded
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.False(t, hostinfo.out.Load())
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
 
 	// Do another traffic check tick, this host should be pending deletion now
-	nc.Out(hostinfo)
+	nc.Out(hostinfo.localIndexId)
-	assert.True(t, hostinfo.out.Load())
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.True(t, hostinfo.pendingDeletion.Load())
+	assert.Contains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.False(t, hostinfo.out.Load())
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 
 	// Do a final traffic check tick, the host should now be removed
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.NotContains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs)
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.NotContains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 	assert.NotContains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 }
 
@@ -146,10 +146,10 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	conf := config.NewC(l)
+	ctx, cancel := context.WithCancel(context.Background())
-	punchy := NewPunchyFromConfig(l, conf)
+	defer cancel()
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
-	nc.intf = ifce
+	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
 	p := []byte("")
 	nb := make([]byte, 12, 12)
 	out := make([]byte, mtu)
@@ -167,129 +167,33 @@ func Test_NewConnectionManagerTest2(t *testing.T) {
 	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
 
 	// We saw traffic out to vpnIp
-	nc.Out(hostinfo)
+	nc.Out(hostinfo.localIndexId)
-	nc.In(hostinfo)
+	nc.In(hostinfo.localIndexId)
-	assert.True(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.vpnAddrs[0])
-	assert.True(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.pendingDeletion.Load())
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 
 	// Do a traffic check tick, should not be pending deletion but should not have any in/out packets recorded
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.False(t, hostinfo.out.Load())
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
 
 	// Do another traffic check tick, this host should be pending deletion now
-	nc.Out(hostinfo)
+	nc.Out(hostinfo.localIndexId)
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.True(t, hostinfo.pendingDeletion.Load())
+	assert.Contains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.False(t, hostinfo.out.Load())
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 
 	// We saw traffic, should no longer be pending deletion
-	nc.In(hostinfo)
+	nc.In(hostinfo.localIndexId)
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.False(t, hostinfo.pendingDeletion.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
-	assert.False(t, hostinfo.out.Load())
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
-	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
-	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
-}
-
-func Test_NewConnectionManager_DisconnectInactive(t *testing.T) {
-	l := test.NewLogger()
-	localrange := netip.MustParsePrefix("10.1.1.1/24")
-	vpnAddrs := []netip.Addr{netip.MustParseAddr("172.1.1.2")}
-	preferredRanges := []netip.Prefix{localrange}
-
-	// Very incomplete mock objects
-	hostMap := newHostMap(l)
-	hostMap.preferredRanges.Store(&preferredRanges)
-
-	cs := &CertState{
-		initiatingVersion: cert.Version1,
-		privateKey:        []byte{},
-		v1Cert:            &dummyCert{version: cert.Version1},
-		v1HandshakeBytes:  []byte{},
-	}
-
-	lh := newTestLighthouse()
-	ifce := &Interface{
-		hostMap:          hostMap,
-		inside:           &test.NoopTun{},
-		outside:          &udp.NoopConn{},
-		firewall:         &Firewall{},
-		lightHouse:       lh,
-		pki:              &PKI{},
-		handshakeManager: NewHandshakeManager(l, hostMap, lh, &udp.NoopConn{}, defaultHandshakeConfig),
-		l:                l,
-	}
-	ifce.pki.cs.Store(cs)
-
-	// Create manager
-	conf := config.NewC(l)
-	conf.Settings["tunnels"] = map[string]any{
-		"drop_inactive": true,
-	}
-	punchy := NewPunchyFromConfig(l, conf)
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
-	assert.True(t, nc.dropInactive.Load())
-	nc.intf = ifce
-
-	// Add an ip we have established a connection w/ to hostmap
-	hostinfo := &HostInfo{
-		vpnAddrs:      vpnAddrs,
-		localIndexId:  1099,
-		remoteIndexId: 9901,
-	}
-	hostinfo.ConnectionState = &ConnectionState{
-		myCert: &dummyCert{version: cert.Version1},
-		H:      &noise.HandshakeState{},
-	}
-	nc.hostMap.unlockedAddHostInfo(hostinfo, ifce)
-
-	// Do a traffic check tick, in and out should be cleared but should not be pending deletion
-	nc.Out(hostinfo)
-	nc.In(hostinfo)
-	assert.True(t, hostinfo.out.Load())
-	assert.True(t, hostinfo.in.Load())
-
-	now := time.Now()
-	decision, _, _ := nc.makeTrafficDecision(hostinfo.localIndexId, now)
-	assert.Equal(t, tryRehandshake, decision)
-	assert.Equal(t, now, hostinfo.lastUsed)
-	assert.False(t, hostinfo.pendingDeletion.Load())
-	assert.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
-
-	decision, _, _ = nc.makeTrafficDecision(hostinfo.localIndexId, now.Add(time.Second*5))
-	assert.Equal(t, doNothing, decision)
-	assert.Equal(t, now, hostinfo.lastUsed)
-	assert.False(t, hostinfo.pendingDeletion.Load())
-	assert.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
-
-	// Do another traffic check tick, should still not be pending deletion
-	decision, _, _ = nc.makeTrafficDecision(hostinfo.localIndexId, now.Add(time.Second*10))
-	assert.Equal(t, doNothing, decision)
-	assert.Equal(t, now, hostinfo.lastUsed)
-	assert.False(t, hostinfo.pendingDeletion.Load())
-	assert.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
-	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
-	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
-
-	// Finally advance beyond the inactivity timeout
-	decision, _, _ = nc.makeTrafficDecision(hostinfo.localIndexId, now.Add(time.Minute*10))
-	assert.Equal(t, closeTunnel, decision)
-	assert.Equal(t, now, hostinfo.lastUsed)
-	assert.False(t, hostinfo.pendingDeletion.Load())
-	assert.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
 	assert.Contains(t, nc.hostMap.Indexes, hostinfo.localIndexId)
 	assert.Contains(t, nc.hostMap.Hosts, hostinfo.vpnAddrs[0])
 }
@@ -360,10 +264,10 @@ func Test_NewConnectionManagerTest_DisconnectInvalid(t *testing.T) {
 	ifce.disconnectInvalid.Store(true)
 
 	// Create manager
-	conf := config.NewC(l)
+	ctx, cancel := context.WithCancel(context.Background())
-	punchy := NewPunchyFromConfig(l, conf)
+	defer cancel()
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
-	nc.intf = ifce
+	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
 	ifce.connectionManager = nc
 
 	hostinfo := &HostInfo{

connection_state.go

@@ -4,13 +4,13 @@ import (
 	"crypto/rand"
 	"encoding/json"
 	"fmt"
-	"sync"
 	"sync/atomic"
 
 	"github.com/flynn/noise"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/noiseutil"
+	"github.com/wadey/synctrace"
 )
 
 const ReplayWindow = 1024
@@ -24,7 +24,7 @@ type ConnectionState struct {
 	initiator      bool
 	messageCounter atomic.Uint64
 	window         *Bits
-	writeLock      sync.Mutex
+	writeLock      synctrace.Mutex
 }
 
 func NewConnectionState(l *logrus.Logger, cs *CertState, crt cert.Certificate, initiator bool, pattern noise.HandshakePattern) (*ConnectionState, error) {
@@ -76,6 +76,7 @@ func NewConnectionState(l *logrus.Logger, cs *CertState, crt cert.Certificate, i
 		initiator: initiator,
 		window:    b,
 		myCert:    crt,
+		writeLock: synctrace.NewMutex("connection-state"),
 	}
 	// always start the counter from 2, as packet 1 and packet 2 are handshake packets.
 	ci.messageCounter.Add(2)

control.go

@@ -26,15 +26,14 @@ type controlHostLister interface {
 }
 
 type Control struct {
-	f                      *Interface
+	f               *Interface
-	l                      *logrus.Logger
+	l               *logrus.Logger
-	ctx                    context.Context
+	ctx             context.Context
-	cancel                 context.CancelFunc
+	cancel          context.CancelFunc
-	sshStart               func()
+	sshStart        func()
-	statsStart             func()
+	statsStart      func()
-	dnsStart               func()
+	dnsStart        func()
-	lighthouseStart        func()
+	lighthouseStart func()
-	connectionManagerStart func(context.Context)
 }
 
 type ControlHostInfo struct {
@@ -64,9 +63,6 @@ func (c *Control) Start() {
 	if c.dnsStart != nil {
 		go c.dnsStart()
 	}
-	if c.connectionManagerStart != nil {
-		go c.connectionManagerStart(c.ctx)
-	}
 	if c.lighthouseStart != nil {
 		c.lighthouseStart()
 	}

control_test.go

@@ -53,7 +53,7 @@ func TestControl_GetHostInfoByVpnIp(t *testing.T) {
 		localIndexId:  201,
 		vpnAddrs:      []netip.Addr{vpnIp},
 		relayState: RelayState{
-			relays:         nil,
+			relays:         map[netip.Addr]struct{}{},
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
@@ -72,7 +72,7 @@ func TestControl_GetHostInfoByVpnIp(t *testing.T) {
 		localIndexId:  201,
 		vpnAddrs:      []netip.Addr{vpnIp2},
 		relayState: RelayState{
-			relays:         nil,
+			relays:         map[netip.Addr]struct{}{},
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},

dns_server.go

@@ -6,12 +6,12 @@ import (
 	"net/netip"
 	"strconv"
 	"strings"
-	"sync"
 
 	"github.com/gaissmai/bart"
 	"github.com/miekg/dns"
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
+	"github.com/wadey/synctrace"
 )
 
 // This whole thing should be rewritten to use context
@@ -21,7 +21,7 @@ var dnsServer *dns.Server
 var dnsAddr string
 
 type dnsRecords struct {
-	sync.RWMutex
+	synctrace.RWMutex
 	l               *logrus.Logger
 	dnsMap4         map[string]netip.Addr
 	dnsMap6         map[string]netip.Addr
@@ -31,6 +31,7 @@ type dnsRecords struct {
 
 func newDnsRecords(l *logrus.Logger, cs *CertState, hostMap *HostMap) *dnsRecords {
 	return &dnsRecords{
+		RWMutex:         synctrace.NewRWMutex("dns-records"),
 		l:               l,
 		dnsMap4:         make(map[string]netip.Addr),
 		dnsMap6:         make(map[string]netip.Addr),

e2e/handshakes_test.go

@@ -506,7 +506,7 @@ func TestReestablishRelays(t *testing.T) {
 	curIndexes := len(myControl.GetHostmap().Indexes)
 	for curIndexes >= start {
 		curIndexes = len(myControl.GetHostmap().Indexes)
-		r.Logf("Wait for the dead index to go away:start=%v indexes, current=%v indexes", start, curIndexes)
+		r.Logf("Wait for the dead index to go away:start=%v indexes, currnet=%v indexes", start, curIndexes)
 		myControl.InjectTunUDPPacket(theirVpnIpNet[0].Addr(), 80, myVpnIpNet[0].Addr(), 80, []byte("Hi from me should fail"))
 
 		r.RouteForAllExitFunc(func(p *udp.Packet, c *nebula.Control) router.ExitType {
@@ -1052,9 +1052,6 @@ func TestRehandshakingLoser(t *testing.T) {
 	t.Log("Stand up a tunnel between me and them")
 	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
 
-	myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false)
-	theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-
 	r.RenderHostmaps("Starting hostmaps", myControl, theirControl)
 
 	r.Log("Renew their certificate and spin until mine sees it")

e2e/helpers_test.go

@@ -29,6 +29,8 @@ type m = map[string]any
 
 // newSimpleServer creates a nebula instance with many assumptions
 func newSimpleServer(v cert.Version, caCrt cert.Certificate, caKey []byte, name string, sVpnNetworks string, overrides m) (*nebula.Control, []netip.Prefix, netip.AddrPort, *config.C) {
+	l := NewTestLogger()
+
 	var vpnNetworks []netip.Prefix
 	for _, sn := range strings.Split(sVpnNetworks, ",") {
 		vpnIpNet, err := netip.ParsePrefix(strings.TrimSpace(sn))
@@ -54,25 +56,6 @@ func newSimpleServer(v cert.Version, caCrt cert.Certificate, caKey []byte, name
 		budpIp[3] = 239
 		udpAddr = netip.AddrPortFrom(netip.AddrFrom16(budpIp), 4242)
 	}
-	return newSimpleServerWithUdp(v, caCrt, caKey, name, sVpnNetworks, udpAddr, overrides)
-}
-
-func newSimpleServerWithUdp(v cert.Version, caCrt cert.Certificate, caKey []byte, name string, sVpnNetworks string, udpAddr netip.AddrPort, overrides m) (*nebula.Control, []netip.Prefix, netip.AddrPort, *config.C) {
-	l := NewTestLogger()
-
-	var vpnNetworks []netip.Prefix
-	for _, sn := range strings.Split(sVpnNetworks, ",") {
-		vpnIpNet, err := netip.ParsePrefix(strings.TrimSpace(sn))
-		if err != nil {
-			panic(err)
-		}
-		vpnNetworks = append(vpnNetworks, vpnIpNet)
-	}
-
-	if len(vpnNetworks) == 0 {
-		panic("no vpn networks")
-	}
-
 	_, _, myPrivKey, myPEM := cert_test.NewTestCert(v, cert.Curve_CURVE25519, caCrt, caKey, name, time.Now(), time.Now().Add(5*time.Minute), vpnNetworks, nil, []string{})
 
 	caB, err := caCrt.MarshalPEM()
@@ -146,109 +129,6 @@ func newSimpleServerWithUdp(v cert.Version, caCrt cert.Certificate, caKey []byte
 	return control, vpnNetworks, udpAddr, c
 }
 
-// newServer creates a nebula instance with fewer assumptions
-func newServer(caCrt []cert.Certificate, certs []cert.Certificate, key []byte, overrides m) (*nebula.Control, []netip.Prefix, netip.AddrPort, *config.C) {
-	l := NewTestLogger()
-
-	vpnNetworks := certs[len(certs)-1].Networks()
-
-	var udpAddr netip.AddrPort
-	if vpnNetworks[0].Addr().Is4() {
-		budpIp := vpnNetworks[0].Addr().As4()
-		budpIp[1] -= 128
-		udpAddr = netip.AddrPortFrom(netip.AddrFrom4(budpIp), 4242)
-	} else {
-		budpIp := vpnNetworks[0].Addr().As16()
-		// beef for funsies
-		budpIp[2] = 190
-		budpIp[3] = 239
-		udpAddr = netip.AddrPortFrom(netip.AddrFrom16(budpIp), 4242)
-	}
-
-	caStr := ""
-	for _, ca := range caCrt {
-		x, err := ca.MarshalPEM()
-		if err != nil {
-			panic(err)
-		}
-		caStr += string(x)
-	}
-	certStr := ""
-	for _, c := range certs {
-		x, err := c.MarshalPEM()
-		if err != nil {
-			panic(err)
-		}
-		certStr += string(x)
-	}
-
-	mc := m{
-		"pki": m{
-			"ca":   caStr,
-			"cert": certStr,
-			"key":  string(key),
-		},
-		//"tun": m{"disabled": true},
-		"firewall": m{
-			"outbound": []m{{
-				"proto": "any",
-				"port":  "any",
-				"host":  "any",
-			}},
-			"inbound": []m{{
-				"proto": "any",
-				"port":  "any",
-				"host":  "any",
-			}},
-		},
-		//"handshakes": m{
-		//	"try_interval": "1s",
-		//},
-		"listen": m{
-			"host": udpAddr.Addr().String(),
-			"port": udpAddr.Port(),
-		},
-		"logging": m{
-			"timestamp_format": fmt.Sprintf("%v 15:04:05.000000", certs[0].Name()),
-			"level":            l.Level.String(),
-		},
-		"timers": m{
-			"pending_deletion_interval": 2,
-			"connection_alive_interval": 2,
-		},
-	}
-
-	if overrides != nil {
-		final := m{}
-		err := mergo.Merge(&final, overrides, mergo.WithAppendSlice)
-		if err != nil {
-			panic(err)
-		}
-		err = mergo.Merge(&final, mc, mergo.WithAppendSlice)
-		if err != nil {
-			panic(err)
-		}
-		mc = final
-	}
-
-	cb, err := yaml.Marshal(mc)
-	if err != nil {
-		panic(err)
-	}
-
-	c := config.NewC(l)
-	cStr := string(cb)
-	c.LoadString(cStr)
-
-	control, err := nebula.Main(c, false, "e2e-test", l, nil)
-
-	if err != nil {
-		panic(err)
-	}
-
-	return control, vpnNetworks, udpAddr, c
-}
-
 type doneCb func()
 
 func deadline(t *testing.T, seconds time.Duration) doneCb {

e2e/router/router.go

@@ -700,7 +700,6 @@ func (r *R) FlushAll() {
 			r.Unlock()
 			panic("Can't FlushAll for host: " + p.To.String())
 		}
-		receiver.InjectUDPPacket(p)
 		r.Unlock()
 	}
 }

e2e/tunnels_test.go

@@ -1,367 +0,0 @@
-//go:build e2e_testing
-// +build e2e_testing
-
-package e2e
-
-import (
-	"fmt"
-	"net/netip"
-	"testing"
-	"time"
-
-	"github.com/slackhq/nebula/cert"
-	"github.com/slackhq/nebula/cert_test"
-	"github.com/slackhq/nebula/e2e/router"
-	"github.com/stretchr/testify/assert"
-	"gopkg.in/yaml.v3"
-)
-
-func TestDropInactiveTunnels(t *testing.T) {
-	// The goal of this test is to ensure the shortest inactivity timeout will close the tunnel on both sides
-	// under ideal conditions
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, myUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "me", "10.128.0.1/24", m{"tunnels": m{"drop_inactive": true, "inactivity_timeout": "5s"}})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServer(cert.Version1, ca, caKey, "them", "10.128.0.2/24", m{"tunnels": m{"drop_inactive": true, "inactivity_timeout": "10m"}})
-
-	// Share our underlay information
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	// Start the servers
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-
-	r.Log("Assert the tunnel between me and them works")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-
-	r.Log("Go inactive and wait for the tunnels to get dropped")
-	waitStart := time.Now()
-	for {
-		myIndexes := len(myControl.GetHostmap().Indexes)
-		theirIndexes := len(theirControl.GetHostmap().Indexes)
-		if myIndexes == 0 && theirIndexes == 0 {
-			break
-		}
-
-		since := time.Since(waitStart)
-		r.Logf("my tunnels: %v; their tunnels: %v; duration: %v", myIndexes, theirIndexes, since)
-		if since > time.Second*30 {
-			t.Fatal("Tunnel should have been declared inactive after 5 seconds and before 30 seconds")
-		}
-
-		time.Sleep(1 * time.Second)
-		r.FlushAll()
-	}
-
-	r.Logf("Inactive tunnels were dropped within %v", time.Since(waitStart))
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestCertUpgrade(t *testing.T) {
-	// The goal of this test is to ensure the shortest inactivity timeout will close the tunnel on both sides
-	// under ideal conditions
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	caB, err := ca.MarshalPEM()
-	if err != nil {
-		panic(err)
-	}
-	ca2, _, caKey2, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-
-	ca2B, err := ca2.MarshalPEM()
-	if err != nil {
-		panic(err)
-	}
-	caStr := fmt.Sprintf("%s\n%s", caB, ca2B)
-
-	myCert, _, myPrivKey, _ := cert_test.NewTestCert(cert.Version1, cert.Curve_CURVE25519, ca, caKey, "me", time.Now(), time.Now().Add(5*time.Minute), []netip.Prefix{netip.MustParsePrefix("10.128.0.1/24")}, nil, []string{})
-	_, myCert2Pem := cert_test.NewTestCertDifferentVersion(myCert, cert.Version2, ca2, caKey2)
-
-	theirCert, _, theirPrivKey, _ := cert_test.NewTestCert(cert.Version1, cert.Curve_CURVE25519, ca, caKey, "them", time.Now(), time.Now().Add(5*time.Minute), []netip.Prefix{netip.MustParsePrefix("10.128.0.2/24")}, nil, []string{})
-	theirCert2, _ := cert_test.NewTestCertDifferentVersion(theirCert, cert.Version2, ca2, caKey2)
-
-	myControl, myVpnIpNet, myUdpAddr, myC := newServer([]cert.Certificate{ca, ca2}, []cert.Certificate{myCert}, myPrivKey, m{})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newServer([]cert.Certificate{ca, ca2}, []cert.Certificate{theirCert, theirCert2}, theirPrivKey, m{})
-
-	// Share our underlay information
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	// Start the servers
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	r.Log("Assert the tunnel between me and them works")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-	r.Log("yay")
-	//todo ???
-	time.Sleep(1 * time.Second)
-	r.FlushAll()
-
-	mc := m{
-		"pki": m{
-			"ca":   caStr,
-			"cert": string(myCert2Pem),
-			"key":  string(myPrivKey),
-		},
-		//"tun": m{"disabled": true},
-		"firewall": myC.Settings["firewall"],
-		//"handshakes": m{
-		//	"try_interval": "1s",
-		//},
-		"listen":  myC.Settings["listen"],
-		"logging": myC.Settings["logging"],
-		"timers":  myC.Settings["timers"],
-	}
-
-	cb, err := yaml.Marshal(mc)
-	if err != nil {
-		panic(err)
-	}
-
-	r.Logf("reload new v2-only config")
-	err = myC.ReloadConfigString(string(cb))
-	assert.NoError(t, err)
-	r.Log("yay, spin until their sees it")
-	waitStart := time.Now()
-	for {
-		assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-		c := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-		if c == nil {
-			r.Log("nil")
-		} else {
-			version := c.Cert.Version()
-			r.Logf("version %d", version)
-			if version == cert.Version2 {
-				break
-			}
-		}
-		since := time.Since(waitStart)
-		if since > time.Second*10 {
-			t.Fatal("Cert should be new by now")
-		}
-		time.Sleep(time.Second)
-	}
-
-	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestCertDowngrade(t *testing.T) {
-	// The goal of this test is to ensure the shortest inactivity timeout will close the tunnel on both sides
-	// under ideal conditions
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	caB, err := ca.MarshalPEM()
-	if err != nil {
-		panic(err)
-	}
-	ca2, _, caKey2, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-
-	ca2B, err := ca2.MarshalPEM()
-	if err != nil {
-		panic(err)
-	}
-	caStr := fmt.Sprintf("%s\n%s", caB, ca2B)
-
-	myCert, _, myPrivKey, myCertPem := cert_test.NewTestCert(cert.Version1, cert.Curve_CURVE25519, ca, caKey, "me", time.Now(), time.Now().Add(5*time.Minute), []netip.Prefix{netip.MustParsePrefix("10.128.0.1/24")}, nil, []string{})
-	myCert2, _ := cert_test.NewTestCertDifferentVersion(myCert, cert.Version2, ca2, caKey2)
-
-	theirCert, _, theirPrivKey, _ := cert_test.NewTestCert(cert.Version1, cert.Curve_CURVE25519, ca, caKey, "them", time.Now(), time.Now().Add(5*time.Minute), []netip.Prefix{netip.MustParsePrefix("10.128.0.2/24")}, nil, []string{})
-	theirCert2, _ := cert_test.NewTestCertDifferentVersion(theirCert, cert.Version2, ca2, caKey2)
-
-	myControl, myVpnIpNet, myUdpAddr, myC := newServer([]cert.Certificate{ca, ca2}, []cert.Certificate{myCert2}, myPrivKey, m{})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newServer([]cert.Certificate{ca, ca2}, []cert.Certificate{theirCert, theirCert2}, theirPrivKey, m{})
-
-	// Share our underlay information
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	// Start the servers
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	r.Log("Assert the tunnel between me and them works")
-	//assertTunnel(t, theirVpnIpNet[0].Addr(), myVpnIpNet[0].Addr(), theirControl, myControl, r)
-	//r.Log("yay")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-	r.Log("yay")
-	//todo ???
-	time.Sleep(1 * time.Second)
-	r.FlushAll()
-
-	mc := m{
-		"pki": m{
-			"ca":   caStr,
-			"cert": string(myCertPem),
-			"key":  string(myPrivKey),
-		},
-		"firewall": myC.Settings["firewall"],
-		"listen":   myC.Settings["listen"],
-		"logging":  myC.Settings["logging"],
-		"timers":   myC.Settings["timers"],
-	}
-
-	cb, err := yaml.Marshal(mc)
-	if err != nil {
-		panic(err)
-	}
-
-	r.Logf("reload new v1-only config")
-	err = myC.ReloadConfigString(string(cb))
-	assert.NoError(t, err)
-	r.Log("yay, spin until their sees it")
-	waitStart := time.Now()
-	for {
-		assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-		c := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-		c2 := myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false)
-		if c == nil || c2 == nil {
-			r.Log("nil")
-		} else {
-			version := c.Cert.Version()
-			theirVersion := c2.Cert.Version()
-			r.Logf("version %d,%d", version, theirVersion)
-			if version == cert.Version1 {
-				break
-			}
-		}
-		since := time.Since(waitStart)
-		if since > time.Second*5 {
-			r.Log("wtf")
-		}
-		if since > time.Second*10 {
-			r.Log("wtf")
-			t.Fatal("Cert should be new by now")
-		}
-		time.Sleep(time.Second)
-	}
-
-	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestCertMismatchCorrection(t *testing.T) {
-	// The goal of this test is to ensure the shortest inactivity timeout will close the tunnel on both sides
-	// under ideal conditions
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version1, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	ca2, _, caKey2, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-
-	myCert, _, myPrivKey, _ := cert_test.NewTestCert(cert.Version1, cert.Curve_CURVE25519, ca, caKey, "me", time.Now(), time.Now().Add(5*time.Minute), []netip.Prefix{netip.MustParsePrefix("10.128.0.1/24")}, nil, []string{})
-	myCert2, _ := cert_test.NewTestCertDifferentVersion(myCert, cert.Version2, ca2, caKey2)
-
-	theirCert, _, theirPrivKey, _ := cert_test.NewTestCert(cert.Version1, cert.Curve_CURVE25519, ca, caKey, "them", time.Now(), time.Now().Add(5*time.Minute), []netip.Prefix{netip.MustParsePrefix("10.128.0.2/24")}, nil, []string{})
-	theirCert2, _ := cert_test.NewTestCertDifferentVersion(theirCert, cert.Version2, ca2, caKey2)
-
-	myControl, myVpnIpNet, myUdpAddr, _ := newServer([]cert.Certificate{ca, ca2}, []cert.Certificate{myCert2}, myPrivKey, m{})
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newServer([]cert.Certificate{ca, ca2}, []cert.Certificate{theirCert, theirCert2}, theirPrivKey, m{})
-
-	// Share our underlay information
-	myControl.InjectLightHouseAddr(theirVpnIpNet[0].Addr(), theirUdpAddr)
-	theirControl.InjectLightHouseAddr(myVpnIpNet[0].Addr(), myUdpAddr)
-
-	// Start the servers
-	myControl.Start()
-	theirControl.Start()
-
-	r := router.NewR(t, myControl, theirControl)
-	defer r.RenderFlow()
-
-	r.Log("Assert the tunnel between me and them works")
-	//assertTunnel(t, theirVpnIpNet[0].Addr(), myVpnIpNet[0].Addr(), theirControl, myControl, r)
-	//r.Log("yay")
-	assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-	r.Log("yay")
-	//todo ???
-	time.Sleep(1 * time.Second)
-	r.FlushAll()
-
-	waitStart := time.Now()
-	for {
-		assertTunnel(t, myVpnIpNet[0].Addr(), theirVpnIpNet[0].Addr(), myControl, theirControl, r)
-		c := theirControl.GetHostInfoByVpnAddr(myVpnIpNet[0].Addr(), false)
-		c2 := myControl.GetHostInfoByVpnAddr(theirVpnIpNet[0].Addr(), false)
-		if c == nil || c2 == nil {
-			r.Log("nil")
-		} else {
-			version := c.Cert.Version()
-			theirVersion := c2.Cert.Version()
-			r.Logf("version %d,%d", version, theirVersion)
-			if version == theirVersion {
-				break
-			}
-		}
-		since := time.Since(waitStart)
-		if since > time.Second*5 {
-			r.Log("wtf")
-		}
-		if since > time.Second*10 {
-			r.Log("wtf")
-			t.Fatal("Cert should be new by now")
-		}
-		time.Sleep(time.Second)
-	}
-
-	r.RenderHostmaps("Final hostmaps", myControl, theirControl)
-
-	myControl.Stop()
-	theirControl.Stop()
-}
-
-func TestCrossStackRelaysWork(t *testing.T) {
-	ca, _, caKey, _ := cert_test.NewTestCaCert(cert.Version2, cert.Curve_CURVE25519, time.Now(), time.Now().Add(10*time.Minute), nil, nil, []string{})
-	myControl, myVpnIpNet, _, _ := newSimpleServer(cert.Version2, ca, caKey, "me     ", "10.128.0.1/24,fc00::1/64", m{"relay": m{"use_relays": true}})
-	relayControl, relayVpnIpNet, relayUdpAddr, _ := newSimpleServer(cert.Version2, ca, caKey, "relay  ", "10.128.0.128/24,fc00::128/64", m{"relay": m{"am_relay": true}})
-	theirUdp := netip.MustParseAddrPort("10.0.0.2:4242")
-	theirControl, theirVpnIpNet, theirUdpAddr, _ := newSimpleServerWithUdp(cert.Version2, ca, caKey, "them   ", "fc00::2/64", theirUdp, m{"relay": m{"use_relays": true}})
-
-	//myVpnV4 := myVpnIpNet[0]
-	myVpnV6 := myVpnIpNet[1]
-	relayVpnV4 := relayVpnIpNet[0]
-	relayVpnV6 := relayVpnIpNet[1]
-	theirVpnV6 := theirVpnIpNet[0]
-
-	// Teach my how to get to the relay and that their can be reached via the relay
-	myControl.InjectLightHouseAddr(relayVpnV4.Addr(), relayUdpAddr)
-	myControl.InjectLightHouseAddr(relayVpnV6.Addr(), relayUdpAddr)
-	myControl.InjectRelays(theirVpnV6.Addr(), []netip.Addr{relayVpnV6.Addr()})
-	relayControl.InjectLightHouseAddr(theirVpnV6.Addr(), theirUdpAddr)
-
-	// Build a router so we don't have to reason who gets which packet
-	r := router.NewR(t, myControl, relayControl, theirControl)
-	defer r.RenderFlow()
-
-	// Start the servers
-	myControl.Start()
-	relayControl.Start()
-	theirControl.Start()
-
-	t.Log("Trigger a handshake from me to them via the relay")
-	myControl.InjectTunUDPPacket(theirVpnV6.Addr(), 80, myVpnV6.Addr(), 80, []byte("Hi from me"))
-
-	p := r.RouteForAllUntilTxTun(theirControl)
-	r.Log("Assert the tunnel works")
-	assertUdpPacket(t, []byte("Hi from me"), p, myVpnV6.Addr(), theirVpnV6.Addr(), 80, 80)
-
-	t.Log("reply?")
-	theirControl.InjectTunUDPPacket(myVpnV6.Addr(), 80, theirVpnV6.Addr(), 80, []byte("Hi from them"))
-	p = r.RouteForAllUntilTxTun(myControl)
-	assertUdpPacket(t, []byte("Hi from them"), p, theirVpnV6.Addr(), myVpnV6.Addr(), 80, 80)
-
-	r.RenderHostmaps("Final hostmaps", myControl, relayControl, theirControl)
-	//t.Log("finish up")
-	//myControl.Stop()
-	//theirControl.Stop()
-	//relayControl.Stop()
-}

examples/config.yml

@@ -338,18 +338,6 @@ logging:
   # after receiving the response for lighthouse queries
   #trigger_buffer: 64
 
-# Tunnel manager settings
-#tunnels:
-  # drop_inactive controls whether inactive tunnels are maintained or dropped after the inactive_timeout period has
-  # elapsed.
-  # In general, it is a good idea to enable this setting. It will be enabled by default in a future release.
-  # This setting is reloadable
-  #drop_inactive: false
-
-  # inactivity_timeout controls how long a tunnel MUST NOT see any inbound or outbound traffic before being considered
-  # inactive and eligible to be dropped.
-  # This setting is reloadable
-  #inactivity_timeout: 10m
 
 # Nebula security group configuration
 firewall:

firewall.go

@@ -10,7 +10,6 @@ import (
 	"reflect"
 	"strconv"
 	"strings"
-	"sync"
 	"time"
 
 	"github.com/gaissmai/bart"
@@ -19,6 +18,7 @@ import (
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/firewall"
+	"github.com/wadey/synctrace"
 )
 
 type FirewallInterface interface {
@@ -76,7 +76,7 @@ type firewallMetrics struct {
 }
 
 type FirewallConntrack struct {
-	sync.Mutex
+	synctrace.Mutex
 
 	Conns      map[firewall.Packet]*conn
 	TimerWheel *TimerWheel[firewall.Packet]
@@ -164,6 +164,7 @@ func NewFirewall(l *logrus.Logger, tcpTimeout, UDPTimeout, defaultTimeout time.D
 
 	return &Firewall{
 		Conntrack: &FirewallConntrack{
+			Mutex:      synctrace.NewMutex("firewall-conntrack"),
 			Conns:      make(map[firewall.Packet]*conn),
 			TimerWheel: NewTimerWheel[firewall.Packet](tmin, tmax),
 		},

firewall_test.go

@@ -68,9 +68,6 @@ func TestFirewall_AddRule(t *testing.T) {
 	ti, err := netip.ParsePrefix("1.2.3.4/32")
 	require.NoError(t, err)
 
-	ti6, err := netip.ParsePrefix("fd12::34/128")
-	require.NoError(t, err)
-
 	require.NoError(t, fw.AddRule(true, firewall.ProtoTCP, 1, 1, []string{}, "", netip.Prefix{}, netip.Prefix{}, "", ""))
 	// An empty rule is any
 	assert.True(t, fw.InRules.TCP[1].Any.Any.Any)
@@ -95,24 +92,12 @@ func TestFirewall_AddRule(t *testing.T) {
 	_, ok := fw.OutRules.AnyProto[1].Any.CIDR.Get(ti)
 	assert.True(t, ok)
 
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	require.NoError(t, fw.AddRule(false, firewall.ProtoAny, 1, 1, []string{}, "", ti6, netip.Prefix{}, "", ""))
-	assert.Nil(t, fw.OutRules.AnyProto[1].Any.Any)
-	_, ok = fw.OutRules.AnyProto[1].Any.CIDR.Get(ti6)
-	assert.True(t, ok)
-
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	require.NoError(t, fw.AddRule(false, firewall.ProtoAny, 1, 1, []string{}, "", netip.Prefix{}, ti, "", ""))
 	assert.NotNil(t, fw.OutRules.AnyProto[1].Any.Any)
 	_, ok = fw.OutRules.AnyProto[1].Any.Any.LocalCIDR.Get(ti)
 	assert.True(t, ok)
 
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	require.NoError(t, fw.AddRule(false, firewall.ProtoAny, 1, 1, []string{}, "", netip.Prefix{}, ti6, "", ""))
-	assert.NotNil(t, fw.OutRules.AnyProto[1].Any.Any)
-	_, ok = fw.OutRules.AnyProto[1].Any.Any.LocalCIDR.Get(ti6)
-	assert.True(t, ok)
-
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	require.NoError(t, fw.AddRule(true, firewall.ProtoUDP, 1, 1, []string{"g1"}, "", netip.Prefix{}, netip.Prefix{}, "ca-name", ""))
 	assert.Contains(t, fw.InRules.UDP[1].CANames, "ca-name")
@@ -132,13 +117,6 @@ func TestFirewall_AddRule(t *testing.T) {
 	require.NoError(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{}, "", anyIp, netip.Prefix{}, "", ""))
 	assert.True(t, fw.OutRules.AnyProto[0].Any.Any.Any)
 
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
-	anyIp6, err := netip.ParsePrefix("::/0")
-	require.NoError(t, err)
-
-	require.NoError(t, fw.AddRule(false, firewall.ProtoAny, 0, 0, []string{}, "", anyIp6, netip.Prefix{}, "", ""))
-	assert.True(t, fw.OutRules.AnyProto[0].Any.Any.Any)
-
 	// Test error conditions
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c)
 	require.Error(t, fw.AddRule(true, math.MaxUint8, 0, 0, []string{}, "", netip.Prefix{}, netip.Prefix{}, "", ""))
@@ -221,82 +199,6 @@ func TestFirewall_Drop(t *testing.T) {
 	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 }
 
-func TestFirewall_DropV6(t *testing.T) {
-	l := test.NewLogger()
-	ob := &bytes.Buffer{}
-	l.SetOutput(ob)
-
-	p := firewall.Packet{
-		LocalAddr:  netip.MustParseAddr("fd12::34"),
-		RemoteAddr: netip.MustParseAddr("fd12::34"),
-		LocalPort:  10,
-		RemotePort: 90,
-		Protocol:   firewall.ProtoUDP,
-		Fragment:   false,
-	}
-
-	c := dummyCert{
-		name:     "host1",
-		networks: []netip.Prefix{netip.MustParsePrefix("fd12::34/120")},
-		groups:   []string{"default-group"},
-		issuer:   "signer-shasum",
-	}
-	h := HostInfo{
-		ConnectionState: &ConnectionState{
-			peerCert: &cert.CachedCertificate{
-				Certificate:    &c,
-				InvertedGroups: map[string]struct{}{"default-group": {}},
-			},
-		},
-		vpnAddrs: []netip.Addr{netip.MustParseAddr("fd12::34")},
-	}
-	h.buildNetworks(c.networks, c.unsafeNetworks)
-
-	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"any"}, "", netip.Prefix{}, netip.Prefix{}, "", ""))
-	cp := cert.NewCAPool()
-
-	// Drop outbound
-	assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp, nil))
-	// Allow inbound
-	resetConntrack(fw)
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
-	// Allow outbound because conntrack
-	require.NoError(t, fw.Drop(p, false, &h, cp, nil))
-
-	// test remote mismatch
-	oldRemote := p.RemoteAddr
-	p.RemoteAddr = netip.MustParseAddr("fd12::56")
-	assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrInvalidRemoteIP)
-	p.RemoteAddr = oldRemote
-
-	// ensure signer doesn't get in the way of group checks
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", netip.Prefix{}, netip.Prefix{}, "", "signer-shasum"))
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", netip.Prefix{}, netip.Prefix{}, "", "signer-shasum-bad"))
-	assert.Equal(t, fw.Drop(p, true, &h, cp, nil), ErrNoMatchingRule)
-
-	// test caSha doesn't drop on match
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", netip.Prefix{}, netip.Prefix{}, "", "signer-shasum-bad"))
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", netip.Prefix{}, netip.Prefix{}, "", "signer-shasum"))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
-
-	// ensure ca name doesn't get in the way of group checks
-	cp.CAs["signer-shasum"] = &cert.CachedCertificate{Certificate: &dummyCert{name: "ca-good"}}
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", netip.Prefix{}, netip.Prefix{}, "ca-good", ""))
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", netip.Prefix{}, netip.Prefix{}, "ca-good-bad", ""))
-	assert.Equal(t, fw.Drop(p, true, &h, cp, nil), ErrNoMatchingRule)
-
-	// test caName doesn't drop on match
-	cp.CAs["signer-shasum"] = &cert.CachedCertificate{Certificate: &dummyCert{name: "ca-good"}}
-	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, &c)
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"nope"}, "", netip.Prefix{}, netip.Prefix{}, "ca-good-bad", ""))
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", netip.Prefix{}, netip.Prefix{}, "ca-good", ""))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
-}
-
 func BenchmarkFirewallTable_match(b *testing.B) {
 	f := &Firewall{}
 	ft := FirewallTable{
@@ -306,10 +208,6 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 	pfix := netip.MustParsePrefix("172.1.1.1/32")
 	_ = ft.TCP.addRule(f, 10, 10, []string{"good-group"}, "good-host", pfix, netip.Prefix{}, "", "")
 	_ = ft.TCP.addRule(f, 100, 100, []string{"good-group"}, "good-host", netip.Prefix{}, pfix, "", "")
-
-	pfix6 := netip.MustParsePrefix("fd11::11/128")
-	_ = ft.TCP.addRule(f, 10, 10, []string{"good-group"}, "good-host", pfix6, netip.Prefix{}, "", "")
-	_ = ft.TCP.addRule(f, 100, 100, []string{"good-group"}, "good-host", netip.Prefix{}, pfix6, "", "")
 	cp := cert.NewCAPool()
 
 	b.Run("fail on proto", func(b *testing.B) {
@@ -341,15 +239,6 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			assert.False(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalAddr: ip.Addr()}, true, c, cp))
 		}
 	})
-	b.Run("pass proto, port, fail on local CIDRv6", func(b *testing.B) {
-		c := &cert.CachedCertificate{
-			Certificate: &dummyCert{},
-		}
-		ip := netip.MustParsePrefix("fd99::99/128")
-		for n := 0; n < b.N; n++ {
-			assert.False(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalAddr: ip.Addr()}, true, c, cp))
-		}
-	})
 
 	b.Run("pass proto, port, any local CIDR, fail all group, name, and cidr", func(b *testing.B) {
 		c := &cert.CachedCertificate{
@@ -363,18 +252,6 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			assert.False(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10}, true, c, cp))
 		}
 	})
-	b.Run("pass proto, port, any local CIDRv6, fail all group, name, and cidr", func(b *testing.B) {
-		c := &cert.CachedCertificate{
-			Certificate: &dummyCert{
-				name:     "nope",
-				networks: []netip.Prefix{netip.MustParsePrefix("fd99::99/128")},
-			},
-			InvertedGroups: map[string]struct{}{"nope": {}},
-		}
-		for n := 0; n < b.N; n++ {
-			assert.False(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 10}, true, c, cp))
-		}
-	})
 
 	b.Run("pass proto, port, specific local CIDR, fail all group, name, and cidr", func(b *testing.B) {
 		c := &cert.CachedCertificate{
@@ -388,18 +265,6 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			assert.False(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalAddr: pfix.Addr()}, true, c, cp))
 		}
 	})
-	b.Run("pass proto, port, specific local CIDRv6, fail all group, name, and cidr", func(b *testing.B) {
-		c := &cert.CachedCertificate{
-			Certificate: &dummyCert{
-				name:     "nope",
-				networks: []netip.Prefix{netip.MustParsePrefix("fd99:99/128")},
-			},
-			InvertedGroups: map[string]struct{}{"nope": {}},
-		}
-		for n := 0; n < b.N; n++ {
-			assert.False(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalAddr: pfix6.Addr()}, true, c, cp))
-		}
-	})
 
 	b.Run("pass on group on any local cidr", func(b *testing.B) {
 		c := &cert.CachedCertificate{
@@ -424,17 +289,6 @@ func BenchmarkFirewallTable_match(b *testing.B) {
 			assert.True(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalAddr: pfix.Addr()}, true, c, cp))
 		}
 	})
-	b.Run("pass on group on specific local cidr6", func(b *testing.B) {
-		c := &cert.CachedCertificate{
-			Certificate: &dummyCert{
-				name: "nope",
-			},
-			InvertedGroups: map[string]struct{}{"good-group": {}},
-		}
-		for n := 0; n < b.N; n++ {
-			assert.True(b, ft.match(firewall.Packet{Protocol: firewall.ProtoTCP, LocalPort: 100, LocalAddr: pfix6.Addr()}, true, c, cp))
-		}
-	})
 
 	b.Run("pass on name", func(b *testing.B) {
 		c := &cert.CachedCertificate{
@@ -593,42 +447,6 @@ func TestFirewall_Drop3(t *testing.T) {
 	require.NoError(t, fw.Drop(p, true, &h1, cp, nil))
 }
 
-func TestFirewall_Drop3V6(t *testing.T) {
-	l := test.NewLogger()
-	ob := &bytes.Buffer{}
-	l.SetOutput(ob)
-
-	p := firewall.Packet{
-		LocalAddr:  netip.MustParseAddr("fd12::34"),
-		RemoteAddr: netip.MustParseAddr("fd12::34"),
-		LocalPort:  1,
-		RemotePort: 1,
-		Protocol:   firewall.ProtoUDP,
-		Fragment:   false,
-	}
-
-	network := netip.MustParsePrefix("fd12::34/120")
-	c := cert.CachedCertificate{
-		Certificate: &dummyCert{
-			name:     "host-owner",
-			networks: []netip.Prefix{network},
-		},
-	}
-	h := HostInfo{
-		ConnectionState: &ConnectionState{
-			peerCert: &c,
-		},
-		vpnAddrs: []netip.Addr{network.Addr()},
-	}
-	h.buildNetworks(c.Certificate.Networks(), c.Certificate.UnsafeNetworks())
-
-	// Test a remote address match
-	fw := NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
-	cp := cert.NewCAPool()
-	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 1, 1, []string{}, "", netip.MustParsePrefix("fd12::34/120"), netip.Prefix{}, "", ""))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
-}
-
 func TestFirewall_DropConntrackReload(t *testing.T) {
 	l := test.NewLogger()
 	ob := &bytes.Buffer{}
@@ -909,21 +727,6 @@ func TestAddFirewallRulesFromConfig(t *testing.T) {
 	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
 	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: netip.Prefix{}, localIp: cidr}, mf.lastCall)
 
-	// Test adding rule with cidr ipv6
-	cidr6 := netip.MustParsePrefix("fd00::/8")
-	conf = config.NewC(l)
-	mf = &mockFirewall{}
-	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "cidr": cidr6.String()}}}
-	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: cidr6, localIp: netip.Prefix{}}, mf.lastCall)
-
-	// Test adding rule with local_cidr ipv6
-	conf = config.NewC(l)
-	mf = &mockFirewall{}
-	conf.Settings["firewall"] = map[string]any{"inbound": []any{map[string]any{"port": "1", "proto": "any", "local_cidr": cidr6.String()}}}
-	require.NoError(t, AddFirewallRulesFromConfig(l, true, conf, mf))
-	assert.Equal(t, addRuleCall{incoming: true, proto: firewall.ProtoAny, startPort: 1, endPort: 1, groups: nil, ip: netip.Prefix{}, localIp: cidr6}, mf.lastCall)
-
 	// Test adding rule with ca_sha
 	conf = config.NewC(l)
 	mf = &mockFirewall{}

go.mod

@@ -1,9 +1,11 @@
 module github.com/slackhq/nebula
 
-go 1.25
+go 1.23.0
+
+toolchain go1.24.1
 
 require (
-	dario.cat/mergo v1.0.2
+	dario.cat/mergo v1.0.1
 	github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be
 	github.com/armon/go-radix v1.0.0
 	github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432
@@ -21,7 +23,8 @@ require (
 	github.com/skip2/go-qrcode v0.0.0-20200617195104-da1b6568686e
 	github.com/stefanberger/go-pkcs11uri v0.0.0-20230803200340-78284954bff6
 	github.com/stretchr/testify v1.10.0
-	github.com/vishvananda/netlink v1.3.1
+	github.com/vishvananda/netlink v1.3.0
+	github.com/wadey/synctrace v0.0.0-20250612192159-94547ef50dfe
 	golang.org/x/crypto v0.37.0
 	golang.org/x/exp v0.0.0-20230725093048-515e97ebf090
 	golang.org/x/net v0.39.0
@@ -40,13 +43,17 @@ require (
 	github.com/beorn7/perks v1.0.1 // indirect
 	github.com/cespare/xxhash/v2 v2.3.0 // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
+	github.com/emirpasic/gods v1.18.1 // indirect
 	github.com/google/btree v1.1.2 // indirect
+	github.com/google/uuid v1.3.0 // indirect
+	github.com/heimdalr/dag v1.4.0 // indirect
 	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
 	github.com/prometheus/client_model v0.6.1 // indirect
 	github.com/prometheus/common v0.62.0 // indirect
 	github.com/prometheus/procfs v0.15.1 // indirect
-	github.com/vishvananda/netns v0.0.5 // indirect
+	github.com/timandy/routine v1.1.5 // indirect
+	github.com/vishvananda/netns v0.0.4 // indirect
 	golang.org/x/mod v0.23.0 // indirect
 	golang.org/x/time v0.5.0 // indirect
 	golang.org/x/tools v0.30.0 // indirect

go.sum

@@ -1,6 +1,6 @@
 cloud.google.com/go v0.34.0/go.mod h1:aQUYkXzVsufM+DwF1aE+0xfcU+56JwCaLick0ClmMTw=
-dario.cat/mergo v1.0.2 h1:85+piFYR1tMbRrLcDwR18y4UKJ3aH1Tbzi24VRW1TK8=
+dario.cat/mergo v1.0.1 h1:Ra4+bf83h2ztPIQYNP99R6m+Y7KfnARDfID+a+vLl4s=
-dario.cat/mergo v1.0.2/go.mod h1:E/hbnu0NxMFBjpMIE34DRGLWqDy0g5FuKDhCb31ngxA=
+dario.cat/mergo v1.0.1/go.mod h1:uNxQE+84aUszobStD9th8a29P2fMDhsBdgRYvZOxGmk=
 github.com/alecthomas/template v0.0.0-20160405071501-a0175ee3bccc/go.mod h1:LOuyumcjzFXgccqObfd/Ljyb9UuFJ6TxHnclSeseNhc=
 github.com/alecthomas/template v0.0.0-20190718012654-fb15b899a751/go.mod h1:LOuyumcjzFXgccqObfd/Ljyb9UuFJ6TxHnclSeseNhc=
 github.com/alecthomas/units v0.0.0-20151022065526-2efee857e7cf/go.mod h1:ybxpYRFXyAe+OPACYpWeL0wqObRcbAqCMya13uyzqw0=
@@ -22,6 +22,8 @@ github.com/cyberdelia/go-metrics-graphite v0.0.0-20161219230853-39f87cc3b432/go.
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/emirpasic/gods v1.18.1 h1:FXtiHYKDGKCW2KzwZKx0iC0PQmdlorYgdFG9jPXJ1Bc=
+github.com/emirpasic/gods v1.18.1/go.mod h1:8tpGGwCnJ5H4r6BWwaV6OrWmMoPhUl5jm/FMNAnJvWQ=
 github.com/flynn/noise v1.1.0 h1:KjPQoQCEFdZDiP03phOvGi11+SVVhBG2wOWAorLsstg=
 github.com/flynn/noise v1.1.0/go.mod h1:xbMo+0i6+IGbYdJhF31t2eR1BIU0CYc12+BNAKwUTag=
 github.com/gaissmai/bart v0.20.4 h1:Ik47r1fy3jRVU+1eYzKSW3ho2UgBVTVnUS8O993584U=
@@ -33,6 +35,8 @@ github.com/go-logfmt/logfmt v0.3.0/go.mod h1:Qt1PoO58o5twSAckw1HlFXLmHsOX5/0LbT9
 github.com/go-logfmt/logfmt v0.4.0/go.mod h1:3RMwSq7FuexP4Kalkev3ejPJsZTpXXBr9+V4qmtdjCk=
 github.com/go-logfmt/logfmt v0.5.0/go.mod h1:wCYkCAKZfumFQihp8CzCvQ3paCTfi41vtzG1KdI/P7A=
 github.com/go-stack/stack v1.8.0/go.mod h1:v0f6uXyyMGvRgIKkXu+yp6POWl0qKG85gN/melR3HDY=
+github.com/go-test/deep v1.1.0 h1:WOcxcdHcvdgThNXjw0t76K42FXTU7HpNQWHpA2HHNlg=
+github.com/go-test/deep v1.1.0/go.mod h1:5C2ZWiW0ErCdrYzpqxLbTX7MG14M9iiw8DgHncVwcsE=
 github.com/gogo/protobuf v1.1.1/go.mod h1:r8qH/GZQm5c6nD/R0oafs1akxWv10x8SbQlK7atdtwQ=
 github.com/gogo/protobuf v1.3.2 h1:Ov1cvc58UF3b5XjBnZv7+opcTcQFZebYjWzi34vdm4Q=
 github.com/gogo/protobuf v1.3.2/go.mod h1:P1XiOD3dCwIKUDQYPy72D8LYyHL2YPYrpS2s69NZV8Q=
@@ -58,6 +62,10 @@ github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/gopacket v1.1.19 h1:ves8RnFZPGiFnTS0uPQStjwru6uO6h+nlr9j6fL7kF8=
 github.com/google/gopacket v1.1.19/go.mod h1:iJ8V8n6KS+z2U1A8pUwu8bW5SyEMkXJB8Yo/Vo+TKTo=
+github.com/google/uuid v1.3.0 h1:t6JiXgmwXMjEs8VusXIJk2BXHsn+wx8BZdTaoZ5fu7I=
+github.com/google/uuid v1.3.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+github.com/heimdalr/dag v1.4.0 h1:zG3JA4RDVLc55k3AXAgfwa+EgBNZ0TkfOO3C29Ucpmg=
+github.com/heimdalr/dag v1.4.0/go.mod h1:OCh6ghKmU0hPjtwMqWBoNxPmtRioKd1xSu7Zs4sbIqM=
 github.com/jpillora/backoff v1.0.0/go.mod h1:J/6gKK9jxlEcS3zixgDgUAsiuZ7yrSoa/FX5e0EB2j4=
 github.com/json-iterator/go v1.1.6/go.mod h1:+SdeFBvtyEkXs7REEP0seUULqWtbJapLOCVDaaPEHmU=
 github.com/json-iterator/go v1.1.10/go.mod h1:KdQUCv79m/52Kvf8AW2vK1V8akMuk1QjK/uOdHXbAo4=
@@ -145,10 +153,14 @@ github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81P
 github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.10.0 h1:Xv5erBjTwe/5IxqUQTdXv5kgmIvbHo3QQyRwhJsOfJA=
 github.com/stretchr/testify v1.10.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
-github.com/vishvananda/netlink v1.3.1 h1:3AEMt62VKqz90r0tmNhog0r/PpWKmrEShJU0wJW6bV0=
+github.com/timandy/routine v1.1.5 h1:LSpm7Iijwb9imIPlucl4krpr2EeCeAUvifiQ9Uf5X+M=
-github.com/vishvananda/netlink v1.3.1/go.mod h1:ARtKouGSTGchR8aMwmkzC0qiNPrrWO5JS/XMVl45+b4=
+github.com/timandy/routine v1.1.5/go.mod h1:kXslgIosdY8LW0byTyPnenDgn4/azt2euufAq9rK51w=
-github.com/vishvananda/netns v0.0.5 h1:DfiHV+j8bA32MFM7bfEunvT8IAqQ/NzSJHtcmW5zdEY=
+github.com/vishvananda/netlink v1.3.0 h1:X7l42GfcV4S6E4vHTsw48qbrV+9PVojNfIhZcwQdrZk=
-github.com/vishvananda/netns v0.0.5/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
+github.com/vishvananda/netlink v1.3.0/go.mod h1:i6NetklAujEcC6fK0JPjT8qSwWyO0HLn4UKG+hGqeJs=
+github.com/vishvananda/netns v0.0.4 h1:Oeaw1EM2JMxD51g9uhtC0D7erkIjgmj8+JZc26m1YX8=
+github.com/vishvananda/netns v0.0.4/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
+github.com/wadey/synctrace v0.0.0-20250612192159-94547ef50dfe h1:dc8Q42VsX+ABr0drJw27f3smvGfcz7eB8rJx+IkVMAo=
+github.com/wadey/synctrace v0.0.0-20250612192159-94547ef50dfe/go.mod h1:F2VCml4UxGPgAAqqm9T0ZfnVRWITrQS1EMZM+KCAm/Q=
 github.com/yuin/goldmark v1.1.27/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 github.com/yuin/goldmark v1.2.1/go.mod h1:3hX8gzYuyVAZsxl0MRgGTJEmQBFcNTphYh9decYSb74=
 golang.org/x/crypto v0.0.0-20180904163835-0709b304e793/go.mod h1:6SG95UA2DQfeDnfUPMdvaQW0Q7yPrPDi9nlGo2tz2b4=

handshake_ix.go

@@ -9,6 +9,7 @@ import (
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/header"
+	"github.com/wadey/synctrace"
 )
 
 // NOISE IX Handshakes
@@ -23,17 +24,13 @@ func ixHandshakeStage0(f *Interface, hh *HandshakeHostInfo) bool {
 		return false
 	}
 
+	// If we're connecting to a v6 address we must use a v2 cert
 	cs := f.pki.getCertState()
 	v := cs.initiatingVersion
-	if hh.initiatingVersionOverride != cert.VersionPre1 {
+	for _, a := range hh.hostinfo.vpnAddrs {
-		v = hh.initiatingVersionOverride
+		if a.Is6() {
-	} else if v < cert.Version2 {
+			v = cert.Version2
-		// If we're connecting to a v6 address we must use a v2 cert
+			break
-		for _, a := range hh.hostinfo.vpnAddrs {
-			if a.Is6() {
-				v = cert.Version2
-				break
-			}
 		}
 	}
 
@@ -167,19 +164,16 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 
 	if remoteCert.Certificate.Version() != ci.myCert.Version() {
 		// We started off using the wrong certificate version, lets see if we can match the version that was sent to us
-		myCertOtherVersion := cs.getCertificate(remoteCert.Certificate.Version())
+		rc := cs.getCertificate(remoteCert.Certificate.Version())
-		if myCertOtherVersion == nil {
+		if rc == nil {
-			if f.l.Level >= logrus.DebugLevel {
+			f.l.WithError(err).WithField("udpAddr", addr).
-				f.l.WithError(err).WithFields(m{
+				WithField("handshake", m{"stage": 1, "style": "ix_psk0"}).WithField("cert", remoteCert).
-					"udpAddr":   addr,
+				Info("Unable to handshake with host due to missing certificate version")
-					"handshake": m{"stage": 1, "style": "ix_psk0"},
+			return
-					"cert":      remoteCert,
-				}).Debug("Might be unable to handshake with host due to missing certificate version")
-			}
-		} else {
-			// Record the certificate we are actually using
-			ci.myCert = myCertOtherVersion
 		}
+
+		// Record the certificate we are actually using
+		ci.myCert = rc
 	}
 
 	if len(remoteCert.Certificate.Networks()) == 0 {
@@ -256,7 +250,8 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 		HandshakePacket:   make(map[uint8][]byte, 0),
 		lastHandshakeTime: hs.Details.Time,
 		relayState: RelayState{
-			relays:         nil,
+			RWMutex:        synctrace.NewRWMutex("relay-state"),
+			relays:         map[netip.Addr]struct{}{},
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
@@ -464,9 +459,9 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 			Info("Handshake message sent")
 	}
 
-	f.connectionManager.AddTrafficWatch(hostinfo)
+	f.connectionManager.AddTrafficWatch(hostinfo.localIndexId)
 
-	hostinfo.remotes.RefreshFromHandshake(vpnAddrs)
+	hostinfo.remotes.ResetBlockedRemotes()
 
 	return
 }
@@ -659,7 +654,7 @@ func ixHandshakeStage2(f *Interface, addr netip.AddrPort, via *ViaSender, hh *Ha
 
 	// Complete our handshake and update metrics, this will replace any existing tunnels for the vpnAddrs here
 	f.handshakeManager.Complete(hostinfo, f)
-	f.connectionManager.AddTrafficWatch(hostinfo)
+	f.connectionManager.AddTrafficWatch(hostinfo.localIndexId)
 
 	if f.l.Level >= logrus.DebugLevel {
 		hostinfo.logger(f.l).Debugf("Sending %d stored packets", len(hh.packetStore))
@@ -674,7 +669,7 @@ func ixHandshakeStage2(f *Interface, addr netip.AddrPort, via *ViaSender, hh *Ha
 		f.cachedPacketMetrics.sent.Inc(int64(len(hh.packetStore)))
 	}
 
-	hostinfo.remotes.RefreshFromHandshake(vpnAddrs)
+	hostinfo.remotes.ResetBlockedRemotes()
 	f.metricHandshakes.Update(duration)
 
 	return false

handshake_manager.go

@@ -8,7 +8,6 @@ import (
 	"errors"
 	"net/netip"
 	"slices"
-	"sync"
 	"time"
 
 	"github.com/rcrowley/go-metrics"
@@ -16,6 +15,7 @@ import (
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/header"
 	"github.com/slackhq/nebula/udp"
+	"github.com/wadey/synctrace"
 )
 
 const (
@@ -45,7 +45,7 @@ type HandshakeConfig struct {
 
 type HandshakeManager struct {
 	// Mutex for interacting with the vpnIps and indexes maps
-	sync.RWMutex
+	synctrace.RWMutex
 
 	vpnIps  map[netip.Addr]*HandshakeHostInfo
 	indexes map[uint32]*HandshakeHostInfo
@@ -66,14 +66,13 @@ type HandshakeManager struct {
 }
 
 type HandshakeHostInfo struct {
-	sync.Mutex
+	synctrace.Mutex
 
-	startTime                 time.Time        // Time that we first started trying with this handshake
+	startTime   time.Time        // Time that we first started trying with this handshake
-	ready                     bool             // Is the handshake ready
+	ready       bool             // Is the handshake ready
-	initiatingVersionOverride cert.Version     // Should we use a non-default cert version for this handshake?
+	counter     int64            // How many attempts have we made so far
-	counter                   int64            // How many attempts have we made so far
+	lastRemotes []netip.AddrPort // Remotes that we sent to during the previous attempt
-	lastRemotes               []netip.AddrPort // Remotes that we sent to during the previous attempt
+	packetStore []*cachedPacket  // A set of packets to be transmitted once the handshake completes
-	packetStore               []*cachedPacket  // A set of packets to be transmitted once the handshake completes
 
 	hostinfo *HostInfo
 }
@@ -105,6 +104,7 @@ func (hh *HandshakeHostInfo) cachePacket(l *logrus.Logger, t header.MessageType,
 
 func NewHandshakeManager(l *logrus.Logger, mainHostMap *HostMap, lightHouse *LightHouse, outside udp.Conn, config HandshakeConfig) *HandshakeManager {
 	return &HandshakeManager{
+		RWMutex:                synctrace.NewRWMutex("handshake-manager"),
 		vpnIps:                 map[netip.Addr]*HandshakeHostInfo{},
 		indexes:                map[uint32]*HandshakeHostInfo{},
 		mainHostMap:            mainHostMap,
@@ -112,7 +112,7 @@ func NewHandshakeManager(l *logrus.Logger, mainHostMap *HostMap, lightHouse *Lig
 		outside:                outside,
 		config:                 config,
 		trigger:                make(chan netip.Addr, config.triggerBuffer),
-		OutboundHandshakeTimer: NewLockingTimerWheel[netip.Addr](config.tryInterval, hsTimeout(config.retries, config.tryInterval)),
+		OutboundHandshakeTimer: NewLockingTimerWheel[netip.Addr]("outbound-handshake-timer", config.tryInterval, hsTimeout(config.retries, config.tryInterval)),
 		messageMetrics:         config.messageMetrics,
 		metricInitiated:        metrics.GetOrRegisterCounter("handshake_manager.initiated", nil),
 		metricTimedOut:         metrics.GetOrRegisterCounter("handshake_manager.timed_out", nil),
@@ -300,8 +300,6 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 						InitiatorRelayIndex: idx,
 					}
 
-					relayFrom := hm.f.myVpnAddrs[0]
-
 					switch relayHostInfo.GetCert().Certificate.Version() {
 					case cert.Version1:
 						if !hm.f.myVpnAddrs[0].Is4() {
@@ -319,13 +317,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 						b = vpnIp.As4()
 						m.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
 					case cert.Version2:
-						if vpnIp.Is4() {
+						m.RelayFromAddr = netAddrToProtoAddr(hm.f.myVpnAddrs[0])
-							relayFrom = hm.f.myVpnAddrs[0]
-						} else {
-							//todo do this smarter
-							relayFrom = hm.f.myVpnAddrs[len(hm.f.myVpnAddrs)-1]
-						}
-						m.RelayFromAddr = netAddrToProtoAddr(relayFrom)
 						m.RelayToAddr = netAddrToProtoAddr(vpnIp)
 					default:
 						hostinfo.logger(hm.l).Error("Unknown certificate version found while creating relay")
@@ -340,7 +332,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 					} else {
 						hm.f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
 						hm.l.WithFields(logrus.Fields{
-							"relayFrom":           relayFrom,
+							"relayFrom":           hm.f.myVpnAddrs[0],
 							"relayTo":             vpnIp,
 							"initiatorRelayIndex": idx,
 							"relay":               relay}).
@@ -366,8 +358,6 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 					InitiatorRelayIndex: existingRelay.LocalIndex,
 				}
 
-				relayFrom := hm.f.myVpnAddrs[0]
-
 				switch relayHostInfo.GetCert().Certificate.Version() {
 				case cert.Version1:
 					if !hm.f.myVpnAddrs[0].Is4() {
@@ -385,14 +375,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 					b = vpnIp.As4()
 					m.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
 				case cert.Version2:
-					if vpnIp.Is4() {
+					m.RelayFromAddr = netAddrToProtoAddr(hm.f.myVpnAddrs[0])
-						relayFrom = hm.f.myVpnAddrs[0]
-					} else {
-						//todo do this smarter
-						relayFrom = hm.f.myVpnAddrs[len(hm.f.myVpnAddrs)-1]
-					}
-
-					m.RelayFromAddr = netAddrToProtoAddr(relayFrom)
 					m.RelayToAddr = netAddrToProtoAddr(vpnIp)
 				default:
 					hostinfo.logger(hm.l).Error("Unknown certificate version found while creating relay")
@@ -407,7 +390,7 @@ func (hm *HandshakeManager) handleOutbound(vpnIp netip.Addr, lighthouseTriggered
 					// This must send over the hostinfo, not over hm.Hosts[ip]
 					hm.f.SendMessageToHostInfo(header.Control, 0, relayHostInfo, msg, make([]byte, 12), make([]byte, mtu))
 					hm.l.WithFields(logrus.Fields{
-						"relayFrom":           relayFrom,
+						"relayFrom":           hm.f.myVpnAddrs[0],
 						"relayTo":             vpnIp,
 						"initiatorRelayIndex": existingRelay.LocalIndex,
 						"relay":               relay}).
@@ -468,13 +451,15 @@ func (hm *HandshakeManager) StartHandshake(vpnAddr netip.Addr, cacheCb func(*Han
 		vpnAddrs:        []netip.Addr{vpnAddr},
 		HandshakePacket: make(map[uint8][]byte, 0),
 		relayState: RelayState{
-			relays:         nil,
+			RWMutex:        synctrace.NewRWMutex("relay-state"),
+			relays:         map[netip.Addr]struct{}{},
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
 	}
 
 	hh := &HandshakeHostInfo{
+		Mutex:     synctrace.NewMutex("handshake-hostinfo"),
 		hostinfo:  hostinfo,
 		startTime: time.Now(),
 	}

hostmap.go

@@ -2,11 +2,8 @@ package nebula
 
 import (
 	"errors"
-	"fmt"
 	"net"
 	"net/netip"
-	"slices"
-	"sync"
 	"sync/atomic"
 	"time"
 
@@ -16,12 +13,15 @@ import (
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/header"
+	"github.com/wadey/synctrace"
 )
 
+// const ProbeLen = 100
 const defaultPromoteEvery = 1000       // Count of packets sent before we try moving a tunnel to a preferred underlay ip address
 const defaultReQueryEvery = 5000       // Count of packets sent before re-querying a hostinfo to the lighthouse
 const defaultReQueryWait = time.Minute // Minimum amount of seconds to wait before re-querying a hostinfo the lighthouse. Evaluated every ReQueryEvery
 const MaxRemotes = 10
+const maxRecvError = 4
 
 // MaxHostInfosPerVpnIp is the max number of hostinfos we will track for a given vpn ip
 // 5 allows for an initial handshake and each host pair re-handshaking twice
@@ -53,22 +53,22 @@ type Relay struct {
 }
 
 type HostMap struct {
-	sync.RWMutex    //Because we concurrently read and write to our maps
+	synctrace.RWMutex //Because we concurrently read and write to our maps
-	Indexes         map[uint32]*HostInfo
+	Indexes           map[uint32]*HostInfo
-	Relays          map[uint32]*HostInfo // Maps a Relay IDX to a Relay HostInfo object
+	Relays            map[uint32]*HostInfo // Maps a Relay IDX to a Relay HostInfo object
-	RemoteIndexes   map[uint32]*HostInfo
+	RemoteIndexes     map[uint32]*HostInfo
-	Hosts           map[netip.Addr]*HostInfo
+	Hosts             map[netip.Addr]*HostInfo
-	preferredRanges atomic.Pointer[[]netip.Prefix]
+	preferredRanges   atomic.Pointer[[]netip.Prefix]
-	l               *logrus.Logger
+	l                 *logrus.Logger
 }
 
 // For synchronization, treat the pointed-to Relay struct as immutable. To edit the Relay
 // struct, make a copy of an existing value, edit the fileds in the copy, and
 // then store a pointer to the new copy in both realyForBy* maps.
 type RelayState struct {
-	sync.RWMutex
+	synctrace.RWMutex
 
-	relays []netip.Addr // Ordered set of VpnAddrs of Hosts to use as relays to access this peer
+	relays map[netip.Addr]struct{} // Set of vpnAddr's of Hosts to use as relays to access this peer
 	// For data race avoidance, the contents of a *Relay are treated immutably. To update a *Relay, copy the existing data,
 	// modify what needs to be updated, and store the new modified copy in the relayForByIp and relayForByIdx maps (with
 	// the RelayState Lock held)
@@ -79,12 +79,7 @@ type RelayState struct {
 func (rs *RelayState) DeleteRelay(ip netip.Addr) {
 	rs.Lock()
 	defer rs.Unlock()
-	for idx, val := range rs.relays {
+	delete(rs.relays, ip)
-		if val == ip {
-			rs.relays = append(rs.relays[:idx], rs.relays[idx+1:]...)
-			return
-		}
-	}
 }
 
 func (rs *RelayState) UpdateRelayForByIpState(vpnIp netip.Addr, state int) {
@@ -129,16 +124,16 @@ func (rs *RelayState) GetRelayForByAddr(addr netip.Addr) (*Relay, bool) {
 func (rs *RelayState) InsertRelayTo(ip netip.Addr) {
 	rs.Lock()
 	defer rs.Unlock()
-	if !slices.Contains(rs.relays, ip) {
+	rs.relays[ip] = struct{}{}
-		rs.relays = append(rs.relays, ip)
-	}
 }
 
 func (rs *RelayState) CopyRelayIps() []netip.Addr {
-	ret := make([]netip.Addr, len(rs.relays))
 	rs.RLock()
 	defer rs.RUnlock()
-	copy(ret, rs.relays)
+	ret := make([]netip.Addr, 0, len(rs.relays))
+	for ip := range rs.relays {
+		ret = append(ret, ip)
+	}
 	return ret
 }
 
@@ -224,7 +219,8 @@ type HostInfo struct {
 	// vpnAddrs is a list of vpn addresses assigned to this host that are within our own vpn networks
 	// The host may have other vpn addresses that are outside our
 	// vpn networks but were removed because they are not usable
-	vpnAddrs []netip.Addr
+	vpnAddrs  []netip.Addr
+	recvError atomic.Uint32
 
 	// networks are both all vpn and unsafe networks assigned to this host
 	networks   *bart.Lite
@@ -254,14 +250,6 @@ type HostInfo struct {
 	// Used to track other hostinfos for this vpn ip since only 1 can be primary
 	// Synchronised via hostmap lock and not the hostinfo lock.
 	next, prev *HostInfo
-
-	//TODO: in, out, and others might benefit from being an atomic.Int32. We could collapse connectionManager pendingDeletion, relayUsed, and in/out into this 1 thing
-	in, out, pendingDeletion atomic.Bool
-
-	// lastUsed tracks the last time ConnectionManager checked the tunnel and it was in use.
-	// This value will be behind against actual tunnel utilization in the hot path.
-	// This should only be used by the ConnectionManagers ticker routine.
-	lastUsed time.Time
 }
 
 type ViaSender struct {
@@ -300,6 +288,7 @@ func NewHostMapFromConfig(l *logrus.Logger, c *config.C) *HostMap {
 
 func newHostMap(l *logrus.Logger) *HostMap {
 	return &HostMap{
+		RWMutex:       synctrace.NewRWMutex("hostmap"),
 		Indexes:       map[uint32]*HostInfo{},
 		Relays:        map[uint32]*HostInfo{},
 		RemoteIndexes: map[uint32]*HostInfo{},
@@ -522,7 +511,6 @@ func (hm *HostMap) QueryVpnAddrsRelayFor(targetIps []netip.Addr, relayHostIp net
 		return nil, nil, errors.New("unable to find host")
 	}
 
-	lastH := h
 	for h != nil {
 		for _, targetIp := range targetIps {
 			r, ok := h.relayState.QueryRelayForByIp(targetIp)
@@ -530,12 +518,10 @@ func (hm *HostMap) QueryVpnAddrsRelayFor(targetIps []netip.Addr, relayHostIp net
 				return h, r, nil
 			}
 		}
-		lastH = h
 		h = h.next
 	}
 
-	//todo no merge
+	return nil, nil, errors.New("unable to find host with relay")
-	return nil, nil, fmt.Errorf("unable to find host with relay: %v", lastH)
 }
 
 func (hm *HostMap) unlockedDisestablishVpnAddrRelayFor(hi *HostInfo) {
@@ -734,6 +720,13 @@ func (i *HostInfo) SetRemoteIfPreferred(hm *HostMap, newRemote netip.AddrPort) b
 	return false
 }
 
+func (i *HostInfo) RecvErrorExceeded() bool {
+	if i.recvError.Add(1) >= maxRecvError {
+		return true
+	}
+	return true
+}
+
 func (i *HostInfo) buildNetworks(networks, unsafeNetworks []netip.Prefix) {
 	if len(networks) == 1 && len(unsafeNetworks) == 0 {
 		// Simple case, no CIDRTree needed

hostmap_test.go

@@ -7,7 +7,6 @@ import (
 	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/test"
 	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
 )
 
 func TestHostMap_MakePrimary(t *testing.T) {
@@ -216,31 +215,3 @@ func TestHostMap_reload(t *testing.T) {
 	c.ReloadConfigString("preferred_ranges: [1.1.1.1/32]")
 	assert.Equal(t, []string{"1.1.1.1/32"}, toS(hm.GetPreferredRanges()))
 }
-
-func TestHostMap_RelayState(t *testing.T) {
-	h1 := &HostInfo{vpnAddrs: []netip.Addr{netip.MustParseAddr("0.0.0.1")}, localIndexId: 1}
-	a1 := netip.MustParseAddr("::1")
-	a2 := netip.MustParseAddr("2001::1")
-
-	h1.relayState.InsertRelayTo(a1)
-	assert.Equal(t, []netip.Addr{a1}, h1.relayState.relays)
-	h1.relayState.InsertRelayTo(a2)
-	assert.Equal(t, []netip.Addr{a1, a2}, h1.relayState.relays)
-	// Ensure that the first relay added is the first one returned in the copy
-	currentRelays := h1.relayState.CopyRelayIps()
-	require.Len(t, currentRelays, 2)
-	assert.Equal(t, a1, currentRelays[0])
-
-	// Deleting the last one in the list works ok
-	h1.relayState.DeleteRelay(a2)
-	assert.Equal(t, []netip.Addr{a1}, h1.relayState.relays)
-
-	// Deleting an element not in the list works ok
-	h1.relayState.DeleteRelay(a2)
-	assert.Equal(t, []netip.Addr{a1}, h1.relayState.relays)
-
-	// Deleting the only element in the list works ok
-	h1.relayState.DeleteRelay(a1)
-	assert.Equal(t, []netip.Addr{}, h1.relayState.relays)
-
-}

inside.go

@@ -288,7 +288,7 @@ func (f *Interface) SendVia(via *HostInfo,
 	c := via.ConnectionState.messageCounter.Add(1)
 
 	out = header.Encode(out, header.Version, header.Message, header.MessageRelay, relay.RemoteIndex, c)
-	f.connectionManager.Out(via)
+	f.connectionManager.Out(via.localIndexId)
 
 	// Authenticate the header and payload, but do not encrypt for this message type.
 	// The payload consists of the inner, unencrypted Nebula header, as well as the end-to-end encrypted payload.
@@ -356,7 +356,7 @@ func (f *Interface) sendNoMetrics(t header.MessageType, st header.MessageSubType
 
 	//l.WithField("trace", string(debug.Stack())).Error("out Header ", &Header{Version, t, st, 0, hostinfo.remoteIndexId, c}, p)
 	out = header.Encode(out, header.Version, t, st, hostinfo.remoteIndexId, c)
-	f.connectionManager.Out(hostinfo)
+	f.connectionManager.Out(hostinfo.localIndexId)
 
 	// Query our LH if we haven't since the last time we've been rebound, this will cause the remote to punch against
 	// all our addrs and enable a faster roaming.

interface.go

@@ -24,23 +24,23 @@ import (
 const mtu = 9001
 
 type InterfaceConfig struct {
-	HostMap            *HostMap
+	HostMap                 *HostMap
-	Outside            udp.Conn
+	Outside                 udp.Conn
-	Inside             overlay.Device
+	Inside                  overlay.Device
-	pki                *PKI
+	pki                     *PKI
-	Cipher             string
+	Firewall                *Firewall
-	Firewall           *Firewall
+	ServeDns                bool
-	ServeDns           bool
+	HandshakeManager        *HandshakeManager
-	HandshakeManager   *HandshakeManager
+	lightHouse              *LightHouse
-	lightHouse         *LightHouse
+	checkInterval           time.Duration
-	connectionManager  *connectionManager
+	pendingDeletionInterval time.Duration
-	DropLocalBroadcast bool
+	DropLocalBroadcast      bool
-	DropMulticast      bool
+	DropMulticast           bool
-	routines           int
+	routines                int
-	MessageMetrics     *MessageMetrics
+	MessageMetrics          *MessageMetrics
-	version            string
+	version                 string
-	relayManager       *relayManager
+	relayManager            *relayManager
-	punchy             *Punchy
+	punchy                  *Punchy
 
 	tryPromoteEvery uint32
 	reQueryEvery    uint32
@@ -157,9 +157,6 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 	if c.Firewall == nil {
 		return nil, errors.New("no firewall rules")
 	}
-	if c.connectionManager == nil {
-		return nil, errors.New("no connection manager")
-	}
 
 	cs := c.pki.getCertState()
 	ifce := &Interface{
@@ -184,7 +181,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 		myVpnAddrsTable:       cs.myVpnAddrsTable,
 		myBroadcastAddrsTable: cs.myVpnBroadcastAddrsTable,
 		relayManager:          c.relayManager,
-		connectionManager:     c.connectionManager,
+
 		conntrackCacheTimeout: c.ConntrackCacheTimeout,
 
 		metricHandshakes: metrics.GetOrRegisterHistogram("handshakes", nil, metrics.NewExpDecaySample(1028, 0.015)),
@@ -201,7 +198,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 	ifce.reQueryEvery.Store(c.reQueryEvery)
 	ifce.reQueryWait.Store(int64(c.reQueryWait))
 
-	ifce.connectionManager.intf = ifce
+	ifce.connectionManager = newConnectionManager(ctx, c.l, ifce, c.checkInterval, c.pendingDeletionInterval, c.punchy)
 
 	return ifce, nil
 }

lighthouse.go

@@ -9,7 +9,6 @@ import (
 	"net/netip"
 	"slices"
 	"strconv"
-	"sync"
 	"sync/atomic"
 	"time"
 
@@ -21,16 +20,16 @@ import (
 	"github.com/slackhq/nebula/header"
 	"github.com/slackhq/nebula/udp"
 	"github.com/slackhq/nebula/util"
+	"github.com/wadey/synctrace"
 )
 
 var ErrHostNotKnown = errors.New("host not known")
-var ErrBadDetailsVpnAddr = errors.New("invalid packet, malformed detailsVpnAddr")
 
 type LightHouse struct {
 	//TODO: We need a timer wheel to kick out vpnAddrs that haven't reported in a long time
-	sync.RWMutex //Because we concurrently read and write to our maps
+	synctrace.RWMutex //Because we concurrently read and write to our maps
-	ctx          context.Context
+	ctx               context.Context
-	amLighthouse bool
+	amLighthouse      bool
 
 	myVpnNetworks      []netip.Prefix
 	myVpnNetworksTable *bart.Lite
@@ -57,7 +56,7 @@ type LightHouse struct {
 	// staticList exists to avoid having a bool in each addrMap entry
 	// since static should be rare
 	staticList  atomic.Pointer[map[netip.Addr]struct{}]
-	lighthouses atomic.Pointer[[]netip.Addr]
+	lighthouses atomic.Pointer[map[netip.Addr]struct{}]
 
 	interval     atomic.Int64
 	updateCancel context.CancelFunc
@@ -97,6 +96,7 @@ func NewLightHouseFromConfig(ctx context.Context, l *logrus.Logger, c *config.C,
 	}
 
 	h := LightHouse{
+		RWMutex:            synctrace.NewRWMutex("lighthouse"),
 		ctx:                ctx,
 		amLighthouse:       amLighthouse,
 		myVpnNetworks:      cs.myVpnNetworks,
@@ -108,7 +108,7 @@ func NewLightHouseFromConfig(ctx context.Context, l *logrus.Logger, c *config.C,
 		queryChan:          make(chan netip.Addr, c.GetUint32("handshakes.query_buffer", 64)),
 		l:                  l,
 	}
-	lighthouses := make([]netip.Addr, 0)
+	lighthouses := make(map[netip.Addr]struct{})
 	h.lighthouses.Store(&lighthouses)
 	staticList := make(map[netip.Addr]struct{})
 	h.staticList.Store(&staticList)
@@ -144,7 +144,7 @@ func (lh *LightHouse) GetStaticHostList() map[netip.Addr]struct{} {
 	return *lh.staticList.Load()
 }
 
-func (lh *LightHouse) GetLighthouses() []netip.Addr {
+func (lh *LightHouse) GetLighthouses() map[netip.Addr]struct{} {
 	return *lh.lighthouses.Load()
 }
 
@@ -307,12 +307,13 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 	}
 
 	if initial || c.HasChanged("lighthouse.hosts") {
-		lhList, err := lh.parseLighthouses(c)
+		lhMap := make(map[netip.Addr]struct{})
+		err := lh.parseLighthouses(c, lhMap)
 		if err != nil {
 			return err
 		}
 
-		lh.lighthouses.Store(&lhList)
+		lh.lighthouses.Store(&lhMap)
 		if !initial {
 			//NOTE: we are not tearing down existing lighthouse connections because they might be used for non lighthouse traffic
 			lh.l.Info("lighthouse.hosts has changed")
@@ -346,37 +347,36 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 	return nil
 }
 
-func (lh *LightHouse) parseLighthouses(c *config.C) ([]netip.Addr, error) {
+func (lh *LightHouse) parseLighthouses(c *config.C, lhMap map[netip.Addr]struct{}) error {
 	lhs := c.GetStringSlice("lighthouse.hosts", []string{})
 	if lh.amLighthouse && len(lhs) != 0 {
 		lh.l.Warn("lighthouse.am_lighthouse enabled on node but upstream lighthouses exist in config")
 	}
-	out := make([]netip.Addr, len(lhs))
 
 	for i, host := range lhs {
 		addr, err := netip.ParseAddr(host)
 		if err != nil {
-			return nil, util.NewContextualError("Unable to parse lighthouse host entry", m{"host": host, "entry": i + 1}, err)
+			return util.NewContextualError("Unable to parse lighthouse host entry", m{"host": host, "entry": i + 1}, err)
 		}
 
 		if !lh.myVpnNetworksTable.Contains(addr) {
-			return nil, util.NewContextualError("lighthouse host is not in our networks, invalid", m{"vpnAddr": addr, "networks": lh.myVpnNetworks}, nil)
+			return util.NewContextualError("lighthouse host is not in our networks, invalid", m{"vpnAddr": addr, "networks": lh.myVpnNetworks}, nil)
 		}
-		out[i] = addr
+		lhMap[addr] = struct{}{}
 	}
 
-	if !lh.amLighthouse && len(out) == 0 {
+	if !lh.amLighthouse && len(lhMap) == 0 {
 		lh.l.Warn("No lighthouse.hosts configured, this host will only be able to initiate tunnels with static_host_map entries")
 	}
 
 	staticList := lh.GetStaticHostList()
-	for i := range out {
+	for lhAddr, _ := range lhMap {
-		if _, ok := staticList[out[i]]; !ok {
+		if _, ok := staticList[lhAddr]; !ok {
-			return nil, fmt.Errorf("lighthouse %s does not have a static_host_map entry", out[i])
+			return fmt.Errorf("lighthouse %s does not have a static_host_map entry", lhAddr)
 		}
 	}
 
-	return out, nil
+	return nil
 }
 
 func getStaticMapCadence(c *config.C) (time.Duration, error) {
@@ -473,6 +473,7 @@ func (lh *LightHouse) QueryServer(vpnAddr netip.Addr) {
 		return
 	}
 
+	synctrace.ChanDebugSend("lighthouse-querychan")
 	lh.queryChan <- vpnAddr
 }
 
@@ -487,7 +488,7 @@ func (lh *LightHouse) QueryCache(vpnAddrs []netip.Addr) *RemoteList {
 	lh.Lock()
 	defer lh.Unlock()
 	// Add an entry if we don't already have one
-	return lh.unlockedGetRemoteList(vpnAddrs) //todo CERT-V2 this contains addrmap lookups we could potentially skip
+	return lh.unlockedGetRemoteList(vpnAddrs)
 }
 
 // queryAndPrepMessage is a lock helper on RemoteList, assisting the caller to build a lighthouse message containing
@@ -520,15 +521,11 @@ func (lh *LightHouse) queryAndPrepMessage(vpnAddr netip.Addr, f func(*cache) (in
 }
 
 func (lh *LightHouse) DeleteVpnAddrs(allVpnAddrs []netip.Addr) {
-	// First we check the static host map. If any of the VpnAddrs to be deleted are present, do nothing.
+	// First we check the static mapping
-	staticList := lh.GetStaticHostList()
+	// and do nothing if it is there
-	for _, addr := range allVpnAddrs {
+	if _, ok := lh.GetStaticHostList()[allVpnAddrs[0]]; ok {
-		if _, ok := staticList[addr]; ok {
+		return
-			return
-		}
 	}
-
-	// None of the VpnAddrs were present. Now we can do the deletes.
 	lh.Lock()
 	rm, ok := lh.addrMap[allVpnAddrs[0]]
 	if ok {
@@ -570,7 +567,7 @@ func (lh *LightHouse) addStaticRemotes(i int, d time.Duration, network string, t
 	am.unlockedSetHostnamesResults(hr)
 
 	for _, addrPort := range hr.GetAddrs() {
-		if !lh.shouldAdd([]netip.Addr{vpnAddr}, addrPort.Addr()) {
+		if !lh.shouldAdd(vpnAddr, addrPort.Addr()) {
 			continue
 		}
 		switch {
@@ -632,30 +629,23 @@ func (lh *LightHouse) addCalculatedRemotes(vpnAddr netip.Addr) bool {
 	return len(calculatedV4) > 0 || len(calculatedV6) > 0
 }
 
-// unlockedGetRemoteList assumes you have the lh lock
+// unlockedGetRemoteList
+// assumes you have the lh lock
 func (lh *LightHouse) unlockedGetRemoteList(allAddrs []netip.Addr) *RemoteList {
-	// before we go and make a new remotelist, we need to make sure we don't have one for any of this set of vpnaddrs yet
+	am, ok := lh.addrMap[allAddrs[0]]
-	for i, addr := range allAddrs {
+	if !ok {
-		am, ok := lh.addrMap[addr]
+		am = NewRemoteList(allAddrs, func(a netip.Addr) bool { return lh.shouldAdd(allAddrs[0], a) })
-		if ok {
+		for _, addr := range allAddrs {
-			if i != 0 {
+			lh.addrMap[addr] = am
-				lh.addrMap[allAddrs[0]] = am
-			}
-			return am
 		}
 	}
-
-	am := NewRemoteList(allAddrs, lh.shouldAdd)
-	for _, addr := range allAddrs {
-		lh.addrMap[addr] = am
-	}
 	return am
 }
 
-func (lh *LightHouse) shouldAdd(vpnAddrs []netip.Addr, to netip.Addr) bool {
+func (lh *LightHouse) shouldAdd(vpnAddr netip.Addr, to netip.Addr) bool {
-	allow := lh.GetRemoteAllowList().AllowAll(vpnAddrs, to)
+	allow := lh.GetRemoteAllowList().Allow(vpnAddr, to)
 	if lh.l.Level >= logrus.TraceLevel {
-		lh.l.WithField("vpnAddrs", vpnAddrs).WithField("udpAddr", to).WithField("allow", allow).
+		lh.l.WithField("vpnAddr", vpnAddr).WithField("udpAddr", to).WithField("allow", allow).
 			Trace("remoteAllowList.Allow")
 	}
 	if !allow {
@@ -710,22 +700,19 @@ func (lh *LightHouse) unlockedShouldAddV6(vpnAddr netip.Addr, to *V6AddrPort) bo
 }
 
 func (lh *LightHouse) IsLighthouseAddr(vpnAddr netip.Addr) bool {
-	l := lh.GetLighthouses()
+	if _, ok := lh.GetLighthouses()[vpnAddr]; ok {
-	for i := range l {
+		return true
-		if l[i] == vpnAddr {
-			return true
-		}
 	}
 	return false
 }
 
-func (lh *LightHouse) IsAnyLighthouseAddr(vpnAddrs []netip.Addr) bool {
+// TODO: CERT-V2 IsLighthouseAddr should be sufficient, we just need to update the vpnAddrs for lighthouses after a handshake
+// so that we know all the lighthouse vpnAddrs, not just the ones we were configured to talk to initially
+func (lh *LightHouse) IsAnyLighthouseAddr(vpnAddr []netip.Addr) bool {
 	l := lh.GetLighthouses()
-	for i := range vpnAddrs {
+	for _, a := range vpnAddr {
-		for j := range l {
+		if _, ok := l[a]; ok {
-			if l[j] == vpnAddrs[i] {
+			return true
-				return true
-			}
 		}
 	}
 	return false
@@ -740,9 +727,11 @@ func (lh *LightHouse) startQueryWorker() {
 		nb := make([]byte, 12, 12)
 		out := make([]byte, mtu)
 
+		synctrace.ChanDebugRecvLock("lighthouse-querychan")
 		for {
 			select {
 			case <-lh.ctx.Done():
+				synctrace.ChanDebugRecvUnlock("lighthouse-querychan")
 				return
 			case addr := <-lh.queryChan:
 				lh.innerQueryServer(addr, nb, out)
@@ -767,7 +756,7 @@ func (lh *LightHouse) innerQueryServer(addr netip.Addr, nb, out []byte) {
 	queried := 0
 	lighthouses := lh.GetLighthouses()
 
-	for _, lhVpnAddr := range lighthouses {
+	for lhVpnAddr := range lighthouses {
 		hi := lh.ifce.GetHostInfo(lhVpnAddr)
 		if hi != nil {
 			v = hi.ConnectionState.myCert.Version()
@@ -885,7 +874,7 @@ func (lh *LightHouse) SendUpdate() {
 	updated := 0
 	lighthouses := lh.GetLighthouses()
 
-	for _, lhVpnAddr := range lighthouses {
+	for lhVpnAddr := range lighthouses {
 		var v cert.Version
 		hi := lh.ifce.GetHostInfo(lhVpnAddr)
 		if hi != nil {
@@ -943,6 +932,7 @@ func (lh *LightHouse) SendUpdate() {
 						V4AddrPorts:   v4,
 						V6AddrPorts:   v6,
 						RelayVpnAddrs: relays,
+						VpnAddr:       netAddrToProtoAddr(lh.myVpnNetworks[0].Addr()),
 					},
 				}
 
@@ -1062,19 +1052,19 @@ func (lhh *LightHouseHandler) handleHostQuery(n *NebulaMeta, fromVpnAddrs []neti
 		return
 	}
 
-	queryVpnAddr, useVersion, err := n.Details.GetVpnAddrAndVersion()
+	useVersion := cert.Version1
-	if err != nil {
+	var queryVpnAddr netip.Addr
+	if n.Details.OldVpnAddr != 0 {
+		b := [4]byte{}
+		binary.BigEndian.PutUint32(b[:], n.Details.OldVpnAddr)
+		queryVpnAddr = netip.AddrFrom4(b)
+		useVersion = 1
+	} else if n.Details.VpnAddr != nil {
+		queryVpnAddr = protoAddrToNetAddr(n.Details.VpnAddr)
+		useVersion = 2
+	} else {
 		if lhh.l.Level >= logrus.DebugLevel {
-			lhh.l.WithField("from", fromVpnAddrs).WithField("details", n.Details).
+			lhh.l.WithField("from", fromVpnAddrs).WithField("details", n.Details).Debugln("Dropping malformed HostQuery")
-				Debugln("Dropping malformed HostQuery")
-		}
-		return
-	}
-	if useVersion == cert.Version1 && queryVpnAddr.Is6() {
-		// this case really shouldn't be possible to represent, but reject it anyway.
-		if lhh.l.Level >= logrus.DebugLevel {
-			lhh.l.WithField("vpnAddrs", fromVpnAddrs).WithField("queryVpnAddr", queryVpnAddr).
-				Debugln("invalid vpn addr for v1 handleHostQuery")
 		}
 		return
 	}
@@ -1083,6 +1073,9 @@ func (lhh *LightHouseHandler) handleHostQuery(n *NebulaMeta, fromVpnAddrs []neti
 		n = lhh.resetMeta()
 		n.Type = NebulaMeta_HostQueryReply
 		if useVersion == cert.Version1 {
+			if !queryVpnAddr.Is4() {
+				return 0, fmt.Errorf("invalid vpn addr for v1 handleHostQuery")
+			}
 			b := queryVpnAddr.As4()
 			n.Details.OldVpnAddr = binary.BigEndian.Uint32(b[:])
 		} else {
@@ -1127,9 +1120,8 @@ func (lhh *LightHouseHandler) sendHostPunchNotification(n *NebulaMeta, fromVpnAd
 			if ok {
 				whereToPunch = newDest
 			} else {
-				if lhh.l.Level >= logrus.DebugLevel {
+				//TODO: CERT-V2 this means the destination will have no addresses in common with the punch-ee
-					lhh.l.WithField("to", crt.Networks()).Debugln("unable to punch to host, no addresses in common")
+				//choosing to do nothing for now, but maybe we return an error?
-				}
 			}
 		}
 
@@ -1188,17 +1180,19 @@ func (lhh *LightHouseHandler) coalesceAnswers(v cert.Version, c *cache, n *Nebul
 				if !r.Is4() {
 					continue
 				}
+
 				b = r.As4()
 				n.Details.OldRelayVpnAddrs = append(n.Details.OldRelayVpnAddrs, binary.BigEndian.Uint32(b[:]))
 			}
+
 		} else if v == cert.Version2 {
 			for _, r := range c.relay.relay {
 				n.Details.RelayVpnAddrs = append(n.Details.RelayVpnAddrs, netAddrToProtoAddr(r))
 			}
+
 		} else {
-			if lhh.l.Level >= logrus.DebugLevel {
+			//TODO: CERT-V2 don't panic
-				lhh.l.WithField("version", v).Debug("unsupported protocol version")
+			panic("unsupported version")
-			}
 		}
 	}
 }
@@ -1208,16 +1202,18 @@ func (lhh *LightHouseHandler) handleHostQueryReply(n *NebulaMeta, fromVpnAddrs [
 		return
 	}
 
-	certVpnAddr, _, err := n.Details.GetVpnAddrAndVersion()
+	lhh.lh.Lock()
-	if err != nil {
+
-		if lhh.l.Level >= logrus.DebugLevel {
+	var certVpnAddr netip.Addr
-			lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).Error("dropping malformed HostQueryReply")
+	if n.Details.OldVpnAddr != 0 {
-		}
+		b := [4]byte{}
-		return
+		binary.BigEndian.PutUint32(b[:], n.Details.OldVpnAddr)
+		certVpnAddr = netip.AddrFrom4(b)
+	} else if n.Details.VpnAddr != nil {
+		certVpnAddr = protoAddrToNetAddr(n.Details.VpnAddr)
 	}
 	relays := n.Details.GetRelays()
 
-	lhh.lh.Lock()
 	am := lhh.lh.unlockedGetRemoteList([]netip.Addr{certVpnAddr})
 	am.Lock()
 	lhh.lh.Unlock()
@@ -1242,24 +1238,27 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 		return
 	}
 
-	// not using GetVpnAddrAndVersion because we don't want to error on a blank detailsVpnAddr
 	var detailsVpnAddr netip.Addr
-	var useVersion cert.Version
+	useVersion := cert.Version1
-	if n.Details.OldVpnAddr != 0 { //v1 always sets this field
+	if n.Details.OldVpnAddr != 0 {
 		b := [4]byte{}
 		binary.BigEndian.PutUint32(b[:], n.Details.OldVpnAddr)
 		detailsVpnAddr = netip.AddrFrom4(b)
 		useVersion = cert.Version1
-	} else if n.Details.VpnAddr != nil { //this field is "optional" in v2, but if it's set, we should enforce it
+	} else if n.Details.VpnAddr != nil {
 		detailsVpnAddr = protoAddrToNetAddr(n.Details.VpnAddr)
 		useVersion = cert.Version2
 	} else {
-		detailsVpnAddr = netip.Addr{}
+		if lhh.l.Level >= logrus.DebugLevel {
-		useVersion = cert.Version2
+			lhh.l.WithField("details", n.Details).Debugf("dropping invalid HostUpdateNotification")
+		}
+		return
 	}
 
-	//Simple check that the host sent this not someone else, if detailsVpnAddr is filled
+	//TODO: CERT-V2 hosts with only v2 certs cannot provide their ipv6 addr when contacting the lighthouse via v4?
-	if detailsVpnAddr.IsValid() && !slices.Contains(fromVpnAddrs, detailsVpnAddr) {
+	//TODO: CERT-V2 why do we care about the vpnAddr in the packet? We know where it came from, right?
+	//Simple check that the host sent this not someone else
+	if !slices.Contains(fromVpnAddrs, detailsVpnAddr) {
 		if lhh.l.Level >= logrus.DebugLevel {
 			lhh.l.WithField("vpnAddrs", fromVpnAddrs).WithField("answer", detailsVpnAddr).Debugln("Host sent invalid update")
 		}
@@ -1273,24 +1272,24 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 	am.Lock()
 	lhh.lh.Unlock()
 
-	am.unlockedSetV4(fromVpnAddrs[0], fromVpnAddrs[0], n.Details.V4AddrPorts, lhh.lh.unlockedShouldAddV4)
+	am.unlockedSetV4(fromVpnAddrs[0], detailsVpnAddr, n.Details.V4AddrPorts, lhh.lh.unlockedShouldAddV4)
-	am.unlockedSetV6(fromVpnAddrs[0], fromVpnAddrs[0], n.Details.V6AddrPorts, lhh.lh.unlockedShouldAddV6)
+	am.unlockedSetV6(fromVpnAddrs[0], detailsVpnAddr, n.Details.V6AddrPorts, lhh.lh.unlockedShouldAddV6)
 	am.unlockedSetRelay(fromVpnAddrs[0], relays)
 	am.Unlock()
 
 	n = lhh.resetMeta()
 	n.Type = NebulaMeta_HostUpdateNotificationAck
-	switch useVersion {
+
-	case cert.Version1:
+	if useVersion == cert.Version1 {
 		if !fromVpnAddrs[0].Is4() {
 			lhh.l.WithField("vpnAddrs", fromVpnAddrs).Error("Can not send HostUpdateNotificationAck for a ipv6 vpn ip in a v1 message")
 			return
 		}
 		vpnAddrB := fromVpnAddrs[0].As4()
 		n.Details.OldVpnAddr = binary.BigEndian.Uint32(vpnAddrB[:])
-	case cert.Version2:
+	} else if useVersion == cert.Version2 {
-		// do nothing, we want to send a blank message
+		n.Details.VpnAddr = netAddrToProtoAddr(fromVpnAddrs[0])
-	default:
+	} else {
 		lhh.l.WithField("useVersion", useVersion).Error("invalid protocol version")
 		return
 	}
@@ -1308,20 +1307,13 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 func (lhh *LightHouseHandler) handleHostPunchNotification(n *NebulaMeta, fromVpnAddrs []netip.Addr, w EncWriter) {
 	//It's possible the lighthouse is communicating with us using a non primary vpn addr,
 	//which means we need to compare all fromVpnAddrs against all configured lighthouse vpn addrs.
+	//maybe one day we'll have a better idea, if it matters.
 	if !lhh.lh.IsAnyLighthouseAddr(fromVpnAddrs) {
 		return
 	}
 
-	detailsVpnAddr, _, err := n.Details.GetVpnAddrAndVersion()
-	if err != nil {
-		if lhh.l.Level >= logrus.DebugLevel {
-			lhh.l.WithField("details", n.Details).WithError(err).Debugln("dropping invalid HostPunchNotification")
-		}
-		return
-	}
-
 	empty := []byte{0}
-	punch := func(vpnPeer netip.AddrPort, logVpnAddr netip.Addr) {
+	punch := func(vpnPeer netip.AddrPort) {
 		if !vpnPeer.IsValid() {
 			return
 		}
@@ -1333,31 +1325,48 @@ func (lhh *LightHouseHandler) handleHostPunchNotification(n *NebulaMeta, fromVpn
 		}()
 
 		if lhh.l.Level >= logrus.DebugLevel {
+			var logVpnAddr netip.Addr
+			if n.Details.OldVpnAddr != 0 {
+				b := [4]byte{}
+				binary.BigEndian.PutUint32(b[:], n.Details.OldVpnAddr)
+				logVpnAddr = netip.AddrFrom4(b)
+			} else if n.Details.VpnAddr != nil {
+				logVpnAddr = protoAddrToNetAddr(n.Details.VpnAddr)
+			}
 			lhh.l.Debugf("Punching on %v for %v", vpnPeer, logVpnAddr)
 		}
 	}
 
 	for _, a := range n.Details.V4AddrPorts {
-		punch(protoV4AddrPortToNetAddrPort(a), detailsVpnAddr)
+		punch(protoV4AddrPortToNetAddrPort(a))
 	}
 
 	for _, a := range n.Details.V6AddrPorts {
-		punch(protoV6AddrPortToNetAddrPort(a), detailsVpnAddr)
+		punch(protoV6AddrPortToNetAddrPort(a))
 	}
 
 	// This sends a nebula test packet to the host trying to contact us. In the case
 	// of a double nat or other difficult scenario, this may help establish
 	// a tunnel.
 	if lhh.lh.punchy.GetRespond() {
+		var queryVpnAddr netip.Addr
+		if n.Details.OldVpnAddr != 0 {
+			b := [4]byte{}
+			binary.BigEndian.PutUint32(b[:], n.Details.OldVpnAddr)
+			queryVpnAddr = netip.AddrFrom4(b)
+		} else if n.Details.VpnAddr != nil {
+			queryVpnAddr = protoAddrToNetAddr(n.Details.VpnAddr)
+		}
+
 		go func() {
 			time.Sleep(lhh.lh.punchy.GetRespondDelay())
 			if lhh.l.Level >= logrus.DebugLevel {
-				lhh.l.Debugf("Sending a nebula test packet to vpn addr %s", detailsVpnAddr)
+				lhh.l.Debugf("Sending a nebula test packet to vpn addr %s", queryVpnAddr)
 			}
 			//NOTE: we have to allocate a new output buffer here since we are spawning a new goroutine
 			// for each punchBack packet. We should move this into a timerwheel or a single goroutine
 			// managed by a channel.
-			w.SendMessageToVpnAddr(header.Test, header.TestRequest, detailsVpnAddr, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
+			w.SendMessageToVpnAddr(header.Test, header.TestRequest, queryVpnAddr, []byte(""), make([]byte, 12, 12), make([]byte, mtu))
 		}()
 	}
 }
@@ -1436,17 +1445,3 @@ func findNetworkUnion(prefixes []netip.Prefix, addrs []netip.Addr) (netip.Addr,
 	}
 	return netip.Addr{}, false
 }
-
-func (d *NebulaMetaDetails) GetVpnAddrAndVersion() (netip.Addr, cert.Version, error) {
-	if d.OldVpnAddr != 0 {
-		b := [4]byte{}
-		binary.BigEndian.PutUint32(b[:], d.OldVpnAddr)
-		detailsVpnAddr := netip.AddrFrom4(b)
-		return detailsVpnAddr, cert.Version1, nil
-	} else if d.VpnAddr != nil {
-		detailsVpnAddr := protoAddrToNetAddr(d.VpnAddr)
-		return detailsVpnAddr, cert.Version2, nil
-	} else {
-		return netip.Addr{}, cert.Version1, ErrBadDetailsVpnAddr
-	}
-}

lighthouse_test.go

@@ -493,123 +493,3 @@ func Test_findNetworkUnion(t *testing.T) {
 	out, ok = findNetworkUnion([]netip.Prefix{fc00}, []netip.Addr{a1, afe81})
 	assert.False(t, ok)
 }
-
-func TestLighthouse_Dont_Delete_Static_Hosts(t *testing.T) {
-	l := test.NewLogger()
-
-	myUdpAddr2 := netip.MustParseAddrPort("1.2.3.4:4242")
-
-	testSameHostNotStatic := netip.MustParseAddr("10.128.0.41")
-	testStaticHost := netip.MustParseAddr("10.128.0.42")
-	//myVpnIp := netip.MustParseAddr("10.128.0.2")
-
-	c := config.NewC(l)
-	lh1 := "10.128.0.2"
-	c.Settings["lighthouse"] = map[string]any{
-		"hosts":    []any{lh1},
-		"interval": "1s",
-	}
-
-	c.Settings["listen"] = map[string]any{"port": 4242}
-	c.Settings["static_host_map"] = map[string]any{
-		lh1:           []any{"1.1.1.1:4242"},
-		"10.128.0.42": []any{"1.2.3.4:4242"},
-	}
-
-	myVpnNet := netip.MustParsePrefix("10.128.0.1/24")
-	nt := new(bart.Lite)
-	nt.Insert(myVpnNet)
-	cs := &CertState{
-		myVpnNetworks:      []netip.Prefix{myVpnNet},
-		myVpnNetworksTable: nt,
-	}
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
-	require.NoError(t, err)
-	lh.ifce = &mockEncWriter{}
-
-	//test that we actually have the static entry:
-	out := lh.Query(testStaticHost)
-	assert.NotNil(t, out)
-	assert.Equal(t, out.vpnAddrs[0], testStaticHost)
-	out.Rebuild([]netip.Prefix{}) //why tho
-	assert.Equal(t, out.addrs[0], myUdpAddr2)
-
-	//bolt on a lower numbered primary IP
-	am := lh.unlockedGetRemoteList([]netip.Addr{testStaticHost})
-	am.vpnAddrs = []netip.Addr{testSameHostNotStatic, testStaticHost}
-	lh.addrMap[testSameHostNotStatic] = am
-	out.Rebuild([]netip.Prefix{}) //???
-
-	//test that we actually have the static entry:
-	out = lh.Query(testStaticHost)
-	assert.NotNil(t, out)
-	assert.Equal(t, out.vpnAddrs[0], testSameHostNotStatic)
-	assert.Equal(t, out.vpnAddrs[1], testStaticHost)
-	assert.Equal(t, out.addrs[0], myUdpAddr2)
-
-	//test that we actually have the static entry for BOTH:
-	out2 := lh.Query(testSameHostNotStatic)
-	assert.Same(t, out2, out)
-
-	//now do the delete
-	lh.DeleteVpnAddrs([]netip.Addr{testSameHostNotStatic, testStaticHost})
-	//verify
-	out = lh.Query(testSameHostNotStatic)
-	assert.NotNil(t, out)
-	if out == nil {
-		t.Fatal("expected non-nil query for the static host")
-	}
-	assert.Equal(t, out.vpnAddrs[0], testSameHostNotStatic)
-	assert.Equal(t, out.vpnAddrs[1], testStaticHost)
-	assert.Equal(t, out.addrs[0], myUdpAddr2)
-}
-
-func TestLighthouse_DeletesWork(t *testing.T) {
-	l := test.NewLogger()
-
-	myUdpAddr2 := netip.MustParseAddrPort("1.2.3.4:4242")
-	testHost := netip.MustParseAddr("10.128.0.42")
-
-	c := config.NewC(l)
-	lh1 := "10.128.0.2"
-	c.Settings["lighthouse"] = map[string]any{
-		"hosts":    []any{lh1},
-		"interval": "1s",
-	}
-
-	c.Settings["listen"] = map[string]any{"port": 4242}
-	c.Settings["static_host_map"] = map[string]any{
-		lh1: []any{"1.1.1.1:4242"},
-	}
-
-	myVpnNet := netip.MustParsePrefix("10.128.0.1/24")
-	nt := new(bart.Lite)
-	nt.Insert(myVpnNet)
-	cs := &CertState{
-		myVpnNetworks:      []netip.Prefix{myVpnNet},
-		myVpnNetworksTable: nt,
-	}
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
-	require.NoError(t, err)
-	lh.ifce = &mockEncWriter{}
-
-	//insert the host
-	am := lh.unlockedGetRemoteList([]netip.Addr{testHost})
-	am.vpnAddrs = []netip.Addr{testHost}
-	am.addrs = []netip.AddrPort{myUdpAddr2}
-	lh.addrMap[testHost] = am
-	am.Rebuild([]netip.Prefix{}) //???
-
-	//test that we actually have the entry:
-	out := lh.Query(testHost)
-	assert.NotNil(t, out)
-	assert.Equal(t, out.vpnAddrs[0], testHost)
-	out.Rebuild([]netip.Prefix{}) //why tho
-	assert.Equal(t, out.addrs[0], myUdpAddr2)
-
-	//now do the delete
-	lh.DeleteVpnAddrs([]netip.Addr{testHost})
-	//verify
-	out = lh.Query(testHost)
-	assert.Nil(t, out)
-}

main.go

@@ -185,7 +185,6 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 
 	hostMap := NewHostMapFromConfig(l, c)
 	punchy := NewPunchyFromConfig(l, c)
-	connManager := newConnectionManagerFromConfig(l, c, hostMap, punchy)
 	lightHouse, err := NewLightHouseFromConfig(ctx, l, c, pki.getCertState(), udpConns[0], punchy)
 	if err != nil {
 		return nil, util.ContextualizeIfNeeded("Failed to initialize lighthouse handler", err)
@@ -221,26 +220,31 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		}
 	}
 
+	checkInterval := c.GetInt("timers.connection_alive_interval", 5)
+	pendingDeletionInterval := c.GetInt("timers.pending_deletion_interval", 10)
+
 	ifConfig := &InterfaceConfig{
-		HostMap:               hostMap,
+		HostMap:                 hostMap,
-		Inside:                tun,
+		Inside:                  tun,
-		Outside:               udpConns[0],
+		Outside:                 udpConns[0],
-		pki:                   pki,
+		pki:                     pki,
-		Firewall:              fw,
+		Firewall:                fw,
-		ServeDns:              serveDns,
+		ServeDns:                serveDns,
-		HandshakeManager:      handshakeManager,
+		HandshakeManager:        handshakeManager,
-		connectionManager:     connManager,
+		lightHouse:              lightHouse,
-		lightHouse:            lightHouse,
+		checkInterval:           time.Second * time.Duration(checkInterval),
-		tryPromoteEvery:       c.GetUint32("counters.try_promote", defaultPromoteEvery),
+		pendingDeletionInterval: time.Second * time.Duration(pendingDeletionInterval),
-		reQueryEvery:          c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
+		tryPromoteEvery:         c.GetUint32("counters.try_promote", defaultPromoteEvery),
-		reQueryWait:           c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
+		reQueryEvery:            c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
-		DropLocalBroadcast:    c.GetBool("tun.drop_local_broadcast", false),
+		reQueryWait:             c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
-		DropMulticast:         c.GetBool("tun.drop_multicast", false),
+		DropLocalBroadcast:      c.GetBool("tun.drop_local_broadcast", false),
-		routines:              routines,
+		DropMulticast:           c.GetBool("tun.drop_multicast", false),
-		MessageMetrics:        messageMetrics,
+		routines:                routines,
-		version:               buildVersion,
+		MessageMetrics:          messageMetrics,
-		relayManager:          NewRelayManager(ctx, l, hostMap, c),
+		version:                 buildVersion,
-		punchy:                punchy,
+		relayManager:            NewRelayManager(ctx, l, hostMap, c),
+		punchy:                  punchy,
+
 		ConntrackCacheTimeout: conntrackCacheTimeout,
 		l:                     l,
 	}
@@ -292,6 +296,5 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		statsStart,
 		dnsStart,
 		lightHouse.StartUpdateWorker,
-		connManager.Start,
 	}, nil
 }

outside.go

@@ -81,7 +81,7 @@ func (f *Interface) readOutsidePackets(ip netip.AddrPort, via *ViaSender, out []
 			// Pull the Roaming parts up here, and return in all call paths.
 			f.handleHostRoaming(hostinfo, ip)
 			// Track usage of both the HostInfo and the Relay for the received & authenticated packet
-			f.connectionManager.In(hostinfo)
+			f.connectionManager.In(hostinfo.localIndexId)
 			f.connectionManager.RelayUsed(h.RemoteIndex)
 
 			relay, ok := hostinfo.relayState.QueryRelayForByIdx(h.RemoteIndex)
@@ -213,7 +213,7 @@ func (f *Interface) readOutsidePackets(ip netip.AddrPort, via *ViaSender, out []
 
 	f.handleHostRoaming(hostinfo, ip)
 
-	f.connectionManager.In(hostinfo)
+	f.connectionManager.In(hostinfo.localIndexId)
 }
 
 // closeTunnel closes a tunnel locally, it does not send a closeTunnel packet to the remote
@@ -254,18 +254,16 @@ func (f *Interface) handleHostRoaming(hostinfo *HostInfo, udpAddr netip.AddrPort
 
 }
 
-// handleEncrypted returns true if a packet should be processed, false otherwise
 func (f *Interface) handleEncrypted(ci *ConnectionState, addr netip.AddrPort, h *header.H) bool {
-	// If connectionstate does not exist, send a recv error, if possible, to encourage a fast reconnect
+	// If connectionstate exists and the replay protector allows, process packet
-	if ci == nil {
+	// Else, send recv errors for 300 seconds after a restart to allow fast reconnection.
+	if ci == nil || !ci.window.Check(f.l, h.MessageCounter) {
 		if addr.IsValid() {
 			f.maybeSendRecvError(addr, h.RemoteIndex)
+			return false
+		} else {
+			return false
 		}
-		return false
-	}
-	// If the window check fails, refuse to process the packet, but don't send a recv error
-	if !ci.window.Check(f.l, h.MessageCounter) {
-		return false
 	}
 
 	return true
@@ -500,7 +498,7 @@ func (f *Interface) decryptToTun(hostinfo *HostInfo, messageCounter uint64, out
 		return false
 	}
 
-	f.connectionManager.In(hostinfo)
+	f.connectionManager.In(hostinfo.localIndexId)
 	_, err = f.readers[q].Write(out)
 	if err != nil {
 		f.l.WithError(err).Error("Failed to write to tun")
@@ -539,6 +537,10 @@ func (f *Interface) handleRecvError(addr netip.AddrPort, h *header.H) {
 		return
 	}
 
+	if !hostinfo.RecvErrorExceeded() {
+		return
+	}
+
 	if hostinfo.remote.IsValid() && hostinfo.remote != addr {
 		f.l.Infoln("Someone spoofing recv_errors? ", addr, hostinfo.remote)
 		return

overlay/tun_linux.go

@@ -293,6 +293,7 @@ func (t *tun) addIPs(link netlink.Link) error {
 
 	//add all new addresses
 	for i := range newAddrs {
+		//TODO: CERT-V2 do we want to stack errors and try as many ops as possible?
 		//AddrReplace still adds new IPs, but if their properties change it will change them as well
 		if err := netlink.AddrReplace(link, newAddrs[i]); err != nil {
 			return err
@@ -360,11 +361,6 @@ func (t *tun) Activate() error {
 		t.l.WithError(err).Error("Failed to set tun tx queue length")
 	}
 
-	const modeNone = 1
-	if err = netlink.LinkSetIP6AddrGenMode(link, modeNone); err != nil {
-		t.l.WithError(err).Warn("Failed to disable link local address generation")
-	}
-
 	if err = t.addIPs(link); err != nil {
 		return err
 	}
@@ -642,11 +638,6 @@ func (t *tun) updateRoutes(r netlink.RouteUpdate) {
 		return
 	}
 
-	if r.Dst == nil {
-		t.l.WithField("route", r).Debug("Ignoring route update, no destination address")
-		return
-	}
-
 	dstAddr, ok := netip.AddrFromSlice(r.Dst.IP)
 	if !ok {
 		t.l.WithField("route", r).Debug("Ignoring route update, invalid destination address")

pkclient/pkclient_cgo.go

@@ -180,7 +180,6 @@ func (c *PKClient) DeriveNoise(peerPubKey []byte) ([]byte, error) {
 		pkcs11.NewAttribute(pkcs11.CKA_DECRYPT, true),
 		pkcs11.NewAttribute(pkcs11.CKA_WRAP, true),
 		pkcs11.NewAttribute(pkcs11.CKA_UNWRAP, true),
-		pkcs11.NewAttribute(pkcs11.CKA_VALUE_LEN, NoiseKeySize),
 	}
 
 	// Set up the parameters which include the peer's public key

pki.go

@@ -100,62 +100,55 @@ func (p *PKI) reloadCerts(c *config.C, initial bool) *util.ContextualError {
 		currentState := p.cs.Load()
 		if newState.v1Cert != nil {
 			if currentState.v1Cert == nil {
-				//adding certs is fine, actually. Networks-in-common confirmed in newCertState().
+				return util.NewContextualError("v1 certificate was added, restart required", nil, err)
-			} else {
-				// did IP in cert change? if so, don't set
-				if !slices.Equal(currentState.v1Cert.Networks(), newState.v1Cert.Networks()) {
-					return util.NewContextualError(
-						"Networks in new cert was different from old",
-						m{"new_networks": newState.v1Cert.Networks(), "old_networks": currentState.v1Cert.Networks(), "cert_version": cert.Version1},
-						nil,
-					)
-				}
-
-				if currentState.v1Cert.Curve() != newState.v1Cert.Curve() {
-					return util.NewContextualError(
-						"Curve in new v1 cert was different from old",
-						m{"new_curve": newState.v1Cert.Curve(), "old_curve": currentState.v1Cert.Curve(), "cert_version": cert.Version1},
-						nil,
-					)
-				}
 			}
+
+			// did IP in cert change? if so, don't set
+			if !slices.Equal(currentState.v1Cert.Networks(), newState.v1Cert.Networks()) {
+				return util.NewContextualError(
+					"Networks in new cert was different from old",
+					m{"new_networks": newState.v1Cert.Networks(), "old_networks": currentState.v1Cert.Networks()},
+					nil,
+				)
+			}
+
+			if currentState.v1Cert.Curve() != newState.v1Cert.Curve() {
+				return util.NewContextualError(
+					"Curve in new cert was different from old",
+					m{"new_curve": newState.v1Cert.Curve(), "old_curve": currentState.v1Cert.Curve()},
+					nil,
+				)
+			}
+
+		} else if currentState.v1Cert != nil {
+			//TODO: CERT-V2 we should be able to tear this down
+			return util.NewContextualError("v1 certificate was removed, restart required", nil, err)
 		}
 
 		if newState.v2Cert != nil {
 			if currentState.v2Cert == nil {
-				//adding certs is fine, actually
+				return util.NewContextualError("v2 certificate was added, restart required", nil, err)
-			} else {
-				// did IP in cert change? if so, don't set
-				if !slices.Equal(currentState.v2Cert.Networks(), newState.v2Cert.Networks()) {
-					return util.NewContextualError(
-						"Networks in new cert was different from old",
-						m{"new_networks": newState.v2Cert.Networks(), "old_networks": currentState.v2Cert.Networks(), "cert_version": cert.Version2},
-						nil,
-					)
-				}
-
-				if currentState.v2Cert.Curve() != newState.v2Cert.Curve() {
-					return util.NewContextualError(
-						"Curve in new cert was different from old",
-						m{"new_curve": newState.v2Cert.Curve(), "old_curve": currentState.v2Cert.Curve(), "cert_version": cert.Version2},
-						nil,
-					)
-				}
 			}
 
-		} else if currentState.v2Cert != nil {
+			// did IP in cert change? if so, don't set
-			//newState.v1Cert is non-nil bc empty certstates aren't permitted
+			if !slices.Equal(currentState.v2Cert.Networks(), newState.v2Cert.Networks()) {
-			if newState.v1Cert == nil {
-				return util.NewContextualError("v1 and v2 certs are nil, this should be impossible", nil, err)
-			}
-			//if we're going to v1-only, we need to make sure we didn't orphan any v2-cert vpnaddrs
-			if !slices.Equal(currentState.v2Cert.Networks(), newState.v1Cert.Networks()) {
 				return util.NewContextualError(
-					"Removing a V2 cert is not permitted unless it has identical networks to the new V1 cert",
+					"Networks in new cert was different from old",
-					m{"new_v1_networks": newState.v1Cert.Networks(), "old_v2_networks": currentState.v2Cert.Networks()},
+					m{"new_networks": newState.v2Cert.Networks(), "old_networks": currentState.v2Cert.Networks()},
 					nil,
 				)
 			}
+
+			if currentState.v2Cert.Curve() != newState.v2Cert.Curve() {
+				return util.NewContextualError(
+					"Curve in new cert was different from old",
+					m{"new_curve": newState.v2Cert.Curve(), "old_curve": currentState.v2Cert.Curve()},
+					nil,
+				)
+			}
+
+		} else if currentState.v2Cert != nil {
+			return util.NewContextualError("v2 certificate was removed, restart required", nil, err)
 		}
 
 		// Cipher cant be hot swapped so just leave it at what it was before
@@ -180,6 +173,7 @@ func (p *PKI) reloadCerts(c *config.C, initial bool) *util.ContextualError {
 
 	p.cs.Store(newState)
 
+	//TODO: CERT-V2 newState needs a stringer that does json
 	if initial {
 		p.l.WithField("cert", newState).Debug("Client nebula certificate(s)")
 	} else {
@@ -365,9 +359,7 @@ func newCertState(dv cert.Version, v1, v2 cert.Certificate, pkcs11backed bool, p
 			return nil, util.NewContextualError("v1 and v2 curve are not the same, ignoring", nil, nil)
 		}
 
-		if v1.Networks()[0] != v2.Networks()[0] {
+		//TODO: CERT-V2 make sure v2 has v1s address
-			return nil, util.NewContextualError("v1 and v2 networks are not the same", nil, nil)
-		}
 
 		cs.initiatingVersion = dv
 	}

relay_manager.go

@@ -190,7 +190,6 @@ func (rm *relayManager) handleCreateRelayResponse(v cert.Version, h *HostInfo, f
 			InitiatorRelayIndex: peerRelay.RemoteIndex,
 		}
 
-		relayFrom := h.vpnAddrs[0]
 		if v == cert.Version1 {
 			peer := peerHostInfo.vpnAddrs[0]
 			if !peer.Is4() {
@@ -208,13 +207,7 @@ func (rm *relayManager) handleCreateRelayResponse(v cert.Version, h *HostInfo, f
 			b = targetAddr.As4()
 			resp.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
 		} else {
-			if targetAddr.Is4() {
+			resp.RelayFromAddr = netAddrToProtoAddr(peerHostInfo.vpnAddrs[0])
-				relayFrom = h.vpnAddrs[0]
-			} else {
-				//todo do this smarter
-				relayFrom = h.vpnAddrs[len(h.vpnAddrs)-1]
-			}
-			resp.RelayFromAddr = netAddrToProtoAddr(relayFrom)
 			resp.RelayToAddr = target
 		}
 
@@ -367,7 +360,7 @@ func (rm *relayManager) handleCreateRelayRequest(v cert.Version, h *HostInfo, f
 			Type:                NebulaControl_CreateRelayRequest,
 			InitiatorRelayIndex: index,
 		}
-		relayFrom := h.vpnAddrs[0]
+
 		if v == cert.Version1 {
 			if !h.vpnAddrs[0].Is4() {
 				rm.l.WithField("relayFrom", h.vpnAddrs[0]).
@@ -384,13 +377,7 @@ func (rm *relayManager) handleCreateRelayRequest(v cert.Version, h *HostInfo, f
 			b = target.As4()
 			req.OldRelayToAddr = binary.BigEndian.Uint32(b[:])
 		} else {
-			if target.Is4() {
+			req.RelayFromAddr = netAddrToProtoAddr(h.vpnAddrs[0])
-				relayFrom = h.vpnAddrs[0]
-			} else {
-				//todo do this smarter
-				relayFrom = h.vpnAddrs[len(h.vpnAddrs)-1]
-			}
-			req.RelayFromAddr = netAddrToProtoAddr(relayFrom)
 			req.RelayToAddr = netAddrToProtoAddr(target)
 		}
 
@@ -401,7 +388,7 @@ func (rm *relayManager) handleCreateRelayRequest(v cert.Version, h *HostInfo, f
 		} else {
 			f.SendMessageToHostInfo(header.Control, 0, peer, msg, make([]byte, 12), make([]byte, mtu))
 			rm.l.WithFields(logrus.Fields{
-				"relayFrom":           relayFrom,
+				"relayFrom":           h.vpnAddrs[0],
 				"relayTo":             target,
 				"initiatorRelayIndex": req.InitiatorRelayIndex,
 				"responderRelayIndex": req.ResponderRelayIndex,

remote_list.go

@@ -7,11 +7,11 @@ import (
 	"slices"
 	"sort"
 	"strconv"
-	"sync"
 	"sync/atomic"
 	"time"
 
 	"github.com/sirupsen/logrus"
+	"github.com/wadey/synctrace"
 )
 
 // forEachFunc is used to benefit folks that want to do work inside the lock
@@ -185,12 +185,12 @@ func (hr *hostnamesResults) GetAddrs() []netip.AddrPort {
 // It serves as a local cache of query replies, host update notifications, and locally learned addresses
 type RemoteList struct {
 	// Every interaction with internals requires a lock!
-	sync.RWMutex
+	synctrace.RWMutex
 
 	// The full list of vpn addresses assigned to this host
 	vpnAddrs []netip.Addr
 
-	// A deduplicated set of underlay addresses. Any accessor should lock beforehand.
+	// A deduplicated set of addresses. Any accessor should lock beforehand.
 	addrs []netip.AddrPort
 
 	// A set of relay addresses. VpnIp addresses that the remote identified as relays.
@@ -201,10 +201,8 @@ type RemoteList struct {
 	// For learned addresses, this is the vpnIp that sent the packet
 	cache map[netip.Addr]*cache
 
-	hr *hostnamesResults
+	hr        *hostnamesResults
-
+	shouldAdd func(netip.Addr) bool
-	// shouldAdd is a nillable function that decides if x should be added to addrs.
-	shouldAdd func(vpnAddrs []netip.Addr, x netip.Addr) bool
 
 	// This is a list of remotes that we have tried to handshake with and have returned from the wrong vpn ip.
 	// They should not be tried again during a handshake
@@ -215,8 +213,9 @@ type RemoteList struct {
 }
 
 // NewRemoteList creates a new empty RemoteList
-func NewRemoteList(vpnAddrs []netip.Addr, shouldAdd func([]netip.Addr, netip.Addr) bool) *RemoteList {
+func NewRemoteList(vpnAddrs []netip.Addr, shouldAdd func(netip.Addr) bool) *RemoteList {
 	r := &RemoteList{
+		RWMutex:   synctrace.NewRWMutex("remote-list"),
 		vpnAddrs:  make([]netip.Addr, len(vpnAddrs)),
 		addrs:     make([]netip.AddrPort, 0),
 		relays:    make([]netip.Addr, 0),
@@ -370,15 +369,6 @@ func (r *RemoteList) CopyBlockedRemotes() []netip.AddrPort {
 	return c
 }
 
-// RefreshFromHandshake locks and updates the RemoteList to account for data learned upon a completed handshake
-func (r *RemoteList) RefreshFromHandshake(vpnAddrs []netip.Addr) {
-	r.Lock()
-	r.badRemotes = nil
-	r.vpnAddrs = make([]netip.Addr, len(vpnAddrs))
-	copy(r.vpnAddrs, vpnAddrs)
-	r.Unlock()
-}
-
 // ResetBlockedRemotes locks and clears the blocked remotes list
 func (r *RemoteList) ResetBlockedRemotes() {
 	r.Lock()
@@ -588,7 +578,7 @@ func (r *RemoteList) unlockedCollect() {
 
 	dnsAddrs := r.hr.GetAddrs()
 	for _, addr := range dnsAddrs {
-		if r.shouldAdd == nil || r.shouldAdd(r.vpnAddrs, addr.Addr()) {
+		if r.shouldAdd == nil || r.shouldAdd(addr.Addr()) {
 			if !r.unlockedIsBad(addr) {
 				addrs = append(addrs, addr)
 			}

sshd/server.go

@@ -5,10 +5,10 @@ import (
 	"errors"
 	"fmt"
 	"net"
-	"sync"
 
 	"github.com/armon/go-radix"
 	"github.com/sirupsen/logrus"
+	"github.com/wadey/synctrace"
 	"golang.org/x/crypto/ssh"
 )
 
@@ -28,7 +28,7 @@ type SSHServer struct {
 	listener    net.Listener
 
 	// Locks the conns/counter to avoid concurrent map access
-	connsLock sync.Mutex
+	connsLock synctrace.Mutex
 	conns     map[int]*session
 	counter   int
 }
@@ -41,6 +41,7 @@ func NewSSHServer(l *logrus.Entry) (*SSHServer, error) {
 		l:           l,
 		commands:    radix.New(),
 		conns:       make(map[int]*session),
+		connsLock:   synctrace.NewMutex("ssh-server-conns"),
 	}
 
 	cc := ssh.CertChecker{

timeout.go

@@ -1,8 +1,9 @@
 package nebula
 
 import (
-	"sync"
 	"time"
+
+	"github.com/wadey/synctrace"
 )
 
 // How many timer objects should be cached
@@ -34,7 +35,7 @@ type TimerWheel[T any] struct {
 }
 
 type LockingTimerWheel[T any] struct {
-	m sync.Mutex
+	m synctrace.Mutex
 	t *TimerWheel[T]
 }
 
@@ -81,8 +82,9 @@ func NewTimerWheel[T any](min, max time.Duration) *TimerWheel[T] {
 }
 
 // NewLockingTimerWheel is version of TimerWheel that is safe for concurrent use with a small performance penalty
-func NewLockingTimerWheel[T any](min, max time.Duration) *LockingTimerWheel[T] {
+func NewLockingTimerWheel[T any](name string, min, max time.Duration) *LockingTimerWheel[T] {
 	return &LockingTimerWheel[T]{
+		m: synctrace.NewMutex(name),
 		t: NewTimerWheel[T](min, max),
 	}
 }

udp/errors.go

@@ -1,5 +0,0 @@
-package udp
-
-import "errors"
-
-var ErrInvalidIPv6RemoteForSocket = errors.New("listener is IPv4, but writing to IPv6 remote")

udp/udp_darwin.go

@@ -3,62 +3,20 @@
 
 package udp
 
+// Darwin support is primarily implemented in udp_generic, besides NewListenConfig
+
 import (
-	"context"
-	"encoding/binary"
-	"errors"
 	"fmt"
 	"net"
 	"net/netip"
 	"syscall"
-	"unsafe"
 
 	"github.com/sirupsen/logrus"
-	"github.com/slackhq/nebula/config"
 	"golang.org/x/sys/unix"
 )
 
-type StdConn struct {
-	*net.UDPConn
-	isV4  bool
-	sysFd uintptr
-	l     *logrus.Logger
-}
-
-var _ Conn = &StdConn{}
-
 func NewListener(l *logrus.Logger, ip netip.Addr, port int, multi bool, batch int) (Conn, error) {
-	lc := NewListenConfig(multi)
+	return NewGenericListener(l, ip, port, multi, batch)
-	pc, err := lc.ListenPacket(context.TODO(), "udp", net.JoinHostPort(ip.String(), fmt.Sprintf("%v", port)))
-	if err != nil {
-		return nil, err
-	}
-
-	if uc, ok := pc.(*net.UDPConn); ok {
-		c := &StdConn{UDPConn: uc, l: l}
-
-		rc, err := uc.SyscallConn()
-		if err != nil {
-			return nil, fmt.Errorf("failed to open udp socket: %w", err)
-		}
-
-		err = rc.Control(func(fd uintptr) {
-			c.sysFd = fd
-		})
-		if err != nil {
-			return nil, fmt.Errorf("failed to get udp fd: %w", err)
-		}
-
-		la, err := c.LocalAddr()
-		if err != nil {
-			return nil, err
-		}
-		c.isV4 = la.Addr().Is4()
-
-		return c, nil
-	}
-
-	return nil, fmt.Errorf("unexpected PacketConn: %T %#v", pc, pc)
 }
 
 func NewListenConfig(multi bool) net.ListenConfig {
@@ -85,116 +43,16 @@ func NewListenConfig(multi bool) net.ListenConfig {
 	}
 }
 
-//go:linkname sendto golang.org/x/sys/unix.sendto
+func (u *GenericConn) Rebind() error {
-//go:noescape
+	rc, err := u.UDPConn.SyscallConn()
-func sendto(s int, buf []byte, flags int, to unsafe.Pointer, addrlen int32) (err error)
-
-func (u *StdConn) WriteTo(b []byte, ap netip.AddrPort) error {
-	var sa unsafe.Pointer
-	var addrLen int32
-
-	if u.isV4 {
-		if ap.Addr().Is6() {
-			return ErrInvalidIPv6RemoteForSocket
-		}
-
-		var rsa unix.RawSockaddrInet6
-		rsa.Family = unix.AF_INET6
-		rsa.Addr = ap.Addr().As16()
-		binary.BigEndian.PutUint16((*[2]byte)(unsafe.Pointer(&rsa.Port))[:], ap.Port())
-		sa = unsafe.Pointer(&rsa)
-		addrLen = syscall.SizeofSockaddrInet4
-	} else {
-		var rsa unix.RawSockaddrInet6
-		rsa.Family = unix.AF_INET6
-		rsa.Addr = ap.Addr().As16()
-		binary.BigEndian.PutUint16((*[2]byte)(unsafe.Pointer(&rsa.Port))[:], ap.Port())
-		sa = unsafe.Pointer(&rsa)
-		addrLen = syscall.SizeofSockaddrInet6
-	}
-
-	// Golang stdlib doesn't handle EAGAIN correctly in some situations so we do writes ourselves
-	// See https://github.com/golang/go/issues/73919
-	for {
-		//_, _, err := unix.Syscall6(unix.SYS_SENDTO, u.sysFd, uintptr(unsafe.Pointer(&b[0])), uintptr(len(b)), 0, sa, addrLen)
-		err := sendto(int(u.sysFd), b, 0, sa, addrLen)
-		if err == nil {
-			// Written, get out before the error handling
-			return nil
-		}
-
-		if errors.Is(err, syscall.EINTR) {
-			// Write was interrupted, retry
-			continue
-		}
-
-		if errors.Is(err, syscall.EAGAIN) {
-			return &net.OpError{Op: "sendto", Err: unix.EWOULDBLOCK}
-		}
-
-		if errors.Is(err, syscall.EBADF) {
-			return net.ErrClosed
-		}
-
-		return &net.OpError{Op: "sendto", Err: err}
-	}
-}
-
-func (u *StdConn) LocalAddr() (netip.AddrPort, error) {
-	a := u.UDPConn.LocalAddr()
-
-	switch v := a.(type) {
-	case *net.UDPAddr:
-		addr, ok := netip.AddrFromSlice(v.IP)
-		if !ok {
-			return netip.AddrPort{}, fmt.Errorf("LocalAddr returned invalid IP address: %s", v.IP)
-		}
-		return netip.AddrPortFrom(addr, uint16(v.Port)), nil
-
-	default:
-		return netip.AddrPort{}, fmt.Errorf("LocalAddr returned: %#v", a)
-	}
-}
-
-func (u *StdConn) ReloadConfig(c *config.C) {
-	// TODO
-}
-
-func NewUDPStatsEmitter(udpConns []Conn) func() {
-	// No UDP stats for non-linux
-	return func() {}
-}
-
-func (u *StdConn) ListenOut(r EncReader) {
-	buffer := make([]byte, MTU)
-
-	for {
-		// Just read one packet at a time
-		n, rua, err := u.ReadFromUDPAddrPort(buffer)
-		if err != nil {
-			if errors.Is(err, net.ErrClosed) {
-				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-				return
-			}
-
-			u.l.WithError(err).Error("unexpected udp socket receive error")
-		}
-
-		r(netip.AddrPortFrom(rua.Addr().Unmap(), rua.Port()), buffer[:n])
-	}
-}
-
-func (u *StdConn) Rebind() error {
-	var err error
-	if u.isV4 {
-		err = syscall.SetsockoptInt(int(u.sysFd), syscall.IPPROTO_IP, syscall.IP_BOUND_IF, 0)
-	} else {
-		err = syscall.SetsockoptInt(int(u.sysFd), syscall.IPPROTO_IPV6, syscall.IPV6_BOUND_IF, 0)
-	}
-
 	if err != nil {
-		u.l.WithError(err).Error("Failed to rebind udp socket")
+		return err
 	}
 
-	return nil
+	return rc.Control(func(fd uintptr) {
+		err := syscall.SetsockoptInt(int(fd), unix.IPPROTO_IPV6, unix.IPV6_BOUND_IF, 0)
+		if err != nil {
+			u.l.WithError(err).Error("Failed to rebind udp socket")
+		}
+	})
 }

udp/udp_generic.go

@@ -1,7 +1,6 @@
-//go:build (!linux || android) && !e2e_testing && !darwin
+//go:build (!linux || android) && !e2e_testing
 // +build !linux android
 // +build !e2e_testing
-// +build !darwin
 
 // udp_generic implements the nebula UDP interface in pure Go stdlib. This
 // means it can be used on platforms like Darwin and Windows.

udp/udp_linux.go

@@ -221,7 +221,7 @@ func (u *StdConn) writeTo6(b []byte, ip netip.AddrPort) error {
 
 func (u *StdConn) writeTo4(b []byte, ip netip.AddrPort) error {
 	if !ip.Addr().Is4() {
-		return ErrInvalidIPv6RemoteForSocket
+		return fmt.Errorf("Listener is IPv4, but writing to IPv6 remote")
 	}
 
 	var rsa unix.RawSockaddrInet4

udp/udp_rio_windows.go

@@ -11,13 +11,13 @@ import (
 	"io"
 	"net"
 	"net/netip"
-	"sync"
 	"sync/atomic"
 	"syscall"
 	"unsafe"
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/config"
+	"github.com/wadey/synctrace"
 	"golang.org/x/sys/windows"
 	"golang.zx2c4.com/wireguard/conn/winrio"
 )
@@ -46,7 +46,7 @@ type ringBuffer struct {
 	iocp       windows.Handle
 	isFull     bool
 	cq         winrio.Cq
-	mu         sync.Mutex
+	mu         synctrace.Mutex
 	overlapped windows.Overlapped
 }
 
@@ -64,7 +64,11 @@ func NewRIOListener(l *logrus.Logger, addr netip.Addr, port int) (*RIOConn, erro
 		return nil, errors.New("could not initialize winrio")
 	}
 
-	u := &RIOConn{l: l}
+	u := &RIOConn{
+		l:  l,
+		rx: ringBuffer{mu: synctrace.NewMutex("rio-rx")},
+		tx: ringBuffer{mu: synctrace.NewMutex("rio-tx")},
+	}
 
 	err := u.bind(&windows.SockaddrInet6{Addr: addr.As16(), Port: port})
 	if err != nil {
@@ -92,25 +96,6 @@ func (u *RIOConn) bind(sa windows.Sockaddr) error {
 	// Enable v4 for this socket
 	syscall.SetsockoptInt(syscall.Handle(u.sock), syscall.IPPROTO_IPV6, syscall.IPV6_V6ONLY, 0)
 
-	// Disable reporting of PORT_UNREACHABLE and NET_UNREACHABLE errors from the UDP socket receive call.
-	// These errors are returned on Windows during UDP receives based on the receipt of ICMP packets. Disable
-	// the UDP receive error returns with these ioctl calls.
-	ret := uint32(0)
-	flag := uint32(0)
-	size := uint32(unsafe.Sizeof(flag))
-	err = syscall.WSAIoctl(syscall.Handle(u.sock), syscall.SIO_UDP_CONNRESET, (*byte)(unsafe.Pointer(&flag)), size, nil, 0, &ret, nil, 0)
-	if err != nil {
-		return err
-	}
-	ret = 0
-	flag = 0
-	size = uint32(unsafe.Sizeof(flag))
-	SIO_UDP_NETRESET := uint32(syscall.IOC_IN | syscall.IOC_VENDOR | 15)
-	err = syscall.WSAIoctl(syscall.Handle(u.sock), SIO_UDP_NETRESET, (*byte)(unsafe.Pointer(&flag)), size, nil, 0, &ret, nil, 0)
-	if err != nil {
-		return err
-	}
-
 	err = u.rx.Open()
 	if err != nil {
 		return err
@@ -141,12 +126,8 @@ func (u *RIOConn) ListenOut(r EncReader) {
 		// Just read one packet at a time
 		n, rua, err := u.receive(buffer)
 		if err != nil {
-			if errors.Is(err, net.ErrClosed) {
+			u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
+			return
-				return
-			}
-			u.l.WithError(err).Error("unexpected udp socket receive error")
-			continue
 		}
 
 		r(netip.AddrPortFrom(netip.AddrFrom16(rua.Addr).Unmap(), (rua.Port>>8)|((rua.Port&0xff)<<8)), buffer[:n])