add a little context to dns

.github/ISSUE_TEMPLATE/config.yml

@@ -1,21 +1,13 @@
 blank_issues_enabled: true
 contact_links:
-  - name: 💨 Performance Issues
-    url: https://github.com/slackhq/nebula/discussions/new/choose
-    about: 'We ask that you create a discussion instead of an issue for performance-related questions. This allows us to have a more open conversation about the issue and helps us to better understand the problem.'
-
-  - name: 📄 Documentation Issues
-    url: https://github.com/definednet/nebula-docs
-    about: "If you've found an issue with the website documentation, please file it in the nebula-docs repository."
-
-  - name: 📱 Mobile Nebula Issues
-    url: https://github.com/definednet/mobile_nebula
-    about: "If you're using the mobile Nebula app and have found an issue, please file it in the mobile_nebula repository."
-
   - name: 📘 Documentation
     url: https://nebula.defined.net/docs/
-    about: 'The documentation is the best place to start if you are new to Nebula.'
+    about: Review documentation.
 
   - name: 💁 Support/Chat
-    url: https://join.slack.com/t/nebulaoss/shared_invite/zt-39pk4xopc-CUKlGcb5Z39dQ0cK1v7ehA
-    about: 'For faster support, join us on Slack for assistance!'
+    url: https://join.slack.com/t/nebulaoss/shared_invite/enQtOTA5MDI4NDg3MTg4LTkwY2EwNTI4NzQyMzc0M2ZlODBjNWI3NTY1MzhiOThiMmZlZjVkMTI0NGY4YTMyNjUwMWEyNzNkZTJmYzQxOGU
+    about: 'This issue tracker is not for support questions. Join us on Slack for assistance!'
+
+  - name: 📱 Mobile Nebula
+    url: https://github.com/definednet/mobile_nebula
+    about: 'This issue tracker is not for mobile support. Try the Mobile Nebula repo instead!'

.github/pull_request_template.md

@@ -1,11 +0,0 @@
-<!--
-Thank you for taking the time to submit a pull request!
-
-Please be sure to provide a clear description of what you're trying to achieve with the change.
-
-- If you're submitting a new feature, please explain how to use it and document any new config options in the example config.
-- If you're submitting a bugfix, please link the related issue or describe the circumstances surrounding the issue.
-- If you're changing a default, explain why you believe the new default is appropriate for most users.
-
-P.S. If you're only updating the README or other docs, please file a pull request here instead: https://github.com/DefinedNet/nebula-docs
--->

.github/workflows/test.yml

@@ -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: v2.1
+        version: v2.0
 
     - name: Test
       run: make test
@@ -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: v2.1
+        version: v2.0
 
     - name: Test
       run: make test

README.md

@@ -4,7 +4,7 @@ It lets you seamlessly connect computers anywhere in the world. Nebula is portab
 It can be used to connect a small number of computers, but is also able to connect tens of thousands of computers.
 
 Nebula incorporates a number of existing concepts like encryption, security groups, certificates,
-and tunneling.
+and tunneling, and each of those individual pieces existed before Nebula in various forms.
 What makes Nebula different to existing offerings is that it brings all of these ideas together,
 resulting in a sum that is greater than its individual parts.
 
@@ -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
 
@@ -28,33 +28,33 @@ Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for
 #### Distribution Packages
 
 - [Arch Linux](https://archlinux.org/packages/extra/x86_64/nebula/)
-    ```sh
-    sudo pacman -S nebula
+    ```
+    $ sudo pacman -S nebula
     ```
 
 - [Fedora Linux](https://src.fedoraproject.org/rpms/nebula)
-    ```sh
-    sudo dnf install nebula
+    ```
+    $ sudo dnf install nebula
     ```
 
 - [Debian Linux](https://packages.debian.org/source/stable/nebula)
-    ```sh
-    sudo apt install nebula
+    ```
+    $ sudo apt install nebula
     ```
 
 - [Alpine Linux](https://pkgs.alpinelinux.org/packages?name=nebula)
-    ```sh
-    sudo apk add nebula
+    ```
+    $ sudo apk add nebula
     ```
 
 - [macOS Homebrew](https://github.com/Homebrew/homebrew-core/blob/HEAD/Formula/n/nebula.rb)
-    ```sh
-    brew install nebula
+    ```
+    $ brew install nebula
     ```
 
 - [Docker](https://hub.docker.com/r/nebulaoss/nebula)
-    ```sh
-    docker pull nebulaoss/nebula
+    ```
+    $ docker pull nebulaoss/nebula
     ```
 
 #### Mobile
@@ -64,10 +64,10 @@ Check the [releases](https://github.com/slackhq/nebula/releases/latest) page for
 
 ## Technical Overview
 
-Nebula is a mutually authenticated peer-to-peer software-defined network based on the [Noise Protocol Framework](https://noiseprotocol.org/).
+Nebula is a mutually authenticated peer-to-peer software defined network based on the [Noise Protocol Framework](https://noiseprotocol.org/).
 Nebula uses certificates to assert a node's IP address, name, and membership within user-defined groups.
 Nebula's user-defined groups allow for provider agnostic traffic filtering between nodes.
-Discovery nodes (aka lighthouses) allow individual peers to find each other and optionally use UDP hole punching to establish connections from behind most firewalls or NATs.
+Discovery nodes allow individual peers to find each other and optionally use UDP hole punching to establish connections from behind most firewalls or NATs.
 Users can move data between nodes in any number of cloud service providers, datacenters, and endpoints, without needing to maintain a particular addressing scheme.
 
 Nebula uses Elliptic-curve Diffie-Hellman (`ECDH`) key exchange and `AES-256-GCM` in its default configuration.
@@ -82,34 +82,28 @@ To set up a Nebula network, you'll need:
 
 #### 2. (Optional, but you really should..) At least one discovery node with a routable IP address, which we call a lighthouse.
 
-Nebula lighthouses allow nodes to find each other, anywhere in the world. A lighthouse is the only node in a Nebula network whose IP should not change. Running a lighthouse requires very few compute resources, and you can easily use the least expensive option from a cloud hosting provider. If you're not sure which provider to use, a number of us have used $6/mo [DigitalOcean](https://digitalocean.com) droplets as lighthouses.
+Nebula lighthouses allow nodes to find each other, anywhere in the world. A lighthouse is the only node in a Nebula network whose IP should not change. Running a lighthouse requires very few compute resources, and you can easily use the least expensive option from a cloud hosting provider. If you're not sure which provider to use, a number of us have used $5/mo [DigitalOcean](https://digitalocean.com) droplets as lighthouses.
+
+  Once you have launched an instance, ensure that Nebula udp traffic (default port udp/4242) can reach it over the internet.
 
-Once you have launched an instance, ensure that Nebula udp traffic (default port udp/4242) can reach it over the internet.
 
 #### 3. A Nebula certificate authority, which will be the root of trust for a particular Nebula network.
 
-```sh
-./nebula-cert ca -name "Myorganization, Inc"
-```
-
-This will create files named `ca.key` and `ca.cert` in the current directory. The `ca.key` file is the most sensitive file you'll create, because it is the key used to sign the certificates for individual nebula nodes/hosts. Please store this file somewhere safe, preferably with strong encryption.
-
-**Be aware!** By default, certificate authorities have a 1-year lifetime before expiration. See [this guide](https://nebula.defined.net/docs/guides/rotating-certificate-authority/) for details on rotating a CA.
+  ```
+  ./nebula-cert ca -name "Myorganization, Inc"
+  ```
+  This will create files named `ca.key` and `ca.cert` in the current directory. The `ca.key` file is the most sensitive file you'll create, because it is the key used to sign the certificates for individual nebula nodes/hosts. Please store this file somewhere safe, preferably with strong encryption.
 
 #### 4. Nebula host keys and certificates generated from that certificate authority
-
 This assumes you have four nodes, named lighthouse1, laptop, server1, host3. You can name the nodes any way you'd like, including FQDN. You'll also need to choose IP addresses and the associated subnet. In this example, we are creating a nebula network that will use 192.168.100.x/24 as its network range. This example also demonstrates nebula groups, which can later be used to define traffic rules in a nebula network.
-```sh
+```
 ./nebula-cert sign -name "lighthouse1" -ip "192.168.100.1/24"
 ./nebula-cert sign -name "laptop" -ip "192.168.100.2/24" -groups "laptop,home,ssh"
 ./nebula-cert sign -name "server1" -ip "192.168.100.9/24" -groups "servers"
 ./nebula-cert sign -name "host3" -ip "192.168.100.10/24"
 ```
 
-By default, host certificates will expire 1 second before the CA expires. Use the `-duration` flag to specify a shorter lifetime.
-
 #### 5. Configuration files for each host
-
 Download a copy of the nebula [example configuration](https://github.com/slackhq/nebula/blob/master/examples/config.yml).
 
 * On the lighthouse node, you'll need to ensure `am_lighthouse: true` is set.
@@ -124,13 +118,10 @@ For each host, copy the nebula binary to the host, along with `config.yml` from
 **DO NOT COPY `ca.key` TO INDIVIDUAL NODES.**
 
 #### 7. Run nebula on each host
-
-```sh
+```
 ./nebula -config /path/to/config.yml
 ```
 
-For more detailed instructions, [find the full documentation here](https://nebula.defined.net/docs/).
-
 ## Building Nebula from source
 
 Make sure you have [go](https://go.dev/doc/install) installed and clone this repo. Change to the nebula directory.
@@ -149,10 +140,8 @@ The default curve used for cryptographic handshakes and signatures is Curve25519
 
 In addition, Nebula can be built using the [BoringCrypto GOEXPERIMENT](https://github.com/golang/go/blob/go1.20/src/crypto/internal/boring/README.md) by running either of the following make targets:
 
-```sh
-make bin-boringcrypto
-make release-boringcrypto
-```
+    make bin-boringcrypto
+    make release-boringcrypto
 
 This is not the recommended default deployment, but may be useful based on your compliance requirements.
 
@@ -160,3 +149,5 @@ This is not the recommended default deployment, but may be useful based on your
 
 Nebula was created at Slack Technologies, Inc by Nate Brown and Ryan Huber, with contributions from Oliver Fross, Alan Lam, Wade Simmons, and Lining Wang.
 
+
+

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/crypto_test.go

@@ -26,21 +26,21 @@ func TestNewArgon2Parameters(t *testing.T) {
 }
 
 func TestDecryptAndUnmarshalSigningPrivateKey(t *testing.T) {
-	passphrase := []byte("DO NOT USE")
+	passphrase := []byte("DO NOT USE THIS KEY")
 	privKey := []byte(`# A good key
 -----BEGIN NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
-CjsKC0FFUy0yNTYtR0NNEiwIExCAgAQYAyAEKiCPoDfGQiosxNPTbPn5EsMlc2MI
-c0Bt4oz6gTrFQhX3aBJcimhHKeAuhyTGvllD0Z19fe+DFPcLH3h5VrdjVfIAajg0
-KrbV3n9UHif/Au5skWmquNJzoW1E4MTdRbvpti6o+WdQ49DxjBFhx0YH8LBqrbPU
-0BGkUHmIO7daP24=
+CjwKC0FFUy0yNTYtR0NNEi0IExCAgIABGAEgBCognnjujd67Vsv99p22wfAjQaDT
+oCMW1mdjkU3gACKNW4MSXOWR9Sts4C81yk1RUku2gvGKs3TB9LYoklLsIizSYOLl
++Vs//O1T0I1Xbml2XBAROsb/VSoDln/6LMqR4B6fn6B3GOsLBBqRI8daDl9lRMPB
+qrlJ69wer3ZUHFXA
 -----END NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
 `)
 	shortKey := []byte(`# A key which, once decrypted, is too short
 -----BEGIN NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
-CjsKC0FFUy0yNTYtR0NNEiwIExCAgAQYAyAEKiAVJwdfl3r+eqi/vF6S7OMdpjfo
-hAzmTCRnr58Su4AqmBJbCv3zleYCEKYJP6UI3S8ekLMGISsgO4hm5leukCCyqT0Z
-cQ76yrberpzkJKoPLGisX8f+xdy4aXSZl7oEYWQte1+vqbtl/eY9PGZhxUQdcyq7
-hqzIyrRqfUgVuA==
+CjwKC0FFUy0yNTYtR0NNEi0IExCAgIABGAEgBCoga5h8owMEBWRSMMJKzuUvWce7
+k0qlBkQmCxiuLh80MuASW70YcKt8jeEIS2axo2V6zAKA9TSMcCsJW1kDDXEtL/xe
+GLF5T7sDl5COp4LU3pGxpV+KoeQ/S3gQCAAcnaOtnJQX+aSDnbO3jCHyP7U9CHbs
+rQr3bdH3Oy/WiYU=
 -----END NEBULA ED25519 ENCRYPTED PRIVATE KEY-----
 `)
 	invalidBanner := []byte(`# Invalid banner (not encrypted)

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")

connection_manager.go

@@ -4,16 +4,13 @@ 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"
 )
 
@@ -30,6 +27,12 @@ const (
 )
 
 type connectionManager struct {
+	in     map[uint32]struct{}
+	inLock *sync.RWMutex
+
+	out     map[uint32]struct{}
+	outLock *sync.RWMutex
+
 	// relayUsed holds which relay localIndexs are in use
 	relayUsed     map[uint32]struct{}
 	relayUsedLock *sync.RWMutex
@@ -37,117 +40,117 @@ type connectionManager struct {
 	hostMap                 *HostMap
 	trafficTimer            *LockingTimerWheel[uint32]
 	intf                    *Interface
+	pendingDeletion         map[uint32]struct{}
 	punchy                  *Punchy
-
-	// Configuration settings
 	checkInterval           time.Duration
 	pendingDeletionInterval time.Duration
-	inactivityTimeout       atomic.Int64
-	dropInactive            atomic.Bool
-
 	metricsTxPunchy         metrics.Counter
 
 	l *logrus.Logger
 }
 
-func newConnectionManagerFromConfig(l *logrus.Logger, c *config.C, hm *HostMap, p *Punchy) *connectionManager {
-	cm := &connectionManager{
-		hostMap:         hm,
-		l:               l,
-		punchy:          p,
+func newConnectionManager(ctx context.Context, l *logrus.Logger, intf *Interface, checkInterval, pendingDeletionInterval time.Duration, punchy *Punchy) *connectionManager {
+	var max time.Duration
+	if checkInterval < pendingDeletionInterval {
+		max = pendingDeletionInterval
+	} else {
+		max = checkInterval
+	}
+
+	nc := &connectionManager{
+		hostMap:                 intf.hostMap,
+		in:                      make(map[uint32]struct{}),
+		inLock:                  &sync.RWMutex{},
+		out:                     make(map[uint32]struct{}),
+		outLock:                 &sync.RWMutex{},
 		relayUsed:               make(map[uint32]struct{}),
 		relayUsedLock:           &sync.RWMutex{},
+		trafficTimer:            NewLockingTimerWheel[uint32](time.Millisecond*500, max),
+		intf:                    intf,
+		pendingDeletion:         make(map[uint32]struct{}),
+		checkInterval:           checkInterval,
+		pendingDeletionInterval: pendingDeletionInterval,
+		punchy:                  punchy,
 		metricsTxPunchy:         metrics.GetOrRegisterCounter("messages.tx.punchy", nil),
+		l:                       l,
 	}
 
-	cm.reload(c, true)
-	c.RegisterReloadCallback(func(c *config.C) {
-		cm.reload(c, false)
-	})
-
-	return cm
+	nc.Start(ctx)
+	return nc
 }
 
-func (cm *connectionManager) reload(c *config.C, initial bool) {
-	if initial {
-		cm.checkInterval = time.Duration(c.GetInt("timers.connection_alive_interval", 5)) * time.Second
-		cm.pendingDeletionInterval = time.Duration(c.GetInt("timers.pending_deletion_interval", 10)) * time.Second
-
-		// We want at least a minimum resolution of 500ms per tick so that we can hit these intervals
-		// pretty close to their configured duration.
-		// The inactivity duration is checked each time a hostinfo ticks through so we don't need the wheel to contain it.
-		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") {
-		old := cm.getInactivityTimeout()
-		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 {
-	return (time.Duration)(cm.inactivityTimeout.Load())
-}
-
-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()
+func (n *connectionManager) In(localIndex uint32) {
+	n.inLock.RLock()
 	// If this already exists, return
-	if _, ok := cm.relayUsed[localIndex]; ok {
-		cm.relayUsedLock.RUnlock()
+	if _, ok := n.in[localIndex]; ok {
+		n.inLock.RUnlock()
 		return
 	}
-	cm.relayUsedLock.RUnlock()
-	cm.relayUsedLock.Lock()
-	cm.relayUsed[localIndex] = struct{}{}
-	cm.relayUsedLock.Unlock()
+	n.inLock.RUnlock()
+	n.inLock.Lock()
+	n.in[localIndex] = struct{}{}
+	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) {
-	in := h.in.Swap(false)
-	out := h.out.Swap(false)
-	if in || out {
-		h.lastUsed = now
-	}
+func (n *connectionManager) getAndResetTrafficCheck(localIndex uint32) (bool, bool) {
+	n.inLock.Lock()
+	n.outLock.Lock()
+	_, in := n.in[localIndex]
+	_, 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.
-// We will continue to monitor the HostInfo until the tunnel is dropped.
-func (cm *connectionManager) AddTrafficWatch(h *HostInfo) {
-	if h.out.Swap(true) == false {
-		cm.trafficTimer.Add(h.localIndexId, cm.checkInterval)
+func (n *connectionManager) AddTrafficWatch(localIndex uint32) {
+	// Use a write lock directly because it should be incredibly rare that we are ever already tracking this index
+	n.outLock.Lock()
+	if _, ok := n.out[localIndex]; ok {
+		n.outLock.Unlock()
+		return
 	}
+	n.out[localIndex] = struct{}{}
+	n.trafficTimer.Add(localIndex, n.checkInterval)
+	n.outLock.Unlock()
 }
 
-func (cm *connectionManager) Start(ctx context.Context) {
-	clockSource := time.NewTicker(cm.trafficTimer.t.tickDuration)
+func (n *connectionManager) Start(ctx context.Context) {
+	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) {
-	decision, hostinfo, primary := cm.makeTrafficDecision(localIndex, now)
+func (n *connectionManager) doTrafficCheck(localIndex uint32, p, nb, out []byte, now time.Time) {
+	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)
-		cm.intf.closeTunnel(hostinfo)
+		n.intf.sendCloseTunnel(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()
-		defer cm.relayUsedLock.Unlock()
+		n.relayUsedLock.Lock()
+		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:
-			switch existing.State {
-			case Established, PeerRequested, Disestablished:
+		case ok && existing.State == Established:
 			// This relay already exists in newhostinfo, then do nothing.
 			continue
-			case Requested:
+		case ok && existing.State == Requested:
 			// The relay exists in a Requested state; re-send the request
 			index = existing.LocalIndex
 			switch r.Type {
 			case TerminalType:
-					relayFrom = cm.intf.myVpnAddrs[0]
+				relayFrom = n.intf.myVpnAddrs[0]
 				relayTo = existing.PeerAddr
 			case ForwardingType:
 				relayFrom = existing.PeerAddr
 				relayTo = newhostinfo.vpnAddrs[0]
 			default:
 				// should never happen
-					panic(fmt.Sprintf("Migrating unknown relay type: %v", r.Type))
-				}
 			}
 		case !ok:
-			cm.relayUsedLock.RLock()
-			if _, relayUsed := cm.relayUsed[r.LocalIndex]; !relayUsed {
+			n.relayUsedLock.RLock()
+			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))
-			cm.l.WithFields(logrus.Fields{
+			n.intf.SendMessageToHostInfo(header.Control, 0, newhostinfo, msg, make([]byte, 12), make([]byte, mtu))
+			n.l.WithFields(logrus.Fields{
 				"relayFrom":           req.RelayFromAddr,
 				"relayTo":             req.RelayToAddr,
 				"initiatorRelayIndex": req.InitiatorRelayIndex,
@@ -318,45 +316,46 @@ func (cm *connectionManager) migrateRelayUsed(oldhostinfo, newhostinfo *HostInfo
 	}
 }
 
-func (cm *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
-	cm.hostMap.RLock()
-	defer cm.hostMap.RUnlock()
+func (n *connectionManager) makeTrafficDecision(localIndex uint32, now time.Time) (trafficDecision, *HostInfo, *HostInfo) {
+	n.hostMap.RLock()
+	defer n.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 {
-			hostinfo.logger(cm.l).
+		if n.l.Level >= logrus.DebugLevel {
+			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.
@@ -364,55 +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)
-			cm.trafficTimer.Add(hostinfo.localIndexId, cm.checkInterval)
+			n.sendPunch(hostinfo)
+			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 {
-			hostinfo.logger(cm.l).
+		if n.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(n.l).
 				WithField("tunnelCheck", m{"state": "testing", "method": "active"}).
 				Debug("Tunnel status")
 		}
@@ -421,33 +411,17 @@ func (cm *connectionManager) makeTrafficDecision(localIndex uint32, now time.Tim
 		decision = sendTestPacket
 
 	} else {
-		if cm.l.Level >= logrus.DebugLevel {
-			hostinfo.logger(cm.l).Debugf("Hostinfo sadness")
+		if n.l.Level >= logrus.DebugLevel {
+			hostinfo.logger(n.l).Debugf("Hostinfo sadness")
 		}
 	}
 
-	hostinfo.pendingDeletion.Store(true)
-	cm.trafficTimer.Add(hostinfo.localIndexId, cm.pendingDeletionInterval)
+	n.pendingDeletion[hostinfo.localIndexId] = struct{}{}
+	n.trafficTimer.Add(hostinfo.localIndexId, n.pendingDeletionInterval)
 	return decision, hostinfo, nil
 }
 
-func (cm *connectionManager) isInactive(hostinfo *HostInfo, now time.Time) (time.Duration, 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 {
+func (n *connectionManager) shouldSwapPrimary(current, primary *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.
@@ -455,80 +429,73 @@ 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 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) {
-	cm.hostMap.Lock()
+func (n *connectionManager) swapPrimary(current, primary *HostInfo) {
+	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 {
-		cm.hostMap.unlockedMakePrimary(current)
+	if n.hostMap.Hosts[current.vpnAddrs[0]] == primary {
+		n.hostMap.unlockedMakePrimary(current)
 	}
-	cm.hostMap.Unlock()
+	n.hostMap.Unlock()
 }
 
 // isInvalidCertificate will check if we should destroy a tunnel if pki.disconnect_invalid is true and
 // the certificate is no longer valid. Block listed certificates will skip the pki.disconnect_invalid
 // 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
 	}
 
-	caPool := cm.intf.pki.GetCAPool()
+	caPool := n.intf.pki.GetCAPool()
 	err := caPool.VerifyCachedCertificate(now, remoteCert)
 	if err == nil {
 		return false
 	}
 
-	if !cm.intf.disconnectInvalid.Load() && err != cert.ErrBlockListed {
+	if !n.intf.disconnectInvalid.Load() && err != cert.ErrBlockListed {
 		// Block listed certificates should always be disconnected
 		return false
 	}
 
-	hostinfo.logger(cm.l).WithError(err).
+	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) {
-	if !cm.punchy.GetPunch() {
+func (n *connectionManager) sendPunch(hostinfo *HostInfo) {
+	if !n.punchy.GetPunch() {
 		// Punching is disabled
 		return
 	}
 
-	if cm.intf.lightHouse.IsAnyLighthouseAddr(hostinfo.vpnAddrs) {
-		// Do not punch to lighthouses, we assume our lighthouse update interval is good enough.
-		// In the event the update interval is not sufficient to maintain NAT state then a publicly available lighthouse
-		// would lose the ability to notify us and punchy.respond would become unreliable.
-		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)
+	if n.punchy.GetTargetEverything() {
+		hostinfo.remotes.ForEach(n.hostMap.GetPreferredRanges(), func(addr netip.AddrPort, preferred bool) {
+			n.metricsTxPunchy.Inc(1)
+			n.intf.outside.WriteTo([]byte{1}, addr)
 		})
 
 	} else if hostinfo.remote.IsValid() {
-		cm.metricsTxPunchy.Inc(1)
-		cm.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
+		n.metricsTxPunchy.Inc(1)
+		n.intf.outside.WriteTo([]byte{1}, hostinfo.remote)
 	}
 }
 
-func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
-	cs := cm.intf.pki.getCertState()
+func (n *connectionManager) tryRehandshake(hostinfo *HostInfo) {
+	cs := n.intf.pki.getCertState()
 	curCrt := hostinfo.ConnectionState.myCert
 	myCrt := cs.getCertificate(curCrt.Version())
 	if curCrt.Version() >= cs.initiatingVersion && bytes.Equal(curCrt.Signature(), myCrt.Signature()) == true {
@@ -536,9 +503,9 @@ func (cm *connectionManager) tryRehandshake(hostinfo *HostInfo) {
 		return
 	}
 
-	cm.l.WithField("vpnAddrs", hostinfo.vpnAddrs).
+	n.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.intf.handshakeManager.StartHandshake(hostinfo.vpnAddrs[0], nil)
 }

connection_manager_test.go

@@ -1,6 +1,7 @@
 package nebula
 
 import (
+	"context"
 	"crypto/ed25519"
 	"crypto/rand"
 	"net/netip"
@@ -63,10 +64,10 @@ func Test_NewConnectionManagerTest(t *testing.T) {
 	ifce.pki.cs.Store(cs)
 
 	// Create manager
-	conf := config.NewC(l)
-	punchy := NewPunchyFromConfig(l, conf)
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
-	nc.intf = ifce
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
+	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.In(hostinfo)
-	assert.False(t, hostinfo.pendingDeletion.Load())
+	nc.Out(hostinfo.localIndexId)
+	nc.In(hostinfo.localIndexId)
+	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.True(t, hostinfo.in.Load())
+	assert.Contains(t, nc.out, 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.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
 
 	// Do another traffic check tick, this host should be pending deletion now
-	nc.Out(hostinfo)
-	assert.True(t, hostinfo.out.Load())
+	nc.Out(hostinfo.localIndexId)
 	nc.doTrafficCheck(hostinfo.localIndexId, p, nb, out, time.Now())
-	assert.True(t, hostinfo.pendingDeletion.Load())
-	assert.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
+	assert.Contains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
+	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)
-	punchy := NewPunchyFromConfig(l, conf)
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
-	nc.intf = ifce
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
+	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.In(hostinfo)
-	assert.True(t, hostinfo.in.Load())
-	assert.True(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.pendingDeletion.Load())
+	nc.Out(hostinfo.localIndexId)
+	nc.In(hostinfo.localIndexId)
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.vpnAddrs[0])
 	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.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
+	assert.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
+	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.False(t, hostinfo.out.Load())
-	assert.False(t, hostinfo.in.Load())
+	assert.Contains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
+	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.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])
-}
-
-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.NotContains(t, nc.pendingDeletion, hostinfo.localIndexId)
+	assert.NotContains(t, nc.out, hostinfo.localIndexId)
+	assert.NotContains(t, nc.in, hostinfo.localIndexId)
 	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)
-	punchy := NewPunchyFromConfig(l, conf)
-	nc := newConnectionManagerFromConfig(l, conf, hostMap, punchy)
-	nc.intf = ifce
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+	punchy := NewPunchyFromConfig(l, config.NewC(l))
+	nc := newConnectionManager(ctx, l, ifce, 5, 10, punchy)
 	ifce.connectionManager = nc
 
 	hostinfo := &HostInfo{

control.go

@@ -34,7 +34,6 @@ type Control struct {
 	statsStart      func()
 	dnsStart        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

@@ -1,6 +1,7 @@
 package nebula
 
 import (
+	"context"
 	"fmt"
 	"net"
 	"net/netip"
@@ -39,7 +40,7 @@ func newDnsRecords(l *logrus.Logger, cs *CertState, hostMap *HostMap) *dnsRecord
 	}
 }
 
-func (d *dnsRecords) Query(q uint16, data string) netip.Addr {
+func (d *dnsRecords) query(q uint16, data string) netip.Addr {
 	data = strings.ToLower(data)
 	d.RLock()
 	defer d.RUnlock()
@@ -57,7 +58,7 @@ func (d *dnsRecords) Query(q uint16, data string) netip.Addr {
 	return netip.Addr{}
 }
 
-func (d *dnsRecords) QueryCert(data string) string {
+func (d *dnsRecords) queryCert(data string) string {
 	ip, err := netip.ParseAddr(data[:len(data)-1])
 	if err != nil {
 		return ""
@@ -122,7 +123,7 @@ func (d *dnsRecords) parseQuery(m *dns.Msg, w dns.ResponseWriter) {
 		case dns.TypeA, dns.TypeAAAA:
 			qType := dns.TypeToString[q.Qtype]
 			d.l.Debugf("Query for %s %s", qType, q.Name)
-			ip := d.Query(q.Qtype, q.Name)
+			ip := d.query(q.Qtype, q.Name)
 			if ip.IsValid() {
 				rr, err := dns.NewRR(fmt.Sprintf("%s %s %s", q.Name, qType, ip))
 				if err == nil {
@@ -135,7 +136,7 @@ func (d *dnsRecords) parseQuery(m *dns.Msg, w dns.ResponseWriter) {
 				return
 			}
 			d.l.Debugf("Query for TXT %s", q.Name)
-			ip := d.QueryCert(q.Name)
+			ip := d.queryCert(q.Name)
 			if ip != "" {
 				rr, err := dns.NewRR(fmt.Sprintf("%s TXT %s", q.Name, ip))
 				if err == nil {
@@ -163,18 +164,18 @@ func (d *dnsRecords) handleDnsRequest(w dns.ResponseWriter, r *dns.Msg) {
 	w.WriteMsg(m)
 }
 
-func dnsMain(l *logrus.Logger, cs *CertState, hostMap *HostMap, c *config.C) func() {
+func dnsMain(ctx context.Context, l *logrus.Logger, cs *CertState, hostMap *HostMap, c *config.C) func() {
 	dnsR = newDnsRecords(l, cs, hostMap)
 
 	// attach request handler func
 	dns.HandleFunc(".", dnsR.handleDnsRequest)
 
 	c.RegisterReloadCallback(func(c *config.C) {
-		reloadDns(l, c)
+		reloadDns(ctx, l, c)
 	})
 
 	return func() {
-		startDns(l, c)
+		startDns(ctx, l, c)
 	}
 }
 
@@ -187,24 +188,24 @@ func getDnsServerAddr(c *config.C) string {
 	return net.JoinHostPort(dnsHost, strconv.Itoa(c.GetInt("lighthouse.dns.port", 53)))
 }
 
-func startDns(l *logrus.Logger, c *config.C) {
+func startDns(ctx context.Context, l *logrus.Logger, c *config.C) {
 	dnsAddr = getDnsServerAddr(c)
 	dnsServer = &dns.Server{Addr: dnsAddr, Net: "udp"}
 	l.WithField("dnsListener", dnsAddr).Info("Starting DNS responder")
 	err := dnsServer.ListenAndServe()
-	defer dnsServer.Shutdown()
+	defer dnsServer.ShutdownContext(ctx)
 	if err != nil {
 		l.Errorf("Failed to start server: %s\n ", err.Error())
 	}
 }
 
-func reloadDns(l *logrus.Logger, c *config.C) {
+func reloadDns(ctx context.Context, l *logrus.Logger, c *config.C) {
 	if dnsAddr == getDnsServerAddr(c) {
 		l.Debug("No DNS server config change detected")
 		return
 	}
 
 	l.Debug("Restarting DNS server")
-	dnsServer.Shutdown()
-	go startDns(l, c)
+	dnsServer.ShutdownContext(ctx)
+	go startDns(ctx, l, c)
 }

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/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

@@ -1,57 +0,0 @@
-//go:build e2e_testing
-// +build e2e_testing
-
-package e2e
-
-import (
-	"testing"
-	"time"
-
-	"github.com/slackhq/nebula/cert"
-	"github.com/slackhq/nebula/cert_test"
-	"github.com/slackhq/nebula/e2e/router"
-)
-
-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()
-}

examples/config.yml

@@ -275,10 +275,6 @@ tun:
   # On linux only, set to true to manage unsafe routes directly on the system route table with gateway routes instead of
   # in nebula configuration files. Default false, not reloadable.
   #use_system_route_table: false
-  # Buffer size for reading routes updates. 0 means default system buffer size. (/proc/sys/net/core/rmem_default).
-  # If using massive routes updates, for example BGP, you may need to increase this value to avoid packet loss.
-  # SO_RCVBUFFORCE is used to avoid having to raise the system wide max
-  #use_system_route_table_buffer_size: 0
 
 # Configure logging level
 logging:
@@ -338,18 +334,6 @@ 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:

examples/go_service/main.go

@@ -5,12 +5,8 @@ import (
 	"fmt"
 	"log"
 	"net"
-	"os"
 
-	"github.com/sirupsen/logrus"
-	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/overlay"
 	"github.com/slackhq/nebula/service"
 )
 
@@ -63,16 +59,7 @@ pki:
 	if err := cfg.LoadString(configStr); err != nil {
 		return err
 	}
-
-	logger := logrus.New()
-	logger.Out = os.Stdout
-
-	ctrl, err := nebula.Main(&cfg, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
-	if err != nil {
-		return err
-	}
-
-	svc, err := service.New(ctrl)
+	svc, err := service.New(&cfg)
 	if err != nil {
 		return err
 	}

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

@@ -5,7 +5,7 @@ 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
@@ -23,7 +23,7 @@ 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
 	golang.org/x/crypto v0.37.0
 	golang.org/x/exp v0.0.0-20230725093048-515e97ebf090
 	golang.org/x/net v0.39.0
@@ -48,7 +48,7 @@ require (
 	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/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 v1.0.2/go.mod h1:E/hbnu0NxMFBjpMIE34DRGLWqDy0g5FuKDhCb31ngxA=
+dario.cat/mergo v1.0.1 h1:Ra4+bf83h2ztPIQYNP99R6m+Y7KfnARDfID+a+vLl4s=
+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=
@@ -145,10 +145,10 @@ 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/vishvananda/netlink v1.3.1/go.mod h1:ARtKouGSTGchR8aMwmkzC0qiNPrrWO5JS/XMVl45+b4=
-github.com/vishvananda/netns v0.0.5 h1:DfiHV+j8bA32MFM7bfEunvT8IAqQ/NzSJHtcmW5zdEY=
-github.com/vishvananda/netns v0.0.5/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
+github.com/vishvananda/netlink v1.3.0 h1:X7l42GfcV4S6E4vHTsw48qbrV+9PVojNfIhZcwQdrZk=
+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/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

@@ -249,7 +249,7 @@ func ixHandshakeStage1(f *Interface, addr netip.AddrPort, via *ViaSender, packet
 		HandshakePacket:   make(map[uint8][]byte, 0),
 		lastHandshakeTime: hs.Details.Time,
 		relayState: RelayState{
-			relays:         nil,
+			relays:         map[netip.Addr]struct{}{},
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},
@@ -457,7 +457,7 @@ 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.ResetBlockedRemotes()
 
@@ -652,7 +652,7 @@ 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))

handshake_manager.go

@@ -450,7 +450,7 @@ func (hm *HandshakeManager) StartHandshake(vpnAddr netip.Addr, cacheCb func(*Han
 		vpnAddrs:        []netip.Addr{vpnAddr},
 		HandshakePacket: make(map[uint8][]byte, 0),
 		relayState: RelayState{
-			relays:         nil,
+			relays:         map[netip.Addr]struct{}{},
 			relayForByAddr: map[netip.Addr]*Relay{},
 			relayForByIdx:  map[uint32]*Relay{},
 		},

hostmap.go

@@ -4,7 +4,6 @@ import (
 	"errors"
 	"net"
 	"net/netip"
-	"slices"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -17,10 +16,12 @@ import (
 	"github.com/slackhq/nebula/header"
 )
 
+// 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
@@ -67,7 +68,7 @@ type HostMap struct {
 type RelayState struct {
 	sync.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)
@@ -78,12 +79,7 @@ type RelayState struct {
 func (rs *RelayState) DeleteRelay(ip netip.Addr) {
 	rs.Lock()
 	defer rs.Unlock()
-	for idx, val := range rs.relays {
-		if val == ip {
-			rs.relays = append(rs.relays[:idx], rs.relays[idx+1:]...)
-			return
-		}
-	}
+	delete(rs.relays, ip)
 }
 
 func (rs *RelayState) UpdateRelayForByIpState(vpnIp netip.Addr, state int) {
@@ -128,16 +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 = append(rs.relays, ip)
-	}
+	rs.relays[ip] = struct{}{}
 }
 
 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
 }
 
@@ -224,6 +220,7 @@ type HostInfo struct {
 	// The host may have other vpn addresses that are outside our
 	// vpn networks but were removed because they are not usable
 	vpnAddrs  []netip.Addr
+	recvError atomic.Uint32
 
 	// networks are both all vpn and unsafe networks assigned to this host
 	networks   *bart.Lite
@@ -253,14 +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 {
@@ -730,6 +719,13 @@ 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

@@ -28,12 +28,12 @@ type InterfaceConfig struct {
 	Outside                 udp.Conn
 	Inside                  overlay.Device
 	pki                     *PKI
-	Cipher             string
 	Firewall                *Firewall
 	ServeDns                bool
 	HandshakeManager        *HandshakeManager
 	lightHouse              *LightHouse
-	connectionManager  *connectionManager
+	checkInterval           time.Duration
+	pendingDeletionInterval time.Duration
 	DropLocalBroadcast      bool
 	DropMulticast           bool
 	routines                int
@@ -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

@@ -21,11 +21,9 @@ import (
 	"github.com/slackhq/nebula/header"
 	"github.com/slackhq/nebula/udp"
 	"github.com/slackhq/nebula/util"
-	"github.com/vishvananda/netlink"
 )
 
 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
@@ -230,7 +228,7 @@ func (lh *LightHouse) reload(c *config.C, initial bool) error {
 				lh.updateCancel()
 			}
 
-			lh.StartUpdateWorkerNetlink()
+			lh.StartUpdateWorker()
 		}
 	}
 
@@ -521,15 +519,11 @@ 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.
-	staticList := lh.GetStaticHostList()
-	for _, addr := range allVpnAddrs {
-		if _, ok := staticList[addr]; ok {
+	// First we check the static mapping
+	// and do nothing if it is there
+	if _, ok := lh.GetStaticHostList()[allVpnAddrs[0]]; ok {
 		return
 	}
-	}
-
-	// None of the VpnAddrs were present. Now we can do the deletes.
 	lh.Lock()
 	rm, ok := lh.addrMap[allVpnAddrs[0]]
 	if ok {
@@ -633,24 +627,16 @@ 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
-	for i, addr := range allAddrs {
-		am, ok := lh.addrMap[addr]
-		if ok {
-			if i != 0 {
-				lh.addrMap[allAddrs[0]] = am
-			}
-			return am
-		}
-	}
-
-	//TODO lighthouse.remote_allow_ranges is almost certainly broken in a multiple-address-per-cert scenario
-	am := NewRemoteList(allAddrs, func(a netip.Addr) bool { return lh.shouldAdd(allAddrs[0], a) })
+	am, ok := lh.addrMap[allAddrs[0]]
+	if !ok {
+		am = NewRemoteList(allAddrs, func(a netip.Addr) bool { return lh.shouldAdd(allAddrs[0], a) })
 		for _, addr := range allAddrs {
 			lh.addrMap[addr] = am
 		}
+	}
 	return am
 }
 
@@ -712,10 +698,14 @@ func (lh *LightHouse) unlockedShouldAddV6(vpnAddr netip.Addr, to *V6AddrPort) bo
 }
 
 func (lh *LightHouse) IsLighthouseAddr(vpnAddr netip.Addr) bool {
-	_, ok := lh.GetLighthouses()[vpnAddr]
-	return ok
+	if _, ok := lh.GetLighthouses()[vpnAddr]; ok {
+		return true
+	}
+	return false
 }
 
+// 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 _, a := range vpnAddr {
@@ -846,65 +836,10 @@ func (lh *LightHouse) StartUpdateWorker() {
 	}()
 }
 
-func (lh *LightHouse) StartUpdateWorkerNetlink() {
-	if lh.amLighthouse {
-		return
-	}
-
-	addrChan := make(chan netlink.AddrUpdate)
-	doneChan := make(chan struct{})
-
-	err := netlink.AddrSubscribe(addrChan, doneChan)
-	if err != nil {
-		lh.l.WithError(err).Error("Failed to subscribe to address updates")
-		return
-	}
-
-	updateCtx, cancel := context.WithCancel(lh.ctx)
-	lh.updateCancel = cancel
-
-	go func() {
-		defer func() {
-			x := struct{}{}
-			doneChan <- x
-		}()
-
-		lh.SendUpdate()
-
-		for {
-			select {
-			case <-updateCtx.Done():
-				return
-			case a := <-addrChan:
-				addr, ok := netip.AddrFromSlice(a.LinkAddress.IP)
-				if !ok {
-					continue
-				}
-				ones, _ := a.LinkAddress.Mask.Size()
-				pfx := netip.PrefixFrom(addr, ones)
-				lh.l.WithFields(logrus.Fields{
-					"LinkAddress": pfx,
-					"LinkIndex":   a.LinkIndex,
-					"Flags":       a.Flags,
-					"Scope":       a.Scope,
-					"NewAddr":     a.NewAddr,
-					"ValidLft":    a.ValidLft,
-					"PreferedLft": a.PreferedLft,
-				}).Warn("Received address update")
-
-				shouldSend := a.PreferedLft != 0 //example criteria
-				if shouldSend {
-					lh.SendUpdate()
-				}
-			}
-		}
-	}()
-}
-
 func (lh *LightHouse) SendUpdate() {
 	var v4 []*V4AddrPort
 	var v6 []*V6AddrPort
-	lh.l.Warn("sending lh update!")
+
 	for _, e := range lh.GetAdvertiseAddrs() {
 		if e.Addr().Is4() {
 			v4 = append(v4, netAddrToProtoV4AddrPort(e.Addr(), e.Port()))
@@ -1113,8 +1048,17 @@ func (lhh *LightHouseHandler) handleHostQuery(n *NebulaMeta, fromVpnAddrs []neti
 		return
 	}
 
-	queryVpnAddr, useVersion, err := n.Details.GetVpnAddrAndVersion()
-	if err != nil {
+	useVersion := cert.Version1
+	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).Debugln("Dropping malformed HostQuery")
 		}
@@ -1172,9 +1116,8 @@ func (lhh *LightHouseHandler) sendHostPunchNotification(n *NebulaMeta, fromVpnAd
 			if ok {
 				whereToPunch = newDest
 			} else {
-				if lhh.l.Level >= logrus.DebugLevel {
-					lhh.l.WithField("to", crt.Networks()).Debugln("unable to punch to host, no addresses in common")
-				}
+				//TODO: CERT-V2 this means the destination will have no addresses in common with the punch-ee
+				//choosing to do nothing for now, but maybe we return an error?
 			}
 		}
 
@@ -1233,17 +1176,19 @@ 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 {
-				lhh.l.WithField("version", v).Debug("unsupported protocol version")
-			}
+			//TODO: CERT-V2 don't panic
+			panic("unsupported version")
 		}
 	}
 }
@@ -1253,16 +1198,18 @@ func (lhh *LightHouseHandler) handleHostQueryReply(n *NebulaMeta, fromVpnAddrs [
 		return
 	}
 
-	certVpnAddr, _, err := n.Details.GetVpnAddrAndVersion()
-	if err != nil {
-		if lhh.l.Level >= logrus.DebugLevel {
-			lhh.l.WithError(err).WithField("vpnAddrs", fromVpnAddrs).Error("dropping malformed HostQueryReply")
-		}
-		return
+	lhh.lh.Lock()
+
+	var certVpnAddr netip.Addr
+	if n.Details.OldVpnAddr != 0 {
+		b := [4]byte{}
+		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()
@@ -1287,13 +1234,24 @@ func (lhh *LightHouseHandler) handleHostUpdateNotification(n *NebulaMeta, fromVp
 		return
 	}
 
-	detailsVpnAddr, useVersion, err := n.Details.GetVpnAddrAndVersion()
-	if err != nil {
+	var detailsVpnAddr netip.Addr
+	useVersion := cert.Version1
+	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 {
+		detailsVpnAddr = protoAddrToNetAddr(n.Details.VpnAddr)
+		useVersion = cert.Version2
+	} else {
 		if lhh.l.Level >= logrus.DebugLevel {
-			lhh.l.WithField("details", n.Details).WithError(err).Debugln("dropping invalid HostUpdateNotification")
+			lhh.l.WithField("details", n.Details).Debugf("dropping invalid HostUpdateNotification")
 		}
+		return
 	}
 
+	//TODO: CERT-V2 hosts with only v2 certs cannot provide their ipv6 addr when contacting the lighthouse via v4?
 	//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) {
@@ -1350,16 +1308,8 @@ func (lhh *LightHouseHandler) handleHostPunchNotification(n *NebulaMeta, fromVpn
 		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
 		}
@@ -1371,31 +1321,48 @@ 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))
 		}()
 	}
 }
@@ -1474,17 +1441,3 @@ 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)
@@ -221,6 +220,9 @@ 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,
 		Inside:                  tun,
@@ -229,8 +231,9 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		Firewall:                fw,
 		ServeDns:                serveDns,
 		HandshakeManager:        handshakeManager,
-		connectionManager:     connManager,
 		lightHouse:              lightHouse,
+		checkInterval:           time.Second * time.Duration(checkInterval),
+		pendingDeletionInterval: time.Second * time.Duration(pendingDeletionInterval),
 		tryPromoteEvery:         c.GetUint32("counters.try_promote", defaultPromoteEvery),
 		reQueryEvery:            c.GetUint32("counters.requery_every_packets", defaultReQueryEvery),
 		reQueryWait:             c.GetDuration("timers.requery_wait_duration", defaultReQueryWait),
@@ -241,6 +244,7 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		version:                 buildVersion,
 		relayManager:            NewRelayManager(ctx, l, hostMap, c),
 		punchy:                  punchy,
+
 		ConntrackCacheTimeout: conntrackCacheTimeout,
 		l:                     l,
 	}
@@ -280,7 +284,7 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 	var dnsStart func()
 	if lightHouse.amLighthouse && serveDns {
 		l.Debugln("Starting dns server")
-		dnsStart = dnsMain(l, pki.getCertState(), hostMap, c)
+		dnsStart = dnsMain(ctx, l, pki.getCertState(), hostMap, c)
 	}
 
 	return &Control{
@@ -291,7 +295,6 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 		sshStart,
 		statsStart,
 		dnsStart,
-		lightHouse.StartUpdateWorkerNetlink,
-		connManager.Start,
+		lightHouse.StartUpdateWorker,
 	}, 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 ci == nil {
+	// If connectionstate exists and the replay protector allows, process packet
+	// 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
 		}
-	// 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
@@ -314,11 +312,12 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 	offset := ipv6.HeaderLen // Start at the end of the ipv6 header
 	next := 0
 	for {
-		if protoAt >= dataLen {
+		if dataLen < offset {
 			break
 		}
-		proto := layers.IPProtocol(data[protoAt])
 
+		proto := layers.IPProtocol(data[protoAt])
+		//fmt.Println(proto, protoAt)
 		switch proto {
 		case layers.IPProtocolICMPv6, layers.IPProtocolESP, layers.IPProtocolNoNextHeader:
 			fp.Protocol = uint8(proto)
@@ -366,7 +365,7 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 
 		case layers.IPProtocolAH:
 			// Auth headers, used by IPSec, have a different meaning for header length
-			if dataLen <= offset+1 {
+			if dataLen < offset+1 {
 				break
 			}
 
@@ -374,7 +373,7 @@ func parseV6(data []byte, incoming bool, fp *firewall.Packet) error {
 
 		default:
 			// Normal ipv6 header length processing
-			if dataLen <= offset+1 {
+			if dataLen < offset+1 {
 				break
 			}
 
@@ -500,7 +499,7 @@ 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 +538,10 @@ 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

outside_test.go

@@ -117,45 +117,6 @@ func Test_newPacket_v6(t *testing.T) {
 	err = newPacket(buffer.Bytes(), true, p)
 	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
 
-	// A v6 packet with a hop-by-hop extension
-	// ICMPv6 Payload (Echo Request)
-	icmpLayer := layers.ICMPv6{
-		TypeCode: layers.ICMPv6TypeEchoRequest,
-	}
-	// Hop-by-Hop Extension Header
-	hopOption := layers.IPv6HopByHopOption{}
-	hopOption.OptionData = []byte{0, 0, 0, 0}
-	hopByHop := layers.IPv6HopByHop{}
-	hopByHop.Options = append(hopByHop.Options, &hopOption)
-
-	ip = layers.IPv6{
-		Version:    6,
-		HopLimit:   128,
-		NextHeader: layers.IPProtocolIPv6Destination,
-		SrcIP:      net.IPv6linklocalallrouters,
-		DstIP:      net.IPv6linklocalallnodes,
-	}
-
-	buffer.Clear()
-	err = gopacket.SerializeLayers(buffer, gopacket.SerializeOptions{
-		ComputeChecksums: false,
-		FixLengths:       true,
-	}, &ip, &hopByHop, &icmpLayer)
-	if err != nil {
-		panic(err)
-	}
-	// Ensure buffer length checks during parsing with the next 2 tests.
-
-	// A full IPv6 header and 1 byte in the first extension, but missing
-	// the length byte.
-	err = newPacket(buffer.Bytes()[:41], true, p)
-	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
-
-	// A full IPv6 header plus 1 full extension, but only 1 byte of the
-	// next layer, missing length byte
-	err = newPacket(buffer.Bytes()[:49], true, p)
-	require.ErrorIs(t, err, ErrIPv6CouldNotFindPayload)
-
 	// A good ICMP packet
 	ip = layers.IPv6{
 		Version:    6,
@@ -327,10 +288,6 @@ func Test_newPacket_v6(t *testing.T) {
 	assert.Equal(t, uint16(22), p.LocalPort)
 	assert.False(t, p.Fragment)
 
-	// Ensure buffer bounds checking during processing
-	err = newPacket(b[:41], true, p)
-	require.ErrorIs(t, err, ErrIPv6PacketTooShort)
-
 	// Invalid AH header
 	b = buffer.Bytes()
 	err = newPacket(b, true, p)

overlay/tun_linux.go

@@ -38,7 +38,6 @@ type tun struct {
 	routeTree       atomic.Pointer[bart.Table[routing.Gateways]]
 	routeChan       chan struct{}
 	useSystemRoutes bool
-	useSystemRoutesBufferSize int
 
 	l *logrus.Logger
 }
@@ -130,7 +129,6 @@ func newTunGeneric(c *config.C, l *logrus.Logger, file *os.File, vpnNetworks []n
 		vpnNetworks:     vpnNetworks,
 		TXQueueLen:      c.GetInt("tun.tx_queue", 500),
 		useSystemRoutes: c.GetBool("tun.use_system_route_table", false),
-		useSystemRoutesBufferSize: c.GetInt("tun.use_system_route_table_buffer_size", 0),
 		l:               l,
 	}
 
@@ -293,6 +291,7 @@ 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 +359,6 @@ 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
 	}
@@ -537,13 +531,7 @@ func (t *tun) watchRoutes() {
 	rch := make(chan netlink.RouteUpdate)
 	doneChan := make(chan struct{})
 
-	netlinkOptions := netlink.RouteSubscribeOptions{
-		ReceiveBufferSize:      t.useSystemRoutesBufferSize,
-		ReceiveBufferForceSize: t.useSystemRoutesBufferSize != 0,
-		ErrorCallback:          func(e error) { t.l.WithError(e).Errorf("netlink error") },
-	}
-
-	if err := netlink.RouteSubscribeWithOptions(rch, doneChan, netlinkOptions); err != nil {
+	if err := netlink.RouteSubscribe(rch, doneChan); err != nil {
 		t.l.WithError(err).Errorf("failed to subscribe to system route changes")
 		return
 	}
@@ -553,14 +541,8 @@ func (t *tun) watchRoutes() {
 	go func() {
 		for {
 			select {
-			case r, ok := <-rch:
-				if ok {
+			case r := <-rch:
 				t.updateRoutes(r)
-				} else {
-					// may be should do something here as
-					// netlink stops sending updates
-					return
-				}
 			case <-doneChan:
 				// netlink.RouteSubscriber will close the rch for us
 				return
@@ -642,11 +624,6 @@ 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")

pki.go

@@ -173,6 +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 {
@@ -358,9 +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] {
-			return nil, util.NewContextualError("v1 and v2 networks are not the same", nil, nil)
-		}
+		//TODO: CERT-V2 make sure v2 has v1s address
 
 		cs.initiatingVersion = dv
 	}

service/service.go

@@ -9,10 +9,13 @@ import (
 	"math"
 	"net"
 	"net/netip"
+	"os"
 	"strings"
 	"sync"
 
+	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula"
+	"github.com/slackhq/nebula/config"
 	"github.com/slackhq/nebula/overlay"
 	"golang.org/x/sync/errgroup"
 	"gvisor.dev/gvisor/pkg/buffer"
@@ -43,7 +46,14 @@ type Service struct {
 	}
 }
 
-func New(control *nebula.Control) (*Service, error) {
+func New(config *config.C) (*Service, error) {
+	logger := logrus.New()
+	logger.Out = os.Stdout
+
+	control, err := nebula.Main(config, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
+	if err != nil {
+		return nil, err
+	}
 	control.Start()
 
 	ctx := control.Context()

service/service_test.go

@@ -5,17 +5,13 @@ import (
 	"context"
 	"errors"
 	"net/netip"
-	"os"
 	"testing"
 	"time"
 
 	"dario.cat/mergo"
-	"github.com/sirupsen/logrus"
-	"github.com/slackhq/nebula"
 	"github.com/slackhq/nebula/cert"
 	"github.com/slackhq/nebula/cert_test"
 	"github.com/slackhq/nebula/config"
-	"github.com/slackhq/nebula/overlay"
 	"golang.org/x/sync/errgroup"
 	"gopkg.in/yaml.v3"
 )
@@ -75,15 +71,7 @@ func newSimpleService(caCrt cert.Certificate, caKey []byte, name string, udpIp n
 		panic(err)
 	}
 
-	logger := logrus.New()
-	logger.Out = os.Stdout
-
-	control, err := nebula.Main(&c, false, "custom-app", logger, overlay.NewUserDeviceFromConfig)
-	if err != nil {
-		panic(err)
-	}
-
-	s, err := New(control)
+	s, err := New(&c)
 	if err != nil {
 		panic(err)
 	}

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)
-	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)
+	return NewGenericListener(l, ip, port, multi, batch)
 }
 
 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
-//go:noescape
-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)
+func (u *GenericConn) Rebind() error {
+	rc, err := u.UDPConn.SyscallConn()
 	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)
+		return err
 	}
 
+	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")
 		}
-
-	return nil
+	})
 }

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

@@ -92,25 +92,6 @@ 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
@@ -141,13 +122,9 @@ 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")
 			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])
 	}