return err

This reverts commit fa098c551a23d055f7ba040f14991b21dd72868f.

cmd/nebula-monitor/nebula-monitor.go

@@ -1,39 +0,0 @@
-package main
-
-import (
-	"fmt"
-	"io"
-	"log"
-	"net/http"
-)
-
-func handlePost(w http.ResponseWriter, r *http.Request) {
-	if r.Method != http.MethodPost {
-		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
-		return
-	}
-
-	// Read the body
-	body, err := io.ReadAll(r.Body)
-	if err != nil {
-		http.Error(w, "Error reading body", http.StatusInternalServerError)
-		return
-	}
-	defer r.Body.Close()
-
-	// Print to console
-	//fmt.Printf("Path: %s\n", r.URL.Path)
-	//fmt.Printf("Headers: %v\n", r.Header)
-	fmt.Printf("%s\n", string(body))
-
-	// Send response
-	w.WriteHeader(http.StatusOK)
-	w.Write([]byte(""))
-}
-
-func main() {
-	http.HandleFunc("/", handlePost)
-
-	fmt.Println("Server starting on :8080")
-	log.Fatal(http.ListenAndServe(":8080", nil))
-}

cmd/nebula-service/main.go

@@ -78,8 +78,16 @@ func main() {
 	}
 
 	if !*configTest {
-		ctrl.Start()
-		ctrl.ShutdownBlock()
+		wait, err := ctrl.Start()
+		if err != nil {
+			util.LogWithContextIfNeeded("Error while running", err, l)
+			os.Exit(1)
+		}
+
+		go ctrl.ShutdownBlock()
+		wait()
+
+		l.Info("Goodbye")
 	}
 
 	os.Exit(0)

cmd/nebula/main.go

@@ -3,6 +3,9 @@ package main
 import (
 	"flag"
 	"fmt"
+	"log"
+	"net/http"
+	_ "net/http/pprof"
 	"os"
 	"runtime/debug"
 	"strings"
@@ -71,10 +74,22 @@ func main() {
 		os.Exit(1)
 	}
 
+	go func() {
+		log.Println(http.ListenAndServe("0.0.0.0:6060", nil))
+	}()
+
 	if !*configTest {
-		ctrl.Start()
+		wait, err := ctrl.Start()
+		if err != nil {
+			util.LogWithContextIfNeeded("Error while running", err, l)
+			os.Exit(1)
+		}
+
+		go ctrl.ShutdownBlock()
 		notifyReady(l)
-		ctrl.ShutdownBlock()
+		wait()
+
+		l.Info("Goodbye")
 	}
 
 	os.Exit(0)

control.go

@@ -1,23 +1,30 @@
 package nebula
 
 import (
-	"bytes"
 	"context"
-	"encoding/json"
-	"io"
-	"net/http"
+	"errors"
 	"net/netip"
 	"os"
 	"os/signal"
+	"sync"
 	"syscall"
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/cert"
-	"github.com/slackhq/nebula/firewall"
 	"github.com/slackhq/nebula/header"
 	"github.com/slackhq/nebula/overlay"
 )
 
+type RunState int
+
+const (
+	Stopped  RunState = 0 // The control has yet to be started
+	Started  RunState = 1 // The control has been started
+	Stopping RunState = 2 // The control is stopping
+)
+
+var ErrAlreadyStarted = errors.New("nebula is already started")
+
 // Every interaction here needs to take extra care to copy memory and not return or use arguments "as is" when touching
 // core. This means copying IP objects, slices, de-referencing pointers and taking the actual value, etc
 
@@ -31,6 +38,9 @@ type controlHostLister interface {
 }
 
 type Control struct {
+	stateLock sync.Mutex
+	state     RunState
+
 	f                      *Interface
 	l                      *logrus.Logger
 	ctx                    context.Context
@@ -54,10 +64,21 @@ type ControlHostInfo struct {
 	CurrentRelaysThroughMe []netip.Addr     `json:"currentRelaysThroughMe"`
 }
 
-// Start actually runs nebula, this is a nonblocking call. To block use Control.ShutdownBlock()
-func (c *Control) Start() {
+// Start actually runs nebula, this is a nonblocking call.
+// The returned function can be used to wait for nebula to fully stop.
+func (c *Control) Start() (func(), error) {
+	c.stateLock.Lock()
+	if c.state != Stopped {
+		c.stateLock.Unlock()
+		return nil, ErrAlreadyStarted
+	}
+
 	// Activate the interface
-	c.f.activate()
+	err := c.f.activate()
+	if err != nil {
+		c.stateLock.Unlock()
+		return nil, err
+	}
 
 	// Call all the delayed funcs that waited patiently for the interface to be created.
 	if c.sshStart != nil {
@@ -76,57 +97,34 @@ func (c *Control) Start() {
 		c.lighthouseStart()
 	}
 
-	go c.firewallEventSender(c.ctx)
-
 	// Start reading packets.
-	c.f.run()
+	c.state = Started
+	c.stateLock.Unlock()
+	return c.f.run()
 }
 
-func (c *Control) firewallEventSender(ctx context.Context) {
-	events := make([]firewall.Event, 1, 100) //todo configurable
-	for {
-		select {
-		//todo exceptionally lazy
-		case e := <-c.f.events:
-			events = append(events, e)
-			for { //slurp up everything into one big post
-				x := false
-				select {
-				case e = <-c.f.events:
-					events = append(events, e)
-				default:
-					x = true
-				}
-				if x {
-					break
-				}
-			}
-			if len(events) >= 5 {
-				out, err := json.Marshal(&events)
-				if err != nil {
-					c.l.WithError(err).Errorf("failed to marshal event")
-				} else {
-					r := io.Reader(bytes.NewBuffer(out))
-					_, err = http.Post("http://127.0.0.1:8080/nebula/event", "application/json", r)
-					if err != nil {
-						c.l.WithError(err).Errorf("failed to post event")
-					}
-				}
-				events = events[:0]
-			}
-
-		case <-ctx.Done():
-			return
-		}
-	}
+func (c *Control) State() RunState {
+	c.stateLock.Lock()
+	defer c.stateLock.Unlock()
+	return c.state
 }
 
 func (c *Control) Context() context.Context {
 	return c.ctx
 }
 
-// Stop signals nebula to shutdown and close all tunnels, returns after the shutdown is complete
+// Stop is a non-blocking call that signals nebula to close all tunnels and shut down
 func (c *Control) Stop() {
+	c.stateLock.Lock()
+	if c.state != Started {
+		c.stateLock.Unlock()
+		// We are stopping or stopped already
+		return
+	}
+
+	c.state = Stopping
+	c.stateLock.Unlock()
+
 	// Stop the handshakeManager (and other services), to prevent new tunnels from
 	// being created while we're shutting them all down.
 	c.cancel()
@@ -135,7 +133,7 @@ func (c *Control) Stop() {
 	if err := c.f.Close(); err != nil {
 		c.l.WithError(err).Error("Close interface failed")
 	}
-	c.l.Info("Goodbye")
+	c.state = Stopped
 }
 
 // ShutdownBlock will listen for and block on term and interrupt signals, calling Control.Stop() once signalled

firewall.go

@@ -45,8 +45,6 @@ type Firewall struct {
 
 	InSendReject  bool
 	OutSendReject bool
-	InLogDrop     bool
-	OutLogDrop    bool
 
 	//TODO: we should have many more options for TCP, an option for ICMP, and mimic the kernel a bit better
 	// https://www.kernel.org/doc/Documentation/networking/nf_conntrack-sysctl.txt
@@ -218,7 +216,6 @@ func NewFirewallFromConfig(l *logrus.Logger, cs *CertState, c *config.C) (*Firew
 	switch inboundAction {
 	case "reject":
 		fw.InSendReject = true
-		fw.InLogDrop = true //todo
 	case "drop":
 		fw.InSendReject = false
 	default:
@@ -230,7 +227,6 @@ func NewFirewallFromConfig(l *logrus.Logger, cs *CertState, c *config.C) (*Firew
 	switch outboundAction {
 	case "reject":
 		fw.OutSendReject = true
-		fw.OutLogDrop = true //todo
 	case "drop":
 		fw.OutSendReject = false
 	default:
@@ -405,11 +401,9 @@ var ErrInvalidRemoteIP = errors.New("remote address is not in remote certificate
 var ErrInvalidLocalIP = errors.New("local address is not in list of handled local addresses")
 var ErrNoMatchingRule = errors.New("no matching rule in firewall table")
 
-type DropHandler func(fp firewall.Packet, incoming bool, h *HostInfo, err error)
-
 // Drop returns an error if the packet should be dropped, explaining why. It
 // returns nil if the packet should not be dropped.
-func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache, onDrop DropHandler) error {
+func (f *Firewall) Drop(fp firewall.Packet, incoming bool, h *HostInfo, caPool *cert.CAPool, localCache firewall.ConntrackCache) error {
 	// Check if we spoke to this tuple, if we did then allow this packet
 	if f.inConns(fp, h, caPool, localCache) {
 		return nil

firewall/event.go

@@ -1,15 +0,0 @@
-package firewall
-
-import (
-	"net/netip"
-	"time"
-)
-
-type Event struct {
-	Packet Packet         `json:"packet,omitempty"`
-	At     time.Time      `json:"at,omitempty"`
-	Remote netip.AddrPort `json:"remote,omitempty"`
-	//todo cert info?
-	//todo connection indexes?
-	//todo underlay info would actually be amazing, for inbounds
-}

firewall/packet.go

@@ -44,7 +44,7 @@ func (fp Packet) MarshalJSON() ([]byte, error) {
 	switch fp.Protocol {
 	case ProtoTCP:
 		proto = "tcp"
-	case ProtoICMP, ProtoICMPv6:
+	case ProtoICMP:
 		proto = "icmp"
 	case ProtoUDP:
 		proto = "udp"

firewall_test.go

@@ -212,44 +212,44 @@ func TestFirewall_Drop(t *testing.T) {
 	cp := cert.NewCAPool()
 
 	// Drop outbound
-	assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp, nil, nil))
+	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, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 	// Allow outbound because conntrack
-	require.NoError(t, fw.Drop(p, false, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, false, &h, cp, nil))
 
 	// test remote mismatch
 	oldRemote := p.RemoteAddr
 	p.RemoteAddr = netip.MustParseAddr("1.2.3.10")
-	assert.Equal(t, fw.Drop(p, false, &h, cp, nil, nil), ErrInvalidRemoteIP)
+	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"}, "", "", "", "", "signer-shasum"))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum-bad"))
-	assert.Equal(t, fw.Drop(p, true, &h, cp, nil, nil), ErrNoMatchingRule)
+	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"}, "", "", "", "", "signer-shasum-bad"))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum"))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	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"}, "", "", "", "ca-good", ""))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good-bad", ""))
-	assert.Equal(t, fw.Drop(p, true, &h, cp, nil, nil), ErrNoMatchingRule)
+	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"}, "", "", "", "ca-good-bad", ""))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good", ""))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 }
 
 func TestFirewall_DropV6(t *testing.T) {
@@ -291,44 +291,44 @@ func TestFirewall_DropV6(t *testing.T) {
 	cp := cert.NewCAPool()
 
 	// Drop outbound
-	assert.Equal(t, ErrNoMatchingRule, fw.Drop(p, false, &h, cp, nil, nil))
+	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, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 	// Allow outbound because conntrack
-	require.NoError(t, fw.Drop(p, false, &h, cp, nil, nil))
+	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, nil), ErrInvalidRemoteIP)
+	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"}, "", "", "", "", "signer-shasum"))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum-bad"))
-	assert.Equal(t, fw.Drop(p, true, &h, cp, nil, nil), ErrNoMatchingRule)
+	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"}, "", "", "", "", "signer-shasum-bad"))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "", "signer-shasum"))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	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"}, "", "", "", "ca-good", ""))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good-bad", ""))
-	assert.Equal(t, fw.Drop(p, true, &h, cp, nil, nil), ErrNoMatchingRule)
+	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"}, "", "", "", "ca-good-bad", ""))
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 0, 0, []string{"default-group"}, "", "", "", "ca-good", ""))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 }
 
 func BenchmarkFirewallTable_match(b *testing.B) {
@@ -536,10 +536,10 @@ func TestFirewall_Drop2(t *testing.T) {
 	cp := cert.NewCAPool()
 
 	// h1/c1 lacks the proper groups
-	require.ErrorIs(t, fw.Drop(p, true, &h1, cp, nil, nil), ErrNoMatchingRule)
+	require.ErrorIs(t, fw.Drop(p, true, &h1, cp, nil), ErrNoMatchingRule)
 	// c has the proper groups
 	resetConntrack(fw)
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 }
 
 func TestFirewall_Drop3(t *testing.T) {
@@ -617,18 +617,18 @@ func TestFirewall_Drop3(t *testing.T) {
 	cp := cert.NewCAPool()
 
 	// c1 should pass because host match
-	require.NoError(t, fw.Drop(p, true, &h1, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h1, cp, nil))
 	// c2 should pass because ca sha match
 	resetConntrack(fw)
-	require.NoError(t, fw.Drop(p, true, &h2, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h2, cp, nil))
 	// c3 should fail because no match
 	resetConntrack(fw)
-	assert.Equal(t, fw.Drop(p, true, &h3, cp, nil, nil), ErrNoMatchingRule)
+	assert.Equal(t, fw.Drop(p, true, &h3, cp, nil), ErrNoMatchingRule)
 
 	// Test a remote address match
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
 	require.NoError(t, fw.AddRule(true, firewall.ProtoAny, 1, 1, []string{}, "", "1.2.3.4/24", "", "", ""))
-	require.NoError(t, fw.Drop(p, true, &h1, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h1, cp, nil))
 }
 
 func TestFirewall_Drop3V6(t *testing.T) {
@@ -666,7 +666,7 @@ func TestFirewall_Drop3V6(t *testing.T) {
 	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{}, "", "fd12::34/120", "", "", ""))
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 }
 
 func TestFirewall_DropConntrackReload(t *testing.T) {
@@ -708,12 +708,12 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 	cp := cert.NewCAPool()
 
 	// Drop outbound
-	assert.Equal(t, fw.Drop(p, false, &h, cp, nil, nil), ErrNoMatchingRule)
+	assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrNoMatchingRule)
 	// Allow inbound
 	resetConntrack(fw)
-	require.NoError(t, fw.Drop(p, true, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, true, &h, cp, nil))
 	// Allow outbound because conntrack
-	require.NoError(t, fw.Drop(p, false, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, false, &h, cp, nil))
 
 	oldFw := fw
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
@@ -722,7 +722,7 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 	fw.rulesVersion = oldFw.rulesVersion + 1
 
 	// Allow outbound because conntrack and new rules allow port 10
-	require.NoError(t, fw.Drop(p, false, &h, cp, nil, nil))
+	require.NoError(t, fw.Drop(p, false, &h, cp, nil))
 
 	oldFw = fw
 	fw = NewFirewall(l, time.Second, time.Minute, time.Hour, c.Certificate)
@@ -731,7 +731,7 @@ func TestFirewall_DropConntrackReload(t *testing.T) {
 	fw.rulesVersion = oldFw.rulesVersion + 1
 
 	// Drop outbound because conntrack doesn't match new ruleset
-	assert.Equal(t, fw.Drop(p, false, &h, cp, nil, nil), ErrNoMatchingRule)
+	assert.Equal(t, fw.Drop(p, false, &h, cp, nil), ErrNoMatchingRule)
 }
 
 func TestFirewall_DropIPSpoofing(t *testing.T) {
@@ -777,7 +777,7 @@ func TestFirewall_DropIPSpoofing(t *testing.T) {
 		Protocol:   firewall.ProtoUDP,
 		Fragment:   false,
 	}
-	assert.Equal(t, fw.Drop(p, true, &h1, cp, nil, nil), ErrInvalidRemoteIP)
+	assert.Equal(t, fw.Drop(p, true, &h1, cp, nil), ErrInvalidRemoteIP)
 }
 
 func BenchmarkLookup(b *testing.B) {
@@ -1184,7 +1184,7 @@ func (c *testcase) Test(t *testing.T, fw *Firewall) {
 	t.Helper()
 	cp := cert.NewCAPool()
 	resetConntrack(fw)
-	err := fw.Drop(c.p, true, c.h, cp, nil, nil)
+	err := fw.Drop(c.p, true, c.h, cp, nil)
 	if c.err == nil {
 		require.NoError(t, err, "failed to not drop remote address %s", c.p.RemoteAddr)
 	} else {

inside.go

@@ -2,7 +2,6 @@ package nebula
 
 import (
 	"net/netip"
-	"time"
 
 	"github.com/sirupsen/logrus"
 	"github.com/slackhq/nebula/firewall"
@@ -54,7 +53,7 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 	})
 
 	if hostinfo == nil {
-		f.rejectInside(fwPacket, packet, out, q)
+		f.rejectInside(packet, out, q)
 		if f.l.Level >= logrus.DebugLevel {
 			f.l.WithField("vpnAddr", fwPacket.RemoteAddr).
 				WithField("fwPacket", fwPacket).
@@ -67,12 +66,12 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 		return
 	}
 
-	dropReason := f.firewall.Drop(*fwPacket, false, hostinfo, f.pki.GetCAPool(), localCache, nil)
+	dropReason := f.firewall.Drop(*fwPacket, false, hostinfo, f.pki.GetCAPool(), localCache)
 	if dropReason == nil {
 		f.sendNoMetrics(header.Message, 0, hostinfo.ConnectionState, hostinfo, netip.AddrPort{}, packet, nb, out, q)
 
 	} else {
-		f.rejectInside(fwPacket, packet, out, q)
+		f.rejectInside(packet, out, q)
 		if f.l.Level >= logrus.DebugLevel {
 			hostinfo.logger(f.l).
 				WithField("fwPacket", fwPacket).
@@ -82,10 +81,7 @@ func (f *Interface) consumeInsidePacket(packet []byte, fwPacket *firewall.Packet
 	}
 }
 
-func (f *Interface) rejectInside(fp *firewall.Packet, packet []byte, out []byte, q int) {
-	if f.firewall.InLogDrop {
-		//todo
-	}
+func (f *Interface) rejectInside(packet []byte, out []byte, q int) {
 	if !f.firewall.InSendReject {
 		return
 	}
@@ -101,16 +97,7 @@ func (f *Interface) rejectInside(fp *firewall.Packet, packet []byte, out []byte,
 	}
 }
 
-func (f *Interface) rejectOutside(fp *firewall.Packet, packet []byte, ci *ConnectionState, hostinfo *HostInfo, nb, out []byte, q int) {
-	if f.firewall.OutLogDrop {
-		e := firewall.Event{
-			Packet: *fp,
-			At:     time.Now(),
-			Remote: hostinfo.remote, //todo if nil capture relay?
-		}
-		f.events <- e
-	}
-
+func (f *Interface) rejectOutside(packet []byte, ci *ConnectionState, hostinfo *HostInfo, nb, out []byte, q int) {
 	if !f.firewall.OutSendReject {
 		return
 	}
@@ -231,7 +218,7 @@ func (f *Interface) sendMessageNow(t header.MessageType, st header.MessageSubTyp
 	}
 
 	// check if packet is in outbound fw rules
-	dropReason := f.firewall.Drop(*fp, false, hostinfo, f.pki.GetCAPool(), nil, nil)
+	dropReason := f.firewall.Drop(*fp, false, hostinfo, f.pki.GetCAPool(), nil)
 	if dropReason != nil {
 		if f.l.Level >= logrus.DebugLevel {
 			f.l.WithField("fwPacket", fp).

interface.go

@@ -6,8 +6,8 @@ import (
 	"fmt"
 	"io"
 	"net/netip"
-	"os"
 	"runtime"
+	"sync"
 	"sync/atomic"
 	"time"
 
@@ -88,13 +88,12 @@ type Interface struct {
 
 	writers []udp.Conn
 	readers []io.ReadWriteCloser
+	wg      sync.WaitGroup
 
 	metricHandshakes    metrics.Histogram
 	messageMetrics      *MessageMetrics
 	cachedPacketMetrics *cachedPacketMetrics
 
-	events chan firewall.Event
-
 	l *logrus.Logger
 }
 
@@ -196,7 +195,6 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 			sent:    metrics.GetOrRegisterCounter("hostinfo.cached_packets.sent", nil),
 			dropped: metrics.GetOrRegisterCounter("hostinfo.cached_packets.dropped", nil),
 		},
-		events: make(chan firewall.Event, 100), //todo configurable
 
 		l: c.l,
 	}
@@ -213,7 +211,7 @@ func NewInterface(ctx context.Context, c *InterfaceConfig) (*Interface, error) {
 // activate creates the interface on the host. After the interface is created, any
 // other services that want to bind listeners to its IP may do so successfully. However,
 // the interface isn't going to process anything until run() is called.
-func (f *Interface) activate() {
+func (f *Interface) activate() error {
 	// actually turn on tun dev
 
 	addr, err := f.outside.LocalAddr()
@@ -241,33 +239,38 @@ func (f *Interface) activate() {
 		if i > 0 {
 			reader, err = f.inside.NewMultiQueueReader()
 			if err != nil {
-				f.l.Fatal(err)
+				return err
 			}
 		}
 		f.readers[i] = reader
 	}
 
-	if err := f.inside.Activate(); err != nil {
+	if err = f.inside.Activate(); err != nil {
 		f.inside.Close()
-		f.l.Fatal(err)
+		return err
 	}
+
+	return nil
 }
 
-func (f *Interface) run() {
+func (f *Interface) run() (func(), error) {
 	// Launch n queues to read packets from udp
 	for i := 0; i < f.routines; i++ {
+		f.wg.Add(1)
 		go f.listenOut(i)
 	}
 
 	// Launch n queues to read packets from tun dev
 	for i := 0; i < f.routines; i++ {
+		f.wg.Add(1)
 		go f.listenIn(f.readers[i], i)
 	}
+
+	return f.wg.Wait, nil
 }
 
 func (f *Interface) listenOut(i int) {
 	runtime.LockOSThread()
-
 	var li udp.Conn
 	if i > 0 {
 		li = f.writers[i]
@@ -282,14 +285,21 @@ func (f *Interface) listenOut(i int) {
 	fwPacket := &firewall.Packet{}
 	nb := make([]byte, 12, 12)
 
-	li.ListenOut(func(fromUdpAddr netip.AddrPort, payload []byte) {
+	err := li.ListenOut(func(fromUdpAddr netip.AddrPort, payload []byte) {
 		f.readOutsidePackets(ViaSender{UdpAddr: fromUdpAddr}, plaintext[:0], payload, h, fwPacket, lhh, nb, i, ctCache.Get(f.l))
 	})
+
+	if err != nil && !f.closed.Load() {
+		f.l.WithError(err).Error("Error while reading packet inbound packet, closing")
+		//TODO: Trigger Control to close
+	}
+
+	f.l.Debugf("underlay reader %v is done", i)
+	f.wg.Done()
 }
 
 func (f *Interface) listenIn(reader io.ReadWriteCloser, i int) {
 	runtime.LockOSThread()
-
 	packet := make([]byte, mtu)
 	out := make([]byte, mtu)
 	fwPacket := &firewall.Packet{}
@@ -300,17 +310,18 @@ func (f *Interface) listenIn(reader io.ReadWriteCloser, i int) {
 	for {
 		n, err := reader.Read(packet)
 		if err != nil {
-			if errors.Is(err, os.ErrClosed) && f.closed.Load() {
-				return
+			if !f.closed.Load() {
+				f.l.WithError(err).Error("Error while reading outbound packet, closing")
+				//TODO: Trigger Control to close
 			}
-
-			f.l.WithError(err).Error("Error while reading outbound packet")
-			// This only seems to happen when something fatal happens to the fd, so exit.
-			os.Exit(2)
+			break
 		}
 
 		f.consumeInsidePacket(packet[:n], fwPacket, nb, out, i, conntrackCache.Get(f.l))
 	}
+
+	f.l.Debugf("overlay reader %v is done", i)
+	f.wg.Done()
 }
 
 func (f *Interface) RegisterConfigChangeCallbacks(c *config.C) {
@@ -487,6 +498,7 @@ func (f *Interface) GetCertState() *CertState {
 func (f *Interface) Close() error {
 	f.closed.Store(true)
 
+	// Release the udp readers
 	for _, u := range f.writers {
 		err := u.Close()
 		if err != nil {
@@ -494,6 +506,13 @@ func (f *Interface) Close() error {
 		}
 	}
 
-	// Release the tun device
-	return f.inside.Close()
+	// Release the tun readers
+	for i, u := range f.readers {
+		err := u.Close()
+		if err != nil {
+			f.l.WithError(err).WithField("i", i).Error("Error while closing tun device")
+		}
+	}
+
+	return nil
 }

lighthouse_test.go

@@ -1,7 +1,6 @@
 package nebula
 
 import (
-	"context"
 	"encoding/binary"
 	"fmt"
 	"net/netip"
@@ -42,14 +41,14 @@ func Test_lhStaticMapping(t *testing.T) {
 	c := config.NewC(l)
 	c.Settings["lighthouse"] = map[string]any{"hosts": []any{lh1}}
 	c.Settings["static_host_map"] = map[string]any{lh1: []any{"1.1.1.1:4242"}}
-	_, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	_, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 
 	lh2 := "10.128.0.3"
 	c = config.NewC(l)
 	c.Settings["lighthouse"] = map[string]any{"hosts": []any{lh1, lh2}}
 	c.Settings["static_host_map"] = map[string]any{lh1: []any{"100.1.1.1:4242"}}
-	_, err = NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	_, err = NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	require.EqualError(t, err, "lighthouse 10.128.0.3 does not have a static_host_map entry")
 }
 
@@ -71,7 +70,7 @@ func TestReloadLighthouseInterval(t *testing.T) {
 	}
 
 	c.Settings["static_host_map"] = map[string]any{lh1: []any{"1.1.1.1:4242"}}
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 	lh.ifce = &mockEncWriter{}
 
@@ -99,7 +98,7 @@ func BenchmarkLighthouseHandleRequest(b *testing.B) {
 	}
 
 	c := config.NewC(l)
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(b.Context(), l, c, cs, nil, nil)
 	require.NoError(b, err)
 
 	hAddr := netip.MustParseAddrPort("4.5.6.7:12345")
@@ -202,7 +201,7 @@ func TestLighthouse_Memory(t *testing.T) {
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
 	}
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	lh.ifce = &mockEncWriter{}
 	require.NoError(t, err)
 	lhh := lh.NewRequestHandler()
@@ -288,7 +287,7 @@ func TestLighthouse_reload(t *testing.T) {
 		myVpnNetworksTable: nt,
 	}
 
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 
 	nc := map[string]any{
@@ -523,7 +522,7 @@ func TestLighthouse_Dont_Delete_Static_Hosts(t *testing.T) {
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
 	}
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 	lh.ifce = &mockEncWriter{}
 
@@ -589,7 +588,7 @@ func TestLighthouse_DeletesWork(t *testing.T) {
 		myVpnNetworks:      []netip.Prefix{myVpnNet},
 		myVpnNetworksTable: nt,
 	}
-	lh, err := NewLightHouseFromConfig(context.Background(), l, c, cs, nil, nil)
+	lh, err := NewLightHouseFromConfig(t.Context(), l, c, cs, nil, nil)
 	require.NoError(t, err)
 	lh.ifce = &mockEncWriter{}
 

main.go

@@ -292,15 +292,15 @@ func Main(c *config.C, configTest bool, buildVersion string, logger *logrus.Logg
 	}
 
 	return &Control{
-		ifce,
-		l,
-		ctx,
-		cancel,
-		sshStart,
-		statsStart,
-		dnsStart,
-		lightHouse.StartUpdateWorker,
-		connManager.Start,
+		f:                      ifce,
+		l:                      l,
+		ctx:                    ctx,
+		cancel:                 cancel,
+		sshStart:               sshStart,
+		statsStart:             statsStart,
+		dnsStart:               dnsStart,
+		lighthouseStart:        lightHouse.StartUpdateWorker,
+		connectionManagerStart: connManager.Start,
 	}, nil
 }
 

outside.go

@@ -494,11 +494,11 @@ func (f *Interface) decryptToTun(hostinfo *HostInfo, messageCounter uint64, out
 		return false
 	}
 
-	dropReason := f.firewall.Drop(*fwPacket, true, hostinfo, f.pki.GetCAPool(), localCache, nil)
+	dropReason := f.firewall.Drop(*fwPacket, true, hostinfo, f.pki.GetCAPool(), localCache)
 	if dropReason != nil {
 		// NOTE: We give `packet` as the `out` here since we already decrypted from it and we don't need it anymore
 		// This gives us a buffer to build the reject packet in
-		f.rejectOutside(fwPacket, out, hostinfo.ConnectionState, hostinfo, nb, packet, q)
+		f.rejectOutside(out, hostinfo.ConnectionState, hostinfo, nb, packet, q)
 		if f.l.Level >= logrus.DebugLevel {
 			hostinfo.logger(f.l).WithField("fwPacket", fwPacket).
 				WithField("reason", dropReason).

overlay/tun.go

@@ -12,6 +12,15 @@ import (
 
 const DefaultMTU = 1300
 
+type NameError struct {
+	Name       string
+	Underlying error
+}
+
+func (e *NameError) Error() string {
+	return fmt.Sprintf("could not set tun device name: %s because %s", e.Name, e.Underlying)
+}
+
 // TODO: We may be able to remove routines
 type DeviceFactory func(c *config.C, l *logrus.Logger, vpnNetworks []netip.Prefix, routines int) (Device, error)
 

overlay/tun_freebsd.go

@@ -266,7 +266,7 @@ func newTun(c *config.C, l *logrus.Logger, vpnNetworks []netip.Prefix, _ bool) (
 		}
 
 		// Set the device name
-		ioctl(fd, syscall.SIOCSIFNAME, uintptr(unsafe.Pointer(&ifrr)))
+		_ = ioctl(fd, syscall.SIOCSIFNAME, uintptr(unsafe.Pointer(&ifrr)))
 	}
 
 	t := &tun{

overlay/tun_linux.go

@@ -112,9 +112,13 @@ func newTun(c *config.C, l *logrus.Logger, vpnNetworks []netip.Prefix, multiqueu
 	if multiqueue {
 		req.Flags |= unix.IFF_MULTI_QUEUE
 	}
-	copy(req.Name[:], c.GetString("tun.dev", ""))
+	nameStr := c.GetString("tun.dev", "")
+	copy(req.Name[:], nameStr)
 	if err = ioctl(uintptr(fd), uintptr(unix.TUNSETIFF), uintptr(unsafe.Pointer(&req))); err != nil {
-		return nil, err
+		return nil, &NameError{
+			Name:       nameStr,
+			Underlying: err,
+		}
 	}
 	name := strings.Trim(string(req.Name[:]), "\x00")
 
@@ -713,6 +717,7 @@ func (t *tun) Close() error {
 
 	if t.ioctlFd > 0 {
 		_ = os.NewFile(t.ioctlFd, "ioctlFd").Close()
+		t.ioctlFd = 0
 	}
 
 	return nil

overlay/tun_windows.go

@@ -74,7 +74,10 @@ func newTun(c *config.C, l *logrus.Logger, vpnNetworks []netip.Prefix, _ bool) (
 		l.WithError(err).Debug("Failed to create wintun device, retrying")
 		tunDevice, err = wintun.CreateTUNWithRequestedGUID(deviceName, guid, t.MTU)
 		if err != nil {
-			return nil, fmt.Errorf("create TUN device failed: %w", err)
+			return nil, &NameError{
+				Name:       deviceName,
+				Underlying: fmt.Errorf("create TUN device failed: %w", err),
+			}
 		}
 	}
 	t.tun = tunDevice.(*wintun.NativeTun)

service/service.go

@@ -44,7 +44,10 @@ type Service struct {
 }
 
 func New(control *nebula.Control) (*Service, error) {
-	control.Start()
+	wait, err := control.Start()
+	if err != nil {
+		return nil, err
+	}
 
 	ctx := control.Context()
 	eg, ctx := errgroup.WithContext(ctx)
@@ -141,6 +144,12 @@ func New(control *nebula.Control) (*Service, error) {
 		}
 	})
 
+	// Add the nebula wait function to the group
+	eg.Go(func() error {
+		wait()
+		return nil
+	})
+
 	return &s, nil
 }
 

udp/conn.go

@@ -16,7 +16,7 @@ type EncReader func(
 type Conn interface {
 	Rebind() error
 	LocalAddr() (netip.AddrPort, error)
-	ListenOut(r EncReader)
+	ListenOut(r EncReader) error
 	WriteTo(b []byte, addr netip.AddrPort) error
 	ReloadConfig(c *config.C)
 	SupportsMultipleReaders() bool
@@ -31,8 +31,8 @@ func (NoopConn) Rebind() error {
 func (NoopConn) LocalAddr() (netip.AddrPort, error) {
 	return netip.AddrPort{}, nil
 }
-func (NoopConn) ListenOut(_ EncReader) {
-	return
+func (NoopConn) ListenOut(_ EncReader) error {
+	return nil
 }
 func (NoopConn) SupportsMultipleReaders() bool {
 	return false

udp/udp_darwin.go

@@ -165,7 +165,7 @@ func NewUDPStatsEmitter(udpConns []Conn) func() {
 	return func() {}
 }
 
-func (u *StdConn) ListenOut(r EncReader) {
+func (u *StdConn) ListenOut(r EncReader) error {
 	buffer := make([]byte, MTU)
 
 	for {
@@ -173,8 +173,7 @@ func (u *StdConn) ListenOut(r EncReader) {
 		n, rua, err := u.ReadFromUDPAddrPort(buffer)
 		if err != nil {
 			if errors.Is(err, net.ErrClosed) {
-				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-				return
+				return err
 			}
 
 			u.l.WithError(err).Error("unexpected udp socket receive error")

udp/udp_generic.go

@@ -10,7 +10,6 @@ package udp
 
 import (
 	"context"
-	"errors"
 	"fmt"
 	"net"
 	"net/netip"
@@ -73,7 +72,7 @@ type rawMessage struct {
 	Len uint32
 }
 
-func (u *GenericConn) ListenOut(r EncReader) {
+func (u *GenericConn) ListenOut(r EncReader) error {
 	buffer := make([]byte, MTU)
 
 	var lastRecvErr time.Time
@@ -82,16 +81,7 @@ func (u *GenericConn) ListenOut(r EncReader) {
 		// Just read one packet at a time
 		n, rua, err := u.ReadFromUDPAddrPort(buffer)
 		if err != nil {
-			if errors.Is(err, net.ErrClosed) {
-				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-				return
-			}
-			// Dampen unexpected message warns to once per minute
-			if lastRecvErr.IsZero() || time.Since(lastRecvErr) > time.Minute {
-				lastRecvErr = time.Now()
-				u.l.WithError(err).Warn("unexpected udp socket receive error")
-			}
-			continue
+			return err
 		}
 
 		r(netip.AddrPortFrom(rua.Addr().Unmap(), rua.Port()), buffer[:n])

udp/udp_linux.go

@@ -24,14 +24,6 @@ type StdConn struct {
 	batch int
 }
 
-func maybeIPV4(ip net.IP) (net.IP, bool) {
-	ip4 := ip.To4()
-	if ip4 != nil {
-		return ip4, true
-	}
-	return ip, false
-}
-
 func NewListener(l *logrus.Logger, ip netip.Addr, port int, multi bool, batch int) (Conn, error) {
 	af := unix.AF_INET6
 	if ip.Is4() {
@@ -122,7 +114,7 @@ func (u *StdConn) LocalAddr() (netip.AddrPort, error) {
 	}
 }
 
-func (u *StdConn) ListenOut(r EncReader) {
+func (u *StdConn) ListenOut(r EncReader) error {
 	var ip netip.Addr
 
 	msgs, buffers, names := u.PrepareRawMessages(u.batch)
@@ -134,8 +126,7 @@ func (u *StdConn) ListenOut(r EncReader) {
 	for {
 		n, err := read(msgs)
 		if err != nil {
-			u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-			return
+			return err
 		}
 
 		for i := 0; i < n; i++ {
@@ -310,6 +301,7 @@ func (u *StdConn) getMemInfo(meminfo *[unix.SK_MEMINFO_VARS]uint32) error {
 }
 
 func (u *StdConn) Close() error {
+	_ = syscall.Shutdown(u.sysFd, syscall.SHUT_RDWR)
 	return syscall.Close(u.sysFd)
 }
 

udp/udp_rio_windows.go

@@ -140,7 +140,7 @@ func (u *RIOConn) bind(l *logrus.Logger, sa windows.Sockaddr) error {
 	return nil
 }
 
-func (u *RIOConn) ListenOut(r EncReader) {
+func (u *RIOConn) ListenOut(r EncReader) error {
 	buffer := make([]byte, MTU)
 
 	var lastRecvErr time.Time
@@ -151,8 +151,7 @@ func (u *RIOConn) ListenOut(r EncReader) {
 
 		if err != nil {
 			if errors.Is(err, net.ErrClosed) {
-				u.l.WithError(err).Debug("udp socket is closed, exiting read loop")
-				return
+				return err
 			}
 			// Dampen unexpected message warns to once per minute
 			if lastRecvErr.IsZero() || time.Since(lastRecvErr) > time.Minute {

udp/udp_tester.go

@@ -6,6 +6,7 @@ package udp
 import (
 	"io"
 	"net/netip"
+	"os"
 	"sync/atomic"
 
 	"github.com/sirupsen/logrus"
@@ -106,11 +107,11 @@ func (u *TesterConn) WriteTo(b []byte, addr netip.AddrPort) error {
 	return nil
 }
 
-func (u *TesterConn) ListenOut(r EncReader) {
+func (u *TesterConn) ListenOut(r EncReader) error {
 	for {
 		p, ok := <-u.RxPackets
 		if !ok {
-			return
+			return os.ErrClosed
 		}
 		r(p.From, p.Data)
 	}