Make Signal Service listen on a standard 443/80 port instead of 10000 (#396)

Right now Signal Service runs the Let'sEncrypt manager on port 80
and a gRPC server on port 10000. There are two separate listeners.
This PR combines these listeners into one with a cmux lib.
The gRPC server runs on either 443 with TLS or 80 without TLS.
Let's Encrypt manager always runs on port 80.

encryption/letsencrypt.go

@@ -8,17 +8,17 @@ import (
 )
 
 // CreateCertManager wraps common logic of generating Let's encrypt certificate.
-func CreateCertManager(datadir string, letsencryptDomain string) *autocert.Manager {
+func CreateCertManager(datadir string, letsencryptDomain string) (*autocert.Manager, error) {
 	certDir := filepath.Join(datadir, "letsencrypt")
 
 	if _, err := os.Stat(certDir); os.IsNotExist(err) {
 		err = os.MkdirAll(certDir, os.ModeDir)
 		if err != nil {
-			log.Fatalf("failed creating Let's encrypt certdir: %s: %v", certDir, err)
+			return nil, err
 		}
 	}
 
-	log.Infof("running with Let's encrypt with domain %s. Cert will be stored in %s", letsencryptDomain, certDir)
+	log.Infof("running with LetsEncrypt (%s). Cert will be stored in %s", letsencryptDomain, certDir)
 
 	certManager := &autocert.Manager{
 		Prompt:     autocert.AcceptTOS,
@@ -26,5 +26,5 @@ func CreateCertManager(datadir string, letsencryptDomain string) *autocert.Manag
 		HostPolicy: autocert.HostWhitelist(letsencryptDomain),
 	}
 
-	return certManager
+	return certManager, nil
 }

go.mod

@@ -38,7 +38,9 @@ require (
 	github.com/patrickmn/go-cache v2.1.0+incompatible
 	github.com/rs/xid v1.3.0
 	github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966
+	github.com/soheilhy/cmux v0.1.5
 	github.com/stretchr/testify v1.7.1
+	golang.org/x/sync v0.0.0-20210220032951-036812b2e83c
 	golang.org/x/term v0.0.0-20220526004731-065cf7ba2467
 )
 
@@ -97,7 +99,6 @@ require (
 	golang.org/x/image v0.0.0-20200430140353-33d19683fad8 // indirect
 	golang.org/x/mod v0.6.0-dev.0.20220106191415-9b9b3d81d5e3 // indirect
 	golang.org/x/net v0.0.0-20220513224357-95641704303c // indirect
-	golang.org/x/sync v0.0.0-20210220032951-036812b2e83c // indirect
 	golang.org/x/text v0.3.8-0.20211105212822-18b340fc7af2 // indirect
 	golang.org/x/tools v0.1.10 // indirect
 	golang.org/x/xerrors v0.0.0-20220411194840-2f41105eb62f // indirect

go.sum

@@ -575,6 +575,8 @@ github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966 h1:JIAuq3EE
 github.com/skratchdot/open-golang v0.0.0-20200116055534-eef842397966/go.mod h1:sUM3LWHvSMaG192sy56D9F7CNvL7jUJVXoqM1QKLnog=
 github.com/smartystreets/assertions v1.13.0 h1:Dx1kYM01xsSqKPno3aqLnrwac2LetPvN23diwyr69Qs=
 github.com/smartystreets/goconvey v1.7.2 h1:9RBaZCeXEQ3UselpuwUQHltGVXvdwm6cv1hgR6gDIPg=
+github.com/soheilhy/cmux v0.1.5 h1:jjzc5WVemNEDTLwv9tlmemhC73tI08BNOIGwBOo10Js=
+github.com/soheilhy/cmux v0.1.5/go.mod h1:T7TcVDs9LWfQgPlPsdngu6I6QIoyIFZDDC6sNE1GqG0=
 github.com/spaolacci/murmur3 v0.0.0-20180118202830-f09979ecbc72/go.mod h1:JwIasOWyU6f++ZhiEuf87xNszmSA2myDM2Kzu9HwQUA=
 github.com/spf13/afero v1.1.2/go.mod h1:j4pytiNVoe2o6bmDsKpLACNPDBIoEAkihy7loJ1B0CQ=
 github.com/spf13/afero v1.2.2/go.mod h1:9ZxEEn6pIJ8Rxe320qSDBk6AsU0r9pR7Q4OcevTdifk=
@@ -740,6 +742,7 @@ golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwY
 golang.org/x/net v0.0.0-20201031054903-ff519b6c9102/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20201110031124-69a78807bb2b/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20201201195509-5d6afe98e0b7/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
+golang.org/x/net v0.0.0-20201202161906-c7110b5ffcbb/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.0.0-20201209123823-ac852fbbde11/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
 golang.org/x/net v0.0.0-20201216054612-986b41b23924/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
 golang.org/x/net v0.0.0-20201224014010-6772e930b67b/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=

infrastructure_files/docker-compose.yml.tmpl

@@ -25,7 +25,7 @@ services:
     volumes:
       - $SIGNAL_VOLUMENAME:/var/lib/netbird
     ports:
-      - 10000:10000
+      - 10000:80
   #     # port and command for Let's Encrypt validation
   #      - 443:443
   #    command: ["--letsencrypt-domain", "$NETBIRD_DOMAIN", "--log-file", "console"]

management/cmd/management.go

@@ -46,6 +46,17 @@ var (
 		Timeout:               2 * time.Second,
 	}
 
+	// TLS enabled:
+	// 		- HTTP 80 for LetsEncrypt
+	//		- if --port not specified gRPC and HTTP servers on 443 (with multiplexing)
+	//		- if --port=X specified then run gRPC and HTTP servers on X (with multiplexing)
+	// 		- if --port=80 forbid this (throw error, otherwise we need to overcomplicate the logic with multiplexing)
+	// TLS disabled:
+	//		- if --port not specified gRPC and HTTP servers on 443 on 80 (with multiplexing)
+	//		- if --port=X specified then run gRPC and HTTP servers on 443 on X (with multiplexing)
+	// Always run gRPC on port 33073 regardless of TLS to be backward compatible
+	// Remove HTTP port 33071 from the configuration.
+
 	mgmtCmd = &cobra.Command{
 		Use:   "management",
 		Short: "start Netbird Management Server",
@@ -97,7 +108,10 @@ var (
 			var httpServer *http.Server
 			if config.HttpConfig.LetsEncryptDomain != "" {
 				// automatically generate a new certificate with Let's Encrypt
-				certManager := encryption.CreateCertManager(config.Datadir, config.HttpConfig.LetsEncryptDomain)
+				certManager, err := encryption.CreateCertManager(config.Datadir, config.HttpConfig.LetsEncryptDomain)
+				if err != nil {
+					log.Fatalf("failed creating Let's Encrypt cert manager: %v", err)
+				}
 				transportCredentials := credentials.NewTLS(certManager.TLSConfig())
 				opts = append(opts, grpc.Creds(transportCredentials))
 

signal/client/grpc.go

@@ -75,6 +75,8 @@ func NewClient(ctx context.Context, addr string, key wgtypes.Key, tlsEnabled boo
 		return nil, err
 	}
 
+	log.Debugf("connected to Signal Service: %v", conn.Target())
+
 	return &GrpcClient{
 		realClient: proto.NewSignalExchangeClient(conn),
 		ctx:        ctx,

signal/cmd/run.go

@@ -4,6 +4,7 @@ import (
 	"errors"
 	"flag"
 	"fmt"
+	"golang.org/x/crypto/acme/autocert"
 	"io"
 	"io/fs"
 	"io/ioutil"
@@ -11,6 +12,7 @@ import (
 	"net/http"
 	"os"
 	"path"
+	"strings"
 	"time"
 
 	"github.com/netbirdio/netbird/encryption"
@@ -29,6 +31,7 @@ var (
 	signalLetsencryptDomain string
 	signalSSLDir            string
 	defaultSignalSSLDir     string
+	tlsEnabled              bool
 
 	signalKaep = grpc.KeepaliveEnforcementPolicy(keepalive.EnforcementPolicy{
 		MinTime:             5 * time.Second,
@@ -44,9 +47,26 @@ var (
 
 	runCmd = &cobra.Command{
 		Use:   "run",
-		Short: "start Netbird Signal Server daemon",
-		Run: func(cmd *cobra.Command, args []string) {
+		Short: "start NetBird Signal Server daemon",
+		PreRun: func(cmd *cobra.Command, args []string) {
+			// detect whether user specified a port
+			userPort := cmd.Flag("port").Changed
+			if signalLetsencryptDomain != "" {
+				tlsEnabled = true
+			}
+
+			if !userPort {
+				// different defaults for signalPort
+				if tlsEnabled {
+					signalPort = 443
+				} else {
+					signalPort = 80
+				}
+			}
+		},
+		RunE: func(cmd *cobra.Command, args []string) error {
 			flag.Parse()
+
 			err := util.InitLog(logLevel, logFile)
 			if err != nil {
 				log.Fatalf("failed initializing log %v", err)
@@ -62,47 +82,120 @@ var (
 			}
 
 			var opts []grpc.ServerOption
-			if signalLetsencryptDomain != "" {
-				if _, err := os.Stat(signalSSLDir); os.IsNotExist(err) {
-					err = os.MkdirAll(signalSSLDir, os.ModeDir)
-					if err != nil {
-						log.Fatalf("failed creating datadir: %s: %v", signalSSLDir, err)
-					}
+			var certManager *autocert.Manager
+			if tlsEnabled {
+				// Let's encrypt enabled -> generate certificate automatically
+				certManager, err = encryption.CreateCertManager(signalSSLDir, signalLetsencryptDomain)
+				if err != nil {
+					return err
 				}
-				certManager := encryption.CreateCertManager(signalSSLDir, signalLetsencryptDomain)
 				transportCredentials := credentials.NewTLS(certManager.TLSConfig())
 				opts = append(opts, grpc.Creds(transportCredentials))
-
-				listener := certManager.Listener()
-				log.Infof("http server listening on %s", listener.Addr())
-				go func() {
-					if err := http.Serve(listener, certManager.HTTPHandler(nil)); err != nil {
-						log.Errorf("failed to serve https server: %v", err)
-					}
-				}()
 			}
 
 			opts = append(opts, signalKaep, signalKasp)
 			grpcServer := grpc.NewServer(opts...)
-
-			lis, err := net.Listen("tcp", fmt.Sprintf(":%d", signalPort))
-			if err != nil {
-				log.Fatalf("failed to listen: %v", err)
-			}
-
 			proto.RegisterSignalExchangeServer(grpcServer, server.NewServer())
-			log.Printf("started server: localhost:%v", signalPort)
-			if err := grpcServer.Serve(lis); err != nil {
-				log.Fatalf("failed to serve: %v", err)
+
+			var compatListener net.Listener
+			if signalPort != 10000 {
+				// The Signal gRPC server was running on port 10000 previously. Old agents that are already connected to Signal
+				// are using port 10000. For compatibility purposes we keep running a 2nd gRPC server on port 10000.
+				compatListener, err = serveGRPC(grpcServer, 10000)
+				if err != nil {
+					return err
+				}
+				log.Infof("running gRPC backward compatibility server: %s", compatListener.Addr().String())
 			}
 
+			var grpcListener net.Listener
+			var httpListener net.Listener
+			if tlsEnabled {
+				httpListener = certManager.Listener()
+				if signalPort == 443 {
+					// running gRPC and HTTP cert manager on the same port
+					serveHTTP(httpListener, certManager.HTTPHandler(grpcHandlerFunc(grpcServer)))
+					log.Infof("running HTTP server (LetsEncrypt challenge handler) and gRPC server on the same port: %s", httpListener.Addr().String())
+				} else {
+					serveHTTP(httpListener, certManager.HTTPHandler(nil))
+					log.Infof("running HTTP server (LetsEncrypt challenge handler): %s", httpListener.Addr().String())
+				}
+			}
+
+			if signalPort != 443 || !tlsEnabled {
+				grpcListener, err = serveGRPC(grpcServer, signalPort)
+				if err != nil {
+					return err
+				}
+				log.Infof("running gRPC server: %s", grpcListener.Addr().String())
+			}
+
+			log.Infof("started Signal Service")
+
 			SetupCloseHandler()
+
 			<-stopCh
-			log.Println("Receive signal to stop running the Signal server")
+			if grpcListener != nil {
+				_ = grpcListener.Close()
+				log.Infof("stopped gRPC server")
+			}
+			if httpListener != nil {
+				_ = httpListener.Close()
+				log.Infof("stopped HTTP server")
+			}
+			if compatListener != nil {
+				_ = compatListener.Close()
+				log.Infof("stopped gRPC backward compatibility server")
+			}
+			log.Infof("stopped Signal Service")
+
+			return nil
 		},
 	}
 )
 
+func grpcHandlerFunc(grpcServer *grpc.Server) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		grpcHeader := strings.HasPrefix(r.Header.Get("Content-Type"), "application/grpc") ||
+			strings.HasPrefix(r.Header.Get("Content-Type"), "application/grpc+proto")
+		if r.ProtoMajor == 2 && grpcHeader {
+			grpcServer.ServeHTTP(w, r)
+		}
+	})
+}
+
+func notifyStop(msg string) {
+	select {
+	case stopCh <- 1:
+		log.Error(msg)
+	default:
+		// stop has been already called, nothing to report
+	}
+}
+
+func serveHTTP(httpListener net.Listener, handler http.Handler) {
+	go func() {
+		err := http.Serve(httpListener, handler)
+		if err != nil {
+			notifyStop(fmt.Sprintf("failed running HTTP server %v", err))
+		}
+	}()
+}
+
+func serveGRPC(grpcServer *grpc.Server, port int) (net.Listener, error) {
+	listener, err := net.Listen("tcp", fmt.Sprintf(":%d", port))
+	if err != nil {
+		return nil, err
+	}
+	go func() {
+		err := grpcServer.Serve(listener)
+		if err != nil {
+			notifyStop(fmt.Sprintf("failed running gRPC server on port %d: %v", port, err))
+		}
+	}()
+	return listener, nil
+}
+
 func cpFile(src, dst string) error {
 	var err error
 	var srcfd *os.File
@@ -191,7 +284,7 @@ func migrateToNetbird(oldPath, newPath string) bool {
 }
 
 func init() {
-	runCmd.PersistentFlags().IntVar(&signalPort, "port", 10000, "Server port to listen on (e.g. 10000)")
+	runCmd.PersistentFlags().IntVar(&signalPort, "port", 80, "Server port to listen on (defaults to 443 if TLS is enabled, 80 otherwise")
 	runCmd.Flags().StringVar(&signalSSLDir, "ssl-dir", defaultSignalSSLDir, "server ssl directory location. *Required only for Let's Encrypt certificates.")
 	runCmd.Flags().StringVar(&signalLetsencryptDomain, "letsencrypt-domain", "", "a domain to issue Let's Encrypt certificate for. Enables TLS using Let's Encrypt. Will fetch and renew certificate, and run the server with TLS")
 }