feat(federation): do not allocate local services for load balancing (#3337)

* refactor: extract proxy into functions

* feat(federation): do not allocate services, directly connect with libp2p

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

core/p2p/federated.go

@@ -40,9 +40,9 @@ func (fs *FederatedServer) RandomServer() string {
 	var tunnelAddresses []string
 	for _, v := range GetAvailableNodes(fs.service) {
 		if v.IsOnline() {
-			tunnelAddresses = append(tunnelAddresses, v.TunnelAddress)
+			tunnelAddresses = append(tunnelAddresses, v.ID)
 		} else {
-			delete(fs.requestTable, v.TunnelAddress) // make sure it's not tracked
+			delete(fs.requestTable, v.ID) // make sure it's not tracked
 			log.Info().Msgf("Node %s is offline", v.ID)
 		}
 	}
@@ -61,8 +61,8 @@ func (fs *FederatedServer) syncTableStatus() {
 
 	for _, v := range GetAvailableNodes(fs.service) {
 		if v.IsOnline() {
-			fs.ensureRecordExist(v.TunnelAddress)
+			fs.ensureRecordExist(v.ID)
-			currentTunnels[v.TunnelAddress] = struct{}{}
+			currentTunnels[v.ID] = struct{}{}
 		}
 	}
 

core/p2p/federated_server.go

@@ -8,16 +8,12 @@ import (
 	"errors"
 	"fmt"
 	"net"
-	"time"
 
 	"github.com/mudler/edgevpn/pkg/node"
-	"github.com/mudler/edgevpn/pkg/protocol"
-	"github.com/mudler/edgevpn/pkg/types"
 	"github.com/rs/zerolog/log"
 )
 
 func (f *FederatedServer) Start(ctx context.Context) error {
-
 	n, err := NewNode(f.p2ptoken)
 	if err != nil {
 		return fmt.Errorf("creating a new node: %w", err)
@@ -29,7 +25,7 @@ func (f *FederatedServer) Start(ctx context.Context) error {
 
 	if err := ServiceDiscoverer(ctx, n, f.p2ptoken, f.service, func(servicesID string, tunnel NodeData) {
 		log.Debug().Msgf("Discovered node: %s", tunnel.ID)
-	}, true); err != nil {
+	}, false); err != nil {
 		return err
 	}
 
@@ -50,21 +46,8 @@ func (fs *FederatedServer) proxy(ctx context.Context, node *node.Node) error {
 		<-ctx.Done()
 		l.Close()
 	}()
-	ledger, _ := node.Ledger()
 
-	// Announce ourselves so nodes accepts our connection
+	nodeAnnounce(ctx, node)
-	ledger.Announce(
-		ctx,
-		10*time.Second,
-		func() {
-			updatedMap := map[string]interface{}{}
-			updatedMap[node.Host().ID().String()] = &types.User{
-				PeerID:    node.Host().ID().String(),
-				Timestamp: time.Now().String(),
-			}
-			ledger.Add(protocol.UsersLedgerKey, updatedMap)
-		},
-	)
 
 	defer l.Close()
 	for {
@@ -82,52 +65,36 @@ func (fs *FederatedServer) proxy(ctx context.Context, node *node.Node) error {
 
 			// Handle connections in a new goroutine, forwarding to the p2p service
 			go func() {
-				tunnelAddr := ""
+				workerID := ""
-
 				if fs.workerTarget != "" {
-					for _, v := range GetAvailableNodes(fs.service) {
+					workerID = fs.workerTarget
-						if v.ID == fs.workerTarget {
-							tunnelAddr = v.TunnelAddress
-							break
-						}
-					}
 				} else if fs.loadBalanced {
 					log.Debug().Msgf("Load balancing request")
 
-					tunnelAddr = fs.SelectLeastUsedServer()
+					workerID = fs.SelectLeastUsedServer()
-					if tunnelAddr == "" {
+					if workerID == "" {
 						log.Debug().Msgf("Least used server not found, selecting random")
-						tunnelAddr = fs.RandomServer()
+						workerID = fs.RandomServer()
 					}
-
 				} else {
-					tunnelAddr = fs.RandomServer()
+					workerID = fs.RandomServer()
 				}
 
-				if tunnelAddr == "" {
+				if workerID == "" {
 					log.Error().Msg("No available nodes yet")
 					return
 				}
 
-				log.Debug().Msgf("Selected tunnel %s", tunnelAddr)
+				log.Debug().Msgf("Selected node %s", workerID)
-
+				nodeData, exists := GetNode(fs.service, workerID)
-				tunnelConn, err := net.Dial("tcp", tunnelAddr)
+				if !exists {
-				if err != nil {
+					log.Error().Msgf("Node %s not found", workerID)
-					log.Error().Err(err).Msg("Error connecting to tunnel")
 					return
 				}
 
-				log.Info().Msgf("Redirecting %s to %s", conn.LocalAddr().String(), tunnelConn.RemoteAddr().String())
+				proxyP2PConnection(ctx, node, nodeData.ServiceID, conn)
-				closer := make(chan struct{}, 2)
-				go copyStream(closer, tunnelConn, conn)
-				go copyStream(closer, conn, tunnelConn)
-				<-closer
-
-				tunnelConn.Close()
-				conn.Close()
-
 				if fs.loadBalanced {
-					fs.RecordRequest(tunnelAddr)
+					fs.RecordRequest(workerID)
 				}
 			}()
 		}

core/p2p/node.go

@@ -14,6 +14,7 @@ type NodeData struct {
 	Name          string
 	ID            string
 	TunnelAddress string
+	ServiceID     string
 	LastSeen      time.Time
 }
 
@@ -39,6 +40,19 @@ func GetAvailableNodes(serviceID string) []NodeData {
 	return availableNodes
 }
 
+func GetNode(serviceID, nodeID string) (NodeData, bool) {
+	if serviceID == "" {
+		serviceID = defaultServicesID
+	}
+	mu.Lock()
+	defer mu.Unlock()
+	if _, ok := nodes[serviceID]; !ok {
+		return NodeData{}, false
+	}
+	nd, exists := nodes[serviceID][nodeID]
+	return nd, exists
+}
+
 func AddNode(serviceID string, node NodeData) {
 	if serviceID == "" {
 		serviceID = defaultServicesID

core/p2p/p2p.go

@@ -66,22 +66,7 @@ func nodeID(s string) string {
 	return fmt.Sprintf("%s-%s", hostname, s)
 }
 
-func allocateLocalService(ctx context.Context, node *node.Node, listenAddr, service string) error {
+func nodeAnnounce(ctx context.Context, node *node.Node) {
-
-	zlog.Info().Msgf("Allocating service '%s' on: %s", service, listenAddr)
-	// Open local port for listening
-	l, err := net.Listen("tcp", listenAddr)
-	if err != nil {
-		zlog.Error().Err(err).Msg("Error listening")
-		return err
-	}
-	go func() {
-		<-ctx.Done()
-		l.Close()
-	}()
-
-	//	ll.Info("Binding local port on", srcaddr)
-
 	ledger, _ := node.Ledger()
 
 	// Announce ourselves so nodes accepts our connection
@@ -97,25 +82,12 @@ func allocateLocalService(ctx context.Context, node *node.Node, listenAddr, serv
 			ledger.Add(protocol.UsersLedgerKey, updatedMap)
 		},
 	)
-
-	defer l.Close()
-	for {
-		select {
-		case <-ctx.Done():
-			return errors.New("context canceled")
-		default:
-			zlog.Debug().Msg("New for connection")
-			// Listen for an incoming connection.
-			conn, err := l.Accept()
-			if err != nil {
-				fmt.Println("Error accepting: ", err.Error())
-				continue
 }
 
-			// Handle connections in a new goroutine, forwarding to the p2p service
+func proxyP2PConnection(ctx context.Context, node *node.Node, serviceID string, conn net.Conn) {
-			go func() {
+	ledger, _ := node.Ledger()
 	// Retrieve current ID for ip in the blockchain
-				existingValue, found := ledger.GetKey(protocol.ServicesLedgerKey, service)
+	existingValue, found := ledger.GetKey(protocol.ServicesLedgerKey, serviceID)
 	service := &types.Service{}
 	existingValue.Unmarshal(service)
 	// If mismatch, update the blockchain
@@ -139,7 +111,7 @@ func allocateLocalService(ctx context.Context, node *node.Node, listenAddr, serv
 	// Open a stream
 	stream, err := node.Host().NewStream(ctx, d, protocol.ServiceProtocol.ID())
 	if err != nil {
-					zlog.Error().Msg("cannot open stream peer")
+		zlog.Error().Err(err).Msg("cannot open stream peer")
 
 		conn.Close()
 		//	ll.Debugf("could not open stream '%s'", err.Error())
@@ -154,7 +126,40 @@ func allocateLocalService(ctx context.Context, node *node.Node, listenAddr, serv
 
 	stream.Close()
 	conn.Close()
-				//	ll.Infof("(service %s) Done handling %s", serviceID, l.Addr().String())
+}
+
+func allocateLocalService(ctx context.Context, node *node.Node, listenAddr, service string) error {
+	zlog.Info().Msgf("Allocating service '%s' on: %s", service, listenAddr)
+	// Open local port for listening
+	l, err := net.Listen("tcp", listenAddr)
+	if err != nil {
+		zlog.Error().Err(err).Msg("Error listening")
+		return err
+	}
+	go func() {
+		<-ctx.Done()
+		l.Close()
+	}()
+
+	nodeAnnounce(ctx, node)
+
+	defer l.Close()
+	for {
+		select {
+		case <-ctx.Done():
+			return errors.New("context canceled")
+		default:
+			zlog.Debug().Msg("New for connection")
+			// Listen for an incoming connection.
+			conn, err := l.Accept()
+			if err != nil {
+				fmt.Println("Error accepting: ", err.Error())
+				continue
+			}
+
+			// Handle connections in a new goroutine, forwarding to the p2p service
+			go func() {
+				proxyP2PConnection(ctx, node, service, conn)
 			}()
 		}
 	}
@@ -258,6 +263,7 @@ var muservice sync.Mutex
 func ensureService(ctx context.Context, n *node.Node, nd *NodeData, sserv string, allocate bool) {
 	muservice.Lock()
 	defer muservice.Unlock()
+	nd.ServiceID = sserv
 	if ndService, found := service[nd.Name]; !found {
 		if !nd.IsOnline() {
 			// if node is offline and not present, do nothing