Merge pull request #22 from Sagit-chu/feat/failover-debug-logging

feat(gost): add debug logging for failover mechanism analysis

.github/workflows/ci-build.yml

@@ -20,11 +20,9 @@ jobs:
         uses: actions/setup-node@v4
         with:
           node-version: '20'
-          cache: 'npm'
-          cache-dependency-path: vite-frontend/package-lock.json
 
       - name: Install dependencies
-        run: npm ci
+        run: npm install --legacy-peer-deps
 
       - name: Build
         run: npm run build

.github/workflows/docker-build.yml

@@ -100,11 +100,11 @@ jobs:
 
       - name: Build GOST binary (AMD64)
         working-directory: ./go-gost
-        run: CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -ldflags="-s -w" -o gost-amd64
+        run: CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -ldflags="-s -w -X main.version=${{ needs.check-version.outputs.version }}" -o gost-amd64
 
       - name: Build GOST binary (ARM64)
         working-directory: ./go-gost
-        run: CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -ldflags="-s -w" -o gost-arm64
+        run: CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -ldflags="-s -w -X main.version=${{ needs.check-version.outputs.version }}" -o gost-arm64
 
       - name: Compress with UPX
         working-directory: ./go-gost

README.md

@@ -16,24 +16,13 @@
 ---
 ### Docker Compose部署
 #### 快速部署
-面板端(稳定版)：
+面板端：
 ```bash
 curl -L https://raw.githubusercontent.com/Sagit-chu/flux-panel/main/panel_install.sh -o panel_install.sh && chmod +x panel_install.sh && ./panel_install.sh
 ```
-节点端(稳定版)：
+节点端：
 ```bash
 curl -L https://raw.githubusercontent.com/Sagit-chu/flux-panel/main/install.sh -o install.sh && chmod +x install.sh && ./install.sh
-
-```
-
-面板端(开发版)：
-```bash
-curl -L https://raw.githubusercontent.com/Sagit-chu/flux-panel/beta/panel_install.sh -o panel_install.sh && chmod +x panel_install.sh && ./panel_install.sh
-```
-节点端(开发版)：
-```bash
-curl -L https://raw.githubusercontent.com/Sagit-chu/flux-panel/beta/install.sh -o install.sh && chmod +x install.sh && ./install.sh
-
 ```
 
 #### 默认管理员账号

go-gost/main.go

@@ -119,7 +119,7 @@ func main() {
 	log := xlogger.NewLogger()
 	logger.SetDefault(log)
 
-	wsReporter := socket.StartWebSocketReporterWithConfig(config.Addr, config.Secret, config.Http, config.Tls, config.Socks, "2.0.2")
+	wsReporter := socket.StartWebSocketReporterWithConfig(config.Addr, config.Secret, config.Http, config.Tls, config.Socks, version)
 	defer wsReporter.Stop()
 	service.SetHTTPReportURL(config.Addr, config.Secret)
 

go-gost/version.go

@@ -1,5 +1,5 @@
 package main
 
 var (
-	version = "3.1.0"
+	version = "dev"
 )

go-gost/x/ctx/value.go

@@ -109,3 +109,23 @@ func LoggerFromContext(ctx context.Context) logger.Logger {
 	v, _ := ctx.Value(keyLogger).(logger.Logger)
 	return v
 }
+
+// excludeNodesKey saves the list of node addresses to exclude during selection.
+// This is used for failover retry logic - when a node fails, it gets added to
+// the exclude list so the next Select() call will skip it.
+type excludeNodesKey struct{}
+
+var (
+	keyExcludeNodes = &excludeNodesKey{}
+)
+
+// ContextWithExcludeNodes returns a context with the list of node addresses to exclude.
+func ContextWithExcludeNodes(ctx context.Context, nodes []string) context.Context {
+	return context.WithValue(ctx, keyExcludeNodes, nodes)
+}
+
+// ExcludeNodesFromContext returns the list of node addresses to exclude from selection.
+func ExcludeNodesFromContext(ctx context.Context) []string {
+	v, _ := ctx.Value(keyExcludeNodes).([]string)
+	return v
+}

go-gost/x/handler/forward/local/handler.go

@@ -176,51 +176,91 @@ func (h *forwardHandler) Handle(ctx context.Context, conn net.Conn, opts ...hand
 		}
 	}
 
-	target := &chain.Node{}
-	if h.hop != nil {
-		target = h.hop.Select(ctx,
-			hop.ProtocolSelectOption(proto),
-		)
-	}
-	if target == nil {
-		err := errors.New("node not available")
-		return err
+	// Determine max retry attempts
+	maxRetries := h.md.maxRetries
+	if maxRetries <= 0 {
+		// Default: try all available nodes
+		if nl, ok := h.hop.(hop.NodeList); ok {
+			maxRetries = len(nl.Nodes())
+		}
+		if maxRetries <= 0 {
+			maxRetries = 1
+		}
 	}
 
-	addr := target.Addr
-	if opts := target.Options(); opts != nil {
-		switch opts.Network {
-		case "unix":
-			network = opts.Network
-		default:
-			if _, _, err := net.SplitHostPort(addr); err != nil {
-				addr += ":0"
+	var triedNodes []string
+	var lastErr error
+	var cc net.Conn
+
+	h.options.Logger.Debugf("[handler.retry] starting retry loop: maxRetries=%d", maxRetries)
+
+	for attempt := 0; attempt < maxRetries; attempt++ {
+		// Select a target node, excluding previously tried nodes
+		selectCtx := ctxvalue.ContextWithExcludeNodes(ctx, triedNodes)
+		var target *chain.Node
+		if h.hop != nil {
+			target = h.hop.Select(selectCtx,
+				hop.ProtocolSelectOption(proto),
+			)
+		}
+		if target == nil {
+			h.options.Logger.Debugf("[handler.retry] attempt=%d target=nil, triedNodes=%v", attempt, triedNodes)
+			if lastErr != nil {
+				return lastErr
+			}
+			return errors.New("node not available")
+		}
+
+		h.options.Logger.Debugf("[handler.retry] attempt=%d selected node=%s addr=%s", attempt, target.Name, target.Addr)
+
+		// Track this node as tried
+		triedNodes = append(triedNodes, target.Addr)
+
+		addr := target.Addr
+		if opts := target.Options(); opts != nil {
+			switch opts.Network {
+			case "unix":
+				network = opts.Network
+			default:
+				if _, _, err := net.SplitHostPort(addr); err != nil {
+					addr += ":0"
+				}
 			}
 		}
-	}
 
-	ro.Network = network
-	ro.Host = addr
+		ro.Network = network
+		ro.Host = addr
 
-	var buf bytes.Buffer
-	cc, err := h.options.Router.Dial(ctxvalue.ContextWithBuffer(ctx, &buf), network, addr)
-	ro.Route = buf.String()
-	if err != nil {
-		// TODO: the router itself may be failed due to the failed node in the router,
-		// the dead marker may be a wrong operation.
-		if marker := target.Marker(); marker != nil {
-			marker.Mark()
+		var buf bytes.Buffer
+		cc, err = h.options.Router.Dial(ctxvalue.ContextWithBuffer(ctx, &buf), network, addr)
+		ro.Route = buf.String()
+		if err != nil {
+			// Mark node as failed for future selections
+			if marker := target.Marker(); marker != nil {
+				marker.Mark()
+				h.options.Logger.Debugf("[handler.retry] attempt=%d dial failed, marked node=%s count=%d err=%v",
+					attempt, target.Addr, marker.Count(), err)
+			}
+			lastErr = err
+			// Try next node
+			continue
 		}
-		return err
-	}
-	if marker := target.Marker(); marker != nil {
-		marker.Reset()
-	}
-	defer cc.Close()
 
-	xnet.Transport(conn, cc)
+		// Success - reset marker and proceed
+		if marker := target.Marker(); marker != nil {
+			marker.Reset()
+		}
+		defer cc.Close()
 
-	return nil
+		xnet.Transport(conn, cc)
+		return nil
+	}
+
+	// All retries exhausted
+	if lastErr != nil {
+		return lastErr
+	}
+	return errors.New("all nodes failed")
 }
 
 func (h *forwardHandler) checkRateLimit(addr net.Addr) bool {

go-gost/x/handler/forward/local/metadata.go

@@ -25,6 +25,12 @@ type metadata struct {
 	privateKey  crypto.PrivateKey
 	alpn        string
 	mitmBypass  bypass.Bypass
+
+	// maxRetries specifies the maximum number of failover retry attempts.
+	// When a target node fails, the handler will try the next available node.
+	// 0 means use the total number of available nodes (try all nodes once).
+	// Default: 0 (try all available nodes)
+	maxRetries int
 }
 
 func (h *forwardHandler) parseMetadata(md mdata.Metadata) (err error) {
@@ -56,5 +62,8 @@ func (h *forwardHandler) parseMetadata(md mdata.Metadata) (err error) {
 	h.md.alpn = mdutil.GetString(md, "mitm.alpn")
 	h.md.mitmBypass = registry.BypassRegistry().Get(mdutil.GetString(md, "mitm.bypass"))
 
+	// maxRetries: 0 means try all available nodes (default behavior)
+	h.md.maxRetries = mdutil.GetInt(md, "maxRetries", "retry.max")
+
 	return
 }

go-gost/x/handler/forward/remote/handler.go

@@ -204,68 +204,101 @@ func (h *forwardHandler) Handle(ctx context.Context, conn net.Conn, opts ...hand
 		}
 	}
 
-	var target *chain.Node
-	if host != "" {
-		target = &chain.Node{
-			Addr: host,
+	// Determine max retry attempts
+	maxRetries := h.md.maxRetries
+	if maxRetries <= 0 {
+		// Default: try all available nodes
+		if nl, ok := h.hop.(hop.NodeList); ok {
+			maxRetries = len(nl.Nodes())
 		}
-	}
-	if h.hop != nil {
-		target = h.hop.Select(ctx,
-			hop.ProtocolSelectOption(proto),
-		)
-	}
-	if target == nil {
-		err := errors.New("node not available")
-		log.Error(err)
-		return err
-	}
-
-	if opts := target.Options(); opts != nil {
-		switch opts.Network {
-		case "unix":
-			network = opts.Network
-		default:
+		if maxRetries <= 0 {
+			maxRetries = 1
 		}
 	}
 
-	ro.Network = network
-	ro.Host = target.Addr
+	var triedNodes []string
+	var lastErr error
+	var cc net.Conn
 
-	log = log.WithFields(map[string]any{
-		"node": target.Name,
-		"dst":  fmt.Sprintf("%s/%s", target.Addr, network),
-	})
+	for attempt := 0; attempt < maxRetries; attempt++ {
+		// Select a target node, excluding previously tried nodes
+		selectCtx := ctxvalue.ContextWithExcludeNodes(ctx, triedNodes)
+		var target *chain.Node
+		if host != "" {
+			target = chain.NewNode("", host)
+		}
+		if h.hop != nil {
+			target = h.hop.Select(selectCtx,
+				hop.ProtocolSelectOption(proto),
+			)
+		}
+		if target == nil {
+			if lastErr != nil {
+				return lastErr
+			}
+			err := errors.New("node not available")
+			log.Error(err)
+			return err
+		}
 
-	log.Debugf("%s >> %s", conn.RemoteAddr(), target.Addr)
+		// Track this node as tried
+		triedNodes = append(triedNodes, target.Addr)
 
-	var buf bytes.Buffer
-	cc, err := h.options.Router.Dial(ctxvalue.ContextWithBuffer(ctx, &buf), network, target.Addr)
-	ro.Route = buf.String()
-	if err != nil {
-		log.Error(err)
-		// TODO: the router itself may be failed due to the failed node in the router,
-		// the dead marker may be a wrong operation.
+		if opts := target.Options(); opts != nil {
+			switch opts.Network {
+			case "unix":
+				network = opts.Network
+			default:
+			}
+		}
+
+		ro.Network = network
+		ro.Host = target.Addr
+
+		targetLog := log.WithFields(map[string]any{
+			"node": target.Name,
+			"dst":  fmt.Sprintf("%s/%s", target.Addr, network),
+		})
+
+		targetLog.Debugf("%s >> %s", conn.RemoteAddr(), target.Addr)
+
+		var buf bytes.Buffer
+		cc, err = h.options.Router.Dial(ctxvalue.ContextWithBuffer(ctx, &buf), network, target.Addr)
+		ro.Route = buf.String()
+		if err != nil {
+			targetLog.Error(err)
+			// Mark node as failed for future selections
+			if marker := target.Marker(); marker != nil {
+				marker.Mark()
+			}
+			lastErr = err
+			// Try next node
+			continue
+		}
+
+		// Success - reset marker and proceed
 		if marker := target.Marker(); marker != nil {
-			marker.Mark()
+			marker.Reset()
 		}
-		return err
-	}
-	defer cc.Close()
-	if marker := target.Marker(); marker != nil {
-		marker.Reset()
+		defer cc.Close()
+
+		cc = proxyproto.WrapClientConn(h.md.proxyProtocol, conn.RemoteAddr(), convertAddr(conn.LocalAddr()), cc)
+
+		t := time.Now()
+		targetLog.Infof("%s <-> %s", conn.RemoteAddr(), target.Addr)
+		xnet.Transport(conn, cc)
+		targetLog.WithFields(map[string]any{
+			"duration": time.Since(t),
+		}).Infof("%s >-< %s", conn.RemoteAddr(), target.Addr)
+
+		return nil
 	}
 
-	cc = proxyproto.WrapClientConn(h.md.proxyProtocol, conn.RemoteAddr(), convertAddr(conn.LocalAddr()), cc)
-
-	t := time.Now()
-	log.Infof("%s <-> %s", conn.RemoteAddr(), target.Addr)
-	xnet.Transport(conn, cc)
-	log.WithFields(map[string]any{
-		"duration": time.Since(t),
-	}).Infof("%s >-< %s", conn.RemoteAddr(), target.Addr)
-
-	return nil
+	// All retries exhausted
+	if lastErr != nil {
+		return lastErr
+	}
+	return errors.New("all nodes failed")
 }
 
 func (h *forwardHandler) checkRateLimit(addr net.Addr) bool {

go-gost/x/handler/forward/remote/metadata.go

@@ -26,6 +26,12 @@ type metadata struct {
 	privateKey  crypto.PrivateKey
 	alpn        string
 	mitmBypass  bypass.Bypass
+
+	// maxRetries specifies the maximum number of failover retry attempts.
+	// When a target node fails, the handler will try the next available node.
+	// 0 means use the total number of available nodes (try all nodes once).
+	// Default: 0 (try all available nodes)
+	maxRetries int
 }
 
 func (h *forwardHandler) parseMetadata(md mdata.Metadata) (err error) {
@@ -57,5 +63,9 @@ func (h *forwardHandler) parseMetadata(md mdata.Metadata) (err error) {
 	}
 	h.md.alpn = mdutil.GetString(md, "mitm.alpn")
 	h.md.mitmBypass = registry.BypassRegistry().Get(mdutil.GetString(md, "mitm.bypass"))
+
+	// maxRetries: 0 means try all available nodes (default behavior)
+	h.md.maxRetries = mdutil.GetInt(md, "maxRetries", "retry.max")
+
 	return
 }

go-gost/x/hop/hop.go

@@ -18,6 +18,7 @@ import (
 	"github.com/go-gost/core/selector"
 	"github.com/go-gost/x/config"
 	node_parser "github.com/go-gost/x/config/parsing/node"
+	ctxvalue "github.com/go-gost/x/ctx"
 	"github.com/go-gost/x/internal/loader"
 )
 
@@ -141,11 +142,28 @@ func (p *chainHop) Select(ctx context.Context, opts ...hop.SelectOption) *chain.
 		return nil
 	}
 
+	// Get list of nodes to exclude (for failover retry)
+	excludeNodes := ctxvalue.ExcludeNodesFromContext(ctx)
+	excludeSet := make(map[string]bool)
+	for _, addr := range excludeNodes {
+		excludeSet[addr] = true
+	}
+
+	// Debug logging for failover analysis
+	log.Debugf("[hop.Select] excludeNodes=%v, totalNodes=%d", excludeNodes, len(p.Nodes()))
+
 	var nodes []*chain.Node
 	for _, node := range p.Nodes() {
 		if node == nil {
 			continue
 		}
+
+		// Skip nodes in the exclude list (failover retry)
+		if excludeSet[node.Addr] || excludeSet[node.Name] {
+			log.Debugf("node %s(%s) excluded for failover retry", node.Name, node.Addr)
+			continue
+		}
+
 		// node level bypass
 		if node.Options().Bypass != nil &&
 			node.Options().Bypass.Contains(ctx, options.Network, options.Addr, bypass.WithHostOpton(options.Host)) {
@@ -177,9 +195,6 @@ func (p *chainHop) Select(ctx context.Context, opts ...hop.SelectOption) *chain.
 	if len(nodes) == 0 {
 		return nil
 	}
-	if len(nodes) == 1 {
-		return nodes[0]
-	}
 
 	sort.Slice(nodes, func(i, j int) bool {
 		return nodes[i].Options().Priority > nodes[j].Options().Priority
@@ -189,9 +204,31 @@ func (p *chainHop) Select(ctx context.Context, opts ...hop.SelectOption) *chain.
 		return nodes[0]
 	}
 
-	if s := p.options.selector; s != nil {
-		return s.Select(ctx, nodes...)
+	// For single-node case: bypass selector/FailFilter to ensure availability.
+	// The marker system still works for metrics, but we don't block requests
+	// based on recent failures - the connection will be attempted regardless.
+	// This matches upstream go-gost/x behavior.
+	if len(nodes) == 1 {
+		return nodes[0]
 	}
+
+	// Multi-node case: use selector with FailFilter for proper failover.
+	// FailFilter will exclude recently-failed nodes, allowing traffic to
+	// be routed to healthy alternatives.
+	if s := p.options.selector; s != nil {
+		log.Debugf("[hop.Select] calling selector.Select with %d nodes", len(nodes))
+		if node := s.Select(ctx, nodes...); node != nil {
+			log.Debugf("[hop.Select] selected node=%s addr=%s", node.Name, node.Addr)
+			return node
+		}
+		// All nodes filtered out by FailFilter - all are marked as failed.
+		// Return nil to signal "no healthy nodes available" to the caller.
+		// The handler's retry loop will handle this appropriately.
+		log.Debugf("all %d nodes filtered out by FailFilter, no healthy nodes available", len(nodes))
+		return nil
+	}
+
+	// Fallback: return first node if no selector configured
 	return nodes[0]
 }
 

go-gost/x/internal/util/forwarder/sniffer.go

@@ -247,64 +247,98 @@ func (h *Sniffer) dial(ctx context.Context, conn net.Conn, req *http.Request, ho
 		}
 	}
 
-	node = &chain.Node{
-		Addr: host,
+	// Determine max retry attempts
+	maxRetries := 1
+	if nl, ok := ho.Hop.(hop.NodeList); ok {
+		maxRetries = len(nl.Nodes())
 	}
-	if ho.Hop != nil {
-		node = ho.Hop.Select(ctx,
-			hop.ClientIPSelectOption(net.ParseIP(ro.ClientIP)),
-			hop.ProtocolSelectOption(sniffing.ProtoHTTP),
-			hop.HostSelectOption(host),
-			hop.MethodSelectOption(req.Method),
-			hop.PathSelectOption(req.URL.Path),
-			hop.QuerySelectOption(req.URL.Query()),
-			hop.HeaderSelectOption(req.Header),
-		)
-	}
-	if node == nil {
-		ho.Log.Warnf("node for %s not found", host)
-		res.StatusCode = http.StatusBadGateway
-		ro.HTTP.StatusCode = res.StatusCode
-		res.Write(conn)
-		return nil, nil, errors.New("node not available")
+	if maxRetries <= 0 {
+		maxRetries = 1
 	}
 
-	ro.Host = node.Addr
-	ho.Log = ho.Log.WithFields(map[string]any{
-		"node": node.Name,
-		"dst":  node.Addr,
-	})
-	ho.Log.Debugf("find node for host %s -> %s(%s)", host, node.Name, node.Addr)
+	var triedNodes []string
+	var lastErr error
 
-	cc, err = dial(ctx, "tcp", node.Addr)
-	if err != nil {
-		// TODO: the router itself may be failed due to the failed node in the router,
-		// the dead marker may be a wrong operation.
-		if marker := node.Marker(); marker != nil {
-			marker.Mark()
+	for attempt := 0; attempt < maxRetries; attempt++ {
+		// Select a node, excluding previously tried nodes
+		selectCtx := ctxvalue.ContextWithExcludeNodes(ctx, triedNodes)
+
+		node = chain.NewNode("", host)
+		if ho.Hop != nil {
+			node = ho.Hop.Select(selectCtx,
+				hop.ClientIPSelectOption(net.ParseIP(ro.ClientIP)),
+				hop.ProtocolSelectOption(sniffing.ProtoHTTP),
+				hop.HostSelectOption(host),
+				hop.MethodSelectOption(req.Method),
+				hop.PathSelectOption(req.URL.Path),
+				hop.QuerySelectOption(req.URL.Query()),
+				hop.HeaderSelectOption(req.Header),
+			)
 		}
-		ho.Log.Warnf("connect to node %s(%s) failed: %v", node.Name, node.Addr, err)
-		res.Write(conn)
-		return
-	}
-	if marker := node.Marker(); marker != nil {
-		marker.Reset()
-	}
-
-	if tlsSettings := node.Options().TLS; tlsSettings != nil {
-		cfg := &tls.Config{
-			ServerName:         tlsSettings.ServerName,
-			InsecureSkipVerify: !tlsSettings.Secure,
+		if node == nil {
+			if lastErr != nil {
+				ho.Log.Warnf("node for %s not found after retries", host)
+				res.StatusCode = http.StatusBadGateway
+				ro.HTTP.StatusCode = res.StatusCode
+				res.Write(conn)
+				return nil, nil, lastErr
+			}
+			ho.Log.Warnf("node for %s not found", host)
+			res.StatusCode = http.StatusBadGateway
+			ro.HTTP.StatusCode = res.StatusCode
+			res.Write(conn)
+			return nil, nil, errors.New("node not available")
 		}
-		tls_util.SetTLSOptions(cfg, &config.TLSOptions{
-			MinVersion:   tlsSettings.Options.MinVersion,
-			MaxVersion:   tlsSettings.Options.MaxVersion,
-			CipherSuites: tlsSettings.Options.CipherSuites,
-			ALPN:         tlsSettings.Options.ALPN,
+
+		// Track this node as tried
+		triedNodes = append(triedNodes, node.Addr)
+
+		ro.Host = node.Addr
+		ho.Log = ho.Log.WithFields(map[string]any{
+			"node": node.Name,
+			"dst":  node.Addr,
 		})
-		cc = tls.Client(cc, cfg)
+		ho.Log.Debugf("find node for host %s -> %s(%s)", host, node.Name, node.Addr)
+
+		cc, err = dial(ctx, "tcp", node.Addr)
+		if err != nil {
+			// Mark node as failed for future selections
+			if marker := node.Marker(); marker != nil {
+				marker.Mark()
+			}
+			ho.Log.Warnf("connect to node %s(%s) failed: %v, trying next node", node.Name, node.Addr, err)
+			lastErr = err
+			continue
+		}
+
+		// Success - reset marker
+		if marker := node.Marker(); marker != nil {
+			marker.Reset()
+		}
+
+		if tlsSettings := node.Options().TLS; tlsSettings != nil {
+			cfg := &tls.Config{
+				ServerName:         tlsSettings.ServerName,
+				InsecureSkipVerify: !tlsSettings.Secure,
+			}
+			tls_util.SetTLSOptions(cfg, &config.TLSOptions{
+				MinVersion:   tlsSettings.Options.MinVersion,
+				MaxVersion:   tlsSettings.Options.MaxVersion,
+				CipherSuites: tlsSettings.Options.CipherSuites,
+				ALPN:         tlsSettings.Options.ALPN,
+			})
+			cc = tls.Client(cc, cfg)
+		}
+		return node, cc, nil
 	}
-	return
+
+	// All retries exhausted
+	ho.Log.Warnf("all nodes failed for host %s", host)
+	res.Write(conn)
+	if lastErr != nil {
+		return nil, nil, lastErr
+	}
+	return nil, nil, errors.New("all nodes failed")
 }
 
 func (h *Sniffer) serveH2(ctx context.Context, conn net.Conn, ho *HandleOptions) error {
@@ -847,74 +881,105 @@ func (h *Sniffer) dialTLS(ctx context.Context, host string, ho *HandleOptions) (
 		return
 	}
 
-	if host != "" {
-		node = &chain.Node{
-			Addr: host,
-		}
-	}
-
 	ro := ho.RecorderObject
-	if ho.Hop != nil {
-		node = ho.Hop.Select(ctx,
-			hop.ClientIPSelectOption(net.ParseIP(ro.ClientIP)),
-			hop.HostSelectOption(host),
-			hop.ProtocolSelectOption(sniffing.ProtoTLS),
-		)
+
+	// Determine max retry attempts
+	maxRetries := 1
+	if nl, ok := ho.Hop.(hop.NodeList); ok {
+		maxRetries = len(nl.Nodes())
 	}
-	if node == nil {
-		err = errors.New("node not available")
-		return
+	if maxRetries <= 0 {
+		maxRetries = 1
 	}
 
-	addr := node.Addr
-	if opts := node.Options(); opts != nil {
-		switch opts.Network {
-		case "unix":
-			ro.Network = opts.Network
-		default:
-			if _, _, err := net.SplitHostPort(addr); err != nil {
-				addr += ":443"
+	var triedNodes []string
+	var lastErr error
+
+	for attempt := 0; attempt < maxRetries; attempt++ {
+		// Select a node, excluding previously tried nodes
+		selectCtx := ctxvalue.ContextWithExcludeNodes(ctx, triedNodes)
+
+		node = nil
+		if host != "" {
+			node = chain.NewNode("", host)
+		}
+		if ho.Hop != nil {
+			node = ho.Hop.Select(selectCtx,
+				hop.ClientIPSelectOption(net.ParseIP(ro.ClientIP)),
+				hop.HostSelectOption(host),
+				hop.ProtocolSelectOption(sniffing.ProtoTLS),
+			)
+		}
+		if node == nil {
+			if lastErr != nil {
+				ho.Log.Warnf("node for %s not found after retries", host)
+				return nil, nil, lastErr
+			}
+			ho.Log.Warnf("node for %s not found", host)
+			return nil, nil, errors.New("node not available")
+		}
+
+		// Track this node as tried
+		triedNodes = append(triedNodes, node.Addr)
+
+		addr := node.Addr
+		if opts := node.Options(); opts != nil {
+			switch opts.Network {
+			case "unix":
+				ro.Network = opts.Network
+			default:
+				if _, _, err := net.SplitHostPort(addr); err != nil {
+					addr += ":443"
+				}
 			}
 		}
-	}
-	ro.Host = addr
+		ro.Host = addr
 
-	ho.Log = ho.Log.WithFields(map[string]any{
-		"host": host,
-		"node": node.Name,
-		"dst":  fmt.Sprintf("%s/%s", addr, ro.Network),
-	})
-	ho.Log.Debugf("find node for host %s -> %s(%s)", host, node.Name, addr)
-
-	cc, err = dial(ctx, ro.Network, addr)
-	if err != nil {
-		// TODO: the router itself may be failed due to the failed node in the router,
-		// the dead marker may be a wrong operation.
-		if marker := node.Marker(); marker != nil {
-			marker.Mark()
-		}
-		ho.Log.Warnf("connect to node %s(%s) failed: %v", node.Name, node.Addr, err)
-		return
-	}
-
-	if marker := node.Marker(); marker != nil {
-		marker.Reset()
-	}
-
-	if tlsSettings := node.Options().TLS; tlsSettings != nil {
-		cfg := &tls.Config{
-			ServerName:         tlsSettings.ServerName,
-			InsecureSkipVerify: !tlsSettings.Secure,
-		}
-		tls_util.SetTLSOptions(cfg, &config.TLSOptions{
-			MinVersion:   tlsSettings.Options.MinVersion,
-			MaxVersion:   tlsSettings.Options.MaxVersion,
-			CipherSuites: tlsSettings.Options.CipherSuites,
-			ALPN:         tlsSettings.Options.ALPN,
+		ho.Log = ho.Log.WithFields(map[string]any{
+			"host": host,
+			"node": node.Name,
+			"dst":  fmt.Sprintf("%s/%s", addr, ro.Network),
 		})
-		cc = tls.Client(cc, cfg)
+		ho.Log.Debugf("find node for host %s -> %s(%s)", host, node.Name, addr)
+
+		cc, err = dial(ctx, ro.Network, addr)
+		if err != nil {
+			// Mark node as failed for future selections
+			if marker := node.Marker(); marker != nil {
+				marker.Mark()
+			}
+			ho.Log.Warnf("connect to node %s(%s) failed: %v, trying next node", node.Name, node.Addr, err)
+			lastErr = err
+			continue
+		}
+
+		// Success - reset marker
+		if marker := node.Marker(); marker != nil {
+			marker.Reset()
+		}
+
+		if tlsSettings := node.Options().TLS; tlsSettings != nil {
+			cfg := &tls.Config{
+				ServerName:         tlsSettings.ServerName,
+				InsecureSkipVerify: !tlsSettings.Secure,
+			}
+			tls_util.SetTLSOptions(cfg, &config.TLSOptions{
+				MinVersion:   tlsSettings.Options.MinVersion,
+				MaxVersion:   tlsSettings.Options.MaxVersion,
+				CipherSuites: tlsSettings.Options.CipherSuites,
+				ALPN:         tlsSettings.Options.ALPN,
+			})
+			cc = tls.Client(cc, cfg)
+		}
+		return node, cc, nil
 	}
-	return
+
+	// All retries exhausted
+	ho.Log.Warnf("all nodes failed for host %s", host)
+	if lastErr != nil {
+		return nil, nil, lastErr
+	}
+	return nil, nil, errors.New("all nodes failed")
 }
 
 func (h *Sniffer) terminateTLS(ctx context.Context, conn, cc net.Conn, clientHello *dissector.ClientHelloInfo, ho *HandleOptions) error {

go-gost/x/registry/AGENTS.md

@@ -0,0 +1,29 @@
+# GO-GOST REGISTRY KNOWLEDGE BASE
+
+**Generated:** Wed Feb 04 2026
+
+## OVERVIEW
+Central registration point for all pluggable GOST components (handlers, listeners, dialers, etc.).
+Allows the configuration system to resolve string types (e.g., "socks5") to actual Go implementations.
+
+## STRUCTURE
+One file per component type, exporting a standard Registry interface.
+```
+go-gost/x/registry/
+├── handler.go    # RegisterHandler(name, newFunc)
+├── listener.go   # RegisterListener(name, newFunc)
+├── dialer.go     # RegisterDialer(name, newFunc)
+└── ...           # Same pattern for auth, bypass, admission
+```
+
+## WHERE TO LOOK
+| Task | Location | Notes |
+|------|----------|-------|
+| Register a new component | `go-gost/x/registry/{type}.go` | Use `Register{Type}(name, creator)` |
+| Component lookup | `go-gost/x/registry/{type}.go` | `Get{Type}(name)` returns the creator function |
+| Default registrations | `go-gost/x/` (init functions) | Most components register themselves in their package `init()` |
+
+## CONVENTIONS
+- Thread-safe maps used for storage.
+- Names are case-sensitive (usually lowercase).
+- Components must be registered *before* the configuration parser runs (usually done via `import _ "..."` in `main.go`).

go-gost/x/selector/filter.go

@@ -2,11 +2,13 @@ package selector
 
 import (
 	"context"
+	"fmt"
 	"time"
 
+	"github.com/go-gost/core/chain"
 	"github.com/go-gost/core/metadata"
-	mdutil "github.com/go-gost/x/metadata/util"
 	"github.com/go-gost/core/selector"
+	mdutil "github.com/go-gost/x/metadata/util"
 )
 
 type failFilter[T any] struct {
@@ -24,6 +26,8 @@ func FailFilter[T any](maxFails int, timeout time.Duration) selector.Filter[T] {
 }
 
 // Filter filters dead objects.
+// For single-node case, skip filtering to ensure availability (matches upstream).
+// For multi-node case, filter out failed nodes to enable failover.
 func (f *failFilter[T]) Filter(ctx context.Context, vs ...T) []T {
 	if len(vs) <= 1 {
 		return vs
@@ -51,8 +55,21 @@ func (f *failFilter[T]) Filter(ctx context.Context, vs ...T) []T {
 
 		if mi, _ := any(v).(selector.Markable); mi != nil {
 			if marker := mi.Marker(); marker != nil {
-				if marker.Count() < int64(maxFails) ||
-					time.Since(marker.Time()) >= failTimeout {
+				count := marker.Count()
+				timeSince := time.Since(marker.Time())
+				passed := count < int64(maxFails) || timeSince >= failTimeout
+
+				// Debug logging for failover analysis
+				nodeName := "unknown"
+				nodeAddr := "unknown"
+				if node, ok := any(v).(*chain.Node); ok {
+					nodeName = node.Name
+					nodeAddr = node.Addr
+				}
+				fmt.Printf("[FailFilter] node=%s addr=%s count=%d maxFails=%d timeSince=%v failTimeout=%v passed=%v\n",
+					nodeName, nodeAddr, count, maxFails, timeSince, failTimeout, passed)
+
+				if passed {
 					l = append(l, v)
 				}
 				continue

vite-frontend/package.json

@@ -54,6 +54,7 @@
     "react-beautiful-dnd": "^13.1.1",
     "react-dom": "18.3.1",
     "react-hot-toast": "^2.5.2",
+    "react-is": "^19.2.4",
     "react-router-dom": "6.23.0",
     "recharts": "^3.1.1",
     "sonner": "^2.0.6",