Merge pull request #90 from Sagit-chu/opencode/lucky-otter

fix(ws): stabilize node connectivity with ping/pong keepalive

.gitignore

@@ -177,6 +177,7 @@ build/
 *.dylib
 your_app.exe
 go-backend/paneld
+go-gost/gost
 
 # Go 测试二进制文件
 *.test

go-backend/internal/store/sqlite/repository.go

@@ -165,7 +165,13 @@ func Open(path string) (*Repository, error) {
 		return nil, err
 	}
 
-	db, err := sql.Open("sqlite", path)
+	// Use _pragma DSN parameters so every connection from the pool gets
+	// the same settings (busy_timeout and synchronous are per-connection).
+	dsn := "file:" + path +
+		"?_pragma=busy_timeout(5000)" +
+		"&_pragma=journal_mode(WAL)" +
+		"&_pragma=synchronous(NORMAL)"
+	db, err := sql.Open("sqlite", dsn)
 	if err != nil {
 		return nil, err
 	}

go-backend/internal/ws/server.go

@@ -54,6 +54,12 @@ type pendingRequest struct {
 	ch     chan CommandResult
 }
 
+const (
+	wsPingPeriod = 15 * time.Second
+	wsPongWait   = 45 * time.Second
+	wsWriteWait  = 5 * time.Second
+)
+
 type CommandResult struct {
 	Type    string                 `json:"type"`
 	Success bool                   `json:"success"`
@@ -120,12 +126,19 @@ func (s *Server) handleAdmin(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 	cw := &connWrap{conn: conn}
+	_ = conn.SetReadDeadline(time.Now().Add(wsPongWait))
+	conn.SetPongHandler(func(string) error {
+		return conn.SetReadDeadline(time.Now().Add(wsPongWait))
+	})
+	done := make(chan struct{})
+	go startKeepalive(cw, done)
 
 	s.mu.Lock()
 	s.admins[cw] = struct{}{}
 	s.mu.Unlock()
 
 	defer func() {
+		close(done)
 		s.mu.Lock()
 		delete(s.admins, cw)
 		s.mu.Unlock()
@@ -145,6 +158,12 @@ func (s *Server) handleNode(w http.ResponseWriter, r *http.Request, nodeID int64
 		return
 	}
 	cw := &connWrap{conn: conn}
+	_ = conn.SetReadDeadline(time.Now().Add(wsPongWait))
+	conn.SetPongHandler(func(string) error {
+		return conn.SetReadDeadline(time.Now().Add(wsPongWait))
+	})
+	done := make(chan struct{})
+	go startKeepalive(cw, done)
 
 	version := r.URL.Query().Get("version")
 	httpVal := parseIntDefault(r.URL.Query().Get("http"), 0)
@@ -165,6 +184,7 @@ func (s *Server) handleNode(w http.ResponseWriter, r *http.Request, nodeID int64
 	s.broadcastStatus(nodeID, 1)
 
 	defer func() {
+		close(done)
 		needOfflineBroadcast := false
 		s.mu.Lock()
 		current, ok := s.nodes[nodeID]
@@ -442,3 +462,27 @@ func parseIntDefault(v string, fallback int) int {
 	}
 	return x
 }
+
+func startKeepalive(cw *connWrap, done <-chan struct{}) {
+	if cw == nil || cw.conn == nil {
+		return
+	}
+	ticker := time.NewTicker(wsPingPeriod)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-done:
+			return
+		case <-ticker.C:
+			cw.mu.Lock()
+			_ = cw.conn.SetWriteDeadline(time.Now().Add(wsWriteWait))
+			err := cw.conn.WriteMessage(websocket.PingMessage, nil)
+			cw.mu.Unlock()
+			if err != nil {
+				_ = cw.conn.Close()
+				return
+			}
+		}
+	}
+}

go-gost/x/socket/websocket_reporter.go

@@ -18,7 +18,6 @@ import (
 	"strconv"
 	"strings"
 	"sync" // 新增：用于管理连接状态的互斥锁
-	"syscall"
 	"time"
 
 	"github.com/go-gost/x/config"
@@ -92,6 +91,11 @@ type TcpPingResponse struct {
 	RequestId    string  `json:"requestId,omitempty"`
 }
 
+const (
+	reporterReadWait  = 60 * time.Second
+	reporterWriteWait = 5 * time.Second
+)
+
 type WebSocketReporter struct {
 	url            string
 	addr           string // 保存服务器地址
@@ -244,6 +248,14 @@ func (w *WebSocketReporter) connect() error {
 
 	w.conn = conn
 	w.connected = true
+	_ = conn.SetReadDeadline(time.Now().Add(reporterReadWait))
+	conn.SetPingHandler(func(appData string) error {
+		_ = conn.SetReadDeadline(time.Now().Add(reporterReadWait))
+		return conn.WriteControl(websocket.PongMessage, []byte(appData), time.Now().Add(reporterWriteWait))
+	})
+	conn.SetPongHandler(func(string) error {
+		return conn.SetReadDeadline(time.Now().Add(reporterReadWait))
+	})
 
 	// 设置关闭处理器来检测连接状态
 	w.conn.SetCloseHandler(func(code int, text string) error {
@@ -384,7 +396,7 @@ func (w *WebSocketReporter) receiveMessages() {
 			}
 
 			// 设置读取超时
-			conn.SetReadDeadline(time.Now().Add(30 * time.Second))
+			conn.SetReadDeadline(time.Now().Add(reporterReadWait))
 
 			messageType, message, err := conn.ReadMessage()
 			if err != nil {
@@ -1043,10 +1055,11 @@ func (w *WebSocketReporter) handleUpgradeAgent(data interface{}) error {
 	w.sendUpgradeProgress("installing", 80, "准备重启...")
 	fmt.Printf("✅ 升级包下载完成 (%d bytes), 准备重启...\n", downloaded)
 
-	// 执行重启脚本（分离进程，不阻塞当前进程）
+	// 执行重启脚本
+	// 使用 systemd-run 在独立的 transient unit 中运行重启脚本，
+	// 避免 systemctl stop 杀死 flux_agent cgroup 内所有进程（包括此脚本自身）导致 mv 未执行。
 	script := fmt.Sprintf("sleep 1 && systemctl stop flux_agent && mv %s %s && systemctl start flux_agent", tmpPath, binaryPath)
-	cmd := exec.Command("/bin/sh", "-c", script)
-	cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
+	cmd := exec.Command("systemd-run", "--quiet", "/bin/sh", "-c", script)
 	if err := cmd.Start(); err != nil {
 		os.Remove(tmpPath)
 		return fmt.Errorf("启动重启脚本失败: %v", err)
@@ -1068,10 +1081,9 @@ func (w *WebSocketReporter) handleRollbackAgent(data interface{}) error {
 
 	fmt.Println("🔄 开始回退到旧版本...")
 
-	// 执行回退脚本
+	// 执行回退脚本（同升级逻辑，使用 systemd-run 避免 cgroup 问题）
 	script := fmt.Sprintf("sleep 1 && systemctl stop flux_agent && cp %s %s && systemctl start flux_agent", backupPath, binaryPath)
-	cmd := exec.Command("/bin/sh", "-c", script)
-	cmd.SysProcAttr = &syscall.SysProcAttr{Setpgid: true}
+	cmd := exec.Command("systemd-run", "--quiet", "/bin/sh", "-c", script)
 	if err := cmd.Start(); err != nil {
 		return fmt.Errorf("启动回退脚本失败: %v", err)
 	}

vite-frontend/src/pages/node.tsx

@@ -125,7 +125,7 @@ const SortableItem = ({
   };
 
   return (
-    <div ref={setNodeRef} style={style} {...attributes}>
+    <div ref={setNodeRef} style={style} {...attributes} className="overflow-hidden">
       {children(listeners)}
     </div>
   );
@@ -1253,7 +1253,7 @@ export default function NodePage() {
                     {(listeners) => (
                     <Card
                       key={node.id}
-                      className="group shadow-sm border border-divider hover:shadow-md transition-shadow duration-200"
+                      className="group shadow-sm border border-divider hover:shadow-md transition-shadow duration-200 overflow-hidden"
                     >
                       <CardHeader className="pb-2">
                         <div className="flex justify-between items-start w-full">
@@ -1514,54 +1514,56 @@ export default function NodePage() {
 
                         {/* 操作按钮 */}
                         <div className="space-y-1.5">
-                          <div className="flex gap-1.5">
+                          {!isRemoteNode && (
+                            <div className="grid grid-cols-3 gap-1.5">
+                              <Button
+                                className="min-h-8"
+                                color="success"
+                                isLoading={node.copyLoading}
+                                size="sm"
+                                variant="flat"
+                                onPress={() => handleCopyInstallCommand(node)}
+                              >
+                                安装
+                              </Button>
+                              <Button
+                                className="min-h-8"
+                                color="warning"
+                                isDisabled={node.connectionStatus !== "online"}
+                                isLoading={node.upgradeLoading}
+                                size="sm"
+                                variant="flat"
+                                onPress={() => openUpgradeModal("single", node.id)}
+                              >
+                                升级
+                              </Button>
+                              <Button
+                                className="min-h-8"
+                                color="secondary"
+                                isDisabled={node.connectionStatus !== "online"}
+                                isLoading={node.rollbackLoading}
+                                size="sm"
+                                variant="flat"
+                                onPress={() => handleRollbackNode(node)}
+                              >
+                                回退
+                              </Button>
+                            </div>
+                          )}
+                          <div className={`grid gap-1.5 ${isRemoteNode ? "grid-cols-1" : "grid-cols-2"}`}>
                             {!isRemoteNode && (
-                              <>
-                                <Button
-                                  className="flex-1 min-h-8"
-                                  color="success"
-                                  isLoading={node.copyLoading}
-                                  size="sm"
-                                  variant="flat"
-                                  onPress={() => handleCopyInstallCommand(node)}
-                                >
-                                  安装
-                                </Button>
-                                <Button
-                                  className="flex-1 min-h-8"
-                                  color="warning"
-                                  isDisabled={node.connectionStatus !== "online"}
-                                  isLoading={node.upgradeLoading}
-                                  size="sm"
-                                  variant="flat"
-                                  onPress={() => openUpgradeModal("single", node.id)}
-                                >
-                                  升级
-                                </Button>
-                                <Button
-                                  className="flex-1 min-h-8"
-                                  color="secondary"
-                                  isDisabled={node.connectionStatus !== "online"}
-                                  isLoading={node.rollbackLoading}
-                                  size="sm"
-                                  variant="flat"
-                                  onPress={() => handleRollbackNode(node)}
-                                >
-                                  回退
-                                </Button>
-                                <Button
-                                  className="flex-1 min-h-8"
-                                  color="primary"
-                                  size="sm"
-                                  variant="flat"
-                                  onPress={() => handleEdit(node)}
-                                >
-                                  编辑
-                                </Button>
-                              </>
+                              <Button
+                                className="min-h-8"
+                                color="primary"
+                                size="sm"
+                                variant="flat"
+                                onPress={() => handleEdit(node)}
+                              >
+                                编辑
+                              </Button>
                             )}
                             <Button
-                              className={`min-h-8 ${isRemoteNode ? "w-full" : "flex-1"}`}
+                              className="min-h-8"
                               color="danger"
                               size="sm"
                               variant="flat"