fix: 提升集群的稳定性；优化路由表；修复节点断开连接不更新的问题

Author: Cyaim

Cyaim.WebSocketServer/Cyaim.WebSocketServer/Infrastructure/Cluster/ClusterManager.cs

@@ -101,6 +101,32 @@ public async Task StartAsync()
             }
         }
 
+        /// <summary>
+        /// Gracefully shutdown the cluster with connection transfer / 优雅关闭集群并转移连接
+        /// </summary>
+        /// <param name="force">Force shutdown without transfer / 强制关闭不转移连接</param>
+        /// <returns>Task / 任务</returns>
+        public async Task ShutdownAsync(bool force = false)
+        {
+            _logger.LogInformation($"Shutting down cluster manager for node {_nodeId} (force: {force})");
+
+            try
+            {
+                if (!force)
+                {
+                    // Graceful shutdown: transfer connections before stopping / 优雅关闭：在停止前转移连接
+                    await _router.GracefulShutdownAsync(_clusterOption);
+                }
+
+                await StopAsync();
+            }
+            catch (Exception ex)
+            {
+                _logger.LogError(ex, $"Error during cluster shutdown for node {_nodeId}");
+                throw;
+            }
+        }
+
         /// <summary>
         /// Stop the cluster
         /// 停止集群

Cyaim.WebSocketServer/Cyaim.WebSocketServer/Infrastructure/Cluster/ClusterRouter.cs

@@ -206,6 +206,22 @@ public async Task ConnectionEntry(HttpContext context, ILogger<WebSocketRouteMid
                         var currentNodeId = Infrastructure.Cluster.GlobalClusterCenter.ClusterContext?.NodeId;
                         _metricsCollector?.RecordConnectionEstablished(currentNodeId, context.Request.Path);
 
+                        // Register connection with cluster manager if cluster is enabled
+                        // 如果启用了集群，向集群管理器注册连接
+                        var clusterManager = Infrastructure.Cluster.GlobalClusterCenter.ClusterManager;
+                        if (clusterManager != null)
+                        {
+                            try
+                            {
+                                await clusterManager.RegisterConnectionAsync(context.Connection.Id, context.Request.Path);
+                                logger.LogDebug($"Registered connection {context.Connection.Id} with cluster manager");
+                            }
+                            catch (Exception ex)
+                            {
+                                logger.LogWarning(ex, $"Failed to register connection {context.Connection.Id} with cluster manager");
+                            }
+                        }
+
                         // 执行BeforeReceivingData管道
                         _ = await InvokePipeline(RequestPipelineStage.Connected, context, webSocket, null, null);
 
@@ -1039,6 +1055,22 @@ private async Task MvcChannel_OnDisconnected(HttpContext context, WebSocketClose
                 if (wsExists)
                 {
                     Clients.TryRemove(context.Connection.Id, out var _);
+                    
+                    // Unregister connection from cluster manager if cluster is enabled
+                    // 如果启用了集群，从集群管理器注销连接
+                    var clusterManager = Infrastructure.Cluster.GlobalClusterCenter.ClusterManager;
+                    if (clusterManager != null)
+                    {
+                        try
+                        {
+                            await clusterManager.UnregisterConnectionAsync(context.Connection.Id);
+                            logger.LogDebug($"Unregistered connection {context.Connection.Id} from cluster manager");
+                        }
+                        catch (Exception ex)
+                        {
+                            logger.LogWarning(ex, $"Failed to unregister connection {context.Connection.Id} from cluster manager");
+                        }
+                    }
                 }
 
                 ParallelForwardLimitSlim?.Dispose();

Cyaim.WebSocketServer/Cyaim.WebSocketServer/Infrastructure/Handlers/MvcHandler/MvcChannelHandler.cs

@@ -88,11 +88,15 @@ public ActionResult<ApiResponse<List<ClientConnectionInfo>>> GetAll([FromQuery]
                     // If not found locally or on remote node, create basic info / 如果未在本地找到或在远程节点，创建基本信息
                     if (clientInfo == null)
                     {
+                        // If connection is in routing table, assume it's Open (disconnected connections are unregistered)
+                        // 如果连接在路由表中，假设它是 Open 状态（断开的连接会被注销）
+                        var state = "Open"; // Assume open if in routing table / 如果在路由表中，假设为 Open
+                        
                         clientInfo = new ClientConnectionInfo
                         {
                             ConnectionId = connectionId,
                             NodeId = targetNodeId,
-                            State = "Unknown", // Remote connections state is unknown / 远程连接状态未知
+                            State = state,
                             BytesSent = 0,
                             BytesReceived = 0,
                             MessagesSent = 0,

Cyaim.WebSocketServer/Dashboard/Cyaim.WebSocketServer.Dashboard/Controllers/ClientController.cs

@@ -108,11 +108,15 @@ public ActionResult<ApiResponse<List<ClientConnectionInfo>>> GetConnections()
                     // If not found locally or on remote node, create basic info / 如果未在本地找到或在远程节点，创建基本信息
                     if (clientInfo == null)
                     {
+                        // If connection is in routing table, assume it's Open (disconnected connections are unregistered)
+                        // 如果连接在路由表中，假设它是 Open 状态（断开的连接会被注销）
+                        var state = "Open"; // Assume open if in routing table / 如果在路由表中，假设为 Open
+                        
                         clientInfo = new ClientConnectionInfo
                         {
                             ConnectionId = connectionId,
                             NodeId = targetNodeId,
-                            State = "Unknown",
+                            State = state,
                             BytesSent = 0,
                             BytesReceived = 0,
                             MessagesSent = 0,

Cyaim.WebSocketServer/Dashboard/Cyaim.WebSocketServer.Dashboard/Controllers/StatisticsController.cs

@@ -23,22 +23,27 @@ public Dictionary<string, string> GetAllClusterConnections()
             var clusterManager = GlobalClusterCenter.ClusterManager;
             var currentNodeId = GlobalClusterCenter.ClusterContext?.NodeId ?? "unknown";
 
-            // Get local connections / 获取本地连接
-            if (MvcChannelHandler.Clients != null)
+            // Priority 1: Get connections from cluster routing table (most authoritative) / 优先级1：从集群路由表获取连接（最权威）
+            // This includes all connections from all nodes in the cluster / 这包括集群中所有节点的所有连接
+            if (clusterManager != null && clusterManager.ConnectionRoutes != null)
             {
-                foreach (var kvp in MvcChannelHandler.Clients)
+                foreach (var kvp in clusterManager.ConnectionRoutes)
                 {
-                    connections[kvp.Key] = currentNodeId;
+                    connections[kvp.Key] = kvp.Value;
                 }
             }
 
-            // Get connections from cluster routing table / 从集群路由表获取连接
-            if (clusterManager != null && clusterManager.ConnectionRoutes != null)
+            // Priority 2: Add local connections that might not be in routing table yet / 优先级2：添加可能尚未在路由表中的本地连接
+            // This ensures we don't miss any local connections that haven't been registered yet / 这确保我们不会遗漏任何尚未注册的本地连接
+            if (MvcChannelHandler.Clients != null)
             {
-                foreach (var kvp in clusterManager.ConnectionRoutes)
+                foreach (var kvp in MvcChannelHandler.Clients)
                 {
-                    // Merge with local connections, cluster routes take precedence / 与本地连接合并，集群路由优先
-                    connections[kvp.Key] = kvp.Value;
+                    // Only add if not already in connections (cluster routes take precedence) / 仅在尚未在连接中时添加（集群路由优先）
+                    if (!connections.ContainsKey(kvp.Key))
+                    {
+                        connections[kvp.Key] = currentNodeId;
+                    }
                 }
             }
 

Cyaim.WebSocketServer/Dashboard/Cyaim.WebSocketServer.Dashboard/Services/DashboardHelperService.cs

@@ -229,21 +229,31 @@
             });
         });
 
-        // 在应用关闭时停止集群
+        // 在应用关闭时停止集群（优雅关闭）
         lifetime.ApplicationStopping.Register(() =>
         {
             Task.Run(async () =>
             {
                 try
                 {
-                    await clusterManager.StopAsync();
-                    logger.LogInformation($"集群节点 {nodeId} 已停止");
+                    // Graceful shutdown: transfer connections before stopping / 优雅关闭：在停止前转移连接
+                    await clusterManager.ShutdownAsync(force: false);
+                    logger.LogInformation($"集群节点 {nodeId} 已优雅关闭");
                 }
                 catch (Exception ex)
                 {
-                    logger.LogError(ex, "停止集群时发生错误");
+                    logger.LogError(ex, "优雅关闭集群时发生错误，尝试强制关闭");
+                    try
+                    {
+                        // Force shutdown if graceful shutdown fails / 如果优雅关闭失败，强制关闭
+                        await clusterManager.ShutdownAsync(force: true);
+                    }
+                    catch (Exception forceEx)
+                    {
+                        logger.LogError(forceEx, "强制关闭集群时发生错误");
+                    }
                 }
-            }).Wait(TimeSpan.FromSeconds(5)); // 等待最多5秒
+            }).Wait(TimeSpan.FromSeconds(30)); // 等待最多30秒（优雅关闭需要更多时间）
         });
     }
     catch (Exception ex)