Socket.IO树莓派5连接频繁断开进入重连循环问题求助

问题背景 #

我正在开展一个项目，使用Socket.IO在外部服务器与树莓派5之间建立低延迟连接。该树莓派同时作为客户端和中继服务器的宿主。连接能正常运行一段时间，但随后会进入断开重连循环。

故障表现日志 #

INFO:socketio.client:Connection failed, new attempt in 5.01 seconds
INFO:geventwebsocket.handler:192.168.1.103 - - [2024-08-20 10:26:38] "GET /socket.io/?transport=polling&EIO=4&sid=BlFj5QtkLH1eZUyYAAAF&t=1724145973.7493463 HTTP/1.1" 200 157 25.001898
DEBUG:geventwebsocket.handler:Initializing WebSocket
DEBUG:geventwebsocket.handler:Validating WebSocket request
DEBUG:geventwebsocket.handler:Can only upgrade connection if using GET method.
INFO:geventwebsocket.handler:192.168.1.103 - - [2024-08-20 10:26:38] "POST /socket.io/?transport=polling&EIO=4&sid=BlFj5QtkLH1eZUyYAAAF HTTP/1.1" 200 143 0.000526
DEBUG:geventwebsocket.handler:Initializing WebSocket
DEBUG:geventwebsocket.handler:Validating WebSocket request
Connection failed, new attempt in 5.46 seconds
INFO:socketio.client:Connection failed, new attempt in 5.46 seconds
Connection failed, new attempt in 4.57 seconds
INFO:socketio.client:Connection failed, new attempt in 4.57 seconds
Received packet PING data 
INFO:engineio.client:Received packet PING data 
Sending packet PONG data 
INFO:engineio.client:Sending packet PONG data 
Connection failed, new attempt in 4.92 seconds
INFO:socketio.client:Connection failed, new attempt in 4.92 seconds
Connection failed, new attempt in 5.00 seconds
INFO:socketio.client:Connection failed, new attempt in 5.00 seconds
Connection failed, new attempt in 4.62 seconds
INFO:socketio.client:Connection failed, new attempt in 4.62 seconds
....
INFO:socketio.client:Connection failed, new attempt in 5.04 seconds
Received packet PING data 
INFO:engineio.client:Received packet PING data 
Sending packet PONG data 
INFO:engineio.client:Sending packet PONG data 
WebSocket connection was closed, aborting
WARNING:engineio.client:WebSocket connection was closed, aborting
Waiting for write loop task to end
INFO:engineio.client:Waiting for write loop task to end
Exiting write loop task
INFO:engineio.client:Exiting write loop task
Engine.IO connection dropped
INFO:socketio.client:Engine.IO connection dropped
2024-08-20 10:29:18,487 INFO: Disconnected from external server
DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): [Url hidden for security]
DEBUG:urllib3.connectionpool: [Url hidden for security] HTTP/1.1" 200 20
INFO:__main__:Disconnected from external server
2024-08-20 10:29:18,984 INFO: Already connected to external server. Skipping reconnection.

当前连接与重连处理代码 #

def connect_to_external_server():
    global failure_count
    connected = False
    while not connected:
        try:
            if not sio.connected:
                logger.info(f"Connecting to external server at {EXTERNAL_SERVER_URL}...")
                sio.connect(EXTERNAL_SERVER_URL, transports=['websocket'])
                connected = True
                logger.info("Successfully connected to external server.")
            else:
                logger.info("Already connected to external server. Skipping reconnection.")
                connected = True
        except socketio.exceptions.ConnectionError as e:
            logger.error(f"Connection to external server failed: {e}. Retrying...")
            failure_count += 1
            if failure_count >= max_failures:
                logger.error(f"Maximum connection failures reached. Restarting script...")
                os.execv(sys.executable, [sys.executable] + sys.argv)
            sleep_time = min(30, 5 * 2 ** (failure_count - 1)) + random.uniform(0, 1)
            logger.info(f"Retrying in {sleep_time} seconds...")
            time.sleep(sleep_time)


# Event handlers for the external server
@sio.event
def connect():
    logger.info('Connected to external server')

@sio.event
def disconnect():
    logger.info('Disconnected from external server')
    connect_to_external_server()

求助问题 #

• 有没有人遇到过类似问题？
• 能否分析该重连循环的原因？
• 欢迎提供提升连接稳定性的建议。

我尝试过自动重启脚本，虽然有效，但这并非理想方案，因为需要处理大量请求。

问题分析与解决建议 #

可能的原因

1. WebSocket升级失败：日志中出现Can only upgrade connection if using GET method，说明从polling协议升级到WebSocket时出现方法不匹配问题，大概率是客户端与服务器的Socket.IO/Engine.IO版本不兼容，或是服务器端WebSocket路由配置错误。
2. 双重重连逻辑冲突：Socket.IO客户端默认开启自动重连机制，而你又在disconnect事件中手动触发connect_to_external_server，两种重连逻辑互相干扰，导致循环重连。
3. 网络与资源问题：树莓派所在网络可能存在间歇性丢包、NAT超时；或是长时间运行后出现内存泄漏、文件句柄耗尽，导致连接无法正常建立。
4. Ping/Pong超时不匹配：客户端与服务器的ping间隔、超时设置不一致，导致服务器判定客户端离线主动断开连接。

优化建议

• 统一重连逻辑：初始化Socket.IO客户端时禁用自动重连，避免与手动逻辑冲突：

  sio = socketio.Client(reconnection=False)
  

• 修复WebSocket升级问题：
  • 检查客户端与服务器的Socket.IO、Engine.IO版本，确保完全匹配；
  • 修改连接代码，允许polling作为备选传输方式：

    sio.connect(EXTERNAL_SERVER_URL, transports=['websocket', 'polling'])
    

• 重置失败计数：在连接成功时重置failure_count，避免后续重连使用过长的退避时间：

  @sio.event
  def connect():
      logger.info('Connected to external server')
      global failure_count
      failure_count = 0
  

• 监控系统资源：在树莓派上定期监控内存、CPU占用情况，排查是否存在资源泄漏；添加连接断开时的资源清理逻辑。
• 检查服务器端配置：确认外部服务器的Socket.IO连接超时、ping/pong间隔设置，避免因超时时间过短误判客户端离线。

内容的提问来源于stack exchange，提问作者dsh33