Flask socketio disconnect, unable to reconnect

[Joe-ChenZ]

Have been having some issues with client trying to reconnect to the flask socketio server. Client is using python socketio. Internet is stable. Client tries to connect every 10s. Transport method is websocket. Reconnection is enabled too. There is no CORS issue. Error I got is transport error.

[miguelgrinberg]

Please provide details and code so that I can see the issue here.

[Joe-ChenZ]

sorry for getting back late. The issue is that whenever i have a deployment of the backend flask server, the python-socketio clients get disconnected and are not able to reconnect later. This seems to only happen to AWS deployment, using a load balancer. I am using Nginx, not sure it is reproducible in other load balancer settings. This does not happen when i run with localhost. Suspect it is a load balancer setup issue.

Using flask socketio 5.5.1, python-socketio 5.13.0

basic client setup:

self.socketio_client = socketio.AsyncClient(handle_sigint=False, logger=client_logger)

 await self.socketio_client.connect(
                        self.backend_url,
                        auth=auth,
                        namespaces=[self.namespace],
                        transports=["websocket"],
                        wait_timeout=5,
                        retry=True,
)

Nginx config

# Elastic Beanstalk Nginx Configuration File

user                    nginx;
error_log               /var/log/nginx/error.log warn;
pid                     /var/run/nginx.pid;
worker_processes        auto;
worker_rlimit_nofile    200000;

events {
    worker_connections  5000;
}

http {
    include       /etc/nginx/mime.types;
    default_type  application/octet-stream;
    types_hash_max_size 2048;
    types_hash_bucket_size 128;

    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                      '$status $body_bytes_sent "$http_referer" '
                      '"$http_user_agent" "$http_x_forwarded_for"';

    include       conf.d/*.conf;

    map $http_upgrade $connection_upgrade {
        default     "upgrade";
    }

    ## Hide Nginx version ##
    server_tokens   off;


    # Add all of our node-balanced gunicorn instances
    upstream scale_nodes {
       least_conn;

       server 127.0.0.1:8000;
       server 127.0.0.1:8001;
       server 127.0.0.1:8002;
       server 127.0.0.1:8003;
       server 127.0.0.1:8004;
       server 127.0.0.1:8005;
       # to scale the app, just add more nodes here! (Update Procfile)
    }
    server {
        listen        80 default_server;
        access_log    /var/log/nginx/access.log main;

        client_header_timeout 60;
        client_body_timeout   60;
        keepalive_timeout     60;
        gzip                  off;
        gzip_comp_level       4;
        gzip_types text/plain text/css application/json application/javascript application/x-javascript text/xml application/xml application/xml+rss text/javascript;

        ## Security headers for Nginx ##
        add_header Strict-Transport-Security "max-age=31536000; includeSubDomains" always;
        add_header X-Content-Type-Options "nosniff" always;
        add_header X-Frame-Options "SAMEORIGIN" always;
        add_header X-Xss-Protection "1; mode=block" always;
        add_header Referrer-Policy  strict-origin-when-cross-origin;
        add_header Feature-policy "accelerometer 'none'; camera 'none'; geolocation 'none'; gyroscope 'none'; magnetometer 'none'; microphone 'none'; payment 'none'; usb 'none'";
        add_header Content-Security-Policy "default-src 'self' http: https: data: blob: ws: 'unsafe-inline' 'unsafe-eval'" always;
        ssl_protocols TLSv1.3;

        # Include the Elastic Beanstalk generated locations
        include conf.d/elasticbeanstalk/*.conf;
    }
}

[miguelgrinberg]

Do you have the logs for your client and your server?

[Joe-ChenZ]

Client logs below. Servers didn't log anything wrong. No errors. No errors in nginx logs in aws either

2025-09-22 13:16:58,439 INFO    Received event "authenticate_connect" [/computers]
2025-09-22 13:16:58,440 INFO    SocketIO authentication complete
2025-09-22 13:16:58,440 INFO    Trying to broadcast_lock_status via socketio to the backend
2025-09-22 13:16:58,498 INFO    Connection to namespace /computers was rejected
2025-09-22 13:16:58,498 ERROR   SocketIO connection error on namespace Unable to connect: ()
2025-09-22 13:16:58,498 INFO    Engine.IO connection dropped
2025-09-22 13:16:58,498 INFO    Disconnected from the backend via socketio, reason: client disconnect
2025-09-22 13:16:58,581  ERROR   Exception thrown when trying to broadcast lock status to the backend socketio room.
Traceback (most recent call last):
  File "/home/user/path/interface/socket_io.py", line 635, in _broadcast_lock_status
    await self.socketio_client.emit(
  File "/home/user/path/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /computers is not a connected namespace.
2025-09-22 13:16:58,582 INFO    Connected to and emitted to broadcast lock status to the backend via socketio!
2025-09-22 13:17:08,591 INFO    Reconnecting after clean socketio disconnection
2025-09-22 13:17:08,898 INFO    Engine.IO connection established
2025-09-22 13:17:13,899 INFO    Engine.IO connection dropped
2025-09-22 13:17:13,962 ERROR   Failed to connect to socketio, waiting 10 seconds and trying again
Traceback (most recent call last):
  File "/home/user/path/interface/socket_io.py", line 345, in _backend_connect
    await self.socketio_client.connect(
  File "/home/user/path/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 175, in connect
    raise exceptions.ConnectionError(
socketio.exceptions.ConnectionError: One or more namespaces failed to connect
2025-09-22 13:17:24,23 INFO    Engine.IO connection established
2025-09-22 13:17:24,323 INFO    Namespace /computers is connected
2025-09-22 13:17:24,323  INFO    SocketIO connected successfully.
2025-09-22 13:17:24,386  INFO    Engine.IO connection dropped
2025-09-22 13:17:24,387 INFO    Disconnected from the backend via socketio, reason: transport error
2025-09-22 13:17:24,387 ERROR   SocketIO transport error
2025-09-22 13:17:24,387 INFO    Connection failed, new attempt in 1.29 seconds

[Joe-ChenZ]

await self.socketio_client.disconnect()
client does call disconnect before calling the connect again.
This is what happens in the server.

2025-09-22 18:52:44,972  DEBUG   Got a disconnection, updating online status appropriately
2025-09-22 18:52:45,007  DEBUG   Disconnect sid:EejcZlj0JE3FNN-QAABY (uuid:15c70a2a-dc32-4b5e-a6b1-0f29ec99fcaa) from namespace /computers with reason client disconnect!

When the client calls disconnect, my backend isn't specifically doing self.disconnect(), that might be the reason?

Otherwise, this seems to be some race conditions. The backend doesn't disconnect fast enough before the client calls connect again. Client thinks it is already connected. Connect triggers an exception. I try except that and retry the whole process

[Joe-ChenZ]

added self.disconnect() in my server in def on_disconnect(). Client still would not reconnect.

2025-09-22 14:15:26,950 INFO    Namespace /computers is connected
2025-09-22 14:15:26,950 INFO    Reconnection successful
2025-09-22 14:15:26,960 INFO    Received event "authenticate_connect" [/computers]
2025-09-22 14:15:26,960 INFO    SocketIO authentication complete
2025-09-22 14:15:26,960 INFO    Trying to broadcast_lock_status via socketio to the backend
2025-09-22 14:15:27,019 INFO    Connection to namespace /computers was rejected
2025-09-22 14:15:27,019 ERROR   SocketIO connection error on namespace Unable to connect: () 
Note, for this line above. the same namespace (a string) that i provide to socket_client.connect() prints out to be `Unable to connect` () is the args provided, which is empty

2025-09-22 14:15:27,019 INFO    Engine.IO connection dropped
2025-09-22 14:15:27,020 INFO    Disconnected from the backend via socketio, reason: client disconnect
2025-09-22 14:15:27,143 ERROR   Exception thrown when trying to broadcast lock status to the backend socketio room.
Traceback (most recent call last):
  File "/home/user/path/interface/socket_io.py", line 635, in _broadcast_lock_status
    await self.socketio_client.emit(
  File "/home/user/path/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /computers is not a connected namespace.
2025-09-22 14:15:27,144 INFO    Connected to and emitted to broadcast lock status to the backend via socketio!
2025-09-22 14:15:37,117 INFO    Reconnecting after clean socketio disconnection
2025-09-22 14:15:37,382 INFO    Engine.IO connection established
2025-09-22 14:15:42,384 INFO    Engine.IO connection dropped
2025-09-22 14:15:42,446 ERROR   Failed to connect to socketio, waiting 10 seconds and trying again
Traceback (most recent call last):
  File "/home//socket_io.py", line 345, in _backend_connect
    await self.socketio_client.connect(
  File "/home/user/path/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 175, in connect
    raise exceptions.ConnectionError(
socketio.exceptions.ConnectionError: One or more namespaces failed to connect

[Joe-ChenZ]

it seems like the connection is initially established, and then transport error occurred almost immediately after, dropping the connection. I use a flag in socketio connect error handler to set disconnect event to true to let the client retry, and the cycle repeats.

[Joe-ChenZ]

I turned on the EngineIO logs on the client side. This is what i got that caused the issue at the end.
Received packet MESSAGE data 4/computers,"Unable to connect"

[miguelgrinberg]

I need to see the logs, please. Your logs shared above are for your own application, not for Socket.IO. The server-side Socket.IO also has a logging option. Please turn Socket.IO logging on both sides and show me the complete sessions.

[Joe-ChenZ]

I have multiple entities interacting with my backend, I am still trying to get the logs for my specific client, but on a high level, in the connect handler, my server emits an event, that requires the client to ack back for a cryptographic challenge. And this is a 4 way communication. Server would need to emit again and wait for another callback from client. I noticed that the packet headers are all "CONNECT" during those communications. Is this what causes the problem? I do see that in _handle_connect, if there is a bidict duplication error, it reutns sid = None, and this is the case for the client seeing "Unable to connect"

[miguelgrinberg]

The connect handler is the wrong place to do this. When the connect handler is invoked the connection isn't established yet. I suspect this is why you see CONNECT packets mixed with your emits. The connect handler's purpose is to approve or reject the client, it shouldn't do anything beyond that, and certainly not waiting for the other side to respond. This auth exchange should be done in regular events that are issued after the connect handler ended and the connection is fully established.

[Joe-ChenZ]

the connect handler we have emits sth to the client and returns True/False. It isn't busy waiting for the client. I pass a callback to the emit and the client will respond to the callback with some args. Although, the callback is in the on_connect handler. Does that make a difference if the callback function is defined in the connect handler or not? Would the packet header change from CONNECT to MESSAGE?

[miguelgrinberg]

I cannot give you advice without seeing code, I'm sorry. But note that packets do not "change" based on how you do things. There is always a CONNECT packet used for the connection, and MESSAGE packets are always used for messages. The problem is that if you do emits in your connect handler then the CONNECT packet is delayed until the handler returns with an acceptance or a rejection. You may want to read about the always_connect option in the docs, maybe that is useful if you intend to continue doing work in the connect handler.

[Joe-ChenZ]

Thank you. I did see the always_connect field. Seems useful in my case. Without that field, how does the client reconnect when the server goes down? for example, in the case of redeployment. It seems like i do get two of the same event the server only emits once.

Attempting WebSocket connection to wss://url/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'OSsnK1O7mPCpCTWaAAAA', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 10000000}
Sending packet MESSAGE data 0/computers,{}}
Exiting write loop task
Received packet MESSAGE data 0/computers,{"sid":"rqk5nkV_LNTZwp80AAAB"}
Sending packet MESSAGE data 2/computers,["authenticate",{"uuid":"21fb09db-c639-4196-a7cc-d295d6420f74"}]
Server sent close packet data 0, aborting
Waiting for write loop task to end



Exiting write loop task
Sending packet MESSAGE data 2/computers,["authenticate",{"uuid":"21fb09db-c639-4196-a7cc-d295d6420f74"}]
Server sent close packet data 0, aborting
Waiting for write loop task to end
Exiting read loop task
Attempting WebSocket connection to wss://url/socket.io/?transport=websocket&EIO=4
Sending packet MESSAGE data 0/computers,{}}
Received packet MESSAGE data 0/computers,{"sid":"3svI46PG0JnVBMXgAAAD"}
Sending packet MESSAGE data 2/computers,["authenticate",{"uuid":"21fb09db-c639-4196-a7cc-d295d6420f74"}]
Received packet MESSAGE data 4/computers,"Unable to connect"


Attempting WebSocket connection to wss://url.com/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'l1wMknivIriDCjSaAAAC', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 10000000}
Sending packet MESSAGE data 0/computers,{}
Received packet MESSAGE data 0/computers,{"sid":"4YpsR3ntakmoy1OgAAAD"}
Received packet MESSAGE data 4/computers,"Unable to connect"
Sending packet MESSAGE data 2/computers,["authenticate",{"uuid":"21fb09db-c639-4196-a7cc-d295d6420f74"}]
Received packet MESSAGE data 51-/computers,2["tee_msg",{"data":{"_placeholder":true,"num":0},"backend_tee":
2025-09-24 17:40:47,821 engineio.client INFO    Received packet MESSAGE data 
Sending packet MESSAGE data 61-/computers,2[{"status":true,"data":{"_placeholder":true,"num":0},"message":null}]
Sending packet MESSAGE data <binary>
Received packet MESSAGE data 51-/computers,4[{}]
Received packet MESSAGE data <binary>
Sending packet MESSAGE data 61-/computers,4[{"status":true,"data":{"_placeholder":true,"num":0},"message":null}]
Sending packet MESSAGE data <binary>
Received packet MESSAGE data 2/computers,["authenticate_connect",{"status":true,"data":null}]
Received packet MESSAGE data 2/computers,["authenticate_connect",{"status":true,"data":null}]

Traceback (most recent call last):
  File "/home/socket_io.py", line 623, in _broadcast_lock_status
    await self.socketio_client.emit(
  File "/home/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /computers is not a connected namespace.
Traceback (most recent call last):
  File "/home//socket_io.py", line 623, in _broadcast_lock_status
    await self.socketio_client.emit(
  File "/home/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /computers is not a connected namespace.

At the end, i get event authenticate twice. It appears that the backend thinks the client is connected. but the client doesn't have the namespace connected

[Joe-ChenZ]

This is the log after i took out the emit from on_connect handler

[Joe-ChenZ]

  async with TaskGroup() as tg:
            result = cls(
                backend_url=backend_url,
                task_group=tg,
            )
            result._setup()
            connect_task = tg.create_task(result._backend_connect())
            try:
                yield result
            finally:
                await result.close()
                connect_task.cancel()


 async def _backend_connect(self) -> None:
        """
        Connect to the backend
        """
        # More precise hardware info for the backend
        device_details = hardware_info_helper()

        auth = {
                 sth here
        }

        logger.info(f"Connecting to backend at {self.backend_url}")
        # Connect to our socketio server
        await self.socketio_client.connect(
            self.backend_url,
            auth=auth,
            namespaces=self.namespace,
            transports=["websocket"],
            retry=True,
            wait_timeout=RETRY_SECONDS,
        )
        logger.info("SocketIO connected successfully.")

looks sth like this for setting up the connection

[miguelgrinberg]

Unfortunately I cannot make sense of your logs. The order of events is strange, and I don't know if it is because there are multiple connections happening at the same time, or because something predating these logs happened that put the client in a bad state.

Let's do this. Please create small client and server examples that I can run here to see your problem and debug it, because without seeing the full code and the full logs I'm not sure I'll be able to tell you what's wrong.

[Joe-ChenZ]

Without that field, how does the client reconnect when the server goes down?

I don't understand this question. The always_connect field has nothing to do with the reconnect feature. It defines how incoming connections are accepted. If always_connect is True, then all connections are accepted before the connect handler is called. With False, the return value of the connect handler is the acceptance or rejection of the connection.

As I said before, the connect handler is not the right place to do any kind of logic aside from authentication. I added the always_connect flag to help people who insisted on not following this rule. :)

I am asking how does the in built socketio reconnection work? What is done under the hood when the client set reconnect to true

[miguelgrinberg]

See the AsyncClient._handle_reconnect method in the code. The logs that you shared do not show that reconnections are attempted. Not sure if you are including complete logs, though, maybe this is shown in parts of the logs that you aren't sharing.

EDIT: actually, there is one in the first log snippets that you shared:

2025-09-22 13:17:24,387 INFO    Connection failed, new attempt in 1.29 seconds

[Joe-ChenZ]

didn't realize those logs don't belong to engineio logs, didn't copy them. Yeah, client is retrying. Does client retry connection when it calls disconnect? I noticed that it requires eio to be connected, what if eio is disconnected? it seems that the client will disconnect the EIO layer when it calls disconnect

[Joe-ChenZ]

according to socketio docs,

Disconnection detection
The Engine.IO connection is considered as closed when:

one HTTP request (either GET or POST) fails (for example, when the server is shutdown)
the WebSocket connection is closed (for example, when the user closes the tab in its browser)
socket.disconnect() is called on the server-side or on the client-side

i guess that would close the connection

[Joe-ChenZ]

Ok. i was able to create a sample application to reproduce this exact error. Here you go. This is the client log

 python examples/test_sio_client.py                                                                                                                  12:05:31
Setting up our callbacks
SocketIO interface initialized successfully
Connecting to backend at http://127.0.0.1:8999
Attempting WebSocket connection to ws://127.0.0.1:8999/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'Q9DmYVsYo4TlM_msAAAA', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Received packet MESSAGE data 0/test,{"sid":"-xwIIE25DQKOv1WoAAAB"}
Namespace /test is connected
Trying to set up auth session with the backend...
Emitting event "authenticate" [/test]
Sending packet MESSAGE data 2/test,["authenticate",{}]
SocketIO connected successfully.
Received packet MESSAGE data 2/test,1["auth_msg",{}]
Received event "auth_msg" [/test]
Sending packet MESSAGE data 3/test,1[{"status":true}]
Received packet MESSAGE data 2/test,["auth_connect",{}]
Received event "auth_connect" [/test]
Emitting event "broadcast_status" [/test]
Sending packet MESSAGE data 2/test,["broadcast_status",{}]
Received packet PING data 
Sending packet PONG data 
packet queue is empty, aborting
Exiting write loop task
Server has stopped communicating, aborting
Waiting for write loop task to end
Engine.IO connection dropped
Disconnected from the backend via socketio, reason: transport error
Connection failed, new attempt in 1.24 seconds
Exiting read loop task
Attempting WebSocket connection to ws://127.0.0.1:8999/socket.io/?transport=websocket&EIO=4




SocketIO connection error on namespace /test with data: ('Connection error',)
Connection failed, new attempt in 1.67 seconds
Attempting WebSocket connection to ws://127.0.0.1:8999/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': '6VbvypcDigdJHGknAAAC', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Exiting write loop task

Sending packet CLOSE data None
Engine.IO connection dropped
Server sent close packet data None, aborting
Waiting for write loop task to end
Exiting read loop task
Connection failed, new attempt in 4.25 seconds
Attempting WebSocket connection to ws://127.0.0.1:8999/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'y1SZ2PYnoYHmoFRUAAAD', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Exiting write loop task
Received packet MESSAGE data 0/test,{"sid":"Y4ftqe8gP_efFPRsAAAE"}
Namespace /test is connected
Trying to set up auth session with the backend...
Emitting event "authenticate" [/test]
Sending packet MESSAGE data 2/test,["authenticate",{}]
Reconnection successful
Server sent close packet data 0, aborting
Waiting for write loop task to end
Engine.IO connection dropped
Disconnected from the backend via socketio, reason: transport error
Connection failed, new attempt in 1.11 seconds
Exiting read loop task
Attempting WebSocket connection to ws://127.0.0.1:8999/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'YGy-FKDBRzkhDXeEAAAF', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Received packet MESSAGE data 0/test,{"sid":"4BmNXhxGc9Vx96itAAAG"}
Namespace /test is connected
Trying to set up auth session with the backend...
Emitting event "authenticate" [/test]
Sending packet MESSAGE data 2/test,["authenticate",{}]
Reconnection successful
Received packet MESSAGE data 4/test,"Unable to connect"
Connection to namespace /test was rejected
SocketIO connection error on namespace /test with data: ('Unable to connect',)
Received packet MESSAGE data 2/test,1["auth_msg",{}]
Received packet MESSAGE data 2/test,2["auth_msg",{}]
Received event "auth_msg" [/test]
Sending packet MESSAGE data 3/test,1[{"status":true}]
Received event "auth_msg" [/test]
Sending packet MESSAGE data 3/test,2[{"status":true}]
Received packet MESSAGE data 2/test,["auth_connect",{}]
Received event "auth_connect" [/test]
Task exception was never retrieved
future: <Task finished name='Task-22' coro=<AsyncClient._handle_eio_message() done, defined at /home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py:558> exception=BadNamespaceError('/test is not a connected namespace.')>
Traceback (most recent call last):
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 575, in _handle_eio_message
    await self._handle_event(pkt.namespace, pkt.id, pkt.data)
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 413, in _handle_event
    r = await self._trigger_event(data[0], namespace, *data[1:])
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 467, in _trigger_event
    ret = await handler(*args)
          ^^^^^^^^^^^^^^^^^^^^
  File "/home/fidotesting/admin-backend/examples/test_sio_client.py", line 60, in _auth_connect
    await self.socketio_client.emit(
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /test is not a connected namespace.
Received packet MESSAGE data 2/test,["auth_connect",{}]
Received event "auth_connect" [/test]
Task exception was never retrieved
future: <Task finished name='Task-23' coro=<AsyncClient._handle_eio_message() done, defined at /home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py:558> exception=BadNamespaceError('/test is not a connected namespace.')>
Traceback (most recent call last):
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 575, in _handle_eio_message
    await self._handle_event(pkt.namespace, pkt.id, pkt.data)
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 413, in _handle_event
    r = await self._trigger_event(data[0], namespace, *data[1:])
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 467, in _trigger_event
    ret = await handler(*args)
          ^^^^^^^^^^^^^^^^^^^^
  File "/home/fidotesting/admin-backend/examples/test_sio_client.py", line 60, in _auth_connect
    await self.socketio_client.emit(
  File "/home/fidotesting/admin-backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /test is not a connected namespace.
Received packet PING data 
Sending packet PONG data 

[Joe-ChenZ]

Server logs

eceived event "authenticate" from -xwIIE25DQKOv1WoAAAB [/test]
client wants to authenticate... ({},)
emitting event "auth_msg" to -xwIIE25DQKOv1WoAAAB [/test]
Q9DmYVsYo4TlM_msAAAA: Sending packet MESSAGE data 2/test,1["auth_msg",{}]
Q9DmYVsYo4TlM_msAAAA: Received packet MESSAGE data 3/test,1[{"status":true}]
received ack from -xwIIE25DQKOv1WoAAAB [/test]
Authentication successful
emitting event "auth_connect" to -xwIIE25DQKOv1WoAAAB [/test]
Q9DmYVsYo4TlM_msAAAA: Sending packet MESSAGE data 2/test,["auth_connect",{}]
Q9DmYVsYo4TlM_msAAAA: Received packet MESSAGE data 2/test,["broadcast_status",{}]
received event "broadcast_status" from -xwIIE25DQKOv1WoAAAB [/test]
client broadcast status: ({},)
Q9DmYVsYo4TlM_msAAAA: Sending packet PING data None
Q9DmYVsYo4TlM_msAAAA: Client is gone, closing socket
Client disconnected -xwIIE25DQKOv1WoAAAB ping timeout
Q9DmYVsYo4TlM_msAAAA: Client is gone, closing socket


6VbvypcDigdJHGknAAAC: Sending packet OPEN data {'sid': '6VbvypcDigdJHGknAAAC', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
6VbvypcDigdJHGknAAAC: Received request to upgrade to websocket
6VbvypcDigdJHGknAAAC: Upgrade to websocket successful
y1SZ2PYnoYHmoFRUAAAD: Sending packet OPEN data {'sid': 'y1SZ2PYnoYHmoFRUAAAD', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
y1SZ2PYnoYHmoFRUAAAD: Received request to upgrade to websocket
y1SZ2PYnoYHmoFRUAAAD: Upgrade to websocket successful
y1SZ2PYnoYHmoFRUAAAD: Received packet MESSAGE data 0/test,{}
Client connected
y1SZ2PYnoYHmoFRUAAAD: Sending packet MESSAGE data 0/test,{"sid":"Y4ftqe8gP_efFPRsAAAE"}
y1SZ2PYnoYHmoFRUAAAD: Received packet CLOSE data 
Client disconnected Y4ftqe8gP_efFPRsAAAE client disconnect
YGy-FKDBRzkhDXeEAAAF: Sending packet OPEN data {'sid': 'YGy-FKDBRzkhDXeEAAAF', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
YGy-FKDBRzkhDXeEAAAF: Received request to upgrade to websocket
YGy-FKDBRzkhDXeEAAAF: Upgrade to websocket successful
YGy-FKDBRzkhDXeEAAAF: Received packet MESSAGE data 0/test,{}
Client connected
YGy-FKDBRzkhDXeEAAAF: Sending packet MESSAGE data 0/test,{"sid":"4BmNXhxGc9Vx96itAAAG"}
YGy-FKDBRzkhDXeEAAAF: Received packet MESSAGE data 2/test,["authenticate",{}]
received event "authenticate" from 4BmNXhxGc9Vx96itAAAG [/test]
YGy-FKDBRzkhDXeEAAAF: Received packet MESSAGE data 0/test,{}
YGy-FKDBRzkhDXeEAAAF: Sending packet MESSAGE data 4/test,"Unable to connect"
client wants to authenticate... ({},)
emitting event "auth_msg" to 4BmNXhxGc9Vx96itAAAG [/test]
YGy-FKDBRzkhDXeEAAAF: Sending packet MESSAGE data 2/test,1["auth_msg",{}]
YGy-FKDBRzkhDXeEAAAF: Received packet MESSAGE data 2/test,["authenticate",{}]
received event "authenticate" from 4BmNXhxGc9Vx96itAAAG [/test]
client wants to authenticate... ({},)
emitting event "auth_msg" to 4BmNXhxGc9Vx96itAAAG [/test]
YGy-FKDBRzkhDXeEAAAF: Sending packet MESSAGE data 2/test,2["auth_msg",{}]
YGy-FKDBRzkhDXeEAAAF: Received packet MESSAGE data 3/test,1[{"status":true}]
received ack from 4BmNXhxGc9Vx96itAAAG [/test]
Authentication successful
emitting event "auth_connect" to 4BmNXhxGc9Vx96itAAAG [/test]
YGy-FKDBRzkhDXeEAAAF: Sending packet MESSAGE data 2/test,["auth_connect",{}]
YGy-FKDBRzkhDXeEAAAF: Received packet MESSAGE data 3/test,2[{"status":true}]
received ack from 4BmNXhxGc9Vx96itAAAG [/test]
Authentication successful
emitting event "auth_connect" to 4BmNXhxGc9Vx96itAAAG [/test]
YGy-FKDBRzkhDXeEAAAF: Sending packet MESSAGE data 2/test,["auth_connect",{}]
YGy-FKDBRzkhDXeEAAAF: Sending packet PING data None
YGy-FKDBRzkhDXeEAAAF: Received packet PONG data 
YGy-FKDBRzkhDXeEAAAF: Sending packet PING data None
YGy-FKDBRzkhDXeEAAAF: Received packet PONG data 
YGy-FKDBRzkhDXeEAAAF: Sending packet PING data None
YGy-FKDBRzkhDXeEAAAF: Received packet PONG data 

[Joe-ChenZ]

Client program

from asyncio import TaskGroup, run
from contextlib import asynccontextmanager
from typing import Any

import socketio

RETRY_SECONDS = 10

class SocketIOInterface:
    def __init__(self):
        self.socketio_client = socketio.AsyncClient(
            handle_sigint=False,
            logger=True,
            engineio_logger=True,
        )
        self.namespace = "/test"
        self.backend_url = "http://127.0.0.1:8999"

    def _setup(self) -> None:
        print("Setting up our callbacks")
        self.socketio_client.on("connect", handler=self._start_authenticate, namespace=self.namespace)
        self.socketio_client.on(
            "disconnect",
            handler=self._socketio_disconnect,
            namespace=self.namespace,
        )
        self.socketio_client.on("auth_msg", handler=self._auth_msg, namespace=self.namespace)
        self.socketio_client.on("auth_connect", handler=self._auth_connect, namespace=self.namespace)
        self.socketio_client.on(
            "connect_error",
            handler=self._socketio_connect_error,
            namespace=self.namespace,
        )

    async def close(self) -> None:
        """Close our socketio client"""
        print("SocketIO interface shutting down")

        # Disconnect from the socketio server
        try:
            await self.socketio_client.disconnect()
        except Exception as e:
            print(f"Failed to disconnect from the backend via socketio {e}")

    async def _start_authenticate(self) -> None:
        """
        Start the authentication protocol
        """
        print(f"Trying to set up auth session with the backend...")
        await self.socketio_client.emit(
            "authenticate",
            {},
            namespace=self.namespace,
        )

    async def _auth_msg(self, *args) -> dict:
        return {"status": True}

    async def _auth_connect(self, *args):
        await self.socketio_client.emit(
            "broadcast_status",
            {},
            namespace=self.namespace,
        )


    async def _socketio_connect_error(self, *args: Any) -> None:
        """
        Connection to a particular namespace was dropped or rejected.
        """
        print(f"SocketIO connection error on namespace {self.namespace} with data: {args}")

    async def _socketio_disconnect(self, reason: Any) -> None:

        print(f"Disconnected from the backend via socketio, reason: {reason}")

    async def _backend_connect(self) -> None:
        """
        Connect to the backend
        """

        print(f"Connecting to backend at {self.backend_url}")
        # Connect to our socketio server
        await self.socketio_client.connect(
            self.backend_url,
            auth={},
            namespaces=self.namespace,
            transports=["websocket"],
            retry=True,
            wait_timeout=RETRY_SECONDS,
        )
        print("SocketIO connected successfully.")

    @classmethod
    @asynccontextmanager
    async def run(
        cls,
    ):
        """
        Runs the SocketIO service.
        """
        async with TaskGroup() as tg:
            result = cls()
            result._setup()
            connect_task = tg.create_task(result._backend_connect())
            try:
                yield result
            finally:
                await result.close()
                connect_task.cancel()

async def main():
    async with SocketIOInterface.run() as socketio_interface:
        print("SocketIO interface initialized successfully")

        # Keep the connection alive for a while to see events
        try:
            # Wait for some time to see connection events
            import asyncio
            await asyncio.sleep(10000)
        except asyncio.CancelledError:
            pass
if __name__ == '__main__':
    run(main())

[Joe-ChenZ]

Server Program:

from gevent import monkey

# Make multiprocessing happy with gevent. This should be at the top
monkey.patch_all()

from threading import Lock
from flask import Flask, request
from flask_socketio import SocketIO, emit

# Set this variable to "threading", "eventlet" or "gevent" to test the
# different async modes, or leave it set to None for the application to choose
# the best option based on installed packages.

app = Flask(__name__)
app.config['SECRET_KEY'] = 'secret!'
socketio = SocketIO(app, async_mode="gevent", logger=True, engineio_logger=True)
thread = None
thread_lock = Lock()


@socketio.on("connect", namespace="/test")
def connect():
    print("Client connected")
    return True

@socketio.on("authenticate", namespace="/test")
def authenticate(*args):
    print(f"client wants to authenticate... {args}")

    emit(
        "auth_msg",
        {},
        to=request.sid,
        namespace="/test",
        callback=auth_success
    )

def auth_success(*args):
    print("Authentication successful")
    emit(
        "auth_connect",
        {},
        to=request.sid,
        namespace="/test",
    )


@socketio.on("broadcast_status", namespace="/test")
def broadcast_status(*args):
    print(f"client broadcast status: {args}")

@socketio.on('disconnect', namespace="/test")
def test_disconnect(reason):
    print('Client disconnected', request.sid, reason)


if __name__ == '__main__':
    socketio.run(app, port=8999, debug=True)

[Joe-ChenZ]

I ran
sudo ifconfig lo down to simulate internet down (which was what caused the BadNamespace error
and sudo ifconfig lo up to turn it back on

[Joe-ChenZ]

what is unusual about the configuration i set up?

[miguelgrinberg]

I said it above:

• You are killing the loopback interface
• For both sides
• At the same time

[Joe-ChenZ]

Ok. Tested the client and server on different machines, connected via a ngrok url. This time i turned off internet from my client machine, and the same thing happened. I added cors_allowed_origins = "*" and changed the application link. Everything else stays the same EDIT(forgot i added a line in the client, see the comment below, and the logs too)

python test_sio_client.py
Setting up our callbacks
SocketIO interface initialized successfully
Connecting to backend at http://2139da52cd45.ngrok-free.app
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'rDvHX_D5IRNuqxHXAAAA', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Received packet MESSAGE data 0/test,{"sid":"y7riqa9HfcG-n8uCAAAB"}
Namespace /test is connected
Trying to set up auth session with the backend...
Emitting event "authenticate" [/test]
Sending packet MESSAGE data 2/test,["authenticate",{}]
SocketIO connected successfully.
Received packet MESSAGE data 2/test,1["auth_msg",{}]
Received event "auth_msg" [/test]
Sending packet MESSAGE data 3/test,1[{"status":true}]
Received packet MESSAGE data 2/test,["auth_connect",{}]
Received event "auth_connect" [/test]
Emitting event "broadcast_status" [/test]
Sending packet MESSAGE data 2/test,["broadcast_status",{}]
Received packet PING data 
Sending packet PONG data 
packet queue is empty, aborting
Exiting write loop task

Server has stopped communicating, aborting
Waiting for write loop task to end
Engine.IO connection dropped
Disconnected from the backend via socketio, reason: transport error
Connection failed, new attempt in 1.00 seconds
Exiting read loop task


Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
SocketIO connection error on namespace /test with data: ('Connection error',)
Connection failed, new attempt in 2.24 seconds
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
SocketIO connection error on namespace /test with data: ('Connection error',)
Connection failed, new attempt in 3.79 seconds

Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
SocketIO connection error on namespace /test with data: ('Connection error',)
Connection failed, new attempt in 4.88 seconds
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
SocketIO connection error on namespace /test with data: ('Connection error',)
Connection failed, new attempt in 4.95 seconds
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
SocketIO connection error on namespace /test with data: ('Connection error',)
Connection failed, new attempt in 5.28 seconds
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'OlxlwWbi3q-vLqd8AAAC', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Exiting write loop task
Sending packet CLOSE data None
Engine.IO connection dropped
Server sent close packet data None, aborting
Waiting for write loop task to end
Exiting read loop task
Connection failed, new attempt in 4.98 seconds
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': 'Tov4iVXZYaUQw2EzAAAD', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Exiting write loop task
Received packet MESSAGE data 0/test,{"sid":"Y9be-ClF6D3ztZvlAAAE"}
Namespace /test is connected
Trying to set up auth session with the backend...
Emitting event "authenticate" [/test]
Sending packet MESSAGE data 2/test,["authenticate",{}]
Reconnection successful
Server sent close packet data 0, aborting
Waiting for write loop task to end
Engine.IO connection dropped
Disconnected from the backend via socketio, reason: transport error
Connection failed, new attempt in 0.65 seconds
Exiting read loop task
Attempting WebSocket connection to ws://2139da52cd45.ngrok-free.app/socket.io/?transport=websocket&EIO=4
WebSocket connection accepted with {'sid': '1_iMvtf3-_rZtqErAAAF', 'upgrades': [], 'pingTimeout': 20000, 'pingInterval': 25000, 'maxPayload': 1000000}
Engine.IO connection established
Sending packet MESSAGE data 0/test,{}
Received packet MESSAGE data 0/test,{"sid":"ygv6H-ynVNL6S7XnAAAG"}
Received packet MESSAGE data 4/test,"Unable to connect"
Namespace /test is connected
Trying to set up auth session with the backend...
Emitting event "authenticate" [/test]
Sending packet MESSAGE data 2/test,["authenticate",{}]
Connection to namespace /test was rejected
SocketIO connection error on namespace /test with data: ('Unable to connect',)
Sending packet MESSAGE data 1/test,
Sending packet CLOSE data None
Engine.IO connection dropped
Exiting write loop task
Reconnection successful
Server sent close packet data None, aborting
Waiting for write loop task to end
Exiting read loop task
^CSocketIO interface shutting down

[Joe-ChenZ]

forgot that i added a client disconnect in connect_error handler. i removed that. Still seeing the BadNamespace error

SocketIO connection error on namespace /test with data: ('Unable to connect',)
Received event "auth_msg" [/test]
Sending packet MESSAGE data 3/test,1[{"status":true}]
Received packet MESSAGE data 2/test,["auth_connect",{}]
Received packet MESSAGE data 2/test,2["auth_msg",{}]
Received event "auth_connect" [/test]
Received event "auth_msg" [/test]
Sending packet MESSAGE data 3/test,2[{"status":true}]
Task exception was never retrieved
future: <Task finished name='Task-74' coro=<AsyncClient._handle_eio_message() done, defined at /Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py:558> exception=BadNamespaceError('/test is not a connected namespace.')>
Traceback (most recent call last):
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 575, in _handle_eio_message
    await self._handle_event(pkt.namespace, pkt.id, pkt.data)
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 413, in _handle_event
    r = await self._trigger_event(data[0], namespace, *data[1:])
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 467, in _trigger_event
    ret = await handler(*args)
          ^^^^^^^^^^^^^^^^^^^^
  File "/Users/backend/test_sio_client.py", line 60, in _auth_connect
    await self.socketio_client.emit(
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /test is not a connected namespace.
Received packet MESSAGE data 2/test,["auth_connect",{}]
Received event "auth_connect" [/test]
Task exception was never retrieved
future: <Task finished name='Task-76' coro=<AsyncClient._handle_eio_message() done, defined at /Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py:558> exception=BadNamespaceError('/test is not a connected namespace.')>
Traceback (most recent call last):
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 575, in _handle_eio_message
    await self._handle_event(pkt.namespace, pkt.id, pkt.data)
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 413, in _handle_event
    r = await self._trigger_event(data[0], namespace, *data[1:])
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 467, in _trigger_event
    ret = await handler(*args)
          ^^^^^^^^^^^^^^^^^^^^
  File "/Users/backend/test_sio_client.py", line 60, in _auth_connect
    await self.socketio_client.emit(
  File "/Users/backend/.venv/lib/python3.12/site-packages/socketio/async_client.py", line 229, in emit
    raise exceptions.BadNamespaceError(
socketio.exceptions.BadNamespaceError: /test is not a connected namespace.
Sending packet CLOSE data None
Engine.IO connection dropped
Exiting write loop task

[miguelgrinberg]

Just so you know, I think I have an idea of what's happening. The client uses a queue to store packets until they are processed. I think an extra element in this queue that was left over by the interrupted connection is causing the reconnect to fail immediately. I'll continue debugging this and let you know when I have a fix for you to test.

[miguelgrinberg]

@Joe-ChenZ If you'd like to give my candidate fix a try, it is in this PR: miguelgrinberg/python-engineio#415. Note that this is for python-engineio, which is a dependency of this package. No direct changes are required in this package.

If you give it a try, let me know if it helps!

[Joe-ChenZ]

The fix seems to work for the sample application so far. Thank you. I am testing it in my real application