0

I've been struggling with a strange problem with my websocket connection. I use socketio 1.4.5 to connect to the server (flask, flask-socketio, gunicorn, nginx).

The problem is that about 1 of 8 connection attempts socketio connects using xhr polling after the websocket connection is established. Somehow this blocks the whole communication for this specific client. When timeout is reached (60seconds in my case) socketio connects again and if no xhr-polling is blocking the next time, the communication is up again.

This i very frustrating. and i have never seen this problem earlier on application with the same setup.

?EIO=3&transport=websocket&sid=997d6d19758641249e3b1232341e9fe2 GET 101 websocket   localhost       websocket   Other   0 B 1.0 min 

?EIO=3&transport=polling&t=LEUZRW7&sid=997d6d19758641249e3b1232341e9fe2 GET 504 http/1.1    localhost   127.0.0.1:443   xhr socket.io-1.4.5.js:1    763 B   1.0 min

The two lines above is what is available on the network tab when analyzing the trafic in chrome. Before the timeout is reached, 504 is replaced with "pending".

I have been trying to find the solution for this for a couple of hours now, and running out of ideas. I hope someone can help me with this..

Easyrider
  • 3,199
  • 5
  • 22
  • 32
  • I think this means that the websocket connection dropped. The client is trying to reconnect, which normally involves connecting via polling first, then upgrading to websocket. – Miguel Grinberg Mar 22 '16 at 14:26
  • usually this reconnect happens instantly. Why does the new polling-connnection need to timeout before everything is OK? Is it somehow possible to get more detailed information from flask-socketio? – Easyrider Mar 22 '16 at 14:45
  • You can add `engineio_logger=True` to your `SocketIO` constructor to have more information dumped to the console – Miguel Grinberg Mar 23 '16 at 04:06

0 Answers0