Call Disconnected bug 2: 15 second timeout edition

[dbkr]

Steps to reproduce

We are seeing bug reports come in with more call disconnections. In all of these cases, the client sends an answer and then 15 seconds later, the call disconnects. Livekit's logging isn't quite good enough to immediately tell whether the client is deliberately closing the connection, but the code on the server is 1005 so this would suggest that.

Client:

2023-11-17T04:29:17.078Z 1 sending answer 
2023-11-17T04:29:32.075Z 4 cannot send signal request before connected, type: ping 
2023-11-17T04:29:32.075Z 4 cannot send signal request before connected, type: pingReq 
2023-11-17T04:29:32.200Z 1 clearing ping interval 
2023-11-17T04:29:32.200Z 1 room event connectionStateChanged {"event":"connectionStateChanged","args":["disconnected"]}
2023-11-17T04:29:32.200Z 1 room event disconnected {"event":"disconnected","args":[null]}
2023-11-17T04:29:32.208Z 2 Leaving call session in room **redacted**
2023-11-17T04:29:32.208Z 2 Session in room **redacted** changed to left

Outcome

What did you expect?

What happened instead?

Operating system

No response

Browser information

No response

URL for webapp

No response

Will you send logs?

Yes

[Ezwen]

Encountering this issue with Firefox under Fedora 39.

[fkwp]

here is a hint towards the 15 seconds https://github.com/livekit/client-sdk-js/blob/a704c1bc56356bf58fb27229b3338bafb58af696/src/connectionHelper/checks/websocket.ts#L20

[Tails]

Also encountering this, only on Firefox though. I thought it was my custom Rust LiveKit SDK integration but it surprisingly just worked with Chrome.