Call ends after 22 seconds and no communication is possible

Maybe we missed a configuration or something.
But the calls plugin isn’t working in our self-hosted mattermost system.

We didn’t change the default config.
If we start a call the call is actually starting and you see that others join and even that they are talking.
But you don’t here anything from them (or see the shared screen).
And after 22 seconds (befor the last update it was 14 seconds) after your join you just get kicked out of the call.

In the server logs is nothing at all, relating the calls plugin.

The browser logs showing this:
image

Further information
Port 8443 for UDP is open in the firewall
Mattermost is running with docker, the port 8065 is routed via traefik to 443 external

Are you on the latest v6.6 version? cc @streamer45

Yes, we use the 6.6.1 docker image.

Hi @MiDniGG ,

bear in mind Calls is currently in closed beta and not officially released yet even though the plugin is available for download/install. This means documentation is not ready, the plugin is not to be considered stable and support will be limited.

With all that said, from the logs you provided it’s most likely a connectivity error causing a timeout on the UDP port. So the client cannot connect to the advertised host at that port.

To help you further I’d need more information on your deployment and how the plugin was configured. Is the IP address showed in the logs the one from the instance running Mattermost?

Hi @streamer45,

thanks for your answer.

Yes that’s the IP of the host.
On it is running a docker stack in which the mattermost container is running.
The mattermost port 8065 is routed via traefik to the domain on ports 80/443
The firewall on the host allows UDP on 8443.

So I think what is missing is the routing of udp port 8443 from the container to the world. For this we have to edit the traefik config. Don’t know yet when we have time for it. As a downtime have to be planned and stuff. ^^
But I will report, when we tried it.

Hey @streamer45 ,

we tried and it’s working now.
As mentioned above, we had to route the udp port via traefik out of the container.

Thanks for the hint.

I am glad to hear that, thanks for letting us know :+1: