Long Running Rooms - Video Stops Working

Hi,

We are running Jitsi on our servers for some time already and we upgraded a few times too. We see one issue that sticks with us.

Basically, we have a long running meeting room for our office. In this room, we have a participant that is screen sharing a Grafana dashboard 24/7. I restart it every hour.

After a day or two (usually every morning), the video streams stop to work. Very often you could see this “mask” icon.
We were able to fix it by kicking everyone out of the meeting for a short period and rejoining. Restarting Jicofo also helps.

Is this something you already have seen in the wild? Should I report it in some issue tracker on your GitHub repos?

Thanks for your hard work :slight_smile:

check the jibri configuration, you most likely want to set usage_timeout to 0
/etc/jitsi/jibri/config.json

        // The amount of time, in minutes, a service is allowed to continue.
        //  Once a service has been running for this long, it will be
        //  stopped (cleanly).  A value of 0 means an indefinite amount
        //  of time is allowed
        "usage_timeout": "0"

Thanks for your interest, but we are not running Jibri. I think I was not clear. We have a participant that is sharing screen that runs Grafana. I’ve edited my original message.

Could it be that nginx/apache is restarted once a day or certbot runs once a day?
I seem to remember reading this in a thread a while ago.

If your users connections passes a NAT router then there might be a 24h NAT timeout in the router.
if you have a linux router check the
nf_conntrack
timeouts.

@xranby I’m not sure if this could be the issue. All other users are disconnect for the night. Only our Mr. Dash Board is reconnecting every hour to the same meeting room. We all join one by one in the morning and the video is not working in this room. It works if we join another room or if all participants leave the room for some time. It’s isolated to a single, long living room. None of us have Linux based routers either, so I’m not able to check the timeout value you suggested.

@Franky1 No crontab was found for any Linux user. Looking at service status, Nginx also runs without interruptions.

I guess I’ll have to look through our logs for something suspicious.