[jitsi-users] CPU load for JVB


#1

I just ran an audio conference in Jitsi Meet on an mac mini Intel I5.

We had 30-31 persons attending.

Average cpu load for JVB was 60-70%, but sometimes the cpu load
rached 99% or maybe more…
Memory was around 15% all time.

Five times during our 1.5 h conference
I got ‘Unfortunately, something went wrong. We’re trying to fix this…’
But could not reconnect without rebooting jvb and nginx.

Why does JVB sometimes increase cpu load?

John


#2

Hi John,

CPU usage spikes are to be expected as various things execute periodically (like health checks, garbage collection, etc) but an average of 65% sounds like a lot to me, especially for a single audio only call... It would be interesting to run some monitoring system (like datadog, zabbix, influxdb, nagios) and look at some graphs.

About the error message that you got, it’d be good to know your stack versions (JVB, Jicofo, Jitsi Meet) and have some logs. Obviously it should not happen, but it’s hard to tell why it happened without more information.

I hope this helps.

Best regards,
George

···

On May 17, 2017, at 4:53 AM, jr@johnroland.eu wrote:

I just ran an audio conference in Jitsi Meet on an mac mini Intel I5.

We had 30-31 persons attending.

Average cpu load for JVB was 60-70%, but sometimes the cpu load
rached 99% or maybe more…
Memory was around 15% all time.

Five times during our 1.5 h conference
I got ‘Unfortunately, something went wrong. We’re trying to fix this…’
But could not reconnect without rebooting jvb and nginx.

Why does JVB sometimes increase cpu load?

John
_______________________________________________
users mailing list
users@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/users