Server CPU peaks making server inaccessible

I have installed Jitsi Meet using the standard apt installation and customized it by changing some images, CSS rules. I am using AWS ec2 t2.mico instance for the same and it has been launched about a month ago. About a week before, I was unable to load the landing page or join any meet so I checked the instance monitoring page and saw CPU was at 99% so i rebooted the instance and everything went fine, Today, again, I faced the same issue. Iwasnt able to ssh into the server so have no idea which program was taking the CPU(using top command).

Here is the CPU utilisation curve which went from about 10% to 99.3% linearly in about 10 mins and 10% elevation prior to that.

I would like to mention that no meet was held on the day prior to this event but a meeting of about an hour was kept the day before.

t2.micro is too small for default configuration. Jicofo and jvb are configured by default to use up to 3GB of ram. You need to adjust it or get a vm wot at least 8gb of ram.

I tried to stress test the installation and was able to manage 10 people in a single meet with about 50% CPU and 50% Memory utilization with 0B swap allocation.
Should I divide ram 50% between JVB and Jicofo ie.
JICOFO_MAX_MEMORY=512m in /etc/jitsi/jicofo/config
VIDEOBRIDGE_MAX_MEMORY=512m in /etc/jitsi/videobridge/config
What I saw was JVB was occupying most of the memory.

:man_shrugging: I had never run it with resources like this.

1 Like