I’m getting a very high CPU use on my Jibri instance(hosted on a different server from the main jitsi-server).
- Platform: AWS EC2
- OS: Ubuntu 20.04
- CPU: 4 Cores (t2.xlarge)
- Memory: 16 GB
For a single user the 1 minute load average was around 3.0 when I joined with another user it shot upto 4.
Here is the screenshot for top command:
When I do the test with a 2 core machine. The load average jumps to 50 and the machine crashes in under a minute!
I had done a similar test about a month back and at that time per user jibri was taking 0.5 cores. The previous setup was jibri + an additional script to stream to multiple platforms.
Is there any recent change in jibri that could cause this spike? I came across some issues on forum and GitHub for memory leak but not on CPU. Has someone else faced this issue?