Recently high memory usage

Infos

  • Used Zammad version: 5.4.1-1683779951.9554bea4.bullseye
  • Used Zammad installation type: package
  • Operating system: Debian
  • Browser + version: —

Expected behavior

  • Use less memory like before.

Actual behavior

  • It is more or less a small instance of Zammad, and I noticed the memory usage increased a lot (i don’t mean ES). I noticed a lot of Zammad Session Jobs Node processes of them idling around consuming a lot of RAM.
    Stopping Zammad (ES is still running) frees up > 30 GB of RAM.

(Temp) fix

Now I set a different limit in etc and naturally the usage is down, but I never noticed the processes before. Something changed - maybe they stopped and respawned than needed or took less RAM.

Steps to reproduce the behavior

  • Run a small instance of Zammad and take a look at the Session Jobs

:wave: Can you please share your performance tuning, especially WEB_CONCURRENCY and ZAMMAD_SESSION_CONCURRENT_JOBS ?

Thank you.

I don’t know if it is really a tuning. Reducing the job to a small number for a small instance. I noticed the massive increase compared to older versions

At the moment

export WEB_CONCURRENCY="4"
export ZAMMAD_SESSION_JOBS_CONCURRENT="5"

I also noticed different instances have different variables. Is there a list? Didn’t find anything on REDIS URL and one instance had it and the second one (both upgraded to v6) it was missing. Same from WEB_CONCURRENCY

You can find all available ENV settings here:
https://docs.zammad.org/en/latest/appendix/configure-env-vars.html

Theoretically you don’t need to do that via a profile file. I hope that it’s configured for the Zammad user.
If you want the active environment of your Zammad instance, you can simply run zammad run rails r "pp ENV; nil"