Performance issues if many agents have tickets open

  • Used Zammad version: 6.3.1
  • Used Zammad installation type: package
  • Operating system: Debian 12
  • Browser + version: different Browsers

Hello,

actually we have some performance issues with our zammad instance. Often (once a day or so) our health status is bad and different errors are occuring. For instance that emails are not processed due to timeouts, LDAP imports stucks some times or the error “scheduler may not run” pops up until the zammad process is restarted.

Many of our agents keep the left sidebar full of opened tickets, even if the tickets itself are closed. Can this lead to such errors like described above?

Hi @richie_77

Can you tell us, what hardware specifications you are running?
When I remember correctly, the doc says you need 10GB of RAM, but our consultant suggested 16GB of RAM. Maybe your server is just undersized?

How is the load? What says “top” when your performance issue appears? Are there any hints in the production.log?

kind regards
Basti

The system is on a vm, has 16G ram and 8 vCPU. Actually there are around 11G ram used and when i open “top”, there is a ruby process with 101% cpu load.

On the system there where around 15 active agents.

This is the output of htop, the ruby processes came from the background worker.

Our setup is pretty similar. 10 active agents, and 16 GB RAM but only 4 CPU’s.
On our system we only have one backupground-worker.rb process. So maybe this is your problem, since the processes are on top of htop i assume they generate the most load.

But I cant tell you if more processes are ok or if that’s really the root of your performance issues.

kind regards
Basti

Whats the value of WEB_CONCURRENCY in your instance?
If 1 you can set it to 2.

And then set ZAMMAD_SESSION_JOBS_CONCURRENT to 2

More here: Configuration via Environment Variables — Zammad System Documentation documentation

Now try again after a reboot