Tune Linux trigger "too many processes"

[Sugarman]

Hello folks

I've run into some trigger alarms on some linux servers lately, claiming "too many processes running". The Zabbix Linux OS template threshold is set at 30, the servers went up to 60, but were nowhere near maximum CPU usage or other limits.

This got me thinking; could the fixed threshold be too low, or at least unsuitable for the task? Googling and forum searching didn't result in something usable.

How do you folks tackle this? I'm not so sure if it is a good idea to monitor the amount of running processes so closely, as many processes not using much ressources can run just fine, two processes clinging to whatever ressource they can get could massively drag a server down - and the process monitor would not fire off at all... granted, other triggers would, but still, can't see the point of solely monitoring the amount of processes. Or am I missing something?

Cheers!

[nick0909]

This is something that is totally up to you, depending on your systems and workload. The Zabbix default templates are just showing you what is possible, I think they are far from good to use in the real world without adjustment. For this case specifically, I don't alert on number of processes, and if I did it would be in the hundreds. I also personally think their polling period is way too short, but that is just my opinion, you have to use it and form your own how it is best to monitor your systems.