0
I was going to setup my Nagios/Icinga to monitor the processes on my machines and in case of a high number warn or error me via email.
I know for monitoring load, you would use something like this:
num of cpu * desired percent / 100
The main question is:
Is there any formula such as: num of cpu * XXX
to get a decent value of the allowed number of processes on a system?
Is this also linked to ulimit
somehow?
1Why should there be a critical number of processes for a system? There isn't. You want to monitor the resources they use, not their number. – Run CMD – 2016-05-13T12:22:19.817
@ClassStacker Is there no such thing as a critical number of processes on any system? – lockdoc – 2016-05-13T12:32:15.297
I thought I already said "no". It means "not per se", not in the sense of load, memory, HDD activity, or similar. There is a kind of an (artificial) limit but it's unlikely that it is relevant for you. At that point, you need to specify what "critical" actually means -- in terms of what your system needs to do -- does it have to spawn thousands of new processes in a short time, and what happens if that doesn't succeed -- we can't know this. But you know how to search the web, don't you? – Run CMD – 2016-05-13T13:09:57.013