hello guys,
I’m interested in monitoring memory usage of my linux servers in my network.
I ran into few plugins on icinga exchange that do just that.
The problem is that many of those monitors have returned high percentage of usage on my servers and therefore sent a critical exit code to my web ui.
The thing is although the memory usage is shown high there are no Implications on server performance, therfore i sort of get false negative from these monitored services.
So what i actually need is an indication on when will the server performance will be impacted on terms of memory performance so i can write my own plugin that will give me a clear indication.
The problem will be when OOM killer is active, it is too late. On the other hand, if the service goes down by that, a process/application check will let you know.
For hosts which run single applications, we typically have memory checks disabled and only monitor the application’s health on its own. E.g. a database consuming 80% memory for caching should be monitored for slow responses and lags, and not overconsuming memory for instance.
Do you have concrete services and applications which are running on these hosts, where memory usage could be of help when analysing problems?
Memory monitoring on its own usually provides very little information due to the way Linux and other OSes use memory. It’s better to make sure swap is not used at all or just a little and then get metrics from the database or the application server itself.