We have a LibreNMS installation in Ubuntu 16.04, which is installed in a Vmware ESXi server, only one of its Xeon processors is assigned to the VM.
The localhost in LibreNMS monitoring shows CPU usage as this
Whereas the ESXi shows the CPU usage as
So is it as low as what the ESXi says; or is it as high as LibreNMS says.
It looks like LibreNMS is picking up those 100% cpu spikes as their constantly between the 5 minute polling time. Your ESXi will be more granular as it’ll be checking more realtime, I wouldn’t say it’s low though,
Thanks for replying Chas
This is from “About” in LibreNMS
and this is in a VM that has only 1GB of RAM and 1 Xeon E3 1220 V3 CPU, and the CPU only spikes every 5 minutes and stays at peak for a minute or two, and idle the rest of the time, isn’t that alright?
It depends, if there is any other problems it could be related to the CPU maxing out. Or was this post more for the explanation of the graph difference?
For example, LibreNMS is IO intensive (especially if using RRDCached), if your VM platform uses a network drive, the CPU may be constantly waiting on writing to disk which impacts performance. You can check this with
Ideally your machine needs more resource for 115 devices.