There's no difference to how the two versions operate when polling these things.
We can only show the data that the SNMPd provides to us. If the number being returned is an instantaneous value instead of a 5 minute average, the act of polling itself will affect the value.
But if you don't poll it, is it even turned on?
Adam.
-----Original Message----- From: observium observium-bounces@observium.org On Behalf Of cody via observium Sent: 07 January 2020 21:52 To: Observium observium@observium.org Cc: cody@mail-server.cc Subject: Re: [Observium] LXC monitoring via SNMP
On 2020-01-07 21:39, cody via observium wrote:
I'm monitoring 3 LXC containers of my new installed homelab server but somehow the CPU values polled via SNMP are completely messy while checking inside the container with TOP they are correct there. Attached is a screenshot of one of three fresh installed LXC containers where they are doing nothing yet showing the CPU utilization via SNMP and TOP. Also the physical server does nothing and is showing the correct values via SNMP and TOP. Anyone know how to fix that?
I've been watching for a while now what happens at every second and it looks like the spikes are generated when the SNMP query is executed. Inside the LXC container the CPU utilization goes up for 1-2 seconds exactly at that time and i assume SNMP fetches then these values. Tried now multiple times with the PRO version in our company and CE version at home and it seems to happen exactly at every 5 minutes when the poller runs.
Attached a screenshot from the CE version at home which also runs as a LXC container when it polls itself - same three cores are showing 100% in TOP for 1 second as in the graph. Doesn't happen in our company with the PRO version running inside a KVM VM.