Description
I'm having a problem with inconsistent instance metrics in the web view.
This is only happening with CPU and network metrics; memory and disk are working fine.
I have two environments, one with ESXi at version 7.0.3.0 and the other with ESXi at version 8.0.3.0.
The vmware.stats.time.window and vm.stats.interval are using the default value in both environments too.
The problem only happens in the environment with ESXi v7, so I think it's something to do with updates.
What did I test?
-
I searched the logs.
I didn't find any errors; in fact, the CPU metrics are being collected, but with the wrong values. -
Collecting metrics directly from vCenter and CloudStack
I used govc to collect real-time metrics from vCenter and then used CloudMonkey to collect metrics from CloudStack.
Interestingly, although cmk was showing some usage, on the web there were gaps in the graphs.
On some VMs, the situation is even worse.
Appreciate for any help!