we've configured monitoring using prometheus and grafana for a web-application running in a 2-cores server with OpenJDK 1.8
We see deeply different values for system cpu usage comparing those returned by prometheus with those printed by 'sar' command, server has a centos distro.
Generally speaking, is it right comparing values for the different metrics (process cpu, system cpu, load average and so on) between those returned by jvm with those returned by OS commands like sar or top?
Thanks
