In my case label instance, using the on(instance) statement (think SQL JOIN ON).
For this on both sides prometheus must have matchin labels, The division is a many-to-one match (if you have more than one process name). So for me just adding up windows_process_cpu_time_total would exclude many other processes on the machine I do not explicitly monitor.Īlso using windows_cpu_time_total is a bad idea, because if the server is under light load, monitored processes would show big percentage of the small total load. I am whitelisting ( ) a few processes to monitor to keep metrics count low. It is important (for my scenario) to use windows_cpu_time_total instead of sum(windows_process_cpu_time_total) for the total load. windows_cpu_time_total vs windows_process_cpu_time_total 100 * windows_process_cpu_time_total / windows_cpu_time_totalĢ.
0 kommentar(er)
