Closed audiocoach closed 3 months ago
We are also negatively affected by this issue after updating the plugins:
All our thresholds are now incorrect and would need manual configuration for each host according to socket count. On setups with thousands of hosts, this is not manageable unfortunately.
We propose that the overall load calculation should be an average of all sockets.
I do also agree with the previous posts. I would expect the overall cpu to be the average of all sockets, as this tends to conflict with a single threshold used for each socket and overall:
Is the overall load supposed to be a maximum of 100%? If so, then the bug does not only apply to systems with multiple sockets. On my 1-socket system with 4 CPU cores, I currently also have a overall load of over 100%.
This is now fixed, as there is another argument that will tell the plugin to report the average of all sockets or the sum.
Both metrics are useful, depending on the environment. Thats why we implemented that with v1.12.0
Hello,
the Invoke-IcingaCheckCPU overall load calculation is wrong if more than one cpu socket ist present. The core total values from the sockets are just summarized but not devided through the number of sockets.
My environment: Icinga 2.14.1 Icinga Web 2.12.1 Icinga Director 1.11 Icinga for Windows 1.11.1