I'm evaluating HQ 3.2.3 more closely right now as part of troubleshooting a pre-production problem with a couple of multi-instance JBoss servers running on Linux. One of the processes consistently exceeds 100% CPU utilization (around 270%) on the dual dual-core servers.
The agent seems to be detecting and recording the CPU utilization correctly, but all of the HQ Charts appear to be stuck at 100%. This makes it difficult to see any variations between 101% and 399% 🙂
Is there an "auto" range for chart axis that I'm missing somewhere? I've searched the forums and haven't found anything like this. This JIRA bug looks close: http://jira.hyperic.com/browse/HHQ-1761, but I don't get any errors in my server.log about the data point value being higher than the ceiling.
Maybe it's a SIGAR thing? IE, system CPU utilization is reported in terms of 100% = "using all 4 cpus" whereas the process utilization is inconsistent--it returns in terms of 400%. I'll try to poke around in the SIGAR forums without cross posting too blatantly 😉
An auto range could be helpful for seeing more detail when the data range is small, too ... Maybe I need to see if you can create charts with Groovy 🙂
Thanks folks 🙂
Added the text about potential SIGAR inconsistency after I got home and thought about it a bit