Over the weekend, we had a situation where we had some Nagios processes fall into a runaway state and spike the CPU on a number of our AIX LPARs. LPAR2RRD captured this spike accurately on both the CPU and CPU OS tab. This morning we cleaned up those processes and the CPU utilization returned to normal (via vmstat, nmon, and topas outputs). When we looked at the LPAR2RRD graphs, the CPU OS (run from the agent) reflected the drop in utilization, but the CPU graph continues to show a high CPU utilization.
I checked the LPAR2RRD error.log and found nothing. I reran the load.sh and it completed without any errors or warnings in the data collection output, yet the CPU graph still reflects incorrect information.
Any ideas on how to correct this issue?