I have PCP (Performance Co-Pilot) recording performance metrics on a RHEL 7 host. The recording cadence is set to Y seconds here:
/etc/pcp/pmlogger/control.d/local
When I copy the performance metrics logs to another machine I'm able to analyze the files using:
pcp atop --interval=Zs -r /path/to/archive/files/
(where Z is a value that may or may not be the same as Y)
When I jump around in the archive using the t and T commands I move by Z seconds, which makes sense.
My question is about the calculated values that pcp atop displays. According to man pcp-atop, values such as CPU utilization percentage and system memory utilization percentage are calculated over the past "interval", but it's unclear whether that interval is the capture interval or the display interval for replayed data.
