SIM 4.2/PMP reports memory events

[Debbie Moore]

PMP reports the following event from every server it manages.

Event Severity Major
Cleared Status Not cleared
Event Source aetml530
Associated System aetml530
Associated System Status Normal
Event Time 19-Jan-2005, 07:32:07 CST
Description This event is created as the result of receiving a performance status change event from the Performance Analysis tool
Assignee
Comments


HP Performance Event Details
Subsystem Name Memory
Current Subsystem Status Major
Previous Subsystem Status Minor
Overall Performance Status Major
Explanation High average page read rate (below 5 is normal), and there is less than 20 MB of available memory.

[David Claypool]

It's showing that you are experiencing a lot of paging to disk which is probably affecting performance. Correlated with the fact that there is less than 20MB of memory to allocate, it appears that you don't have adequate memory for the workloads you have.

[Brent Neste]

I get these all the time as well. I have PMP running on five servers and every one of them generates this alert at least 5 times a week. Three of the servers have a very low load on them and no "strange" software installed. When reviewing the playback of the performance data it always shows up as a single spike in the graphs. I am more inclined to think that this is either a bug in PMP or an hp driver or agent. I cannot reliably send out performance alerts because I get so many of these, and they don't appear to be valid. Just my 2 cents. :)

[Rob Buxton]

Spikes in Page usage are caused when an application places a demand for additional memory. Pages are faulted in and the process continues.
In some instances the process could be re-engineered to pre-allocate memory, thereby not having the spikes as it allocates and then sheds memory.
I agree it does seem a bit sensitive at times, but disagree that it's a bug.