Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations gkittelson on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Swap-in reported in nmon, but no paging space paging activity visible

Status
Not open for further replies.

kingesox

Technical User
Aug 23, 2007
3
FR
Hello,

The PROC tab in my nmon trace reports some "swap-in" activity (albeit low: avg 1.5 swap-in per 10 second interval). My understanding is that this counter reports the number of threads once swapped out, but now ready to be swapped back into memory.

What mesmerizes me is that I can't see any paging space paging activity (pi/po = 0) in the trace. And there's 7+ GB of memory available on the box at all times.

My server is not suffering any performance issue at the moment, so the question is principally out of curiosity: what causes these threads to be swapped out in the first place? Or am I misinterpreting the "swap-in" value in nmon?

Thanks in advance for the pointers.
 
If the amount of computational, or program, memory exceeds (Real Memory - maxperm), the computational pages become eligible for paging out by the virtual memory manager. The swap-ins you see might be processes or modules that see very little activity.

I don't use nmon trace all that often, but I've got a similar situation in one I had at hand. The swap-in column on PROC had a single non-zero entry, but there was no corresponding activity on the PAGE tab's pi/po graph. There was, however, a peak on the fsin/fsout graph at the same point in time.

Another possibility would be a glitch in either nmon or nmon_analyzer.xls.

- Rod


IBM Certified Advanced Technical Expert pSeries and AIX 5L
CompTIA Linux+
CompTIA Security+

A Simple Code for Posting on the Web
 
Rod,

Thanks for your reply. I observed numperm to be higher than maxperm, so I'm not sure a computational page may have become eligible for replacement during the test run, especially with 7GB+ free memory. By the way, how come I see so much free memory (7GB out of 16GB)? Isn't AIX supposed to use the maximum available memory?

Glitch in nmon or analyser cannot be excluded, I'll check their website for similar postings.

Thanks again,
jm
 
kingesox said:
Isn't AIX supposed to use the maximum available memory?

It should, but it can take a while to max out. Did your test run read more than 16GB-computational pages of file data?

- Rod
 
No probably not, I don't think it had that much to read in. It's only got Oracle running, and the machine had been rebooted some time before. (I can't track what the uptime was anymore because it was rebooted again in the meantime). So yes, that's probably the reason it wasn't using all. Thanks again for your reply.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top