I would be guessing you have lots of Oracle threads - all of which will have
the same common code mapped. And counted by the summation.
These fields only count resident memory - swap usage will be (sort of)
irrelevant. Pages that are swap cached (as distinct from only swapped out)
reside in real pages, and so can affect the pool of available page frames.
Determining this number is not simple. Recently the memory manager has
started using compressed page cache to ameliorate this effect somewhat.
Refer to my previous (flippant) comment.
Shane ...
On Sat, Jul 24th, 2010 at 8:05 AM, Tom Duerbusch wrote:
> I tried your script on some of my images. Works fine, except when Oracle
> is involved.
>
> linux62:~ # ps -eo pmem | awk '{pmem += $1};END {print "pmem ="pmem"%"}';
> pmem =1347.1%
> I do have a lot of swap blocks allocated. This is due to a batch type
> run, that is run off hours. During the day, when users are on, we swap
> very little.
>
> So if this does include swap pages, I don't think the script would give me
> what I need, during normal processing. Do you agree? Or am off track
> here?
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/
