[CentOS] oom-killer.

Hi all,

Just watching for my first time the 'oom-killer' call in a CentOS 4.2
workstation. System was running normally but with high load of nice process,
iow, rosetta and seti, others appart, nautilus, etc... i realized that after a
night of standby running normally, the next moorning after few manual operations
from X-session in local console, it started a massive swapping, this time not
recovered as previously other time, finishing in a oom-killer call.

Finally inspected the syslog and realize the oom-killer was called to leave the
system stable having killed firefox-bin, 3 x rosetta, xmms and one nautilus.

It's a PIII 800 Mhz, 40GB IDE, 512 Mb RAM, 1024 Swap space, up for 4 days.

I'm not very xperienced in tunning the memmory system nor the swap system, help
appreciated to avoid this memory 'trashing' situtation (14 virtual desktops
openned at the same time).

Is it a normal situation? Is it a correct protecction measure? It's due a low
bandwitch IDE HD? how can i tune it to avoid this situations? is it a high load
for this workstation? I think the oom-killer was called correctly to avoid the
system collapsed, but i think there was a problem in the swapping subsystem or a
memmory trashing situation to avoid.

Help and hints appreciated,

Attached you'll the oom-killer log,

TIA

J.J.











-- 
Linux smarteyebox.stigmatedbrain.net 2.6.9-22.0.1.EL i686 GNU/Linux
11:55:01 up 1 day, 12:06, 9 users, load average: 1.20, 1.30, 1.57
---------------------------------------------------------
"Rome wasn't burned in a day. "
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: oom-killer.call.txt
URL: <http://lists.centos.org/pipermail/centos/attachments/20051214/b1233651/attachment.txt>