Out of memory: kill process 4387 (gnome-session)

[Maaz]

Hi Court Campbell
>Can you post the info from /proc/sys/vm/overcommit_memory?

# cat /proc/sys/vm/swappiness
60
# cat /proc/sys/vm/overcommit_memory
0
# cat /proc/sys/vm/overcommit_ratio
50

Hi SEP
>Based on the recent data, especially the fact the system is not paging, it seems
>like an unresolved problem in nautilus.
OK.

>If nautilus is crashing consistently with out of memory and the system has
>plenty of memory or is not paging, the finger should be pointed to either the
>failing application or general kernel memory management.
a, It happend just the once
b, nautilus was not crashed(I mean no crash message was appeared) but the the small window that shows the copy progress/status was automatically closed/quit .. the copy/paste progress/status window/dialog-box was closed as like that the copy operation was completed.(e.g you start copying via mouse and once all data copied the progress window/dialog-box disappears/closes).
c, now this machine is a file server(samba) and is working quite normal/fine, i.e no crash or performance issue. And I never copied such big directories in X(GNOME) again.

>I would look now enable core dumps to disk, duplicate the problem again and
>send the core dump to Novell.
I dont have this opportunity, because I have no spear/additional box, and I cant practice this on the machine in question(because this machine is in production)

Regards
Maaz

[Steven E. Protter]

So Maaz,

This rather involved thread is due to a problem that occurred once. We didn't get memory data off the system because it was production and rebooted, presumably to get it back into service.

Since the problem can not be easily duplicated, allow me to present the possibility:

The error message was accurate, due to some process with a memory leak, the system really was out of memory, and nautilus crashed exactly why it said it crashed.

I therefore recommend that you run the memory leak detector an hour a week to identify processes that may be leaky. I further recommend you enable core dumps to disk and use the system normally until it happens again.

Good Luck with this.

It was a fun thread.

SEP

[Court Campbell]

Since you don't like overcommiting you can echo the number 2 to /proc/sys/vm/overcommit_memory. I am sure this can be set permanently, but I can not sure where to do that on SUSE. This will disable overcommiting memory. Now if a process can't claim any pages it will just fail.

My only other suggestion would be to use a shell. Forget about window managers. They just add bloat. I'll take a simple cp command, or an rsync command over the extra bloat added by doing a copy through nautilis.