On 09/14/2012 03:58 AM, Winnie Lacesso wrote:
> Server has 16GB RAM & 20GB swap; problem is, it's a compute server used by
> many. So one user (later mortified & remorseful, but) causes grief for
> many. And a forced reset is always a concern re: possible filesystem
> corruption.
>
> Thank you all very much for enlightenment + advice.


The elephant in the room is "what was the kernel panic" (backtrace, dmesg, /var/log/messages, etc)?

This is obviously a 'should not happen' event and the panic may help in resolving the kernel bug or open the door to more options for workarounds by tuning your VM system or other drivers.

My systems do not generally panic when they run out of memory.  I often reboot at the next earliest convenience as the OOM killer may have killed off other processes it unwisely chose to kill.