On Thu, 2 Mar 2000, Tom Rini wrote:
> On Thu, Mar 02, 2000 at 10:16:01PM +0000, Alan Cox wrote:
>
> > Your ulimits are applied first. This only kicks in at the point something
> > has to die and we are totally out of memory
> >
> > It scores down root processes, stuff that is touching hw etc. Basically it
> > means when the shit hits the fan we shoot the user not their X server 8)
>
> Would there be any way to try and say "kill these procs first?"
> So far every time I've seen the OOM stuff kick in (admitidly on
> 2.2.14) it never seemed to kill the rapant process (squid), just
> everything else (bye-bye init).
Please try 2.2.15pre<new> and tell me if the OOM killer
kills the wrong process. I am very much interested in
tests of the OOM killer on as many different workloads
and systems as possible.
Although the killer has been tested and killed the 'right'
process almost all of the time, there might be workloads
where it doesn't kill the right process (at least, in
theory ... although I can't really think of one right now).
If you suspect you may have one of those workloads, please
try an out-of-memory situation in 2.2.15pre>=10 and tell
me if it breaks and how.
regards,
Rik
-- The Internet is not a network of computers. It is a network of people. That is its real strength.- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:15 EST