Kernel 2.4.18 and strange OOM Killer behaveness

Kernel 2.4.18 and strange OOM Killer behaveness

[Marc-Christian Petersen]

Hi there,

i just opened an 800mb movie file with pico/nano, waited for the OOM, and here 
it is, have a look:

Apr 26 16:10:45 codeman kernel: Out of Memory: Killed process 19789 (apache).
Apr 26 16:10:46 codeman kernel: Out of Memory: Killed process 27828 (apache).
Apr 26 16:10:48 codeman kernel: Out of Memory: Killed process 14554 (apache).
Apr 26 16:10:50 codeman kernel: Out of Memory: Killed process 19893 (apache).
Apr 26 16:10:52 codeman kernel: Out of Memory: Killed process 32036 (apache).
Apr 26 16:10:54 codeman kernel: Out of Memory: Killed process 8370 (apache).
Apr 26 16:10:56 codeman kernel: Out of Memory: Killed process 680 (mysqld).
Apr 26 16:10:56 codeman kernel: Out of Memory: Killed process 7638 (mysqld).
Apr 26 16:10:56 codeman kernel: Out of Memory: Killed process 26780 (mysqld).
Apr 26 16:10:56 codeman kernel: Out of Memory: Killed process 26038 (mysqld).
Apr 26 16:11:01 codeman kernel: Out of Memory: Killed process 3914 (pico).
Apr 26 16:11:01 codeman kernel: VM: killing process pico
Apr 26 16:11:04 codeman kernel: Out of Memory: Killed process 20471 (squid).

So, you guess, apache, mysqld, squid and the causer pico are killed, but NO, 
ONLY, and i mean ONLY pico was killed, all the other Processes listed above 
are running fine, accepting connections, short: works fine!!
And yes, its reproduceable !!

The above is a kernel without rmap!

I also tried this with rmap enabled kernel, no OOM Messages appears, but the 
system freezes. It does NOT accept any input on keyboard, mouse or via the 
network (e.g ping, traceroute, telnet to smtp port etc.). Also, if the system 
does not accept any input, the harddisk is doing something, i've waited ~ 45 
minutes, system was still not accepting anything, harddisk was doing 
anything. This was rmap12h with 2.4.18.

Kinda strange, isn't it?! :-)

I like it, cause only the causer get killed, but why those syslog messages?!
Syslogd lies like a trooper ;)

-- 
Kind regards
	Marc-Christian Petersen

http://sourceforge.net/projects/wolk

PGP/GnuPG Key: 1024D/569DE2E3DB441A16
Fingerprint: 3469 0CF8 CA7E 0042 7824  080A 569D E2E3 DB44 1A16
Key available at wwwkeys.pgp.net.   Encrypted e-mail preferred.

Re: Kernel 2.4.18 and strange OOM Killer behaveness

[Dieter Nützel]

Marc-Christian Petersen wrote:

> Apr 26 16:10:56 codeman kernel: Out of Memory: Killed process 26038
> (mysqld).
> Apr 26 16:11:01 codeman kernel: Out of Memory: Killed process 3914 (pico).
> Apr 26 16:11:01 codeman kernel: VM: killing process pico
> Apr 26 16:11:04 codeman kernel: Out of Memory: Killed process 20471 (squid).
>
> So, you guess, apache, mysqld, squid and the causer pico are killed, but NO, 
> ONLY, and i mean ONLY pico was killed, all the other Processes listed above 
> are running fine, accepting connections, short: works fine!!
> And yes, its reproduceable !!
>
> The above is a kernel without rmap!

Try with -AA (splitted vm33), latest ist 2.4.19pre7aa2.
It works for "ages".

Regards,
	Dieter

BTW Have anyone a copy of these nice "test.c" for the OOM, handy?

-- 
Dieter Nützel
Graduate Student, Computer Science

University of Hamburg
Department of Computer Science
@home: Dieter.Nuetzel@hamburg.de

Re: Kernel 2.4.18 and strange OOM Killer behaveness

[Diego Calleja]

On Fri, 26 Apr 2002 16:38:08 +0200
Marc-Christian Petersen <mcp@linux-systeme.de> escribió:

> I also tried this with rmap enabled kernel, no OOM Messages appears, but the 
> system freezes. It does NOT accept any input on keyboard, mouse or via the 
> network (e.g ping, traceroute, telnet to smtp port etc.). Also, if the system 
> does not accept any input, the harddisk is doing something, i've waited ~ 45 
> minutes, system was still not accepting anything, harddisk was doing 
> anything. This was rmap12h with 2.4.18.

I guess it should be better to OOM here...?

> -- 
> Kind regards
> 	Marc-Christian Petersen
> 
> http://sourceforge.net/projects/wolk
> 
> PGP/GnuPG Key: 1024D/569DE2E3DB441A16
> Fingerprint: 3469 0CF8 CA7E 0042 7824  080A 569D E2E3 DB44 1A16
> Key available at wwwkeys.pgp.net.   Encrypted e-mail preferred.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: Kernel 2.4.18 and strange OOM Killer behaveness

[Andrea Arcangeli]

On Fri, Apr 26, 2002 at 07:46:14PM +0200, Dieter Nützel wrote:
> Marc-Christian Petersen wrote:
> 
> > Apr 26 16:10:56 codeman kernel: Out of Memory: Killed process 26038
> > (mysqld).
> > Apr 26 16:11:01 codeman kernel: Out of Memory: Killed process 3914 (pico).
> > Apr 26 16:11:01 codeman kernel: VM: killing process pico
> > Apr 26 16:11:04 codeman kernel: Out of Memory: Killed process 20471 (squid).
> >
> > So, you guess, apache, mysqld, squid and the causer pico are killed, but NO, 
> > ONLY, and i mean ONLY pico was killed, all the other Processes listed above 
> > are running fine, accepting connections, short: works fine!!
> > And yes, its reproduceable !!
> >
> > The above is a kernel without rmap!
> 
> Try with -AA (splitted vm33), latest ist 2.4.19pre7aa2.
> It works for "ages".

I second the suggestion :)

Andrea

Re: Kernel 2.4.18 and strange OOM Killer behaveness

[Alvaro Figueroa]

> So, you guess, apache, mysqld, squid and the causer pico are killed, but NO, 
> ONLY, and i mean ONLY pico was killed, all the other Processes listed above 
> are running fine, accepting connections, short: works fine!!

I'm probably talking nonsense here, but apache, mysql and squid; all of
them run there services with several processes. Could it be that the OOM
Killer only killed a couple of childs and not the parent of them all?

-- 
Alvaro Figueroa