no OOM-Killer at high RAM and Swapusage

[Jens Kubieziel]

Hi,

I recently had a problem with my Sun Fire v40z (24 GB RAM and four
dual core Opterons). This machine runs on a plain 2.6.15 kernel. A
user started Java-processes (java -Xmx 20000m ...). However after some
time I realised that this machine was unresponsive (screen session
didn't respond and no new SSH connection). A htop gave me the
following information:

load: 12.78/11.09/8.25
MEM:  23616/23738MB
Swap: 2000/2000MB
Tasks: 402 total, 17 running

The machine was rebootet because it went unresponsive.

I would normally expect that the OOM-killer start at some point and
kills processes. Obviously this didn't happen here. Could you tell why
this didn't happen? What information could I provide furthermore? What
criteria are there for starting the OOM-killer (links to docs are
appreciated)?

Thanks for any hints
Jens
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/