Red Hat Bugzilla – Bug 130489
kernel kills db2 processes because of OOM error on RHEL Update2 and Update3
Last modified: 2007-11-30 17:07:03 EST
Glen, I think this is a NUMA issue, please add "numa=off" to the
command line, reboot and and let me know if the problem goes away.
----- Additional Comments From firstname.lastname@example.org 2004-08-23 18:43 -------
The problem goes away after rebooting with "numa=off"
----- Additional Comments From email@example.com 2004-08-24 11:07 -------
Question for RedHat,
Is numa=on by default on RHEL 3 U3 ? If this does not work, it will be better
if this is off by default i.e. mayne until RHEL 4 where 2.6 kernel has more
stable numa code ?? Thhanks.
For RHEL3-U3 NUMA is ON by default and its too late to change it now.
For RHEL3-U4 wa are investigating what changes would be necessary to
allow it to remain ON by default and still work correctly under high
memory load situations.
----- Additional Comments From firstname.lastname@example.org 2004-08-24 13:29 -------
I have added you to cc list. Please see if you are hitting this on p-series
with NUMA=on. Thanks.
----- Additional Comments From AVenkat@us.ibm.com 2004-08-24 13:32 -------
Thanks for letting me know. Will check and let you know. --Ananda
I think this problem is restricted to Opteron.
----- Additional Comments From email@example.com 2004-08-24 13:25 -------
Based on response from Larry, I am going to put target milesstone in as
RHEL 3 QU4. We will leave problem open until numa=on works. Thanks.
*** This bug has been marked as a duplicate of 131295 ***
----- Additional Comments From AVenkat@us.ibm.com 2004-12-01 22:19 EDT -------
I haven't used numa feature so far on pSeries. Probably, I might not be using
the large pages since the squadron boxes that I use have only 2GB (SF2) or 4GB
(L4). However, I will try to borrow one box with a larger memory.
It has been decided that x86_64 RHEL3 kernels should continue to enable
NUMA by default. However, if an OOM kill occurs on a NUMA system, an
extra message will be printed by the kernel suggesting that using the
"numa=off" boot option might be a good way to work around the issue.
The exact message is:
OOM kill occurred on an x86_64 NUMA system!
The numa=off boot option might help avoid this.
This change was committed to the RHEL3 U5 patch pool on 9-Feb-2005 (in
kernel version 2.4.21-27.12.EL).
A fix for this problem has just been committed to the RHEL3 U7
patch pool this evening (in kernel version 2.4.21-37.12.EL).
To enable an improved NUMA-friendly page allocation policy, please
set /proc/sys/vm/numa_memory_allocator via the "sysctl" command
(or put "vm.numa_memory_allocator = 1" in /etc/sysctl.conf).
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.