CentOS 6.4
/var/log/messages-20131013:Oct 9 03:16:36 vixen kernel: EMT: page allocation failure. order:4, mode:0xd0 /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page allocation failure. order:2, mode:0x20 /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page allocation failure. order:2, mode:0x20 /var/log/messages-20131027:Oct 20 16:00:47 vixen kernel: sshd: page allocation failure. order:2, mode:0x20 /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page allocation failure. order:2, mode:0x20 /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: swapper: page allocation failure. order:1, mode:0x20 /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page allocation failure. order:2, mode:0x20 /var/log/messages-20131027:Oct 23 10:22:41 vixen kernel: httpd: page allocation failure. order:1, mode:0x20
I know It's been discussed about a year ago but I can't view the bug reports mentioned in the thread. It seems to require an active RHEL subscription. What was that workaround again?
Thanks
----- Original Message ----- | CentOS 6.4 | | /var/log/messages-20131013:Oct 9 03:16:36 vixen kernel: EMT: page | allocation failure. order:4, mode:0xd0 | /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:00:47 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: swapper: | page | allocation failure. order:1, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 23 10:22:41 vixen kernel: httpd: page | allocation failure. order:1, mode:0x20 | | I know It's been discussed about a year ago but I can't view the bug | reports mentioned in the thread. It seems to require an active RHEL | subscription. What was that workaround again? | | Thanks
How much memory is in the system? Is it a NUMA machine? Can you tell me what memory looks like during these errors? It looks like a memory problem, but I'll need more details.
James A. Peltier писал 2013-10-28 03:02:
----- Original Message ----- | CentOS 6.4 | | /var/log/messages-20131013:Oct 9 03:16:36 vixen kernel: EMT: page | allocation failure. order:4, mode:0xd0 | /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:00:47 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: swapper: | page | allocation failure. order:1, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 23 10:22:41 vixen kernel: httpd: page | allocation failure. order:1, mode:0x20 | | I know It's been discussed about a year ago but I can't view the bug | reports mentioned in the thread. It seems to require an active RHEL | subscription. What was that workaround again? | | Thanks
How much memory is in the system? Is it a NUMA machine? Can you tell me what memory looks like during these errors? It looks like a memory problem, but I'll need more details.
RAM: 8GB No NUMA configuration found
dmesg and monthly memory usage chart: http://www.mediafire.com/folder/wc55gwc17dpp4/vixen
I don't see anything suspicious there. I had an impression this is a known problem with an easy fix.
Александр Кириллов nevis2us@infoline.su a écrit :
I don't see anything suspicious there. I had an impression this is a known problem with an easy fix.
echo never >/sys/kernel/mm/redhat_transparent_hugepage/defrag echo no > /sys/kernel/mm/redhat_transparent_hugepage/khugepaged/defrag is something I got on several boxes here. If I remember correctly, it « fixes » the problem.
HTH, Laurent.
Laurent Wandrebeck писал 2013-10-29 13:54:
echo never >/sys/kernel/mm/redhat_transparent_hugepage/defrag echo no > /sys/kernel/mm/redhat_transparent_hugepage/khugepaged/defrag is something I got on several boxes here. If I remember correctly, it « fixes » the problem.
I will try that, thanks. Any refs to the source of the solution?
Александр Кириллов nevis2us@infoline.su a écrit :
Laurent Wandrebeck писал 2013-10-29 13:54:
echo never >/sys/kernel/mm/redhat_transparent_hugepage/defrag echo no > /sys/kernel/mm/redhat_transparent_hugepage/khugepaged/defrag is something I got on several boxes here. If I remember correctly, it « fixes » the problem.
I will try that, thanks. Any refs to the source of the solution?
hmmmm I'm pretty sure it comes from CentOS ml but I can't find the ref right now. Anyway, the problem was back then close to this one: http://bugs.centos.org/view.php?id=5716
HTH, Laurent.
echo never >/sys/kernel/mm/redhat_transparent_hugepage/defrag echo no > /sys/kernel/mm/redhat_transparent_hugepage/khugepaged/defrag is something I got on several boxes here. If I remember correctly, it « fixes » the problem.
I will try that, thanks. Any refs to the source of the solution?
hmmmm I'm pretty sure it comes from CentOS ml but I can't find the ref right now. Anyway, the problem was back then close to this one: http://bugs.centos.org/view.php?id=5716
Thanks Laurent,
But are you sure that's a cure for the problem I'm having? Occasional page allocation failures? I don't see anything of the sort in this thread.
----- Original Message ----- | CentOS 6.4 | | /var/log/messages-20131013:Oct 9 03:16:36 vixen kernel: EMT: page | allocation failure. order:4, mode:0xd0 | /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131020:Oct 14 13:15:11 vixen kernel: httpd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:00:47 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: swapper: | page | allocation failure. order:1, mode:0x20 | /var/log/messages-20131027:Oct 20 16:14:28 vixen kernel: sshd: page | allocation failure. order:2, mode:0x20 | /var/log/messages-20131027:Oct 23 10:22:41 vixen kernel: httpd: page | allocation failure. order:1, mode:0x20 | | I know It's been discussed about a year ago but I can't view the bug | reports mentioned in the thread. It seems to require an active RHEL | subscription. What was that workaround again? | | Thanks
BTW, please don't try to "clean up" the output. Dump a dmesg into pastebin and send us the link. There are likely other errors that we're not able to see to determine the cause of the problem.