OpenVZ Forum


Home » General » Support » OOM didn't save the machine
Re: OOM didn't save the machine [message #35518 is a reply to message #35501] Wed, 01 April 2009 15:59 Go to previous messageGo to previous message
maratrus is currently offline  maratrus
Messages: 1495
Registered: August 2007
Location: Moscow
Senior Member
I have to admit that I'm a little bit confused. Sad
The only assumption for now is the following.
The memory freeing process takes place in two steps:
- uncharging beancounters
- freeing memory with kfree via RCU.

If some process is stuck inside kernel space, RCU will never be scheduled and thus the second step will never be done. Hence, user_beancounters output shows nothing interesting (everything is ok) but slab cache consumes a huge amount of memory.

So, we have to single out the process that is stuck inside kernel space. You mentioned that you'd look at the wchan output but it is not the reliable way of debugging.

So, the ideal situation would be a serial console
http://wiki.openvz.org/Remote_console_setup
and alt-sysrq-
- p (twice the number of CPUs)
- w (several times)
- t (for all processes calltrace. this is a time consuming operation)

Report message to a moderator

[Message index]
 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: 2.6.27 - how's it coming?
Next Topic: vzdump fix: ERROR: wrong lvm mount point
Goto Forum:
  

[ Syndicate this forum (XML) ] [ RSS ] [ PDF ]

Current Time: Mon Sep 02 10:08:27 GMT 2024

Total time taken to generate the page: 0.05258 seconds
Powered by FUDforum Powered by Virtuozzo Containers