On Fri, Jul 24, 2009 at 7:02 AM, Herbert Poetzl<herbert@13thfloor.at> wrote:
> On Thu, Jul 23, 2009 at 08:34:55PM +0800, Jeff Jansen wrote:
>> When I've set an rss limit on RAM in a guest and the guest hits the
>> limit, then the kernel has to kill some processes. So I see something
>> like this in the logs:
>
>> Jul 23 15:05:18 s1 kernel: [2458402.479963] VM: killing process
>> couriertcpd
>
>> Is there any way from that message to know *which* vserver guest hit
>> the limit?
>
> yep, by checking the /proc/virtual/*/limits
>
>> Does the "[2458402.479963]" tell me something I can use?
>
> what kernel is that (and what Linux-VServer patch)?
>
> TIA,
> Herbert
>
>> TIA
>>
>> Jeff Jansen
>
If you have cacti I do have limit graphs in my vserver graphing package.
http://www.jointhegrid.com/vserver/
http://www.jointhegrid.com/svn/vserver-cacti-jtg/trunk
I included a sample.
You can graph all memory usage over time.
