Re: [PATCH] badness() dramatically overcounts memory

[David Rientjes]

On Wed, 6 Feb 2008, KOSAKI Motohiro wrote:

> > Andrea Arcangeli has patches pending which change this to the RSS.  
> > Specifically:
> > 
> > 	http://marc.info/?l=linux-mm&m=119977937126925
> 
> I agreed with you that RSS is better :)
> 
> 
> 
> but..
> on many node numa, per zone rss is more better..
> 

It depends on how your applications are taking advantage of NUMA 
optimizations.  If they're constrained by mempolicies to a subset of nodes 
then the badness scoring isn't even used: the task that triggered the OOM 
condition is the one that is automatically killed.

At this point, I think you're going to need to present an actual case 
study where Andrea's patch isn't sufficient for selecting the appropriate 
task on large NUMA machines.

		David
--
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/