Re: oom killer rewrite

[KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>]

On Wed, 19 May 2010 15:14:42 -0700 (PDT)
David Rientjes <rientjes@google.com> wrote:

> KOSAKI,
> 
> I've been notified that my entire oom killer rewrite has been dropped from 
> -mm based solely on your feedback.  The problem is that I have absolutely 
> no idea what issues you have with the changes that haven't already been 
> addressed (nobody else does, either, it seems).
> 

I've pointed out that "normalized" parameter doesn't seem to work well in some
situaion (in cluster). I hope you'll have an extra interface as

	echo 3G > /proc/<pid>/oom_indemification

to allow users have "absolute value" setting.
(If the admin know usual memory usage of an application, we can only
 add badness to extra memory usage.)

To be honest, I can't fully understand why we need _normalized_ parameter. Why
oom_adj _which is now used_ is not enough for setting "relative importance" ?

Does google guys controls importance of processes in very small step ?

And, IIRC, Nick pointed out that "don't remove _used_ interfaces just because
you hate it or it seems not clean". So, I recommend you to drop sysctl changes.

I think the whole concept of your patch series is good and I like it.
But changes in interfaces seem not very sensible. 

Don't take my word very serious but I don't like changes in interface.

Cheers,
-Kame


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>