Re: [PATCH V3 3/3] memcg: add memory.numastat api for numa statistics

From: Ying Han <yinghan@google.com>
To: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
	Minchan Kim <minchan.kim@gmail.com>,
	Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>,
	Balbir Singh <balbir@linux.vnet.ibm.com>,
	Tejun Heo <tj@kernel.org>, Pavel Emelyanov <xemul@openvz.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Li Zefan <lizf@cn.fujitsu.com>, Mel Gorman <mel@csn.ul.ie>,
	Christoph Lameter <cl@linux.com>,
	Johannes Weiner <hannes@cmpxchg.org>,
	Rik van Riel <riel@redhat.com>, Hugh Dickins <hughd@google.com>,
	Michal Hocko <mhocko@suse.cz>,
	Dave Hansen <dave@linux.vnet.ibm.com>,
	Zhu Yanhai <zhu.yanhai@gmail.com>,
	linux-mm@kvack.org
Subject: Re: [PATCH V3 3/3] memcg: add memory.numastat api for numa statistics
Date: Thu, 19 May 2011 18:31:41 -0700	[thread overview]
Message-ID: <BANLkTimtH6r09w8Em1gCh4VvWHV9P89cmQ@mail.gmail.com> (raw)
In-Reply-To: <20110520092424.1f1b514f.kamezawa.hiroyu@jp.fujitsu.com>

[-- Attachment #1: Type: text/plain, Size: 2156 bytes --]

On Thu, May 19, 2011 at 5:24 PM, KAMEZAWA Hiroyuki <
kamezawa.hiroyu@jp.fujitsu.com> wrote:

> On Thu, 19 May 2011 17:11:49 -0700
> Ying Han <yinghan@google.com> wrote:
>
> > On Thu, May 19, 2011 at 4:51 PM, KAMEZAWA Hiroyuki <
> > kamezawa.hiroyu@jp.fujitsu.com> wrote:
> >
> > > On Thu, 19 May 2011 10:32:40 -0700
> > > Ying Han <yinghan@google.com> wrote:
> > >
> > > > The new API exports numa_maps per-memcg basis. This is a piece of
> useful
> > > > information where it exports per-memcg page distribution across real
> numa
> > > > nodes.
> > > >
> > > > One of the usecase is evaluating application performance by combining
> > > this
> > > > information w/ the cpu allocation to the application.
> > > >
> > > > The output of the memory.numastat tries to follow w/ simiar format of
> > > numa_maps
> > > > like:
> > > >
> > > > total=<total pages> N0=<node 0 pages> N1=<node 1 pages> ...
> > > > file=<total file pages> N0=<node 0 pages> N1=<node 1 pages> ...
> > > > anon=<total anon pages> N0=<node 0 pages> N1=<node 1 pages> ...
> > > >
> > > > $ cat /dev/cgroup/memory/memory.numa_stat
> > > > total=246594 N0=18225 N1=72025 N2=26378 N3=129966
> > > > file=221728 N0=15030 N1=60804 N2=23238 N3=122656
> > > > anon=21120 N0=2937 N1=7733 N2=3140 N3=7310
> > > >
> > >
> > > Hmm ? this doesn't seem consistent....Isn't this log updated ?
> > >
> >
> > Nope. This is the V3 i posted w/ updated testing result.
> >
>
> Did you get this log while applications are running and LRU are changing ?
> See N1, 72505 != 60804 + 7733. big error.
>

Could you clarify why total != file + anon ?
> Does the number seems consistent when the system is calm ?
>

 That is because the total includes "unevictable" which is not listed here
as "file" and "anon"

>
>
> BTW, I wonder why unevictable is not shown...
> mem_cgroup_node_nr_lru_pages() counts unevictable into it because of
> for_each_lru().
>
> There are 2 ways.
>  1. show unevictable
>  2. use for_each_evictable_lru().
>
> I vote for 1.
>

Sounds good to me, I can add the "unevictable" following the "file" and
"anon" on the next post.

Thanks for the review

--Ying

>
>
> Thanks,
> -Kame
>
>
>

[-- Attachment #2: Type: text/html, Size: 3933 bytes --]