[PATCH 0/4] swapcgroup(v2)

[Rik van Riel]

On Fri, 23 May 2008 12:10:27 +0900
KAMEZAWA Hiroyuki <kamezawa.hiroyu at jp.fujitsu.com> wrote:
> On Thu, 22 May 2008 22:26:55 -0400
> Rik van Riel <riel at redhat.com> wrote:
> 
> > Even worse is that a cgroup has NO CONTROL over how much
> > of its memory is kept in RAM and how much is swapped out.
> Could you explain "NO CONTROL" ? cgroup has LRU....
> 'how mucch memory should be swapped out from memory' is well controlled
> in the VM besides LRU logic ?

The kernel controls what is swapped out.  The userland
processes in the cgroup can do nothing to reduce their
swap usage.

> Consider following system. (and there is no swap controller.) 
> Memory 4G. Swap 1G. with 2 cgroups A, B.
> 
> state 1) swap is not used.
>   A....memory limit to be 1G  no swap usage memory_usage=0M
>   B....memory limit to be 1G  no swap usage memory_usage=0M
> 
> state 2) Run a big program on A.
>   A....memory limit to be 1G and try to use 1.7G. uses 700MBytes of swap.
>        memory_usage=1G swap_usage=700M
>   B....memory_usage=0M
> 
> state 3) A some of programs ends in 'A'
>   A....memory_usage=500M swap_usage=700M
>   B....memory_usage=0M.
> 
> state 4) Run a big program on B.
>   A...memory_usage=500M swap_usage=700M.
>   B...memory_usage=1G   swap_usage=300M
> 
> Group B can only use 1.3G because of unfair swap use of group A.
> But users think why A uses 700M of swap with 500M of free memory....
> 
> If we don't have limitation to swap, we'll have to innovate a way to move swap
> to memory in some reasonable logic.

OK, I see the use case.

In the above example, it would be possible for cgroup A
to have only 800MB of anonymous memory total, in addition
to 400MB of page cache.  The page cache could push the
anonymous memory into swap, indirectly penalizing how much
memory cgroup B can use.

Of course, it could be argued that the system should just
be run with enough swap space, but that is another story :)

-- 
All rights reversed.