Date:	Thu, 25 Aug 2011 15:02:04 -0300
From:	Glauber Costa <glommer@...allels.com>
To:	"Eric W. Biederman" <ebiederm@...ssion.com>
CC:	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	Linux Containers <containers@...ts.osdl.org>,
	<netdev@...r.kernel.org>, David Miller <davem@...emloft.net>,
	Pavel Emelyanov <xemul@...allels.com>
Subject: Re: [RFC] per-containers tcp buffer limitation

On 08/24/2011 11:16 PM, Eric W. Biederman wrote:
> KAMEZAWA Hiroyuki<kamezawa.hiroyu@...fujitsu.com>  writes:
>
>> On Wed, 24 Aug 2011 22:28:59 -0300
>> Glauber Costa<glommer@...allels.com>  wrote:
>>
>>> On 08/24/2011 09:35 PM, Eric W. Biederman wrote:
>>>> Glauber Costa<glommer@...allels.com>   writes:
>>> Hi Eric,
>>>
>>> Thanks for your attention.
>>>
>>> So, this that you propose was my first implementation. I ended up
>>> throwing it away after playing with it for a while.
>>>
>>> One of the first problems that arise from that, is that the sysctls are
>>> a tunable visible from inside the container. Those limits, however, are
>>> to be set from the outside world. The code is not much better than that
>>> either, and instead of creating new cgroup structures and linking them
>>> to the protocol, we end up doing it for net ns. We end up increasing
>>> structures just the same...
>
> You don't need to add a netns member to sockets.
But then you have to grow the netns structure itself somehow.
>
> But I do agree that there are odd permission issues with using the
> existing sysctls and making them per namespace.
>
> However almost everything I have seen with memory limits I have found
> very strange.  They all seem like a very bad version of disabling memory
> over commits.

More or less. At least from our perspective, the only thing we're really 
interested in capping are non-swappable resources. So you could not 
overcommit anyway.

For the sockets/tcp case, it is an even easier case. The code as it is 
today already allow you to define soft and hard memory limits: I am just 
making it container-wide, instead of system-wide.

>>> Also, since we're doing resource control, it seems more natural to use
>>> cgroups. Now, the fact that there are no correlation whatsoever between
>>> cgroups and namespaces does bother me. But that's another story, much
>>> more broader and general than this patch.
>>>
>>
>> I think using cgroup makes sense. A question in mind is whehter it is
>> better to integrate this kind of 'memory usage' controls to memcg or
>> not.
>
> Maybe.  When sockets start getting a cgroup member I start wondering,
> how many cgroup members will sockets potentially belong to.
>
>> How do you think ? IMHO, having cgroup per class of object is messy.
>> ...
>> How about adding
>> 	memory.tcp_mem
>> to memcg ?
>>
>> Or, adding kmem cgroup ?
>>
>>> About overhead, since this is the first RFC, I did not care about
>>> measuring. However, it seems trivial to me to guarantee that at least
>>> that it won't impose a significant performance penalty when it is
>>> compiled out. If we're moving forward with this implementation, I will
>>> include data in the next release so we can discuss in this basis.
>>>
>>
>> IMHO, you should show performance number even if RFC. Then, people will
>> see patch with more interests.
>
> And also compiled out doesn't really count.  Cgroups are something you
> want people to compile into distributions for the common case, and you
> don't want to impose a noticeable performance penalty for the common
> case.
Absolutely agreed.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists