Re: [PATCH 1/2] Traffic control cgroups subsystem

[Paul Menage]

On Wed, Sep 10, 2008 at 4:24 PM, Thomas Graf <tgraf@xxxxxxx> wrote:
> Without a) this whole feature is very limited. It requires a process to
> be registered to the cgroup before it creates any sockets.

I think that it's a bit more flexible than that - any newly created
sockets (or newly accepted sockets) will inherit the correct classid,
so it's only existing connections that don't get reclassified.

But the common case is that applications are going to be dropped in
the correct cgroup *before* they start, via mechanisms such as pam
login modules or job-control middleware.

> Otherwise
> these sockets will not have the proper classid value and traffic from
> and to this sockets will not be classified. I don't see how this is
> practical since many applications create their sockets when the
> application is started. F.e. a web browser is causing a bulk data
> transfer, admin/user notices this and wants to put it in a restricted
> cgroup, won't work.
>

Yes, for this particular case it doesn't work. But isn't it much more
likely that the admin/user will know that web-browsers tend to trigger
bulk data transfers and configure the system so that the web browser
always starts in its own cgroup, rather than trying to jump on it
after the fact?

Paul
--
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/