[Devel] Network namespace status

[Eric W. Biederman]

Benjamin Thery <benjamin.thery at bull.net> writes:

> Eric,
>
> Thanks a lot for the updated patchset.
>
> I haven't look at the details yet, but I've already integrated it to your
> network namespaces patchset and ported the whole thing to 2.6.22-rc4-mm2 today.
> It is far easier to do when you don't have to think about the sysfs stuff :)

Agreed.  Which is a big part of motivation for picking something on the
sysfs side that works and getting it merged.

> I only performed some basic testing on it so far as I'm home today, but it
> already seems to work better than the patchset I ported to 2.6.21-mm2. It solves
> the issues I still had with sysfs.
>
> Thanks again.

Welcome.

I ported the rest of my patchset so I could test it (which is part of
what took me so long).  I will have to sync up with you and Daniel
and see what our differences are.

There are also a couple of other things we need to integrate with.
- The recent work of Patrick McHardy <kaber at trash.net>
  * netlink support for creating new virtual devices.
  * The macvlan driver
- Which ever version etun or veth that gets merged into the
  kernel.

Given that we don't have a measurable performance loss when the original
devices is in the network namespace it will be interesting to see if
the macvlan code gives us native performance, from my quick review it
should be trivial to ensure we don't unnecessary packet copies going on.

That said the only really big piece left is to pick the basic level 2
network namespace infrastructure and getting the bye off from the
maintainers of the networking code.  So I'm going to see if I can
split that part of the conversation into comprehensible chunks and
see if we can things moving there.

Eric
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers