[Lxc-users] veth interface not deleted?

[Serge Hallyn]

Quoting Fajar A. Nugraha (list at fajar.net):
> On Thu, Sep 19, 2013 at 9:34 PM, Serge Hallyn <serge.hallyn at ubuntu.com>wrote:
> 
> > Quoting Fajar A. Nugraha (list at fajar.net):
> > > On Fri, Sep 13, 2013 at 10:16 PM, Fajar A. Nugraha <list at fajar.net>
> > wrote:
> >
> 
> 
> > Is there some kind of hook where I can add the "ip link del" command
> > > manually when the container shuts down?
> >
> > lxc.network.script.down
> >
> >
> Hmmm ... I immediately got "lxc-start: failed to read configuration file"
> when adding this line
> 
> lxc.network.script.down = /var/lib/lxc/news/network_down.sh
> 
> comment it making it work again. Does that directive require lxc >= 0.9?

Oh, yeah, it might.

For me on 1.0 with a script that just echoes its arguments to a file, I
get

I am running with .u1 net down veth vethQL83W9.

> I'm still with 0.8.0~rc1-4ubuntu39.12.10.2~ubuntu12.04.1 at the moment.

The patch to introduce it was pretty simple if you're able to cherrypick
into your package.  But you probably can't...

> with 0.9 (when it was available) and 1.0 from ubuntu-lxc/daily ppa, I got
> even weirder errors (e.g. lxc would ALWAYS list containers as down, even
> when it's actually up) because the commands (e.g. lxc-ls, lxc-stop) would
> check on /var/lib/*lxc*/<container_name>/command, while "lsof" shows the
> open socket is actually on /var/lib/*lxc_anon*/<container_name>/command.

Eh what?

Can you open a bug with details about how you set up your containers so
I can try and reproduce?

> Any ideas how to fix this? If this works, I can try and test the hook on
> 1.0.
> 
> -- 
> Fajar