[lxc-users] Full network device access in the container

[Serge Hallyn]

Quoting Piotr R (tytus64 at gmail.com):
> The host is running stripped down version of Red Hat 6.4 and container is
> running a bit different stripped down version of Red Hat 6.4. Both guest
> and the host are customized quite substantially so it would be difficult to
> list all the modifications. The container was created by extracting root
> file system from an ISO. The root file system is distributed in an ISO in
> an image file. I extracted the image file from the ISO, loopback mounted
> the image an copied entire file system to a new directory. I did not use
> lxc-create.
> 
> Following your advice I attempted to disable udev startup by the following
> change:
> --- /etc/rc.d/rc.sysinit.bak 2013-12-20 19:35:29.000000000 +0000
> +++ /etc/rc.d/rc.sysinit 2014-01-16 19:08:34.527904279 +0000
> @@ -137,7 +137,6 @@
>  nashpid=$(pidof nash 2>/dev/null)
>  [ -n "$nashpid" ] && kill $nashpid >/dev/null 2>&1
>  unset nashpid
> -/sbin/start_udev
> 
>  # Load other user-defined modules
>  for file in /etc/sysconfig/modules/*.modules ; do
> 
> 
> This change get rid of udevd error message as expected but I still can't
> get console prompt. I would greatly appreciate any suggestions about where
> look for the problem.
> 
> BTW I don't think I understand how all the mounts are done in the
> container. For example at which point and in what context the fstab file
> specified by lxc.mount is used? Are the mounts done before or after
> chrooting to the container?

The fstab entries specified in lxc.mount are done before the container
init starts.

-serge