Re: [atomic-devel] systemd in spc, lvm in spc

[Daniel J Walsh <dwalsh redhat com>]

On 03/07/2016 04:06 AM, Tobias Florek wrote:
> Hi!
>
> Sorry for getting back so late.
>
> I am trying to run cluster-native glusterfs on openshift/kubernetes.
> Gluster uses LVM snapshots to implement Gluster volume snapshots.
>
> I can (and in the past did) run a systemd-nspawn container to run
> systemd (and gluster), but that won't work for kubernetes. RunC is still
> a bit away from a RHEL atomic or CentOS atomic release. I don't think
> rkt is coming at all, so I am pretty much focused on using docker.
>
>
> My preferred way is running a separate lvm within the container. I don't
> think that's possible though.
>
>
> Using lvm using `chroot /host lvm` does only work with hostIPC afaict.
> At least it hangs when not using it. Using --noudevsync does not make it
> work either.
>
>
> I want to let systemd manage the container, because I don't need to run
> a different boot manager and there are quiet some processes to manage
> when running Gluster.
>
>
> Thank you for your support,
>   Tobias Florek
Where does lvm within the host fail?  Are you mounting in the /dev into 
the container?
Probably would work better with docker-1.10 release, since there have 
been fixes for sharing
of /dev into the container.