Re: [atomic-devel] mount gluster volume on atomic host

[Clayton Coleman <ccoleman redhat com>]

To add to the usecase - for Kube and OpenShift we eventually want to run the kubelet in a container.  The kubelet expects to mount volumes and drives into the mount table, and the in some cases access them from the filesystem.  So the kubelet would need to see the hosts mounts somehow (whether though the hosts mount namespace or direct kernel calls).


> On Mar 6, 2015, at 10:34 AM, Stephen C. Tweedie <sct redhat com> wrote:
> 
> Hi,
> 
>> On Thu, 2015-03-05 at 17:26 -0500, Daniel J Walsh wrote:
>> Could we make this a general purpose tool or add this as an option to
>> standard mount.
> 
> I'd love to see "mount --namespace=/proc/$n/ns/mnt" work as standard, it
> seems far preferable to a preload hack.
> 
> But, as long as we've got so _many_ different mount binaries (with
> separate ones for gluster, fuse, nfs, etc etc), that's definitely going
> to be a longer-term effort.
> 
> --Stephen
> 
>