[Lxc-users] zfs support

Serge Hallyn serge.hallyn at ubuntu.com
Wed Apr 3 21:18:04 UTC 2013


Quoting Papp Tamas (tompos at martos.bme.hu):
> On 03/05/2013 02:46 PM, Serge Hallyn wrote:
> >
> >Quoting Scott Moser (smoser at ubuntu.com):
> >>On Tue, 5 Mar 2013, Papp Tamas wrote:
> >>
> >>>On 03/05/2013 05:27 AM, Scott Moser wrote:
> >>>>
> >>>>I wrote the btrfs clone stuff, I just cloned the rootfs directory as it
> >>>>was the only thing I thought would be of any significant size.  The other
> >>>>things I think just get copied).
> >>>>
> >>>>Its probably me thats missing something obvious, but whats the benefit to
> >>>>having all of $lxc_path/$lxc_name ?
> >>>
> >>>They belong to each other.
> >>>If a user wants to backup a container, everything have to be saved and the
> >>>same is true, if the container has to be relocated.
> >>>If it's cloned as a template, maybe the config and fstab files are useless,
> >>>but I guess it's OK, if they are there, they can be removed or even used as
> >>>template config.
> >>
> >>Nah, it sounds reasonable.  I hadn't thought about backing up, or
> >>persistent snapshots, that does make sense.  I can't think of anything
> >>that would hurt from the additional handful of small text files.
> >>
> >>My driving purpose was only fast cloning.
> >
> >Thanks guys.
> >
> >So the only trick will be we'll need some code (at least in lxc-clone)
> >to make sure that we detect older sites where only the rootfs is on the
> >subvolume, and do the right thing.
> >
> 
> I reopened this project in my queue.
> 
> And I'm wondering, whether you're planning to convert these shell
> scripts to python. If so, I'm not sure, it's worth to hack them
> now...

I'm working right now on clone in the c api.  The basics were simple
enough, now to tackle the blockdev copy/snapshotting...

-serge




More information about the lxc-users mailing list