[libvirt-users] unattended cloning

[Adam King]

----- Original Message -----
> From: "Eric Blake" <eblake at redhat.com>
> To: "Adam King" <kinga at sghs.org.uk>, "libvirt-users" <libvirt-users at redhat.com>
> Sent: Monday, 22 September, 2014 5:34:52 PM
> Subject: Re: [libvirt-users] unattended cloning
> 
> On 09/16/2014 01:19 AM, Adam King wrote:
> > Morning,
> > 
> > I have a KVM guest running Win 2012 with MS SQL 2012.
> > In order to provide a quick method of restoring the service should
> > the live server die, we've decided to clone it to a preserved
> > state.
> > Ideally, this clone should also be kept up to date and cloning
> > should be done on a regular basis.
> > 
> > Is there any reason not to do unattended cloning? As in, when I
> > leave on a Friday afternoon, start a script to clone the guest and
> > auto start the live guest?
> 
> Sounds more like you are interested in snapshots than actually
> cloning
> your guest, where you could resume execution from the point of the
> snapshot.  And with external snapshots, it is indeed possible to
> automate the periodic capture of a snapshot of a running guest; where
> we
> still lack polish is the steps for reverting back to the state in a
> saved snapshot (it's doable, but requires effort outside of a simple
> libvirt API call).
> 
> --
> Eric Blake   eblake redhat com    +1-919-301-3266
> Libvirt virtualization library http://libvirt.org
> 
> 

Thanks Eric & Kashyap. 
Maybe I'm missing something with snapshots, my understanding is that if I did something like this:
virsh snapshot-create-as VM snap1-vm  --diskspec vda,file=/export/vmimgs/snap1-testvm.qcow2

And then do the same to create snap2-vm, I will get 2 snapshots, where snap2 has snap1 as its backing file. 
OK, so doing a virsh blockpull can pull out snap1 but it looks to me as if cleaning up old snapshots is quite tricky. 

Thanks

Adam