[ovirt-users] Delete disk references without deleting the disk
Johan Kooijman
mail at johankooijman.com
Mon Nov 23 03:10:27 EST 2015
One weird thing though: when I try to remove the VM itself, it won't let me
uncheck the "Remove disks" checkbox.
On Sun, Nov 22, 2015 at 9:00 PM, Nir Soffer <nsoffer at redhat.com> wrote:
> On Sun, Nov 22, 2015 at 6:14 PM, Johan Kooijman <mail at johankooijman.com>
> wrote:
> > Hi all,
> >
> > I have about 100 old VM's in my cluster. They're powered down, ready for
> > deletion. What I want to do is delete the VM's including disks without
> > actually deleting the disk images from the storage array itself. Is that
> > possible?
>
> Select the vm, click "remove", in the confirmation dialog, uncheck the
> "Delete disks"
> checkbox, confirm.
>
> > At the end I want to be able to delete the storage domain (which
> > then should not hold any data, as far as ovirt is concerned).
>
> Ovirt deleted the vms, but is keeping the disks, so the storage domain
> does hold all
> the disks.
>
> >
> > Reason for this: it's a ZFS pool with dedup enabled, deleting the images
> one
> > by one will kill the array with 100% iowa for some time.
>
> So what do you need is to destroy the storage domain, which will
> remove all the entities
> associated with it, but will keep the storage without any change.
>
> Do this:
> 1. Select the storage tab
> 2. select the domain
> 3. In the data center sub tab, click "maintenance"
> 4. When domain is in maintenance, click "detach"
> 5. Right click the domain and choose "destroy"
>
> This will remove the storage domain from engine database, leaving
> the contents of the domain.
>
> You can now delete the contents using favorite system tools.
>
> Now, if we want to add support for this in ovirt, how would you delete
> the entire domain in a more efficient way?
>
> Nir
>
--
Met vriendelijke groeten / With kind regards,
Johan Kooijman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20151123/89b5e07c/attachment.html>
More information about the Users
mailing list