On Fri, Aug 25, 2017 at 07:52:25AM -0400, John Ferlan wrote:
On 08/25/2017 05:44 AM, Martin Kletzander wrote:
> On Thu, Aug 24, 2017 at 06:28:27PM -0400, John Ferlan wrote:
>> Alter wipeVol to do same refresh operation as pool refresh would do.
>>
>
> I think we should rather keep the format as it is. Did I miss something
> here or isn't the source of the problem just the fact that we wipe the
> volume without keeping the format?
>
Once you "wipe" the file/image the format that was there (such as qcow2
as noted in the bz from patch2) is no longer there.
Consider the following sequence:
virsh vol-create-as default bz 10M --format qcow2
virsh vol-dumpxml bz default | grep format
<format type='qcow2'/>
qemu-img info /home/vm-images/bz
image: /home/vm-images/bz
file format: qcow2
virtual size: 10M (10485760 bytes)
disk size: 196K
cluster_size: 65536
Format specific information:
compat: 0.10
refcount bits: 16
virsh vol-wipe bz default
qemu-img info /home/vm-images/bz
image: /home/vm-images/bz
file format: raw
virtual size: 196K (200704 bytes)
disk size: 196K
virsh vol-dumpxml bz default | grep format
<format type='qcow2'/>
(without the patch in 2/2 of course)
BTW: I did consider just changing the format to RAW regardless, but
figured that was just too simple and may not be the right answer for
every case.
Sure, but that's not my point. My point is that instead of rewriting
the data with zeros, we could re-initialize it. Anyway, from what I see
in the docs, we "fixed" it by documenting this behaviour already, so my
point is moo:
https://libvirt.org/html/libvirt-libvirt-storage.html#virStorageVolWipe
(I guess I should know that when I pushed that documentation into the tree)
So ACK series.