Re: [PATCH 2/4] qemustatusxml2xmldata: backup-pull: Add private data for scratch image
On Mon, Nov 01, 2021 at 18:32:19 +0200, Nir Soffer wrote:
On Mon, Nov 1, 2021 at 4:48 PM Peter Krempa
<pkrempa(a)redhat.com> wrote:
>
> Signed-off-by: Peter Krempa <pkrempa(a)redhat.com>
> ---
> tests/qemustatusxml2xmldata/backup-pull-in.xml | 9 ++++++++-
> 1 file changed, 8 insertions(+), 1 deletion(-)
>
> diff --git a/tests/qemustatusxml2xmldata/backup-pull-in.xml
b/tests/qemustatusxml2xmldata/backup-pull-in.xml
> index 95afd3a51f..59c934d4f7 100644
> --- a/tests/qemustatusxml2xmldata/backup-pull-in.xml
> +++ b/tests/qemustatusxml2xmldata/backup-pull-in.xml
> @@ -257,7 +257,14 @@
> <server transport='tcp' tls='yes' name='localhost'
port='10809'/>
> <disks>
> <disk name='vda' backup='yes' state='running'
type='file' index='123'>
> - <scratch file='/path/to/file/'/>
> + <scratch file='/path/to/file/'>
> + <privateData>
> + <nodenames>
> + <nodename type='storage'
name='libvirt-1337-storage'/>
> + <nodename type='format'
name='libvirt-1337-format'/>
Shouldn't this be libvirt-123-*, matching the index=?
It does normally, but here the job data (above out of context) have
dummy nodenames.
But that doesn't matter that much.
Since we have the index entry, and libvirt knows that it has
libvirt-N-storage and libvirt-N-format
for every disk, why do we need to keep this info?
Storing nodenames explicitly allows us to see which objects are in use
actually. E.g. I'm removing the use of the 'raw' driver if it's just
dummy to improve performance, and it comes in handy with upgrades.
But yes, it is actually redundant.
Finally, is this exposed in backup-dumpxml?
The index is. The rest of the private data is not, that's just for
libvirt state storage in the status XML file which is not to be used by
anybody else.
> + </nodenames>
> + </privateData>
> + </scratch>
> </disk>
> </disks>
> <privateData>
> --
> 2.31.1
>