I can refer again to what we do in the ovirt-system-tests:

    testlib.assert_true_within_long(
        lambda:
        vm1_snapshots_service.list()[-1].snapshot_status == types.SnapshotStatus.OK
    )


Which tests whether the status change to desirable one within a given period of time (I think ten minutes in case of assert_true_within_long), the assert code itself is in ovirtlago:
 https://github.com/lago-project/lago-ost-plugin/blob/130bed27a04c9b63161d6fc9cd3e68cd7b54d0c6/ovirtlago/testlib.py#L224

So instead of just asserting you can modify it to do whatever you need

On Thu, Jun 21, 2018 at 4:06 PM Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
On Thu, Jun 21, 2018 at 2:00 PM, Benny Zlotnik <bzlotnik@redhat.com> wrote:
You could something like this (IIUC):
    dead_snap1_params = types.Snapshot(
        description=SNAPSHOT_DESC_1,
        persist_memorystate=False,
        disk_attachments=[
            types.DiskAttachment(
                disk=types.Disk(
                    id=disk.id
                )
            )
        ]
    )

Taken from ovirt-system-tests[1]




Hi, thanks for your input!
It seems I was able to reach my target

What I've done

# Locate the service that manages the disk attachments of the virtual
# machine:
disk_attachments_service = vm_service.disk_attachments_service()

# Retrieve the list of disk attachments and then the bootable disk
disk_attachments = disk_attachments_service.list()

bootdisk = None
for disk_attachment in disk_attachments:
    disk = connection.follow_link(disk_attachment.disk)
    if disk_attachment.bootable == True:
        bootdisk = connection.follow_link(disk_attachment.disk)
        break

snaps_service = vm_service.snapshots_service()
snap = snaps_service.add(
    snapshot=types.Snapshot(
        description=snap_description,
        persist_memorystate=False,
        disk_attachments=[
            types.DiskAttachment(
                disk=types.Disk(
                    id=bootdisk.id
                )
            )
        ]
    ),
)
    'Sent request to create snapshot \'%s\', the id is \'%s\'.',
    snap.description, snap.id,
)


It seems also the monitor function already present in backup.py of the sdk examples linked in my previous message is working ok, 

# Poll and wait till the status of the snapshot is 'ok', which means
# that it is completely created:
snap_service = snaps_service.snapshot_service(snap.id)
while snap.snapshot_status != types.SnapshotStatus.OK:
        'Waiting till the snapshot is created, the status is now \'%s\'.',
        snap.snapshot_status,
    )
    time.sleep(1)
    snap = snap_service.get()
logging.info('The snapshot is now complete.')


In fact in the log file I have

INFO:root:Sent request to create snapshot 'padnpro_imp-backup-0e0c7064-bec5-429b-9ad7-cd8d1e5b25be', the id is '4135e8cb-87e8-4f09-82f5-b9ad0ed2f5be'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:Waiting till the snapshot is created, the status is now 'locked'.
INFO:root:The snapshot is now complete.

What if I would like to emit an event if for any reason the creation of the snapshot doesn't complete in a predefined elapsed time and manage it?
I think I have also to manage the case when for any reason no disk is marked as bootable inside the VM I'm backing up...

Gianluca