Tests: Add check for refreshing additional snapshots

[roosterfish]

With this change we actually test if a delta of snapshots is transferred accordingly for both local and remote VM copies.

Before the refresh was only testing if the volume itself can be refreshed as the number of snapshots on both source and target volume was identical due to the copy operation.

Part of canonical/lxd#13305.

[roosterfish]

The storage-vm tests on zfs (e.g. https://github.com/canonical/lxd-ci/actions/runs/8644234403/job/23698965959?pr=138) are expected to fail due to canonical/lxd#13304.

[tomponline]

Thanks!

Can we add similar tests for the storage volume CI tests too (as a separate PR)?

[roosterfish]

Can we add similar tests for the storage volume CI tests too (as a separate PR)?

Absolutely. I'll also add the cross pool migration checks. Thanks for the hint, looks like snapshots of custom volumes are completely missing there until now.

[roosterfish]

I have force pushed once more as the LVM thin pool test was exceeding the disk space. So the test now removes the second snapshot after every check to save disk space.

Pipeline looks clean now.

[simondeziel]

The v1/snap1 is apparently identical to v1/snap0 since v1 is stopped the whole time. Should we power up v1 in between to have different snap content to refresh? Or maybe we could grow v1's rootfs in between and see if v2 has the new size after the refresh?

@roosterfish what do you think of ^? I'd be happy to add it if that makes sense to you (I know very little about snapshot refreshes).

[roosterfish]

The v1/snap1 is apparently identical to v1/snap0 since v1 is stopped the whole time.

We start v1 here https://github.com/canonical/lxd-ci/blob/main/tests/storage-vm#L52 and leave it running throughout the copy/refresh tests. Is it actually stopped somewhere in the middle?

I'd be happy to add it if that makes sense to you (I know very little about snapshot refreshes).

I like the idea. Maybe just putting files would be easier as we wouldn't need exceptions for the PowerFlex driver in regards to volume size :)

[tomponline]

I like the idea. Maybe just putting files would be easier as we wouldn't need exceptions for the PowerFlex driver in regards to volume size :)

Sounds good to me!