Restore Cinder Volume Snapshots to a New Disk

IssuesSep 1

Firstly, apologies if this functionality already exists and we have misunderstood the current workflow. If so, please feel free to point us in the right direction.

Our current understanding is that OSIE does not provide a straightforward way to take an existing Cinder volume snapshot and restore/convert it into a new Cinder volume/disk?

At present, the recovery workflow appears to require us to create/deploy a VM using the snapshot/image, rather than simply converting the snapshot back into a usable disk?

If so, for our customers and Service Operations teams, this makes what should be a relatively simple disk rollback or recovery operation unnecessarily complicated.

Ideally, the workflow would be:

Create a snapshot of the required Cinder-backed VM disk.
If a rollback is later required, select the snapshot and choose something such as Create Disk from Snapshot.
OSIE creates a new Cinder volume from that snapshot.
Shut down the affected VM.
Detach the existing disk from the VM.
Attach the newly created disk from the snapshot.
Power the VM back on.

We have not currently been able to test whether the underlying OpenStack behaviour differs depending on the Cinder backend. We suspect this workflow may behave differently with Ceph-backed Cinder storage, compared with our current storage platform?

Potential Future Enhancement

A further enhancement could be to automate the full rollback workflow within OSIE.

For example, a Revert Disk to Snapshot action could:

Prompt the user to select the disk and snapshot.
Shut down the VM safely.
Create a new Cinder volume from the selected snapshot.
Detach the current disk.
Attach the replacement disk in the same position/device mapping.
Power the VM back on.

With appropriate confirmation and warning prompts, this could turn a multi-step recovery process into a much simpler and safer operation for both customers and Service Operations.

The primary request, however, is simply to expose a straightforward snapshot → new Cinder volume → disk swap workflow within OSIE.

Posted by Kevin.Pannell · Sep 1

Completed3

Alin

Team•

Sep 7

Thanks for the feedback. We will improve the workflow

Kevin.Pannell

•

Sep 7

Aha, thank you! I’ve just tested this and have successfully restored a new volume from a snapshot.

A bit of feedback from testing the workflow:

I think the functionality is perhaps just a little tucked away and not immediately obvious. Would it be possible to add a “Create Volume from Snapshot” action directly within the snapshot list and/or on the snapshot details page itself? That would make the recovery workflow much more intuitive for customers and our Service Operations teams.

One other thing I noticed while testing:

In the current method, if a server has multiple disks, it doesn’t seem possible to easily identify which original disk/volume the snapshot relates to from the Create Volume screen. It would be useful if this showed some additional source information, for example the original volume name/ID, attached server and device mapping. Even just showing the disk size would help where available.

The automated “Revert Disk to Snapshot” workflow mentioned in the original request could still be a useful future enhancement, but even simply adding the Create Volume from Snapshot action and clearer source-volume information would make the existing functionality much easier to discover and use.

Thanks again for pointing me in the right direction!