Snapshots with Ceph

(Spencer) #1


I have an issue:
I can fill the disk on my VM with dd, snapshot the disk, delete the dd file, and then repeat until the storage pool is full. OpenNebula currently has no quota for snapshot, and I don’t think ceph is reporting any storage consumption by the snapshot to OpenNebula.

What can I do to avoid this maluse of storage?

Additionally, ONe creates a clone of the rbd which includes all snapshots between revert commands. Thus if I make 2 snapshots, revert, and then take 3rd snapshot, the 2 snapshots will be associated with a clone, and the 3rd snapshot will be associated with the original volume until another revert. This makes grabbing the differential file between the 2nd and 3rd snap not possible, as the rbd export-diff --from-snap command does not work across two different rbd volumes.