Description of problem:
Snapshot with the description "Auto-generated for Live Storage Migration" has ballooned to 2.5 TB, causing the VM to not be able to be started
Version-Release number of selected component (if applicable):
RHEV 3.2
How reproducible:
Unknown at this time. The snapshot is Pre Allocated rather than Thin Provisioned. This provides the assumption that this should've been thin-provisioned instead?
Steps to Reproduce:
1. Unknown at this time
Actual results:
The snapshot for generated via LSM has grown large enough that the VM is unable to start and the entire storage domain is full.
Expected results:
The snapshot does not take up the entire storage domain and hinder the ability of the VM to run
Additional info:
Comment 4Federico Simoncelli
2013-10-06 15:45:07 UTC
We still have to little information to understand what happened but the first thing that comes to my mind is that this could this be a duplicate of bug 998443.
What you're actually asking for here is to shrink the size of the backing file.
To do this we'd need to be able to introspect on the content of the disk (since it is raw and we have no idea what blocks are already in use), move blocks around internally so that they're sequential on disk and then shrink the LV to the actual size used.
This may be possible to do offline with something like libguestfs