Repository navigation
LINSTOR thin volumes become full-size volumes after reverting from a snapshot. #11765
Description
Activity
cc @rp-
Can you get the log of the qemu command on the agent that is "restoring" the snapshot?
I think it can be seen with DEBUG loggingThank you for your reply. If I understand you correctly, you are referring to this command:
2025-10-02T16:16:16.888305+03:00 node3 java[3762555]: DEBUG [utils.script.Script] (agentRequest-Handler-2:) (logid:abd91fb9) Successfully executed process [3767985] for command [qemu-img convert -O raw -U --image-opts driver=qcow2,file.filename=/mnt/382be04c-7123-3002-bca6-4751cf92136b/snaptest /dev/drbd1195 ].But just in case, I am attaching the full log output from the agent concerning two different virtual machines where the reverting is being performed. The disks of both virtual machines occupy 4.2 GB before reverting and 20 GB after. I am checking the disk sizes via linstor v l
Yeah this looks like it doesn't use the sparse/discard options for converting the image.
I'll check if this wrong in the driver part or somewhere within CloudStack itselfI found and fixed the issue if you are brave, you can give it a try with this .jar:
@rp- Thank you so much for this fix! I think I'll wait for the release, but just in case I change my mind, how do I upgrade plugin? Do I need to extract it to the /usr/share/cloudstack-management/lib folder, replacing what's there, and then restart the management server? Is this correct procedure?
Hi @DaanHoogland Is there a chance this fix will be included in version 4.20.2?🙏@rp- Thank you so much for this fix! I think I'll wait for the release, but just in case I change my mind, how do I upgrade plugin? Do I need to extract it to the /usr/share/cloudstack-management/lib folder, replacing what's there, and then restart the management server? Is this correct procedure? Hi @DaanHoogland Is there a chance this fix will be included in version 4.20.2?🙏
This fix is actually for the agent side, so you would need to replace it in
/usr/share/cloudstack-agent/lib, but this .jar has also additional fixes, not yet in 4.19.3.0, which also use management server code.Reacted by Andrey SviridovHi @DaanHoogland Is there a chance this fix will be included in version 4.20.2?🙏
We must merge the 4.19 branch forward to get it in (we usually don’t forget 😉 )
Reacted by Andrey Sviridovfixed in #11787
- moved this from Todo to Done in Apache CloudStack BugFest - Issues
on Oct 31, 2025
problem
As was discussed in #9950 After reverting volume from snapshot it allways became a full size volume instead of thin. Provisioning not changing, it's stays thin.
After fixes in the discussion, issue with creating new disks from snapshots has been resolved. New disks are created as thin volumes. However, the problem with disks during reverting from snapshots remains—they become thick volumes.
I am using version 4.19.3. I haven’t tried it, but I think this is also relevant for version 4.20.
versions
Ubuntu 24.04
ACS 4.19.3
Linstor 1.32.1
LVM-THIN
The steps to reproduce the bug
What to do about it?
After restoring disk from snapshot, disk must remains thin and not full-sized.