Damans227 opened a new issue, #14030:
URL: https://github.com/apache/cloudstack/issues/14030

   Summary:
   When deleting a VM snapshot on KVM for a VM with more than one disk, each 
disk's data gets folded back into its real file one at a time, but CloudStack 
only gets a single "succeeded or failed" answer for the whole set. If an 
earlier disk's fold finishes for real but a later disk's then fails or times 
out, the whole thing is reported as failed, so CloudStack never updates its 
record for the disk that actually finished. Its database is left pointing to a 
file that no longer exists, with nothing to catch or fix this later. The VM 
then fails to start with "Can't find volume:<uuid>", and the only current fix 
is to manually correct the database to match the real file.
   
   Steps to reproduce:
   1. Create a VM with two or more disks on KVM.
   2. Take a VM snapshot, then delete it while the VM has enough disk activity 
that the merge takes a while (or induce a timeout/communication failure partway 
through the multi-disk merge).
   3. If one disk's merge completes on the host before another disk's merge 
fails/times out, the completed disk's volumes.path is left stale.
   4. Attempt to start the VM. It fails looking for the old file.
   
   Environment where this was observed: KVM, disk-only VM snapshots, VM with 2 
disks (ROOT + DATA), primary storage on NFS.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to