OneStor: VM block device badly deifined if recovery_snapshot_create_live fails

[xorel]

Description When recovery_snapshot_create_live fails

/var/tmp/one/tm/ssh/recovery_snap_create_live

command: virsh -c qemu:///system snapshot-create-as one-66 recovery_snap --diskspec /var/lib/one//datastores/101/66/disk.0,file=/var/lib/one//datastores/101/66/disk.0.snap/rs_tmp --diskspec /var/lib/one//datastores/101/66/disk.0.snap/rs_tmp,snapshot=no --diskspec /var/lib/one//datastores/101/66/disk.1,snapshot=no --disk-only --atomic --no-metadata

error:

error: unsupported configuration: too many disk snapshot requests for domain

VM block device is already updated to rs_tmp

virsh # domblklist one-89
 Target   Source
--------------------------------------------------------------
 vda      /var/lib/one//datastores/101/89/disk.0.snap/rs_tmp
 hda      /var/lib/one//datastores/101/89/disk.1

But the rs_tmp file is empty, vda should be instead point tot disk.0 like before the operation.

To Reproduce Force the recovery_snap_create_live to fail, vda points to rs_tmp

Expected behavior Even if the recovery_snap_create_live fails, VM should be always in correct state.

Details

• Affected Component: OneStor
• Hypervisor: kvm

Progress Status

• [ ] Code committed
• [ ] Testing - QA
• [ ] Documentation (Release notes - resolved issues, compatibility, known issues)