Additional information about the issue:

- This is not a regression bug, it is reproducible also in 4.4.5
- It doesn't occur on hosted engine env since it is not supported yet, see the bug : (https://bugzilla.redhat.com/show_bug.cgi?id=1959490).
- The removal MBS domain is as basic as adding MBS domain.
  In addition, maintenance of the MBS domain works when attaching its disk to the VM and never running the VM, which is not likely 
  to do, so in my opinion, it is a blocker.

Sophie, please add full engine and cinderlib logs for these bugs


Additionally, you can't move the SD to maintenance at all? Or just for some time, and it is possible later?

(In reply to Benny Zlotnik from comment #3)
> Sophie, please add full engine and cinderlib logs for these bugs
> 
> 
> Additionally, you can't move the SD to maintenance at all? Or just for some
> time, and it is possible later?
Not possible to set the SD to maintenance at all, even not later.

Couple of notes:
1. There is no need to run and poweroff the VM
2. The problem is that RemoveVm's callback is not invoked for MBS disks[1]



An incorrect run looks like this:
2021-08-17 06:55:26,738-04 INFO  [org.ovirt.engine.core.bll.RemoveVmCommand] (default task-5) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] Lock Acquired to object 'EngineLock:{exclusiveLocks='[ff5a87ff-4f3b-441f-b488-be49c240ac17=VM]', sharedLocks=''}'
2021-08-17 06:55:26,838-04 INFO  [org.ovirt.engine.core.bll.RemoveVmCommand] (EE-ManagedThreadFactory-engine-Thread-51) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] Running command: RemoveVmCommand internal: false. Entities affected :  ID: ff5a87ff-4f3b-441f-b488-be49c240ac17 Type: VMAction group DELETE_VM with role type USER
2021-08-17 06:55:26,846-04 INFO  [org.ovirt.engine.core.vdsbroker.SetVmStatusVDSCommand] (EE-ManagedThreadFactory-engine-Thread-51) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] START, SetVmStatusVDSCommand( SetVmStatusVDSCommandParameters:{vmId='ff5a87ff-4f3b-441f-b488-be49c240ac17', status='ImageLocked', exitStatus='Normal'}), log id: 23850106
2021-08-17 06:55:26,868-04 INFO  [org.ovirt.engine.core.vdsbroker.SetVmStatusVDSCommand] (EE-ManagedThreadFactory-engine-Thread-51) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] FINISH, SetVmStatusVDSCommand, return: , log id: 23850106
2021-08-17 06:55:26,886-04 INFO  [org.ovirt.engine.core.bll.RemoveVmCommand] (EE-ManagedThreadFactory-engine-Thread-51) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] Lock freed to object 'EngineLock:{exclusiveLocks='[ff5a87ff-4f3b-441f-b488-be49c240ac17=VM]', sharedLocks=''}'
2021-08-17 06:55:26,900-04 INFO  [org.ovirt.engine.core.bll.RemoveMemoryVolumesCommand] (EE-ManagedThreadFactory-engine-Thread-51) [31fd4cf6] Running command: RemoveMemoryVolumesCommand internal: true.
2021-08-17 06:55:27,009-04 INFO  [org.ovirt.engine.core.bll.storage.disk.managedblock.RemoveAllManagedBlockStorageDisksCommand] (EE-ManagedExecutorService-commandCoordinator-Thread-2) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] Running command: RemoveAllManagedBlockStorageDisksCommand internal: true.
2021-08-17 06:55:27,054-04 INFO  [org.ovirt.engine.core.bll.storage.disk.managedblock.RemoveManagedBlockStorageDiskCommand] (EE-ManagedExecutorService-commandCoordinator-Thread-3) [3b42eaa5-86d8-4e98-b797-b47922e4eca2] Running command: RemoveManagedBlockStorageDiskCommand internal: true.



A correct run would look like:
2021-08-17 10:14:57,077-04 INFO  [org.ovirt.engine.core.bll.ConcurrentChildCommandsExecutionCallback] (EE-ManagedScheduledExecutorService-engineScheduledThreadPool-Thread-48) [f538ea2b-dfbf-4e60-8e07-9ce8c0f45eb0] Command 'RemoveManagedBlockStorageDisk' id: '10398245-6556-4ab4-8fcf-bfb7e287411d' child commands '[]' executions were completed, status 'SUCCEEDED'
2021-08-17 10:14:58,143-04 INFO  [org.ovirt.engine.core.bll.RemoveVmCommand] (EE-ManagedScheduledExecutorService-engineScheduledThreadPool-Thread-95) [f538ea2b-dfbf-4e60-8e07-9ce8c0f45eb0] Ending command 'org.ovirt.engine.core.bll.RemoveVmCommand' successfully.


The difference is that in the second command we can see an indication that the endAction sequence is going to be invoked for RemoveVm, after RemoveManagedBlockStorageDisk completed


[1] https://github.com/oVirt/ovirt-engine/blob/a4c08581377d669300ee35d1cd6406f0b37f25f0/backend/manager/modules/bll/src/main/java/org/ovirt/engine/core/bll/RemoveVmCommand.java#L237

Verified.

Versions:
vdsm-4.50.0.5-1.el8ev
ovirt-engine-4.5.0-582.gd548206.185.el8ev

Steps to Reproduce:
1.Create MBS domain via rest API and attaching it to the datacenter.
2.Create a new MB disk under the MBS domain that has been created.
3.Create a new VM and attach the disk as bootable.
4.Run the VM
5.power off the VM
6.Remove the VM with the disk
7.Storage-> Data centers-> data center that VM was attached to -> maintenance the MBS domain
 

Actual results:
The MBD was moved to maintenance mode successfully.
There was no problem detaching it afterward and removing it eventually.

Was verified on Upstream.

This bugzilla is included in oVirt 4.5.0 release, published on April 20th 2022.

Since the problem described in this bug report should be resolved in oVirt 4.5.0 release, it has been closed with a resolution of CURRENT RELEASE.

If the solution does not work for you, please open a new bug report.