>> This may be by design, but I'll say it here and let the compute team
>> decide on the correct behavior.
>>
>> On some failures, like the one from step #1, the refresh script leaves
>> the instance in a locked state instead of clearing it.
>
> ya that kind of a bug.
> we put it in the locked state to make sure the end user cannot make any action like hard rebooting the instace
> while we are messing with the db. that is also why we require the vm to be off so that they cant power it off
> by sshing in.
>
> regardless of the success or failure the reshsh command shoudl restore the lock state
>
> so if it was locked before leave it locked and if it was unlocked leave it unlocked.
> so this sound like a bug in our error handeling and clean up

ref-bug: https://bugzilla.redhat.com/show_bug.cgi?id=2178506

The thread where Gorka explains his feedback is at [1]. I'll try to break it down into specific fixes/work items in subsequent comments in this BZ.
[1] https://lists.corp.redhat.com/archives/rhos-compute/2022-December/000883.html