Comment 16 for bug 1817833

The more I re-read the bug report in comment 0, the more I don't understand how the bug report could be correct. I also couldn't reproduce the reported behavior on stable/queens.

The bug reporter shows that the resource_providers table is empty after the service deletion (even while the nova-compute process is running).

So, when starting the nova-compute process again, how could it possibly get a collision in the resource_providers table and raise ResourceProviderCreationFailed?