Instances are only created in cell databases once they are scheduled to a host, so we can know which specific cell DB to call for a new instance record.
When the scheduler passes a NoValidHost, then given there wouldn't be a host for that instance, the instance record is rather created to a specific cell0 database.
Now, it looks like you had a problem with your DB given the above log. When the scheduler tried to look at all the compute nodes, it got a SQL exception hence why it didn't provided a valid host.
Looks to me that specific problem created some issue hence why you weren't possible to delete that instance.
Instances are only created in cell databases once they are scheduled to a host, so we can know which specific cell DB to call for a new instance record.
When the scheduler passes a NoValidHost, then given there wouldn't be a host for that instance, the instance record is rather created to a specific cell0 database.
See https:/ /docs.openstack .org/nova/ latest/ user/cellsv2- layout. html#concepts for more explanations.
Now, it looks like you had a problem with your DB given the above log. When the scheduler tried to look at all the compute nodes, it got a SQL exception hence why it didn't provided a valid host.
Looks to me that specific problem created some issue hence why you weren't possible to delete that instance.
Can you please try to reproduce that problem ?