Description of problem: In our nova-compute.log file old messages of failed migrations keep popping up in the logs. 2019-01-24 15:13:07.894 35643 WARNING nova.compute.resource_tracker [req-4b64e6c1-73b2-45e0-8fdc-22a37b4a9772 - - - - -] [instance: ee5af528-0085-46e5-af12-53b55318c840] Instance not resizing, skipping migration. 2019-01-24 15:13:07.928 35643 WARNING nova.compute.resource_tracker [req-4b64e6c1-73b2-45e0-8fdc-22a37b4a9772 - - - - -] [instance: a162d866-3762-440f-a6cf-5890f4e4c323] Instance not resizing, skipping migration. These are similar to reports referenced in https://access.redhat.com/solutions/2859371 whic was supposed to be fixed in OSP9. We are running OSP 13 and are still seeing the errors even though the instances are active and we can migrate them around. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: log message every 60 seconds Expected results: no messages Additional info:
This happens because there's an old migration record for the instance in some 'in-progress' state, but the task_state of the instance itself indicates that this is no longer the case. These messages are harmless, and can be ignored. They should not be warnings. Firstly, we should convert these to debug messages. Secondly, it would be interesting to log what the status of the migration record is that it's was returned. Lastly, ideally we would change the migration datamodel such that there can only be 1, unambiguos migration for an instance. After all, it isn't possible to migrate an instance more than once concurrently. However, I've proposed this upstream before without gaining any traction.
I am closing this bug as it has not been addressed for a very long time. Please feel free to reopen if it is still relevant.