Description of problem: The cinder-backup service is no longer able to back up a volume in a different AZ. How reproducible: Always Steps to Reproduce: 1. Prepare a deployment with a cinder-volume service in one AZ, and the backup service in a different AZ. 2. Create a volume 3. Create a cross-AZ backup of that volume Actual results: The backup fails, with an error message in the cinder-scheduler log: 2023-08-07 12:52:34.964 2 ERROR cinder.scheduler.manager [req-08d6c888-8ee0-4d95-92ab-735bb723b790 256db7de35ee451b946003930aef5c99 262b573d07f2425ba7eb3718825e5893 - - -] Service not found for creating backup.: cinder.exception.ServiceNotFound: Service cinder-backup could not be found. Expected results: The backup succeeds. Additional info: The issue was reported upstream a while ago, and there's a patch proposed to fix it. The patch needs to be updated (it's in merge-conflict).
Regression was introduced in upstream ussuri by change Ie2afb57c4861c4, so this should not need to go to train.