Description of problem:
Please correct me if I am wrong, but it looks like at this point only one cinder-backup backend can be configured at a time and its availability zone is obtained from CONF.storage_availability_zone configuration parameter.
At the same time, cinder-backup service will throw ServiceNotFound exception when user tries to create a backup for volume from availability zones that doesn't match cinder-backup service's AZ and user doesn't specify correct AZ using --availability-zone argument. The reason is that cinder-backup will obtain information about desired AZ from volume's metadata.
I am not sure if obtaining information about AZ from volume's metadata is reasonable if there is only one instance of cinder-backup service running in specific AZ anyway?
Version-Release number of selected component (if applicable):
RHOSP 16.2
How reproducible:
$ cinder create --volume-type nfs_def-01 --name "test_backup_vol" 1
$ cinder --os-volume-api-version 3.51 backup-create test_backup_vol --name "backup_from_lvm_vol_no_az"
ERROR: Service cinder-backup could not be found. (HTTP 503) (Request-ID: req-5ec98710-d229-4847-839f-1c8f23fb2239)
Additional info:
Latest set of sosreports contains logs for specified request ID.
- external trackers