I've commented on https://review.openstack.org/#/c/525873 to this effect; enabling support for the CG API should be driven based on the storage backends that form part of the deployment, rather than being a blind configuration option that the user has to enable.
For example, CG's don't have context for our default storage backend (ceph) so it makes no sense to even allow a user to attempt to enable them in this use-case; if say we had a cinder-netapp subordinate charm it *would* make sense to expose CG support on the cinder-netapp charm and use its relation to cinder to instruct the cinder charm to enable the required policy changes.
I've commented on https:/ /review. openstack. org/#/c/ 525873 to this effect; enabling support for the CG API should be driven based on the storage backends that form part of the deployment, rather than being a blind configuration option that the user has to enable.
For example, CG's don't have context for our default storage backend (ceph) so it makes no sense to even allow a user to attempt to enable them in this use-case; if say we had a cinder-netapp subordinate charm it *would* make sense to expose CG support on the cinder-netapp charm and use its relation to cinder to instruct the cinder charm to enable the required policy changes.