Comment 54 for bug 1546507

I also wanted to add some related comments from my side:

* I do not think these issues should affect Glance v2's ready status.

* image-locations were always designed to be admin oriented as it's impossible for an end user to determine which all types of location schema can be supplied to glance to let it comprehend the same.

* There is no discovery added to it.

* The use case of image-locations has been for operator to specify the most optimal location to pull image from -- say if it's a massively distributed cloud and Nova or like services need to be aware the best location to use data from.

* The default configuration results to 'False' or disallowed usage of this feature.

* The feature in Nova (rbd snapshots) use it but the testing is being done on a separate gate job and the default (glance) pull (say on devstack) won't enable the operator to use glance with it.

* Currently, the documentation suggests only direct_url config. Once the glance-v2 is used by default, multiple locations would need to be enabled.

* The security impact of the feature would need to be updated to reflect the same and the severity of the impact may increase http://specs.openstack.org/openstack/nova-specs/specs/mitaka/implemented/rbd-instance-snapshots.html#security-impact

* Nevertheless, the effort to port to Glance v2 should not be halted or changed. Glance v2 is on way to be the default version and we need to work around proper documentation or best practices. Some of this information can go in operator manual where the ops would need to carefully deploy private glance nodes to enable multiple locations for ceph case -- but all that is a non-standard anyway.