Cannot clone a volume from the image within the same ceph cluster

[Eric Xie]

Reproduce steps:
* Deploy glance with 'cinder' in CONF.enabled_backends
* Deploy cinder with CONF.allowed_direct_url_schemes = cinder
* Create one volume with the image img01
* Upload the volume to image: img02
* Create one new volume with img02

When creating the volume from img02, the log was as below:
2021-12-17 10:42:12,143.143 18647 INFO cinder.image.image_utils [req-daf59b27-8772-4855-b309-ba6bd700ad43 ee3e5834aca1440c84e2e29446552dcb e3a44f9bc1dc410a97b65ba2705e46fd - Federated default] Image download 40960.00 MB at 101.95 MB/s
2021-12-17 10:43:57,000.000 18647 INFO cinder.image.image_utils [req-daf59b27-8772-4855-b309-ba6bd700ad43 ee3e5834aca1440c84e2e29446552dcb e3a44f9bc1dc410a97b65ba2705e46fd - Federated default] Converted 40960.00 MB image at 398.43 MB/s

The store backend of both glance and cinder is same ceph cluster, but the volume creation flow using downloading operation that generate additional capacity and traffic occupancy. Also the clone operation will use the backend's ability.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/cinder/+/822278

[Sofia Enriquez (lsofia-enriquez)]

If I'm understanding correctly, you're proposing an optimization to avoid downloading the image multiple times. If that's the case, there's already a spec approved regarding this https://review.opendev.org/c/openstack/cinder-specs/+/810363 . If that's not the case, please add the log error you're facing and steps to reproduce the problem.
Thanks in advance