Manage existing volume failures does not clean quota usage

[Tobias Urdin]

We are using the openstacksdk and are trying to manage an existing volume in our ceph (rbd backend).

The manage fails in cinder-volume because the image does not exist in backend which is expected.
http://paste.openstack.org/show/783021/

The issue here is that the quota becomes wrong and it increases the `volumes` and `volumes_ssd` (assuming since our volume type is named ssd) resources in the `quota_usages` database table but when we issue a delete on the failed volume this quota is not released.

This causes the quota to be full when doing this over and over, I've not been able to pin-point it to an exact location where it doesn't remove the quota (or why).

[Tobias Urdin (tobias-urdin)]

Cinder version is rocky
python2-cinderclient-4.0.1-1.el7.noarch
python-cinder-13.0.0-1.el7.noarch
openstack-cinder-13.0.0-1.el7.noarch

[Tobias Urdin (tobias-urdin)]

Partially related? https://bugs.launchpad.net/cinder/+bug/1408763

[Tobias Urdin (tobias-urdin)]

closing as it's old and maybe solved