Gathering of thin provisioning stats gets allocated (rather than used) size

[Arne Wiebalck]

In Ocata, the provisioning stats gathering is getting the allocated size of the volumes, rather than the space actually used:

https://github.com/openstack/cinder/blob/stable/ocata/cinder/volume/drivers/rbd.py#L366

In order to warn about a pool's fill status in case of over-subscription, the actual usage of the pool should be gathered.

[Gorka Eguileor (gorka)]

https://review.openstack.org/#/c/483092/

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/ocata
Review: https://review.openstack.org/483298

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/483298
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=e72dead5ce085a6ba66f7aad2ff58061842f43d2
Submitter: Jenkins
Branch: stable/ocata

commit e72dead5ce085a6ba66f7aad2ff58061842f43d2
Author: Jon Bernard <email address hidden>
Date: Wed Jul 12 13:35:19 2017 -0400

    Revert "Correct RBD Provision stats&fix a perf problem"

    This reverts commit b907450d7d7500fbcb9e4040efc2d0e42f4a78f9.

    Even though this approach should perform better, the value it calculates
    is not what the scheduler expects, we should be returning allocated
    size.

    We'll address the performance issue in a follow up patch

    Closes-Bug: #1704104
    Change-Id: I01b8fb2302884c3ed398f9553bb8e6a2aeb307c7
    (cherry picked from commit 86e8cf948aaa100c78d2fd9378c5c837b3f7304f)

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/cinder 10.0.4 release.