NetApp NFS driver for Cinder may over-provision pools in an active-active configuration

[Michael Arndt]

Even with the patch at https://review.opendev.org/c/openstack/cinder/+/798198 to get up to date
information with each Cinder provisioning request, users can still run into race conditions where multiple
management nodes will provision to the same pool (FlexVol) simultaneously, and end up over-provisioning
Cinder volumes on a FlexVol when active-active management nodes are being used.

This bug request is to ask for an option that allows for a pool to be locked for each Cinder volume
creation, either by the scheduler or the driver, so that nearly simultaneous provisioning requests to
Cinder do not result in simultaneous Cinder volume creations on the same pool.