NFS volumes not sparse after migration

[Eric Harney]

Migration of NFS volumes which are sparse results in a fully allocated file after migration. Additionally, the migration process manually copies the entire size rather than only the needed parts, making migration more resource-intensive than it should be.

We should support copying data in an efficient way that preserves sparseness.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/183633

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/183633
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=4e3180c76629215d315921436ef19788ed49d2ab
Submitter: Jenkins
Branch: master

commit 4e3180c76629215d315921436ef19788ed49d2ab
Author: Eric Harney <email address hidden>
Date: Fri May 15 13:23:08 2015 -0400

    Add ability for drivers to copy data preserving sparseness

    Drivers can set self._sparse_copy_volume_data to opt in
    to copying data in a way that preserves sparseness.

    This is likely to be desired elsewhere, but currently this
    patch only enables it for the NFS driver and provides a
    method for other drivers to opt in, in case there are
    unexpected issues in some drivers.

    On the NFS driver, this prevents volume migration from
    changing a volume file from sparse to fully allocated.

    Change-Id: Ib74b2fb4015b805d33de42dfe60c8a97cc970b84
    Closes-Bug: #1455595