live migration fails when using non-image backed disk

[Dirk Mueller]

running live migration with --block-migrate fails if the disk was resized before (aka detached from the cow image). This is because nova.virt.libvirt.driver.py uses disk_size, not virt_disk_size for re-creating the qcow2 file on the destination host. in the case of qcow2 files, qemu-img however needs to get the virt_disk size passed down, otherwise the block migration step will not be able to convert all blocks.

[Matt Riedemann (mriedem)]

Patch is here: https://review.openstack.org/#/c/59771/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/59771
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=c61c67c5afd100749b7b55b251bd4e4e3bb556a2
Submitter: Jenkins
Branch: master

commit c61c67c5afd100749b7b55b251bd4e4e3bb556a2
Author: Dirk Mueller <email address hidden>
Date: Tue Dec 3 17:16:58 2013 +0100

    Setup destination disk from virt_disk_size

    When running live-migration --block-migrate on a qcow2 backed
    VM without cow image, the destination qcow2 file should be created
    with the virtual disk size. For raw images, the virt_disk_size
    is set to disk_size to ensure that virt_disk_size is always the
    size of the disk that should be re-created.

    Update unit tests to be more strict and check for sizes to be correct.

    Closes-Bug: #1257355

    Change-Id: Ie3be46024f06b9f59af92f5e3918a1958386d4f1

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/havana
Review: https://review.openstack.org/67795