Snapshots can't be force deleted, even with force-delete API

[Goutham Pacha Ravi]

When using snapshot-delete API, if the share driver logs an exception (Ex: snapshot is busy), the status of the snapshot is set to 'error_deleting'. If an administrator wants to remove this snapshot, he/she could use snapshot-force-delete API. However, if the share driver still cannot delete the resource, it will remain in 'error_deleting' state.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/297205

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/297205
Committed: https://git.openstack.org/cgit/openstack/manila/commit/?id=35648819aa00afe2cdd7ebba63f42def8c43e971
Submitter: Jenkins
Branch: master

commit 35648819aa00afe2cdd7ebba63f42def8c43e971
Author: Goutham Pacha Ravi <email address hidden>
Date: Mon Mar 21 13:28:16 2016 -0400

    Fix force-delete on snapshot resource

    When using snapshot-delete API, if the share driver
    logs an exception, the status of the snapshot is set
    to 'error_deleting'. If an administrator wants to remove
    this snapshot, he/she would expect to use the
    snapshot-force-delete API. However, if the share
    driver still cannot delete the resource, it will remain
    in 'error_deleting' state; leaving the administrator
    with no choice but to resort to making changes to
    the snapshot record in Manila's database.

    Propagate the force flag to the share manager and log
    driver exceptions, while ignoring them to obliterate
    the snapshot on Manila's database if the delete
    operation is forced.

    Also fix data being sent in create/delete snapshot
    driver interfaces and refactor corresponding unit tests.

    Change-Id: Ic2494e1f9eb9104f96ef3a50f12d0f304735ee4f
    Closes-Bug: #1560119

[Thierry Carrez (ttx)]

This issue was fixed in the openstack/manila 3.0.0.0b1 development milestone.

[Srinivas Tammali (srinivas-tammali)]

I see the condition mentioned in this BUG in my environment which was configured last week. I am unable to delete the snapshot with manila snapshot-force-delete and it just stays with error_deleting status.

Openstack version: Mitaka all latest just did an yum update as well.
OS on Physical Nodes: CENTOS 7
manila version is 1.8.1 on Shared FS node.

I am new to openstack got 2 physical servers, one configure as CONTROLLER, IDENTITY, DASHBOARD & NETWORK node and the other has NOVA, CINDER, SWIFT & MANILA.

CINDER is configured to use SSD; i got shares created and tested; however last evening the ssd crashed when i tried "manila extend" command to extend the share size.

Any directions on downloading & applying the fix mentioned in this bug report would be of great help.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/mitaka
Review: https://review.openstack.org/349750

[Goutham Pacha Ravi (gouthamr)]

Srinivas: I cherry-picked the change to stable/mitaka: https://review.openstack.org/349750

I'll fish for reviews and get this merged. Thanks for pointing this out. If it gets merged, it will be part of the next milestone release of the stable/mitaka branch.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/349750
Committed: https://git.openstack.org/cgit/openstack/manila/commit/?id=eba4586ef3db648bdda9cf0ffeb76398f0a761ea
Submitter: Jenkins
Branch: stable/mitaka

commit eba4586ef3db648bdda9cf0ffeb76398f0a761ea
Author: Goutham Pacha Ravi <email address hidden>
Date: Mon Mar 21 13:28:16 2016 -0400

    Fix force-delete on snapshot resource

    When using snapshot-delete API, if the share driver
    logs an exception, the status of the snapshot is set
    to 'error_deleting'. If an administrator wants to remove
    this snapshot, he/she would expect to use the
    snapshot-force-delete API. However, if the share
    driver still cannot delete the resource, it will remain
    in 'error_deleting' state; leaving the administrator
    with no choice but to resort to making changes to
    the snapshot record in Manila's database.

    Propagate the force flag to the share manager and log
    driver exceptions, while ignoring them to obliterate
    the snapshot on Manila's database if the delete
    operation is forced.

    Also fix data being sent in create/delete snapshot
    driver interfaces and refactor corresponding unit tests.

    Change-Id: Ic2494e1f9eb9104f96ef3a50f12d0f304735ee4f
    Closes-Bug: #1560119
    (cherry picked from commit 35648819aa00afe2cdd7ebba63f42def8c43e971)