VolumeIsBusy and SnapshotIsBusy shouldn't be ignored

[Ivan Kolodyazhny]

During volume or snapshot deletion we catch VolumeIsBusy [1] and SnapshotIsBusy [2] and ignore them setting a 'available' state.

At least, we need to try once more to run delete action.

Maybe, setting state to 'error-deleting' is also acceptable if N tries failed.

[1] https://github.com/openstack/cinder/blob/ae5a7d5ea31fa14317696859b1e6e900ff740f26/cinder/volume/manager.py#L469
[2] https://github.com/openstack/cinder/blob/ae5a7d5ea31fa14317696859b1e6e900ff740f26/cinder/volume/manager.py#L612

[John Griffith (john-griffith)]

I agree, some things in progress I'd like to clean a lot of this up and have started with snapshot create. If I can get to the bottom of the seg-fault this introduces in the storwize tests I think we have a good general pattern to use to update all of the LVM calls.

[Ivan Kolodyazhny (e0ne)]

John, this is related for all backends. I found this issue with RDB (Ceph) driver

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/209156

[Huang Zhiteng (zhiteng-huang)]

Setting volume to 'error_deleting' means the volume won't be usable anymore (can't be attached, even deleted), it would require admin's intervention to do something about this volume. So I don't think it's a good idea to set a volume to 'error_deleting' just because it's been busy.

[Jordan Pittier (jordan-pittier)]

Yeah, the "error deleting" status is really a pain to deal with...

[Yuriy Nesenenko (ynesenenko)]

Add retries to delete volume or snapshot https://review.openstack.org/#/c/209156/
Now after N tries of volume (or snapshot) deletion failed it's setting state to 'available' if it's
been busy.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Yuriy Nesenenko (<email address hidden>) on branch: master
Review: https://review.openstack.org/209156
Reason: Retries added in the RBD driver https://review.openstack.org/#/c/217746/ So such issues need to be addressed in the drivers themeselves.

[javeme (javaloveme)]

This problem would cause the user confused: we did not delete the volume/snapshot successfully, but it did not report any errors and not record any state.

it is far from enough to just print an error log!

[Sean McGinnis (sean-mcginnis)]

Unassigning due to no activity for > 6 months.

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.openstack.org/589951

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.openstack.org/597028

[OpenStack Infra (hudson-openstack)]

Change abandoned by Rajat Dhasmana (<email address hidden>) on branch: master
Review: https://review.openstack.org/597028
Reason: Functionality not required.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/589951
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=df41cd5173f4854f3dbd9b4460bc0813560aceb8
Submitter: Zuul
Branch: master

commit df41cd5173f4854f3dbd9b4460bc0813560aceb8
Author: whoami-rajat <email address hidden>
Date: Wed Aug 8 21:10:39 2018 +0530

    Add retries to delete a volume in the Nimble Driver

    Retrying to delete a volume for N tries and if
    failed we raise NimbleAPIException.

    Change-Id: I937147b7a4caf8480a8b7429effa545cd8926a8d
    Related-Bug: #1413750