OS::Nova::Server does not check_delete_complete properly

[Randall Burt]

The OS::Nova::Server resource does not implement check_delete_complete and instead uses a co-routine inside handle_create. This generally works, except the implementation does not account for error statuses. Since this internal coroutine isn't run with a timeout, this can result in this routine running forever.

[Clint Byrum (clint-fewbar)]

The word "forever" gets a bump to High IMO. It has the potential to waste billed resources.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/99800

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/102637

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/102638

[Steven Hardy (shardy)]

I've just approved the first two patches in the series, but there's some discussion re tests in https://review.openstack.org/#/c/102638/, I don't want to presume to overrule asalkeld's -1., so Randall if it's OK with you I'll bump this bug to J3 and we can land that final patch early in J3 after you've reached agreement re the tests with Angus?

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/102637
Committed: https://git.openstack.org/cgit/openstack/heat/commit/?id=eb4956aa17da7c575ede3856b2cfc8c91ba78355
Submitter: Jenkins
Branch: master

commit eb4956aa17da7c575ede3856b2cfc8c91ba78355
Author: Randall Burt <email address hidden>
Date: Tue Jun 24 22:30:54 2014 -0500

    Account for status when deleting servers

    nova_utils.delete did not account for error state.

    Partial-Bug: #1329453
    Change-Id: I162a2441c41ae6331d6e150385a5e8aacc56167e

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/99800
Committed: https://git.openstack.org/cgit/openstack/heat/commit/?id=3942548450238d4ac79866bb358a27c9d6a2ef6d
Submitter: Jenkins
Branch: master

commit 3942548450238d4ac79866bb358a27c9d6a2ef6d
Author: Randall Burt <email address hidden>
Date: Wed Jun 25 12:54:22 2014 -0500

    Refactor OS::Nova::Server delete handling

    OS::Nova::Server used an internal coroutine to handle_delete instead of
    implementing check_delete_complete. While this mostly worked, the coroutine
    didn't account for errors in deletion from Nova nor did it use a timeout. This
    would lead to infinite execution when servers didn't delete properly. This
    patch refactors delete to use the standard handle->check lifecycle methodology
    and accounts for error statuses during delete.

    Change-Id: I6b13c1201e46c0e73aa11e78825045b345931fb7
    Partial-Bug: #1329453

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/102638
Committed: https://git.openstack.org/cgit/openstack/heat/commit/?id=2bbc908b003d969821b2a0c491dbb2a81c718aa2
Submitter: Jenkins
Branch: master

commit 2bbc908b003d969821b2a0c491dbb2a81c718aa2
Author: Randall Burt <email address hidden>
Date: Wed Jun 25 15:39:20 2014 -0500

    Refactor AWS::EC2::Instance delete handling

    Refactor instance to use the nova_utils delete function.

    Closes-Bug: #1329453
    Change-Id: Ie84fb3c5c9f6381e4eea3a5e99d62266a8abd023