Comment 21 for bug 1836754

Reviewed: https://review.opendev.org/c/openstack/nova/+/688802
Committed: https://opendev.org/openstack/nova/commit/c09d98dadb6cd69e294420ba7ecea0f9b9cfcd71
Submitter: "Zuul (22348)"
Branch: master

commit c09d98dadb6cd69e294420ba7ecea0f9b9cfcd71
Author: Matt Riedemann <email address hidden>
Date: Tue Oct 15 15:49:55 2019 -0400

    Add force kwarg to delete_allocation_for_instance

    This adds a force kwarg to delete_allocation_for_instance which
    defaults to True because that was found to be the most common use case
    by a significant margin during implementation of this patch.
    In most cases, this method is called when we want to delete the
    allocations because they should be gone, e.g. server delete, failed
    build, or shelve offload. The alternative in these cases is the caller
    could trap the conflict error and retry but we might as well just force
    the delete in that case (it's cleaner).

    When force=True, it will DELETE the consumer allocations rather than
    GET and PUT with an empty allocations dict and the consumer generation
    which can result in a 409 conflict from Placement. For example, bug
    1836754 shows that in one tempest test that creates a server and then
    immediately deletes it, we can hit a very tight window where the method
    GETs the allocations and before it PUTs the empty allocations to remove
    them, something changes which results in a conflict and the server
    delete fails with a 409 error.

    It's worth noting that delete_allocation_for_instance used to just
    DELETE the allocations before Stein [1] when we started taking consumer
    generations into account. There was also a related mailing list thread
    [2].

    Closes-Bug: #1836754

    [1] I77f34788dd7ab8fdf60d668a4f76452e03cf9888
    [2] http://lists.openstack.org/pipermail/openstack-dev/2018-August/133374.html

    Change-Id: Ife3c7a5a95c5d707983ab33fd2fbfc1cfb72f676