Comment 11 for bug 1334345

Reviewed: https://review.openstack.org/103161
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=9a01e62693a28a73120544b27ee2104558e0250e
Submitter: Jenkins
Branch: master

commit 9a01e62693a28a73120544b27ee2104558e0250e
Author: Matt Riedemann <email address hidden>
Date: Fri Jun 27 06:46:10 2014 -0700

    Enforce task_state is None in ec2 create_image stop instance wait loop

    We're hitting races in the gate where the instance.vm_state is STOPPED
    but the task_state is POWERING_OFF so when the compute_api.start method
    is called, we're in an invalid task state and fail.

    The compute manager's stop_instance method is correctly setting the
    vm_state to STOPPED and the task_state to None when the instance is
    powered off via the virt driver, so we must be hitting this from races
    in the ec2 API as noted in the TODO above the method definition.

    This change simply checks the task_state in addition to the vm_state
    in the wait loop before continuing. The error message is also updated
    for context by including the instance uuid, vm_state and task_state,
    and removes the timeout value in the message since it was in
    milliseconds, not seconds, to begin with.

    There is already a unit test that covers this change (which was racing,
    hence the bug). There are no changes to that unit test since it's really
    an integration test that's running through the compute API and compute
    manager code, so the fix tests itself.

    Closes-Bug: #1334345

    Change-Id: I13f0c743cadda6439ae15607a9ef6e4e4985626d