Comment 2 for bug 1785568

Reviewed: https://review.openstack.org/588479
Committed: https://git.openstack.org/cgit/openstack/tripleo-upgrade/commit/?id=fe8ede8f4ad8992b3f9ce05cef2db55d02a4a14e
Submitter: Zuul
Branch: stable/pike

commit fe8ede8f4ad8992b3f9ce05cef2db55d02a4a14e
Author: Yurii Prokulevych <email address hidden>
Date: Fri Aug 3 10:59:29 2018 +0200

    Don't invoke multiple migration requests for single vm.

    This is Pike only fix. If there are multiple migration requests
    for the same instance, the placement inventory might
    not be cleaned up in time, leading to further migration requests
    to this compute to fail:
      <snip>
      Allocation for VCPU on resource provider
      b6d1973f-bc39-4ea0-8d28-15f988f762e1 violates min_unit, max_unit,
        or step_size. Requested: 5, min_unit: 1, max_unit: 4, step_size: 1
      Placement API returning an error response:
       Unable to allocate inventory:
        Unable to create allocation for 'VCPU' on resource provider
         The requested amount would violate inventory constraints

    Change migration timeout to 180 seconds.
    Both nova-compute and neutron-openvswitch-agent have to be up to
    have live-migration working. Added check for neutron's part.

    Closes-Bug: 1785568
    Change-Id: If46c43aee9a409b59f2dcb139d739106bc9a562f