Comment 7 for bug 1781648

Reviewed: https://review.openstack.org/597571
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=4fe4fbe7a43ac7e971b0aabc8a9079c7701cae55
Submitter: Zuul
Branch: master

commit 4fe4fbe7a43ac7e971b0aabc8a9079c7701cae55
Author: Matt Riedemann <email address hidden>
Date: Wed Aug 29 11:58:12 2018 -0400

    Restart scheduler in TestNovaManagePlacementHealAllocations

    TestNovaManagePlacementHealAllocations tests rely on the
    CachingScheduler specifically because it does not use
    placement and therefore does not create allocations during
    scheduling, which gives us instances that need to heal
    allocations.

    However, we have a race in the test setup where the scheduler
    is started before the compute services. The CachingScheduler
    runs a periodic task on startup to fetch the hosts from the
    DB to initialize it's cache, and then uses that during scheduling
    until the periodic runs again. So what we can hit is the
    scheduler starts, loads an empty cache, then we start the computes
    and try to create an instance but because of the empty cache we
    fail with a NoValidHost error.

    This restarts and resets the CachingScheduler cache *after* we have
    started the computes and asserted they are available in the API.

    Change-Id: I32f607a436e9851a96877123ae3d1fe51f444f73
    Closes-Bug: #1781648