Comment 18 for bug 1494363

Reviewed: https://review.openstack.org/278501
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=a5a7b892df46f5d48840f600f03e7be2db999a1d
Submitter: Jenkins
Branch: master

commit a5a7b892df46f5d48840f600f03e7be2db999a1d
Author: Jakub Libosvar <email address hidden>
Date: Wed Feb 10 16:35:30 2016 +0000

    fullstack: Gracefully stop neutron-server process

    There is possible scenario that neutron-server loses connection to AMQP
    bus and that can cause timeouts when hanging on rpc calls. On the other
    hand, we should be able to stop service gracefully even if AMQP server
    isn't running.

    During teardown there were still neutron-server orphaned api workers
    processes running that had open connection to database which caused
    unexpected failures.

    Note that this change applies only for neutron-server because previous
    investigations showed agents using oslo service can hang on rpc causing
    other failures. Next step should be making sure rabbitmq is stopped as
    last or decrease rpc timeouts in agents.

    Change-Id: I3d793ba924dc3fd229d8588f4be1e943614f22bb
    Related-bug: 1494363
    Closes-bug: 1541742