Comment 3 for bug 1726203

Reviewed: https://review.openstack.org/514103
Committed: https://git.openstack.org/cgit/openstack/barbican/commit/?id=71829664b2ac4547c54940ad62bd8d0123f6e422
Submitter: Zuul
Branch: master

commit 71829664b2ac4547c54940ad62bd8d0123f6e422
Author: Dave McCowan <email address hidden>
Date: Sun Oct 22 22:19:26 2017 -0400

    Fix Race Condition in get_or_create_project()

    get_or_create_project() first calls a get (query of the database).
    If that fails, then it calls create. The issues is when
    there are two threads and both fail the original get, then
    each will call create. One of those creates will fail, resulting
    in the traceback.

    One way to protect this critical region is to call create first,
    then if that fails, call get.

    In this case, get_or_create_project() is called for every
    Barbican API operation. The create action (which is slow), happens
    once per project for the life of the cloud. The get action happens
    every other time. For performance gains, I left call get first,
    since that call succeeds almost always.

    Change-Id: I2f13b495a6b45e226e0742291f76ee88dbcf52df
    Closes-bug: #1726203