corner case in nsx api_client code

[Aaron Rosen]

There is a corner case that the nsx api_client code does not handle today where the nsx controller can return a 307 in order to redirect the request to another controller. At this point neutron-server issues this request to the redirected controller and usually this works fine. Though in the case that the session cookie has expired we'll issue the request and get a 401 and clear the cookie from the request. Then we'll retry the request and get the same 307 again which will result in a 401 as the session cookie was never renewed.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/106494
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=0b7a7687309e8c740ccf79fbbbcabbc64b6f5fd1
Submitter: Jenkins
Branch: master

commit 0b7a7687309e8c740ccf79fbbbcabbc64b6f5fd1
Author: Aaron Rosen <email address hidden>
Date: Fri Jul 11 13:56:15 2014 -0700

    NSX: Correct default timeout params

    Previously, req_timeout and http_timeout were set to the same value
    which is not correct. req_timeout is the total time limit for a cluster
    request and http_timeout is the time allowed before aborting a request on
    an unresponsive controller. Since the default configuration allows 2
    retries req_timeout should be double that of http_timeout because of this
    this patch goes ahead and removes req_timeout as this should just be
    http_timeout * retries.

    Because prevouly req_timeout and http_timeout were the same this exposed
    a corner case that when the nsx controller returned a 307 we would issue
    the request against the redirected controller but in the case where the
    session cookie had expire when the request was issued we would get a 401
    response back and never retry the request. Now that the default values are
    corrected this issue should no longer occur as the next time time we issue
    the request we'll fetch a new auth cookie for the redirected controller.

    This patch also bumps the timeout values to be higher. We've seen
    more and more timeouts occur in our CI system largely because our
    cloud is overloaded so increasing the default timeouts will *hopefully*
    help reduce test failures.

    DocImpact

    Closes-bug: 1340969
    Closes-bug: 1338846

    Change-Id: Id7244cd4d9316931f4f7df1c3b41b3a894f2909a