DB Duplicate error while scheduling distributed routers

[Armando Migliaccio]

I observed this error here:

http://logs.openstack.org/77/108177/10/experimental/check-tempest-dsvm-neutron-dvr/9e67d95/logs/screen-q-svc.txt.gz?level=TRACE#_2014-07-31_15_05_48_864

And in few other places. This seems to be triggered (for instance) during the following testcase:

tempest.scenario[...]test_server_connectivity_pause_unpause

It looks like the scheduling process fails because of a duplicate entry to the agent/router binding table. This might be an effect of fix:

https://review.openstack.org/#/c/73234/

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/111143

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/111143
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=0b4e42fe11bf918e18ea8f240d9055b3967b60bb
Submitter: Jenkins
Branch: master

commit 0b4e42fe11bf918e18ea8f240d9055b3967b60bb
Author: armando-migliaccio <email address hidden>
Date: Thu Jul 31 19:20:00 2014 -0700

    Fix DB Duplicate error when scheduling distributed routers

    The error was caused by binding the router to an agent
    candidate that was already selected during the scheduling
    process.

    A DB lookup was also saved by passing the router object
    around; this led to a minor style cleanup.

    Closes-bug: #1351123

    Change-Id: Ib71a0140c8a7fbd5b230609d33487f8adba252e7