L3 scheduler: make RouterL3AgentBinding always concurrently safe

[John Schwarz]

Changeset I3447ea5bcb7c57365c6f50efe12a1671e86588b3 added a binding_index column to the RouterL3AgentBinding table, which is unique with the router_id. However, the current logic isn't concurrent safe as some concurrent cases can raise a DBDuplicateEntry (if the same binding_index is being used by 2 different workers).

[Brian Haley (brian-haley)]

https://review.openstack.org/#/c/365326/ is up for review

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/365326
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=1e195afaf12cbb426a356a11cacfe45f04bd43a2
Submitter: Jenkins
Branch: master

commit 1e195afaf12cbb426a356a11cacfe45f04bd43a2
Author: John Schwarz <email address hidden>
Date: Sun Sep 4 13:48:58 2016 +0300

    L3 scheduler: add retry indicators for bind_router

    Change I3447ea5bcb7c57365c6f50efe12a1671e86588b3 added a binding_index
    to the RouterL3AgentBinding table. In certain (concurrent) cases, a row
    with the same binding_index might be used twice, which will raise
    DBDuplicateEntry. However, that change didn't retry on this case at all
    code-paths, so this patch rectifies this issue.

    Closes-Bug: #1633042
    Change-Id: I228b0084a8e7c48e78bc2ea6a3fccf7437210e73

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/neutron 10.0.0.0b2 development milestone.