[RFE] Add support for DBDeadlock handling.

[Łukasz Leszczuk]

Current implementation lacks proper handling of DBDeadlock exception, which is happening quite often on Galera clusters under load.

Proper DBDeadlock handling is important from stability perspective.

[Grzegorz Grasza (xek)]

nova does this with @oslo_db_api.wrap_db_retry(max_retries=10, retry_on_deadlock=True) on all write operations on the DB

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/394413

[Ruby Loo (rloo)]

Seems like this is more of a bug than a feature, but I'm fine either way, I think this is a good idea :)

[Ruby Loo (rloo)]

jroll and JayF are good with this too [1]

[1] http://eavesdrop.openstack.org/irclogs/%23openstack-ironic/%23openstack-ironic.2016-11-09.log.html#t2016-11-09T21:48:34

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/394413
Committed: https://git.openstack.org/cgit/openstack/ironic/commit/?id=3428cb74f0a6fe05c23cf00dc66da4c1d9766409
Submitter: Jenkins
Branch: master

commit 3428cb74f0a6fe05c23cf00dc66da4c1d9766409
Author: Joanna Taryma <email address hidden>
Date: Mon Feb 27 13:39:37 2017 -0800

    Add support for DBDeadlock handling

    Wrap all DB write operations with oslo_db decorator to retry on
    deadlock. DBDeadlock exception can be raised quite often on Galera
    clusters under load.
    This also fixes an issue which causes conductor's periodic
    tasks to stop executing.

    Closes-Bug: #1637210
    Closes-Bug: #1639338
    Co-Authored-By: Vladyslav Drok <email address hidden>
    Co-Authored-By: Joanna Taryma <email address hidden>
    Change-Id: I61db83637adfd98a5394d1f570f3de4302c93497

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/ironic 8.0.0 release.