Handle upgrading cinder-volume under pacemaker
authorAlan Bishop <abishop@redhat.com>
Tue, 23 May 2017 14:42:24 +0000 (10:42 -0400)
committerAlan Bishop <abishop@redhat.com>
Fri, 2 Jun 2017 15:56:40 +0000 (11:56 -0400)
commit862585bbbcf5a8b9bd171ada44cda90d3600f2d5
tree852cd4a73ad823c5752cfcd0892fe189d91f8fc4
parent0b6ce86e7ae59f25a6502269b216f16e2189708a
Handle upgrading cinder-volume under pacemaker

Add upgrade tasks for cinder-volume when it's controlled by pacemaker:

o Stop the service before the entire pacemaker cluster is stopped.
  This ensures the service is stopped before infrastructure services
  (e.g. rabbitmq) go away.
o Migrate the cinder DB prior to restarting the service. This covers
  the situation when puppet-cinder (who otherwise would handle the db
  sync) isn't managing the service.
o Start the service after the rest of the pacemaker cluster has been
  started.

Closes-Bug: #1691851
Change-Id: I5874ab862964fadb68320d5c4de39b20f53dc25c
(cherry picked from commit c4e3bbe039135f32f0e198365e704b3dbfd00290)
puppet/services/pacemaker/cinder-volume.yaml