rabbitmq fail to restart on image update/upgrade

[Vladislav Belogrudov]

If one runs upgrade from release 0.1 to 0.2 rabbitmq data container will restart
and will lose cookie. Bootstrap container won't recreate it because the check for
cookie is done before the data container restarts. This results in failure of
rabbitmq start-up.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/251434

[Vladislav Belogrudov (vlad-belogrudov)]

The problem is much more serious because all data containers restart and remove any state. Only in HA environment one can do such update which essentially means host wipe-out-reinstall

[Steven Dake (sdake)]

Note in the bug description is that the cookie should be recreated - this is not accurate. Once the data container has been created once, it should never be created again.

[Vladislav Belogrudov (vlad-belogrudov)]

one could set pull policy to missing in bootstrap tasks to avoid data container restart as workaround

[Steven Dake (sdake)]

This is resolved with the upgrade action in mitaka.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Vladislav Belogrudov (<email address hidden>) on branch: master
Review: https://review.openstack.org/251434
Reason: no need