Description of problem:
When rebooting controllers one by one, rabbitmq gets in a splitbrain state and nothing work any longer until rabbitmq-bundle is restarted and nova-computes containers are restarted.

Version-Release number of selected component (if applicable):
(undercloud) [stack@undercloud-0-rhosp13 ~]$ cat local_registry_images.yaml  | grep rabbit
- imagename: registry.access.redhat.com/rhosp13/openstack-rabbitmq:13.0-74


How reproducible:
Always

Steps to Reproduce:
1. Reboot controllers one by one
2. Try creating a VM
3.

Actual results:
Fails

Expected results:
Succeds

Additional info:
This appears to be consistent for many customers.