Use Rabbitmq clusterer

Registered by Sam Yaple

Currently we run rabbitmq through the load balancer. This was never ideal but it was the only setup that would allow dynamically growing the rabbitmq nodes. Using clusterer [1] we can hot reload the cluster in much the same way we do haproxy. Additionally, this will improve the stability of the cluster especially when forming the initial cluster.

Of note, this requires rabbitmq-server>=3.5.4 which we already have in our environments.

[1] https://github.com/rabbitmq/rabbitmq-clusterer

Blueprint information

Status:
Complete
Approver:
Sam Yaple
Priority:
Essential
Drafter:
Sam Yaple
Direction:
Approved
Assignee:
Sam Yaple
Definition:
Approved
Series goal:
Accepted for mitaka
Implementation:
Implemented
Milestone target:
milestone icon mitaka-1
Started by
Sam Yaple
Completed by
Sam Yaple

Related branches

Sprints

Whiteboard

Gerrit topic: https://review.openstack.org/#q,topic:bp/rabbitmq-clusterer,n,z

Addressed by: https://review.openstack.org/235756
    Download, install, and enable rabbitmq_cluster

Addressed by: https://review.openstack.org/235757
    Use rabbitmq-clusterer when deploying

Addressed by: https://review.openstack.org/235777
    Remove vip for rabbitmq

Addressed by: https://review.openstack.org/242381
    Download, install, and enable rabbitmq_cluster

Addressed by: https://review.openstack.org/242382
    Use rabbitmq-clusterer when deploying

Addressed by: https://review.openstack.org/242384
    Remove vip for rabbitmq

(?)

Work Items

Work items:
Enable clusterer plugin for rabbitmq: DONE
Update Ansible with the new flow and configs: DONE
Update configs using rabbitmq with the appropriate servers list: DONE
Add hot-reload to rabbitmq cluster config: TODO

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.