Ensure rabbitmq users exist when processing access requests

Bug #1782330 reported by Alvaro Uria
14
This bug affects 2 people
Affects Status Importance Assigned to Milestone
OpenStack RabbitMQ Server Charm
Triaged
Medium
Unassigned

Bug Description

Initial status:
* 3 rmq units clustered and related to OpenStack services
* rmq users created when the application was related to other OpenStack applications (nova-cloud-controller, neutron-api, etc.)

Step to reproduce failure:
* juju remove-unit <all 3 units> # ie. because mnesia db is corrupted and we are unable to restart the cluster
* juju add-unit --to lxd:host1
* juju add-unit --to lxd:host2
* juju add-unit --to lxd:host3

Problem:
* New rmq cluster does not have rmq users (amqp-relation-changed hook does not recreate them)

Workaround:
* juju remove-relation rabbitmq-server:amqp <openstackservice>:amqp
* juju add-relation rabbitmq-server:amqp <openstackservice>:amqp

Request:
* Juju action to trigger rmq user recreation (ie. by relation or "--all"), to avoid the remove/add relation dance

Thank you.

Revision history for this message
James Page (james-page) wrote :

I really think the charm should just do the right thing; when the new units appear and one is elected leader, it should figure out that the users don't exist and re-create them (I suspect the check is done on presence of data in the leader database provided by juju).

summary: - Juju action to (re)create amqp users
+ Ensure rabbitmq users exist when processing access requests
James Page (james-page)
Changed in charm-rabbitmq-server:
status: New → Triaged
importance: Undecided → Medium
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.