TTL pattern does not match leaking queues

[Gabriel Samfira]

The PR:

https://opendev.org/openstack/charm-heat/commit/de88ad5344c4411dd81c600f724b760a0e8cf03f

allowed setting a TTL policy on heat queues. However, it still leaks quite a fair number of queues after each restart (18 for a 3 node heat deployment).

The following diff sets a pattern that will set a TTL on all heat queues (fanout included):

diff --git a/hooks/heat_relations.py b/hooks/heat_relations.py
index fa6d9ce..3c937e7 100755
--- a/hooks/heat_relations.py
+++ b/hooks/heat_relations.py
@@ -225,7 +225,7 @@ def amqp_joined(relation_id=None):
     relation_set(relation_id=relation_id,
                  username=config('rabbit-user'), vhost=config('rabbit-vhost'),
                  ttlname='heat_expiry',
- ttlreg='heat-engine-listener|engine_worker',
+ ttlreg='^heat-engine-listener.*|^engine_worker.*|^engine_fanout.*',
                  ttl=config('ttl'))

This, coupled with the change in:

https://bugs.launchpad.net/charm-rabbitmq-server/+bug/1939681

made the queues disapear after TTL expired.

[Billy Olsen (billy-olsen)]

Triaged by looking at the regex provided in the diff. Seems pretty straightforward; adjust the regex to account for queues that are prefixed with the queuenames and adds the engine_fanout queue.