transport_url not being used while doing pool update

[Eduardo Gonzalez]

RabbitMQ options in oslo_messaging_rabbit section are deprecated to make use of transport_url in DEFAULT section.

However when doing designate-manage pool update, transport_url option is ignored and still using old rabbitmq ones. As those options are not used any more, are using default values (127.0.0.1).

After updating pools using rabbitmq options, I've removed those options only leaving transport_url and all request work fine with no error at logs and recordset properly created automatically when launching instances. Guess is only an issue with designate-manage.

Using CentOS from RDO packages for pike (delorean)

Updating Pools Configuration
****************************

2017-03-16 09:34:16.687 17 ERROR oslo.messaging._drivers.impl_rabbit [designate-manage - - - - -] [924446d7-b32d-4245-9491-482d011d7324] AMQP server on 127.0.0.1:5672 is unreachable: [Errno 111] ECONNREFUSED. Trying again in 1 seconds. Client port: None
2017-03-16 09:34:17.697 17 ERROR oslo.messaging._drivers.impl_rabbit [designate-manage - - - - -] [924446d7-b32d-4245-9491-482d011d7324] AMQP server on 127.0.0.1:5672 is unreachable: [Errno 111] ECONNREFUSED. Trying again in 2 seconds. Client port: None
2017-03-16 09:34:19.740 17 ERROR oslo.messaging._drivers.impl_rabbit [designate-manage - - - - -] [924446d7-b32d-4245-9491-482d011d7324] AMQP server on 127.0.0.1:5672 is unreachable: [Errno 111] ECONNREFUSED. Trying again in 4 seconds. Client port: None

Deprecation Messages
********************

2017-03-16 09:53:08.349 5 WARNING oslo_config.cfg [-] Option "rabbit_hosts" from group "oslo_messaging_rabbit" is deprecated for removal. Its value may be silently ignored in the future.
2017-03-16 09:53:08.349 5 WARNING oslo_config.cfg [-] Option "rabbit_password" from group "oslo_messaging_rabbit" is deprecated for removal. Its value may be silently ignored in the future.
2017-03-16 09:53:08.349 5 WARNING oslo_config.cfg [-] Option "rabbit_userid" from group "oslo_messaging_rabbit" is deprecated for removal. Its value may be silently ignored in the future.

[Johannes Kulik (jkulik)]

I've debugged that a little: The problem is, that `rpc.init(CONF)` is called in the `__init__()` of the respecting command class (e.g. `PoolCommand` https://github.com/openstack/designate/blob/master/designate/manage/pool.py#L41), this happens, before the config has been parsed - or rather, it seems, while the config is parsed (here: https://github.com/openstack/designate/blob/master/designate/cmd/manage.py#L56).

A possible solution might be letting each command init and create the rpc client or add a decorator to each action to call an init method initializing/creating the rpc client. The former definitely uses `transport_url`, as that's my hotfix for now.