Description of problem: What problem/issue/behavior are you having trouble with? What do you expect to see? Openstack services always connect to the first message broker defined in the rabbit_hosts list. This issue was solved in Icehouse but was introduced again during the refactor done here: https://github.com/openstack/oslo.messaging/commit/973301aa70527171749fa34897276c43898aeeb2 This issue impacts in the load balancing of rabbitmq connections and introduces a degradation of the service when the first controller is down (as Oslo always try to connect to the Stopped broker, and after the failure, it connects to the second one) The patch has been submitted upstream: https://review.openstack.org/#/c/231039/ We would like to backport it downstream as soon as possible Version-Release number of selected component (if applicable): rabbitmq-server-3.3.5-4.el7
*** This bug has been marked as a duplicate of bug 1267736 ***