Description of problem: All neutron agent were noticed as down. No other openstack components were noticed as effected. =ERROR REPORT==== 22-Aug-2017::10:04:58 === Discarding message {'$gen_call',{<0.1053.0>,#Ref<0.0.3.72786>},{basic_consume,false,<5644.21200.217>,<5644.28503.217>,false,0,<<"1">>,false,[],undefined}} from <0.1053.0> to <0.2208.0> in an old incarnation (3) of this node (1) =ERROR REPORT==== 22-Aug-2017::10:04:59 === Channel error on connection <0.26111.0> (172.22.178.74:43108 -> 172.22.178.74:5672, vhost: '/', user: 'guest'), channel 1: operation queue.declare caused a channel exception not_found: "failed to perform operation on queue 'q-reports-plugin' in vhost '/' due to timeout" No alarms noticed Version-Release number of selected component (if applicable): rabbitmq-server-3.6.3-6.el7ost.noarch RHOS9 python-oslo-messaging-4.5.0-2.el7ost.noarch How reproducible: On customer environment , until rabbitmq was restarted. Steps to Reproduce: 1. 2. 3. Actual results: all neutron agents were noticed as down in neutron agent-list Restarting individual components does not fix the issue. Restarting rabbitmq server across all controller fixed the issue. Expected results: Additional info: Looks similar to the following bug, except that its supposed to fixed in above version. https://bugzilla.redhat.com/show_bug.cgi?id=1418668
At a glance, this looks like https://github.com/rabbitmq/rabbitmq-server/issues/959 but will require some more log analysis to be sure.