will just
- start some docker containers
- build a cluster
- configure users
- generate some load
- kill a node (container)
- check if broken queues exist
created to allow rabbitmq list to (hopefully) debug the issue.
to use check_rabbitmq_unroutable_msg
- run check_rabbitmq_unroutable_msg for the first time, this will setup some queues, bindings, ...
- create an fanout-exchange "unroutable"
- create a binding from unroutable-exchange to unroutable-queue
- change your policy to configure an "alternate exchange" => "unroutable"
- enable the "unroutable msg counter" feature in rabbitmq
check_rabbitmq_unroutable_msg will now trigger an alert if:
- the unroutable msg counter increases
- msgs got placed in "unroutable"
to fix detected errors
ERROR:root:dropped msg in 10.77.16.106:41036 -> 10.78.22.11:5672 (1) 'neutron'
- run wip-detect-broken-bindings, write output into file
- execute the output-file (contains bash-cli-cmds)
ERROR:root:Exchange: openstack, RoutingKey: cinder-scheduler could not be routed
- open your rabbitmq-mngt-webgui
- navigate to the "openstack" exchange
- click on the queue using the "cinder-scheduler"-routing-key
- click on "delete queue"