Closed estahn closed 6 years ago
There is no such metric (yet). If I understand how cli works correctly, this unresponsive queues list produces by iterating over all (local in our case) queues and querying queue process (https://github.com/rabbitmq/rabbitmq-server/blob/master/src/rabbit_amqqueue.erl#L762 and https://github.com/rabbitmq/rabbitmq-server/blob/master/src/rabbit_amqqueue.erl#L700) and waiting for a timeout.
While it's probably ok for occasional manual use I doubt it's healthy to do regularly because 1) it could add load 2) it can take too much time and prometheus will just give up and mark endpoint as a failed one.
Having this in mind, it still can be added. But this metric will be disabled by default and perhaps timeout will be configurable too.
@deadtrickster Fair enough. I will reopen if I need something specific.
Is there a metric per queue available whether the queue is responsive or not (e.g.
rabbitmq_queue_unresponsive
true|false)?If not, can we add one?