RabbitMQ keeps failing and we fix it by restarting the service but we don't know why it fails. Has anybody else experienced these issues?
|
feedback
|
|
What are the symptoms when RabbitMQ fails? Is it consuming all CPU and memory? Looking through the logs should give you a good idea of what the problem is. We've had a similar problem at our production site and it was due to the fact that we were throwing way too many jobs at RabbitMQ and it simply could not keep up. After many months of tweaking, we decided to dump RabbitMQ in favour of Resque (plus Redis). It seems to be doing a much better job than RabbitMQ. If that's not a viable option, you could try clustering it. RabbitMQ supports this natively. Of course, this assumes RabbitMQ is crashing due to load. | |||
|
feedback
|
|
Did you look in /var/log/rabbitmq/* ? If the startup_log complains about persister errors you are probably running an old version of Rabbit and should upgrade to 2.4.1 Are messages multiplying inside your system? we had one process that wrote a message to a completed queue and to an error queue, and both queues had no realtime listeners. Run top and see what Rabbit's memory usage is. When it gets to 40% Rabbit has problems, especially if it has a large number of persistent messages in its queues. Did someone run some experiments with persistent queues in another vhost, and then abandon the experiment with the queues full? | |||
|
feedback
|
