[rabbitmq-discuss] RabbitMQ Aliveness Test vs Monitoring RabbitMQ Process

[Emile Joubert]

Hi,

On 17/07/12 08:21, Saravana.Prabhu at emc.com wrote:
> I would like to know if we can have scenarios where Broker Process will
> be up and running, but it could not be able to publish and receive
> messages due to unknown reasons.

The API documentation has an explanation of what it does:

http://hg.rabbitmq.com/rabbitmq-management/raw-file/default/priv/www/api/index.html

The aliveness test sends and receives a message. This is a more
meaningful (and more expensive) test than checking an OS PID. It only
makes sense to monitor the OS PID if you expect a crash so dramatic that
the entire Erlang VM crashes. This can happen e.g. in an out-of-memory
scenario, but there are many failure modes that could cause the broker
to stop working while the OS PID of the Erlang VM remains running. So I
suggest using the aliveness test instead.


-Emile