[rabbitmq-discuss] detecting publish failure across restart

[jesse at wavemarket.com]

I'm running RabbitMQ 1.7.2 from package on Ubuntu 10.04 LTS and using the
Java Client (also 1.7.2).

I'm using a single durable, non-auto delete exchange, a single durable
queue, and a single direct binding for a single routing key.

I'm testing calls to Channel.basicPublish using persistent, mandatory (but
non-immediate) messages. I have another process that consumes the messages,
but it's not relevant for this test.

My test case was:
 - Publish ~100K messages
 - In the middle, restart RabbitMQ
 - Count the number of messages in the queue according to both my Java
client and RabbitMQ (via "rabbitmqctl list_queues")

I consistently see that the Java client sends some messages that do not
cause an exception and that do not make into the queue. I have a registered
ReturnListener, but it does not record any messages being returned.

I understand that publishing a message is asynchronous, but surely there
must be a way to discover that message was not successfully published. What
am I missing?

Thanks,

Jesse
-- 
View this message in context: http://old.nabble.com/detecting-publish-failure-across-restart-tp28712608p28712608.html
Sent from the RabbitMQ mailing list archive at Nabble.com.