[rabbitmq-discuss] .NET API - Using PublishConfirms to get Reliability

[Michael Klishin]

On 26 Feb 2014, at 11:48, Marty Wasznicky <marty.wasznicky at neudesic.com> wrote:

> Yet I'm still getting sometimes (actually ) always losing messages.  I'm still trying to track it down.

May I point out that you still haven’t specified what your expectations are.

“Losing messages” is a broad claim and the root cause can vary quite a bit:

 * Messages never make it to the server because of an issue in the client
 * Data is in flight
 * Server acks too early
 * Messages are not published as persistent or queues are not durable
 * Message recovery issues
 * Peer failure is not detected adequately by the client

and so on. Some of these issues are pretty unlikely in a product that is 7 years old
but to investigate this we need to clarify *expectations*. Moreover, the reasons
will vary between “losing messages” on the publisher end vs. on the consumer end.
Potential issues in the two so far have been lumped together in this thread.

Can you please explain what your expectations are, separately for publishers and consumers,
in the event of one node and full cluster failure, as detailed as possible?
--
MK

Software Engineer, Pivotal/RabbitMQ