[rabbitmq-discuss] Question about configuration of queues

Wed Nov 5 00:45:02 GMT 2008

William,

On Wed, Nov 5, 2008 at 12:08 AM, Beene, William R.
<beenewilliamr at gmail.com> wrote:
> I am not entirely sure what you mean by queue depth. This is my first
> time dealing with
> AMQP.

Queue depth refers to the number of undelivered messages in a queue at
any point in time.

> Our scenario is that we have several producers of data and
> several consumers.
> All consumers want to get all of the data from all of the producers
> for the most part.

This could achieved by declaring a queue per consumer and setting up
your routing rules so that a copy of every message that gets published
is sent to each queue.

You *may* be interested in this background article about consumer
driven messaging:
http://hopper.squarespace.com/blog/2008/10/7/consumer-driven-messaging.html

> I don't want to worry about start up order and I also don't want to
> worry about how to
> handle a consumer losing the connection and missing several messages
> when it reconnects.

Unless you declare the queues to be auto-delete, the broker will
buffer up the messages for each client until they reconnect.

So I can I infer from this statement that you don't actually have a
requirement to complete a batch of production strictly before begin
consumption?

Futhermore, you may be interested in some of the FAQs:
https://dev.rabbitmq.com/wiki/FrequentlyAskedQuestions

HTH,

Ben