[rabbitmq-discuss] Examining Queue Failover Behavior

[Ben Hood]

Jason,

On Wed, Feb 11, 2009 at 4:43 PM, Jason J. W. Williams
<jasonjwwilliams at gmail.com> wrote:
> Everyone seems to think I expected the content to migrate. :-)

Life is tough and unfair :-)

> I only
> expected the queue to migrate (sans content). :-) Otherwise, messages
> published to the exchange get blackholed and persistence doesn't mean
> a whole lot.

This is true. There are two issues here (which are unlikely to
resolved quickly):

1. Specifying persistence as a message level attribute is
questionable. There is a school of thought that says this this should
an attribute of a queue. This would be a discussion for either
evolving the spec or looking into whether a broker could handle this
as a QOS feature.

2. Rabbit does not automatically failover - currently it only
addresses the ability to recover.

As indicated in my previous post from 20 mins ago, there may be a way
to mitigate this, but the overall lack of automatic failover may be a
KO for you.

Ben