[rabbitmq-discuss] Pathological dead queue behavior in a cluster.

[Matthias Radestock]

Matthew Sackman wrote:
> Binding recreation is dependent on both the exchange and queue being
> durable, even if the exchange survives due to the fact it's part of a
> cluster.

Simon and I were discussing that very scenario today and concluded that 
rabbit is probably exhibiting the wrong behaviour here. I have filed a bug.

An interesting related issue is whether a client should be able to 
add/remove bindings for such queues while they are unavailable. I reckon 
the answer is 'no'.


Matthias.