[rabbitmq-discuss] Query related to Rabbitmq Clustering.

[Simon MacMullen]

On 13/12/12 08:15, Vikrant Sayeewal wrote:
> Could you please let me know if in any of the cluster 2 rabbitmq are
> running and suddenly both the rabbitmq node are down and then any of the
> node come back then there will be a messages loss or data loss?

Assuming that you have published persistent messages to durable queues, 
you will not lose any messages for which your publisher has received 
confirmations. Non-persistent messages and non-durable queues will be 
lost, and any messages which were in flight (i.e. you have not received 
confirmations) may or may not be lost depending on where they were when 
the cluster went down.

Note also that if node B goes down first, *then* node A, then node A 
will have to come back before node B will work (since node B cannot 
trust its copy of the cluster state; other things may have happened 
while it was down). If node A is not going to come back you can use 
"rabbitmqctl forget_cluster_node".

> Or application become down until we restart the whole cluster?

Not quite sure what you mean here, hopefully my answer above helps.

Cheers, Simon

-- 
Simon MacMullen
RabbitMQ, VMware