[rabbitmq-discuss] Cluster Recovery Modes

[Tim Watson]

On 24 Sep 2013, at 00:25, Cordell, Ron wrote:
> I meant to ask in this same message about the heartbeat interval set to 0 and how it is different for the cluster 'ignore' mode.
> 

AMQP heartbeats are a feature to enable client and server to detect connection failures in a timely fashion), whereas clustering operates over a completely different network transport and uses a different protocol. So there's no connection between these two things.

> One of the cluster recovery configurations is "ignore":
[snip]
> Does this mode effectively turn off the heartbeat?

No - as mentioned above, AMQP heartbeats and clustering are orthogonal.

> Is a node detected as partitioned but nothing done?

Yes, that's pretty much what this recover mode does. The idea behind 'ignore' is to avoid any further loss of service. The other recovery modes can lead to one or more of the remaining (unpartitioned) nodes restarting in order to heal the cluster (see 'autoheal' mode) or total loss of service on some nodes if they reside in a minority island (see 'pause_minority' mode). 

Cheers,
Tim

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20130924/b3a5009a/attachment.htm>