[rabbitmq-discuss] Identify master node
girish at giri.sh
Tue Jul 16 15:28:33 BST 2013
I ran the command and here's what I see -
I was expecting to see the name of the queues and the master node listed
next to it. That's not quite what I see on the screen though.
Also, does a master node in this case refer to the node the queue was
created on? And if yes, does this mean if there are multiple queues in an
HA cluster (each created on a different node), there could multiple masters
at any given time in a mirrored queue set up?
On Tue, Jul 16, 2013 at 6:26 AM, Emile Joubert <emile at rabbitmq.com> wrote:
> On 15/07/13 22:19, Girish Gangadharan wrote:
> > Can somebody please explain to me how does one go about identifying the
> > slave vs master nodes in a cluster? The idea is to bring one down and
> > see how the clients handle them, especially the Consumer Cancellation
> > notifications (for testing high availability setup). Also, is that a
> > valid way to test HA? Or the fact that these nodes sit behind a
> > load-balancer make it an unnecessary, redundant test?
> There is no unique master node in a cluster, since the concept of master
> is defined per HA queue. You can determine the master node for each
> queue with "rabbitmqctl list_queues pid"
> Stopping the master node of a queue sounds like a valid test regardless
> of load-balancer.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the rabbitmq-discuss