[rabbitmq-discuss] Cluster and memory usage

[Matthias Radestock]

On 11/11/11 14:26, Cheremukhin Andrey wrote:
> I set up cluster within two nodes - rabbitmq 2.6.1.
> And I am seeing
>
> node1$
> PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
> 31196 rabbitmq 51 50 0 2990M 2170M ucond 10 1:52 170.90% beam.smp
>
> node2$
> 22915 rabbitmq 51 103 0 442M 402M CPU11 11 7:49 495.65% beam.smp
>
> All queues live on on node2.

What is node1 doing? Are publishers connecting to it? If so the memory 
might simply be taken up by message backlogs, i.e. the producers are 
producing messages at a faster rate than rabbit can sustain. Eventually 
this would result in a memory alarm and producer connections getting 
blocked until the backlog has been reduced.

Regards,

Matthias.