[rabbitmq-discuss] RabbitMQ cluster with potentially millions of connections

[Matthias Radestock]

Roman,

On 26/01/13 11:58, Romanas S. wrote:
> Each connections has it's own exchange and a temporary direct queue,

The meta data for exchanges and queues is held in RAM *on all nodes* in 
a cluster. Shouldn't be more than a few hundred bytes, but still. It's 
quite unusual to create one exchange per connection. You may want to try 
eliminating that.

> However, after scaling to that many connections, the cluster seems to
> grind to a halt and eventually nodes become unresponsive. The
> management API takes upwards of 5 (FIVE!) minutes to retrieve a
> server summary report.

Make sure you are running the latest version of RabbitMQ, i.e. 3.0.1. 
Recent release contain some significant improvements to behaviour under 
load.

You may also want to try disabling the management plug-in.

Regards,

Matthias.