[rabbitmq-discuss] Ingress does not work for me and the RMQ goes down

Cermak, Marek Marek.Cermak at Honeywell.com
Fri Apr 5 13:48:42 BST 2013


Hi,
well does this flow_control mechanism help against the broker crash? Is it handled by the broker using socket blocking you mentioned in case of TCP connection?

thanks
marek

> -----Original Message-----
> From: Simon MacMullen [mailto:simon at rabbitmq.com]
> Sent: 05 April, 2013 14:37
> To: Cermak, Marek
> Cc: Discussions about RabbitMQ
> Subject: Re: [rabbitmq-discuss] Ingress does not work for me and the RMQ goes
> down
> 
> On 05/04/13 13:18, Cermak, Marek wrote:
> > Hi Simon,
> > yes, I use direct for both consumer and publisher. They run in one node,
> RMQ is in another.
> 
> In that case you'll need to participate in the internal flow control scheme
> (which is not related to register_flow_handler) on the publisher side. This
> is slightly more fiddly:
> 
> * Use amqp_channel:cast_flow/3 instead of amqp_channel:cast/3 to publish.
> * Be prepared to receive {bump_credit, Msg}, and pass that Msg to
> credit_flow:handle_bump_msg(Msg).
> * Use credit_flow:blocked/0 to determine if you are allowed to publish.
> 
> That's if you have a single (class of) processes which publish. If you have a
> pipeline of processes through which messages pass then you might want to
> extend the same flow control mechanism through the pipeline. See for example
> the pipeline
> 
> rabbit_stomp_reader -> rabbit_stomp_processor -> amqp_channel.
> 
> Cheers, Simon
> 
> --
> Simon MacMullen
> RabbitMQ, VMware


More information about the rabbitmq-discuss mailing list