[rabbitmq-discuss] Connection Pooling Options to solve FD Consumption on RabbitMQ.

Thu Jul 21 06:14:49 BST 2011

Yes . currently I am creating a connection on each Thrift (similar to HTTP
request).

I am assuming each channel is opening a TCP connection - Is it correct ?
Also should I create a channel from a globale connections instead of
creating connection each time?

-Abhishek Kona

On Thu, Jul 21, 2011 at 1:12 AM, Michael Klishin <
michael.s.klishin at gmail.com> wrote:

> 2011/7/20 Abhishek K <abhishek.kona at gmail.com>
>
>> On each request, I create a new Connection and Channel. During high load
>> times, this is causing the Rabbit to go unresponsive (due to a large no. of
>> open file Descriptor's).
>>
>> This would be solved if I could connection pool on the client side, what
>> are the strategies and API's suggested for that.
>>
>
> Why do you need a new connection on each (HTTP?) request? This is what
> channels are for, if you want to absolutely separate data flowing between on
> each individual request.
> AMQP was designed with the idea of keeping # of TCP connections as low as #
> of applications.
>
> --
> MK
>
> http://github.com/michaelklishin
> http://twitter.com/michaelklishin
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20110721/06313ee9/attachment.htm>