[rabbitmq-discuss] Name of particular mercurial branch for basic.qos feature?

[Brian Noguchi]

Hi Matthias,

To answer your questions:

Is the number of consumers constant?
No, I am trying to adjust the number of consumers to accommodate increases
in the rate of publishing.

Can you arrange for all consumers to subscribe before messages are being
published?
Before I publish any messages, I make sure the consumers are subscribed.
However, I cannot do this for new consumers I want to add when I am trying
to match the publishing rate.

Are all your consumers operating at approximately the same rate?
Each consumer is identical, so they should all be operating at approximately
the same rate

Is the total rate of consumption greater than the rate of publication? (It
better be, or you will run out of memory eventually)
My goal is to increase the consumers to match the queue length (and
therefore the rate of publication).

What are your suggestions?

Thanks,
--Brian

On Sun, Nov 30, 2008 at 2:48 PM, Matthias Radestock <matthias at lshift.net>wrote:

> Brian,
>
> Brian Noguchi wrote:
>
>> If I don't use the branch, what strategy would you suggest in order to
>> make sure that several consumers of a queue are consuming an equal number of
>> messages without all the messages piling up into the local buffer of a
>> consumer?  Any other alternatives than just doing a Basic.get periodically?
>>
>
> That depends on the use case.
>
> Is the number of consumers constant?
>
> Can you arrange for all consumers to subscribe before messages are being
> published?
>
> Are all your consumers operating at approximately the same rate?
>
> Is the total rate of consumption greater than the rate of publication? (It
> better be, or you will run out of memory eventually)
>
>
> Depending on the answers to the above, the normal round-robin distribution
> mechanisms employed by queues may suffice.
>
>
> Matthias
> PS: please keep the list cc'ed.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20081130/7644bbfc/attachment.htm