<div dir="ltr">Basically, we delay the ack by adding a timeout/callback to the event loop after processing a message from the 'silver' queue. The delay reflects the desired rate. Crude but it works. We use pika w the libev adapter I wrote.<div>
<br></div><div>BTW re node restart: we do restart our Cassandra nodes if they fail. They discover and automatically sync up w the other Cassandra nodes. This has happened in production. We also do rolling restarts of the production Cassandra cluster to upgrade to new versions.</div>
</div><div class="gmail_extra"><br><br><div class="gmail_quote">On Wed, Mar 5, 2014 at 10:54 AM, Ben Hood <span dir="ltr"><<a href="mailto:0x6e6562@gmail.com" target="_blank">0x6e6562@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="">On Wed, Mar 5, 2014 at 12:51 PM, Laing, Michael<br>
<<a href="mailto:michael.laing@nytimes.com">michael.laing@nytimes.com</a>> wrote:<br>
> Some of our fabrik apps apply flow control. For example, we apply flow<br>
> control to the ingress of 'silver' service messages so that 'gold' service<br>
> takes precedence.<br>
<br>
</div>How do you apply flow control in your infrastructure?<br>
<div class="HOEnZb"><div class="h5">_______________________________________________<br>
rabbitmq-discuss mailing list<br>
<a href="mailto:rabbitmq-discuss@lists.rabbitmq.com">rabbitmq-discuss@lists.rabbitmq.com</a><br>
<a href="https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss" target="_blank">https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss</a><br>
</div></div></blockquote></div><br></div>