[rabbitmq-discuss] Federation Issues: Disconnects and Failures on Queue Bind - 2012-01-20

Tue Jan 24 18:37:08 GMT 2012

Hi Simon,

On a side note; the performance I am seeing between cluster is very 
unexpected.
Publishing to the federated exchange seems to consume about ~4 times the 
CPU as a typical exchange.
The downstream cluster also sees about the same level of cpu usage.

Setup:
Cluster(001) dc001-exchange-u (fanout)
Cluster(002) dc001-exchange-d (fanout) -> dc001-queue-d

Test 1:
Messages: ~3K/sec
Network:  8.5KB/sec
Cluster(001) CPUs: 45% consumed
Cluster(002) CPUs: 47% consumed

Test 2:
Messages: ~11K/sec
Network: 38KB/sec
Cluster(001) CPUs: 97% consumed
Cluster(002) CPUs: 97% consumed

What I was hoping to see that Cluster(002) would be about half the cpu 
consumption as Cluster(001).
Or something like the scenario of publishing to a cluster on host 1 with 
the queue on host 2.
Host 1 would take on the message handling from the publisher and hand it 
off to host 2 which writes to the queue (spreads the load across the 
cluster, but adds a little overhead).

Please let me know if I have missed something, or if you can think of 
any tricks to reduce the cpu usage.
Thanks

On 1/24/12 12:19 PM, DawgTool wrote:
> Hi Simon,
>
> It appears there was a delay on dc001-exchange-u binding to 
> federation: dc001........
>
> Not sure the exact time it took before I could send messages to that 
> exchange (complained about no route), but eventually it started working.
> I'll run some additional test and respond back to this thread.
>
> Thank you.
>
>
> On 1/24/12 6:43 AM, Simon MacMullen wrote:
>> On 23/01/12 19:46, DawgTool wrote:
>>> ok, created the exchange before starting the downstream server.
>>> Federation exchange is created, but it is not bound to the
>>> dc001-exchange-u.
>>> Am I missing something?
>>
>> What do you mean by not bound? The exchanges should be on different 
>> machines so they can't be bound in the usual sense of the word. What 
>> were you expecting to see, and what did you see instead?
>>
>> Cheers, Simon
>>
>
> _______________________________________________
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss