[rabbitmq-discuss] is anybody using tx?

[Matthias Radestock]

Al,

On 27/06/11 18:29, Alessandro Ranellucci wrote:
> I have daemons processing external events and writing to a RDMS,
> filesystem and RabbitMQ. If something fails -sort of a distributed
> transaction- I can just rollback everything on RabbitMQ.

Interesting. What guarantees are you getting out of this? Presumably you 
cannot (easily) roll back the file system write. And since the RDBMS 
transaction and RabbitMQ transaction are separate, how do you undo the 
first one when the second one fails?

More broadly, what is the recovery strategy, i.e. what do you after 
you've done the rolling back? Retry again after a while?

> Plus, what Tim Fox said about atomicity.

Are you relying on tx atomicity at the moment? If so, can you elaborate 
on the specific use case? Also is the "only atomic for a single queue" 
restriction of AMQP's tx ok for your app?


Regards,

Matthias.