On Thu, Jun 25, 2009 at 6:11 AM, Robert Godfrey<rob.j.godf...@gmail.com> wrote: > 2009/6/25 Martin Ritchie <ritch...@apache.org> > >> Rajith, >> >> I think we need to have a think about what we want from Failover. It >> is so poorly tested just now having it disabled by default is probably >> a very good idea. >> >> > I am 100% in agreement that we should not have failover (retry) switched on > by default. the behaviour is not what you would expect from a JMS client.
Totally agreed. There were some nasty issues with some end users due to the java client retrying by default and also retrying more than expected. So by experience I also agree that we shouldn't retry by default. I also agree that we need to differentiate between failover and retry. > If people understand the retry mechanism and its limitations it is fine for > them to switch it on (if it works). We should also think about how we > differentiate between retries/failover where no state is lost (which is > possible in 0-10) and retries/failover where state is lost - they are most > definitely not the same thing, and their presentation to the client > application should be different (i.e. where no state is lost the failover > can be invisible to the client application. If state is potentially lost > then the mechanism by which the client is informed of this should be made > obvious. > > -- Rob > -- Regards, Rajith Attapattu Red Hat http://rajith.2rlabs.com/ --------------------------------------------------------------------- Apache Qpid - AMQP Messaging Implementation Project: http://qpid.apache.org Use/Interact: mailto:dev-subscr...@qpid.apache.org