Yes, both stateless and clustered stateful components could benefit from this behavior...
Cheers, Guillaume Nodet On 4/11/06, Jamie McCrindle <[EMAIL PROTECTED]> wrote: > Jira has been created: > > http://issues.apache.org/activemq/browse/SM-395 > > > > If it's a stateful process engine that sent out the InOut exchange and > > > it's clustered to multiple nodes, any one of them should be able to > > > accept the returning InOut exchange. > > > > I guess you mean stateless ? > > In the case of BpmScript, clustered statefulness. I assume the other > clustered BPE's will find InOut failover useful too. > > Regards, > jamie. > > On 4/11/06, Guillaume Nodet <[EMAIL PROTECTED]> wrote: > > On 4/11/06, Jamie McCrindle <[EMAIL PROTECTED]> wrote: > > > Hi, > > > > > > I've been doing some testing in a cluster and it turns out (quite > > > reasonably) that an InOut exchange can't reply if it's source > > > component is not available even if there are multiple instances of > > > that component in a cluster. > > > > Yeah, right. The main reason is that the component can be statefull and / > > or > > use a sendSync. > > > > > > > > If it's a stateful process engine that sent out the InOut exchange and > > > it's clustered to multiple nodes, any one of them should be able to > > > accept the returning InOut exchange. > > > > I guess you mean stateless ? > > > > > > > > Is there a way that servicemix could allow this kind of failover? > > > > This is not supported at the moment, but please raise a JIRA for this issue. > > > > > > > > Regards, > > > Jamie. > > > > > >
