Failover after tx has reached a server

[brian.stansberry]

Galder Zamarreño just pointed out to me that ClusterChooserInterceptor does not include any of the logic that the old detached invoker proxies had that prevents a failover once a tx has reached a target server. Was this just an oversight?

[galder.zamarreno]

Could someone from the EJB3 team please confirm or deny my suspicion? This is important in the scope of http://jira.jboss.com/jira/browse/JBAS-4455 and more specifically, its implementation for EJB3.

Cheers :)

[brian.stansberry]

Give it another 24 hours. If no one gives you a reason, assume it's a bug. :)

[galder.zamarreno]

Even more important, if the suspicion is confirmed, it could be considered an important bug because you'd be allowing failovers when you shouldn't

[galder.zamarreno]

Give it another 24 hours. If no one gives you a reason, assume it's a bug. :)

k :)