We have a JBoss 4.2.3 cluster successfully set up and failing over properly, except for one thing: Quartz Cron Triggers don't resume on the node we fail over to.
We have to manually resume the Job through through our SLSB to get it going again.
We upgraded to Quartz 1.6.5 and Quartz is running as a JBoss service. We have set the RequestsRecovery flag to true on the Quartz Job. In the Quartz database, the QRTZ_PAUSED_TRIGGER_GRPS is empty when we failover which surprised us because resuming the Job manually works.
As a possible workaround, is there any way to detect when a failover has happened so that we can programatically resume the Quartz Job?
To solve this issue you need to add a dependency on the barrier controller like this:
@Depends({"jboss.ha:service=HASingletonDeployer,type=Barrier"})
and resume the Job somewhere in the start() lifecycle method of an MDB.