Figure: Recovering the Stardust Runtime Environment
If an EJB server crashes, it may leave non-interactive activities in an uncompleted state. Their branch in an activity-transition-network "hangs". Stardust users can always restart interactive activities.
Also, it is possible that non-interactive activities, as well as some interactive ones, fail to be completed due to unexpected errors, either because of programming errors or invalid state of the application, in which case the activities are marked as interrupted.
You can start a recovery from the Administration perspective launch panel in the Administrative Actions section. To "reanimate" problematic activities invoke the option Recovery.
Figure: Recovering the Stardust Runtime Environment
A dialog opens to confirm the start of the recovery process.
Figure: Confirm the Recovery Process
The Stardust runtime environment does not use any other persistent and/or transactional resource beside the audit trail database. The only backup task for the runtime environment is the backup of the audit trail database. If Stardust is used to implicitly access other transactional resources than the audit trail database (e.g. via Entity Beans), Stardust attempts to use JTA/JTS (Java Transaction API/Java Transaction Services) to synchronize the access. Recovery is then performed according to the current transaction monitor.