Database Reference
In-Depth Information
This is also the best way to handle individual faults needing recovery action,
providing a degree of fault recovery granularity not available elsewhere. The
icon indicates a fault being marked for manual recovery with any one of the
available
Recovery Actions
from the drop-down list:
•
Retry:
This action immediately retries the instance from the point of failure.
•
Abort:
Terminates the entire instance and marks the instance as
Termin-
ated
.
•
Replay:
Replays again the entire scope activity in which the fault occurred.
•
Rethrow:
Rethrows the fault and propagates it to BPEL fault handlers (that
is, catch branches). By default, all exceptions are caught by the fault man-
agement framework unless an explicit rethrow fault policy is provided.
•
Continue:
Ignores the fault and continues processing (marks the faulted
activity as successful).
Managing BPEL component faults
The following screenshot shows a faulted instance of the
OrderProcessing
BPEL component that failed due to a remote exception (endpoint unavailable)
and was marked as
Recoverable
. Once the target system is available, the fault
can be recovered by retrying it. The
Retry Recovery Action
will let the instance
retry and complete its execution. There is a certain limitation with this approach
as the recovery is limited to an individual faulted instance. As such bulk fault re-
coveries cannot be performed at this level.