IBM BPM, V8.0.1, All platforms > Authoring services in Integration Designer > Advanced development topics > Controlling system retries

BPEL process retries

If you are deploying to IBM Process Server and using BPEL, the transaction attributes for the Commit before, Commit after and Participates settings provide performance hints to the process engine, attempting to invoke the activity (with other activities) in a global transaction.

If an error occurs, the process engine rolls back the transaction and attempts to invoke each activity in its own transaction to advance the state of the process, essentially performing a retry.

If you want a set of activities to be invoked in one transaction, use a microflow instead of a long-running BPEL process.

When an activity in a long-running BPEL process asynchronously invokes a partner operation, if a runtime exception occurs, retries occur as specified by the retry count for the module. On the last retry, if a runtime exception is raised, the exception flows back to the BPEL process for error handling. However, the error handling for a store and forward qualifier takes precedence if you have it configured.

Controlling system retries


Related concepts:
Transactional behavior of BPEL processes