Exception Handling, Logging and Retry Patterns in SAP Cloud Integration

Keep this: job aid

Cloud Integration error-handling cheat sheet

• Classify first: transient (retry), permanent technical (alert integration team), functional (route to the business). Never retry functional errors.
• Exception subprocess: Error Start, then End Message = Completed (set SAP_MessageProcessingLogCustomStatus), Error End = Failed, Escalation End = Escalated.
• Read errors with ${exception.message}, ${exception.stacktrace}, CamelExceptionCaught (script) and, with Throw Exception on Failure off, ${header.CamelHttpResponseCode}.
• Handle local process errors inside the local process; share one handler across flows via ProcessDirect.
• JMS sender: Retry Interval (default 1 min), Exponential Backoff, Maximum Retry Interval (default 60). No max-retries field: use SAPJMSRetries (non-exclusive only).
• The Dead-Letter Queue option blocks node-crashing messages (release in Message Locks); it is not a retries-exhausted queue. Park on purpose instead.
• Idempotent Process Call keyed on a business ID (PO number); IDs kept 90 days; still make the receiver reject duplicates.
• Log levels: Info in production; Trace only briefly (10 min, content 1 h); None hides messages from Cloud ALM; Error hides successful retries.
• Search by SAP_ApplicationID, SAP_MessageType, SAP_Sender/Receiver, custom headers. No personal data in attachments or custom headers.
• AIF forwarding: Error_Propagation_Request_In (push) or transfer job (pull); IDoc and web service interfaces only; status A; cancel only.