How to handle recoverable or transient or connectivity errors in SAP Integration flows

[pavanKollaSAP]

In SAP Cloud Platform Integration (CPI), custom retry mechanisms handle transient errors (e.g., network issues, HTTP 503) while avoiding retries for persistent errors (e.g., mapping issues). Below is a concise overview of retry approaches and configurations.

Retry Mechanisms
1. JMS Queues
- Stores failed messages in queues for retries.
- Setup: Enable JMS (Enterprise Edition). Use JMS receiver to store messages, JMS sender to poll (e.g., every 10 mins). Set max 5 retries, exponential backoff. Move to dead-letter queue or notify after max retries.
- Use Case: Retry HTTP 503 errors.
- Pros: Reliable, visible in Queue Monitor.
- Cons: 10 GB limit, polling may pause.
- Example: Failed message retried 5 times, then emailed.

2. Data Store
- Stores messages in CPI database (32 GB) for retries.
- Setup: Passthrough iFlow to business logic iFlow. On failure, store in Data Store (“DS_RetryAllError”). Schedule retry iFlow to poll (e.g., 5 mins). Retry transient errors, move persistent to another Data Store.
- Use Case: High-volume, limited JMS.
- Pros: No JMS dependency, manual reprocessing.
- Cons: 32 GB limit, no bulk retrieval.
- Example: Connection error retried daily, mapping errors to “DS_ManualReview”.

3. HTTP Adapter Retry
- Retries HTTP errors (400–599) since CPI 7.18.
- Setup: Enable “Retry” in HTTP adapter. Set 1–3 retries, 5–60 sec intervals. Log retries.
- Use Case: Server unavailability.
- Pros: Simple, no custom logic.
- Cons: In-memory, max 3 retries.
- Example: HTTP 503 retried 3 times, fails to Exception Subprocess.

4. SuccessFactors OData V2 Retry
- Retries HTTP 429, 502, 504 errors.
- Setup: Enable “Retry on Failure”. Fixed 5 retries, 3-min intervals.
- Use Case: SuccessFactors network issues.
- Pros: Built-in.
- Cons: Fixed settings.
- Example: 502 error retried 5 times.

5. TPM Retry
- JMS-based retries for B2B since TPM 2.2.0.
- Setup: Set max retries (3–25) in Configuration Manager. Manual retry in B2B Monitor (v2.3.7).
- Use Case: B2B/EDI delivery.
- Pros: Native, manual option.
- Cons: Needs TPM 2.2.0+.
- Example: B2B message retried 3 times, manually retried.

6. XI Adapter Custom Retry
- Custom retry logic in Exception Subprocess.
- Setup: Use Local Integration Process, track retries (e.g., SAPJMSRetries). Router for retry (max 6) or alternative action.
- Use Case: Complex scenarios.
- Pros: Flexible.
- Cons: Complex setup.
- Example: Notify after 3 retries, redirect after 6.

Performance
- JMS: Locks with high volumes; monitor queues.
- Data Store: 32 GB limit; avoid large payloads.
- HTTP Retry: Lightweight, non-persistent.
- Latency: Long intervals add delays.