A Step Functions task is retried after a network timeout.
If the first call succeeded, do not create the same date partition twice.
Which configuration is most appropriate?
Use an execution ID and logical date as an idempotency key with conditional writes or MERGE.
Detailed explanation
The output can detect the same retry and safely reuse or update the existing result.
The output can detect the same retry and safely reuse or update the existing result.
Removing audit records does not remove duplicate output.
Removing audit records does not remove duplicate output.
Transient failures cannot recover, and a successful-but-unknown first call remains unresolved.
Transient failures cannot recover, and a successful-but-unknown first call remains unresolved.
Each retry creates another dataset and can cause downstream double counting.
Each retry creates another dataset and can cause downstream double counting.
Check official documentation
Compare the following points with the official documentation.
Step Functions再試行、S3条件付き書き込み、Redshift MERGEの公式説明を確認する。
What to confirm
再試行・タイムアウト・冪等性の関係を説明できる。
Key points
- Idempotency
- Execution ID
- Conditional write
- MERGE
Notes
- Environment: AWS公式DEA-C01試験ガイドと各サービスの公式ドキュメントの確認
- Confirm current AWS service behavior in the official DEA-C01 documentation.
Foundation review
At-least-once execution
A timeout can leave the first result unknown, so the task may run again.
Output uniqueness
Logical date, input version, and run identity define whether an output already exists.