idempotency-handling

Idempotency-key handling prevents duplicate processing by intercepting requests with a unique key, caching responses in Redis, and checking the cache before executing mutations. If the key exists, the cached response is returned; otherwise, the operation executes and the response is stored for future retries.

Yes. Idempotency keys are designed for critical mutations like payments and webhook handling where duplicate processing causes issues. The pattern uses Redis caching with a 24-hour TTL and a durable database table to ensure deterministic responses and prevent race conditions across retries.

The workflow reads an idempotency-key from request headers, checks Redis for a prior response, and returns the cached result if found. If not cached, the operation executes, the response stores in Redis, and state records in a relational table with atomic processing to prevent race conditions.

Yes. Idempotency keys enable safe retries by recording operation state in both Redis and a database table. Status updates upon success or failure, so repeated requests with the same key return the cached outcome without re-executing the mutation.

Redis caching uses a 24-hour TTL by default for idempotency keys. The Skill provides clear TTL guidance and cleanup strategies; the durable database-backed key table persists state independently, ensuring atomicity even after cache expiration.

Redis provides fast response caching with a 24-hour TTL for frequent retries, while the relational database table durably records operation state and keys. Together they ensure both performance and atomicity, preventing duplicates across distributed system retries.