Question

A developer is building an order-processing microservice that consumes messages from an Amazon SQS standard queue. During testing, the team notices that some orders occasionally get processed more than once, resulting in duplicate database inserts. The processing logic itself is correct. What is the MOST appropriate way to address this behavior?

Accepted Answer

Design the consumer to be idempotent, for example by using a unique order ID as a conditional write key so duplicate deliveries have no additional effect.

Answer

Increase the queue's visibility timeout to a very large value to guarantee each message is delivered only once.

Answer

Disable long polling on the queue so messages are received individually and cannot be duplicated.

Answer

Enable a dead-letter queue so duplicate messages are automatically routed away from the main queue.