A consumer application reads from an Amazon SQS queue. Some messages have an invalid format and always fail processing. They are retried repeatedly and consume consumer capacity. What is the best way to prevent these "poison pill" messages from blocking normal processing?

Enable long polling and increase the maximum message retention to 30 days.

Long polling affects how long the consumer waits for messages during ReceiveMessage, not how many times a failing message is retried. Increasing retention only keeps problematic messages available longer; it does not automatically isolate them.

Switch the queue to FIFO and disable retries in the consumer code.

FIFO affects ordering, not the automated quarantine behavior for repeated failures. Disabling retries in code may reduce retries temporarily, but it does not provide the built-in maxReceiveCount-based routing to a DLQ that systematically isolates poison pills.

Delete the main queue and recreate it after every failure.

Recreating the queue is disruptive and does not address the underlying processing issue. It also risks data loss, operational complexity, and losing the ability to analyze problematic messages.

What does this SAA-C03 question test?

Static NAT maps one inside address to one outside address.

What is the correct answer to this question?

The correct answer is: Configure a dead-letter queue (DLQ) with a redrive policy and a maxReceiveCount. — Use an SQS dead-letter queue (DLQ) with a redrive policy. Set a maxReceiveCount value so that when a message fails processing repeatedly (each failure results in the message becoming visible again after the visibility timeout), SQS eventually routes that message to the DLQ instead of letting it keep circulating in the main queue. This prevents poison-pill messages from consuming consumer capacity, while still preserving the failed messages for later inspection, correction, and possible replay. A is wrong because long polling and retention settings do not stop repeated failed processing cycles. C is wrong because FIFO and disabling retries in code do not provide SQS’s managed maxReceiveCount quarantine mechanism. D is wrong because deleting and recreating the queue is not a resilience pattern and can interrupt processing and lose diagnostic information.

What should I do if I get this SAA-C03 question wrong?

Then try more questions from the same exam bank and focus on understanding why the wrong options are tempting.