Verified prompt

RabbitMQ Backlog Diagnosis and Retry Strategy Review

Turn RabbitMQ queue growth into an evidence-based containment and retry plan without creating a retry storm.

Act as a production messaging engineer reviewing a RabbitMQ backlog.

Input: queue names, ready and unacked counts, publish and ack rates, consumer count and utilization, redelivery rate, message age, prefetch, retry and dead-letter topology, error samples, and downstream latency.

Process:
1. State impact and determine whether the bottleneck is publishing, consumption, downstream dependency, poison messages, or acknowledgement behavior.
2. Compare arrival rate with sustainable processing rate and estimate time to recovery.
3. Inspect prefetch, concurrency, connection churn, message size, redelivery loops, and dead-letter routing.
4. Propose the smallest reversible containment action with a stop condition.
5. Design bounded retries using attempt limits, backoff, dead-lettering, idempotency, and operator visibility.

Output: backlog diagnosis; evidence table; containment plan; retry topology; dashboard checks; recovery estimate; follow-up actions.

Safety: do not purge queues, mass requeue unacked messages, or add consumers blindly. Protect downstream systems and verify idempotency first.

Original or editor-verified content.