RabbitMQ Messaging

Instructions

Apply these patterns for reliable, scalable message-driven systems with RabbitMQ.

Core concepts:

• Exchanges route messages to queues via bindings and routing keys
• Exchange types: direct (exact key match), topic (wildcard patterns *.error, audit.#), fanout (broadcast to all bound queues), headers (match on message headers)
• Queues store messages until consumed; bind queues to exchanges with routing keys
• Use durable queues and persistent messages (deliveryMode: 2) for data that must survive broker restarts

Consumer design:

• Manual acknowledgment (autoAck: false) — ack only after successful processing to prevent message loss
• Set prefetch count (basic.qos) to control how many unacked messages a consumer holds; start with 1 for ordered processing, increase for throughput
• Implement idempotent consumers: messages may be redelivered on failure, so design handlers to tolerate duplicates
• Use consumer tags to identify and cancel specific consumers gracefully

Dead-letter exchanges (DLX):

• Configure x-dead-letter-exchange and x-dead-letter-routing-key on the source queue
• Messages route to the DLX on rejection (basic.nack with requeue: false), TTL expiry, or queue length overflow
• Monitor dead-letter queues and build alerting for accumulation — these represent processing failures

Publisher confirms:

• Enable publisher confirms (confirm.select) for reliable publishing
• Wait for broker ack before considering a message sent; implement retry with backoff on nack
• Combine with mandatory flag to detect unroutable messages

High availability:

• Use quorum queues (x-queue-type: quorum) for replicated, Raft-consensus-based durability
• Quorum queues replace classic mirrored queues — they handle network partitions and node failures safely
• Deploy clusters with an odd number of nodes (3 or 5) for proper Raft leader election

Monitoring:

• Enable the rabbitmq_management plugin for the web UI and HTTP API on port 15672
• Track queue depth, consumer count, message rates, and unacked message count
• Alert on queue depth growth without corresponding consumer activity

Docker deployment:

• Use the rabbitmq:3-management image for built-in management UI
• Mount a volume for /var/lib/rabbitmq to persist data across container restarts
• Set RABBITMQ_DEFAULT_USER and RABBITMQ_DEFAULT_PASS environment variables; rotate credentials in production