rabbitmq-expert_skill

Overview

This skill is an expert RabbitMQ administrator and developer focused on designing, operating, and optimizing message broker systems for production. I help design exchange patterns, tune throughput, implement high-availability clusters, and build reliable failure-handling and observability. The goal is predictable message delivery, secure deployments, and measurable performance under load.

How this skill works

I inspect architecture choices (exchanges, queues, bindings), broker configuration (memory, disk, policies), and client patterns (prefetch, confirms, batching). I evaluate HA options (quorum vs mirrored queues), DLX and retry strategies, and provide actionable tuning for throughput and latency. I also verify security posture (TLS, authz) and observability (metrics, alerts, management API).

When to use it

• Designing a new message-driven system or refactoring existing queues
• Troubleshooting message loss, ack/nack behavior, or DLX routing
• Optimizing throughput: prefetch, batching, confirms, and pooling
• Planning or operating RabbitMQ clusters and HA strategies
• Hardening broker security and auditability for production

Best practices

• Follow TDD for message flows: unit test consumers and integration test confirms/DLX
• Tune prefetch by consumer latency: high for fast consumers, low for slow ones
• Enable durability + publisher confirms to avoid silent message loss
• Use DLX, TTL, and retries to separate permanent and transient failures
• Prefer quorum queues for consistency, lazy queues for large disk-backed backlogs
• Instrument queues, consumer lag, and node health; alert on memory/disk thresholds

Example use cases

• Implement a pub/sub architecture with topic exchanges and consumer groups
• Migrate classic mirrored queues to quorum queues for safer failover
• Design DLX-based retry choreography and dead-letter monitoring
• Optimize a high-throughput publisher using batching and async confirms
• Set up connection pooling and consumer prefetch tuning to reduce resource churn