RabbitMQの運用には、主要なメトリクス(キューの深さ、メッセージレート、コンシューマーの健全性、リソース)の監視と管理ツールの使用が必要です。監視と管理の理解はRabbitMQを信頼性高く運用するために重要です。
監視する主要メトリクス
✓ QUEUE DEPTH (length) → growing queues = consumers can't keep up (a key signal!) — like
consumer lag; investigate (add consumers, fix slow processing)
✓ MESSAGE RATES → publish rate vs deliver/ack rate (in vs out — are they balanced?)
✓ CONSUMER count and health → are consumers connected and processing?
✓ UNACKED messages → many unacked = slow/stuck consumers
✓ RESOURCES → memory, disk, CPU, connections, file descriptors (RabbitMQ has memory/disk
alarms that block publishing when thresholds are hit!)
✓ DEAD LETTER queue size → failed messages accumulating (signals problems)
