高可用性在 RabbitMQ 中通过集群(多个节点)和复制队列(镜像或仲裁队列)实现,使系统能够在节点故障时幸存并且不会丢失消息。理解 HA 对于可靠的生产 RabbitMQ 很重要。
Clustering
RabbitMQ CLUSTER → multiple RabbitMQ nodes working together as one logical broker:
→ nodes share metadata (queues, exchanges, bindings, users)
→ distributes connections and load across nodes
⚠️ by default, a queue lives on ONE node → if that node fails, the queue is unavailable
(clustering alone doesn't make queues highly available → need replication)
