Skip to content
🎉 Launch sale:50% off everything over $22 — automatically applied at checkout· ends Aug 2Shop the sale →
DevOps AI ToolKit
Newsletter
All troubleshooting hubs Troubleshooting hub

RabbitMQ Cluster, Queue & Resource Errors

The failures that degrade the broker itself rather than a single connection: a Mnesia network partition (split brain), a quorum queue that cannot elect a leader, a node that will not rejoin the cluster, memory or disk high-watermark alarms that block publishers, flow control kicking in, messages that are unroutable or dead-lettered, and streams or consumers that stall. Each guide opens with the direct cause and the `rabbitmqctl`, `rabbitmq-diagnostics`, and `rabbitmq-queues` commands to confirm it. Start with your symptom below, or paste the broker log into the assistant.

Fix your error now

Paste the error or logs and let the Incident Assistant diagnose it, or run a guided triage prompt with AI — no copy-paste.

Diagnose with the Incident Assistant →

Every guide in this cluster

64 guides

RabbitMQ AI prompts

Copy-paste, production-safe prompts for this stack.

RabbitMQ command center

Top errors, validators, and runbooks for the whole stack.

Frequently asked questions

How do I recover from a RabbitMQ Mnesia network partition (split brain)?
Nodes lost contact and each side kept running, so state diverged. Decide which partition is authoritative, then restart the losing nodes so they re-sync — never both. Check rabbitmq-diagnostics cluster_status and review your cluster_partition_handling mode. See Mnesia network partition.
Why is my quorum queue stuck with "cannot reach majority"?
A quorum queue needs a majority of its members online to elect a leader; with too many member nodes down it goes read-only. Bring the down nodes back or grow the cluster to restore majority — do not delete the queue. See quorum lost majority.
Why are my publishers blocked with a vm_memory_high_watermark alarm?
RabbitMQ raised a memory (or disk) alarm and is applying back-pressure by blocking publishing connections until usage drops. Free memory/disk, raise the watermark deliberately, or drain the backlog — blocking is protection, not a bug. See memory high-watermark blocked.