101 ways to break your RabbitMQ
summary
Migrating on‑prem monoliths to cloud microservices uses RabbitMQ with AMQP and MQTT and exposes specific failure modes: default Linux 1024 file descriptors breaks large deployments and triggers an Erlang VM crash until fixed; clusters do not replicate queues so choose quorum queues or mirrored HA queues and avoid even‑node clusters; misused connections/channels spawn thousands of channels; prefetch with buggy framework threads stalls consumers; long queues kill throughput so enforce TTLs and DLQs; enable rabbitmq_prometheus plugin and Grafana and run the management UI on multiple nodes; a Python tool rebuilds the topology graph.
AI-written summary · about
chapters (10)
- 0:0002:16 - Introduction
- 2:1713:30 - Let's brake RabbitMQ
- 13:3113:51 - RabbitMQ Usage
- 13:5216:40 - Security
- 16:4117:46 - new connections love TCP packets
- 17:4718:42 - Connections / Channels
- 18:4323:22 - Prefetch
- 23:2325:27 - Observe
- 25:2826:19 - Conclusion
- 26:2039:23 - QnA
related talks
18mHow to Break RabbitMQ
21mWhy does RabbitMQ stop and what can be done about it
22mLessons Learned from Scaling RabbitMQ in a High-Growth (...)
34mUpdate RabbitMQ now!
24mMigration from On-Prem Messaging System to The Cloud
48mA Beginner's Guide to Inter-Service Messaging
40mScaling RabbitMQ
35mHow to Safely and Reliably (...) While Maintaining 100% Uptime