Debuggin a Large Node Cluster
summary
Debugging a large Node cluster is hard because bugs arise from accumulated distributed state not reproducible in dev; the concrete remedy shown is running a live reppel over a Unix socket (use socat) to eval JavaScript into running processes, expose state on an object (example: MJR) and flip handlers or wrap module exports to trace specific requests without restarting; also use aggregated logs with node IDs and timestamps and tooling like node reddis and node inspector to narrow faults in Voxer scale.
AI-written summary · about
chapters (8)
- 0:00Intro
- 2:20What about other CPUs
- 5:00Debugging in Node
- 6:00Starting out
- 6:40Two different places
- 7:25Multiple states
- 13:55Mirroring
- 16:40Testing
related talks
57mTroubleshooting Node.js Applications Under the Hood
21mFoundations of Debugging Node.js
25mNode js applications diagnostics under the hood
50mPost-Mortem Debugging in Node.js
28mPost mortem NodeJS debugging
50mDebugging Node.js in Production
23mThrow Me a Lifebuoy: Debugging Node.js in Production with Diagnostic Reports
31mConsole.loggers Anonymous: A Debugging Roadmap from Local to Production