Debugging Production: No Re-deploys, No Downtime, No Risk
summary
Explains challenges of debugging Java applications in production versus local debugging, comparing traditional log-and-reproduce methods and Hot Code Replacement using java.lang.instrument.Instrumentation and their limitations. Describes using the lower-level JVMTI interface and a commercial JVMTI-based product as an alternative for safe, non-disruptive production debugging without stopping the JVM.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:26Debugging production problems
- 6:28Why not use a debugger
- 8:18HCR and GMR intro
- 12:11Hot Code Replacement demo
- 17:36Testing vs production issues
- 20:43Safe debugging with agents
- 27:40Data architecture and streaming
- 30:08Lightrun alternatives and solution
- 31:11Q&A 1
- 35:32Q&A 2
- 40:29Q&A 3
- 44:09Conclusion
related talks
58mDebugging #Java
58mPractical Java Problem Solving: A Hands-On Guide to Debugging
25mDebugging at Scale in Production with kubectl debug
26mUn-breaking production bugs with non-breaking breakpoints
48mIntelliJ Debugger Tips & Tricks
44mDiagnosing Problems in Java Applications
59mAn Introduction to JVM Performance
46mDiagnosing Production Outages — Why Applications Stop Working