Production Is on Fire? No Stress, I Have strace (and Friends)
summary
This talk demonstrates using low-level OS tools like /proc inspection and strace (on WSL) to diagnose production incidents when observability is lacking. It shows live examples tracing processes for Java, Python, etc., to identify issues quickly, gather useful indicators, and improve later telemetry without deploying heavy monitoring stacks.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:12Intervention mindset
- 2:17Finding vital info
- 6:31Mastering /proc
- 10:23Analyzing load average
- 12:31Detecting fake proxy
- 19:26Why strace dives to kernel
- 21:54Tracing Java live
- 29:30Filtering syscalls
- 33:51Decrypting file and socket calls
- 37:40strace risks and best practices
- 40:49Tracing in Docker
- 43:19Why /proc and kernel help
- 44:20Q&A
related talks
30mA Well-Maintained Production Application
40mProduction Detectives: Solve the Case Before the Crash
28mKnocking On A Process’s Door: Introspecting Live Python Processes
53mSource Instrumentation for Monitoring C++ in Production
32mListen to your Engine
31mTracing, Profiling & Debugging in Production (eBPF)
51mProduction Profiling: What, Why and How
43mSelf-Aware Applications: Automatic Production Monitoring