Solving an Internal Real-World SRE Issue with Eclipse Trace Compass
summary
A multi‑gigabyte production log (about 5 GB/day and three years retained) is loaded into Trace Compass classic, a regex parser is built and shared (takes under 10 minutes), and a custom analysis surfaces a spike of 14 million requests in one hour caused by 50,000 copied jobs polling a help file 300 times/second; the file change from 800 B to 300 KB amplifies traffic to 80% of the service. Parsers, analyses and a small timezone patch are exported and the root cause is fixed.
AI-written summary · about
chapters (10)
- 0:00Start
- 2:29Problem Statement
- 5:00Eclipse Trace Compass
- 7:36A Side Note About Systemic Thinking
- 9:15An Example
- 10:28Second real-world example
- 12:16Create a parser
- 19:51Sharing the recipes
- 26:10Modifying Trace Compass
- 29:06Reminders
related talks
37mEclipse Trace Compass – Past, Present and Future
27mShipping Software with an SRE Mindset
41mScaling Observability: From Chaos to Clarity with Open Source Tools
47mHow I Solved Production Issues with OpenTelemetry (And How You Can Too)
43mThe Dangers of Being Overly-Paranoid
33mDistributed Tracing - All The Warning Signs Were Out There!
21mLeveraging Continuous Production Profiling for Providing Insights into...
40mLog Story Short: Chopping Through Forests Of Data