K8s incidents: make logs work for you in real-world troubleshooting
recording made private on YouTubehttps://youtu.be/kVoJwOUWQuY
summary
Logs alone don’t find root cause; adding concrete context does. A real incident traces a frontend gRPC error to the card service which has lost Redis connectivity, and a live topology view plus metrics and deployment history narrows root cause to a recent config/image rollout. Uses OpenTelemetry-style traces, automated topology/auto-instrumentation and metrics correlation to jump from noisy logs to the source; the approach resolves ~90% of cases quickly and exposes blind spots in partial instrumentation.
AI-written summary · about
related talks
15mBeyond Root Cause Analysis for K8s: What Your Logs Should Be Telling You
14mLeveraging Distributed Tracing in Kubernetes & Containerized Environments
30mStreamlined Troubleshooting in Kubernetes
34mReduce Complexity with Distributing Tracing
32mAutomatic logs, traces and metrics in K8s using OpenTelemetry & SigNoz
35mA story about debugging of one Go application
42mDistributed Monitoring: How to understand the Chaos
18mK8s Troubleshooting Demystified: Five Best Practices