Seeing the Invisible Observability with
summary
Live demo shows a k3d cluster running Linkerd with Linkerd Viz and the service profile CRD to locate and fix failures. The Emoji Vote app yields repeated grpc status 2 errors for donut votes; profiles are generated from protobufs or by observing 10s of traffic. Applying retries (marking a HEAD route retryable) boosts the effective success rate from ~60% toward 100% so users are unaffected while developers investigate the root cause.
AI-written summary · about
chapters (7)
- 0:00Introduction to Linker D
- 2:33Cluster environment setup
- 3:57Debugging the emoji app
- 9:18Configuring service profiles
- 11:36Dashboard analysis
- 13:41Mitigating errors with retries
- 18:50Conclusion and resources
related talks
29mDebugging an app with a service mesh
05mLightning Talk: Locating and Debugging Failures with Linkerd...
29mObservability and Debugging Cloud-Native Apps using Linkerd & Conduit
36mEmissary + Linkerd Resilience Patterns: Rate Limits, Retries & Timeouts
01h 10mService Mesh Introduction
37mLinkerd End User Panel: Case Studies from Production
21mTap Tap, Debugging an App with your Service Mesh
36mWhat's going on in my cluster?