Navigating the Distributed Systems Execution Maze With OpenTracing
summary
Kasten adds OpenTracing and Jaeger to a Go Kubernetes app (image-catalog + image-store) and instruments middleware, a custom HTTP transport, and a few MongoDB call sites by adding begin/finish span lines and tags. Deployed with Helm, Jaeger surfaces chained traces that reveal a 14s catalog call and a 30s store read that logs and metrics miss. Notes: context propagation is required, some client APIs lack context so manual spans are needed, and sampling or conditional tracing limits overhead.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 0:44Why OpenTracing
- 2:01Kubernetes Ecosystem
- 2:26Demo
- 7:20What is tracing
- 8:58OpenTracing
- 10:16OpenTracing Go
- 11:33Adding Tracing
- 15:29Can We Always Use The Trick
- 19:20Should We Use Tracing
- 26:30Conditional Tracing
- 27:30Programming Languages
- 28:30Example
related talks
34mReduce Complexity with Distributing Tracing
51mPractical introduction to OpenTelemetry tracing
47mI'm Tracing You!
34mWould You Like Some Tracing With Your Monitoring?
31mDistributed Tracing Integration with OpenTelemetry and Knative
38mPractical introduction to OpenTelemetry tracing
37mOpenTelemetry & Distributed Tracing: A Jaeger v2 Deep Dive
42mObservable Applications Using OpenTelemetry