JaegerとTempoによる分散トレーシングを実装して、マイクロサービス間のリクエストを追跡し、パフォーマンスボトルネックを特定します。マイクロサービスのデバッグ、リクエストフローの分析、分散システムの可観測性実装時に使用します。
The distributed-tracing skill implements end-to-end request tracking using Jaeger and Tempo for microservices. It enables the visibility of request flows between services, making it easier to identify performance bottlenecks, debug latency issues, and understand system dependencies. This skill is particularly useful when debugging complex distributed systems, improving the observability of microservices, and enhancing troubleshooting capabilities for distributed architectures.
You can deploy Jaeger using Kubernetes or Docker Compose. For Kubernetes, you can use the Jaeger Operator and deploy a Jaeger instance with the provided YAML configuration. Alternatively, for Docker Compose, you can use the 'jaegertracing/all-in-one' image to start the Jaeger services on your local machine.
Yes, this skill supports OpenTelemetry for tracing. You can instrument your applications in Python using Flask or in Node.js using Express by setting up the necessary OpenTelemetry SDK and configuring it to export traces to Jaeger.
The main components of distributed tracing are traces, spans, context, tags, and logs. A trace represents the end-to-end request journey, spans represent individual operations within a trace, context contains metadata propagated across services, tags help with filtering spans, and logs capture timestamped events during a span's execution.
Quick Setup:
.claude/skills/Repository
amurata/cc-tools