distributed-tracing

What This Skill Does

The Distributed Tracing skill provides a robust framework for implementing end-to-end request visibility in microservice architectures using Jaeger and Tempo. By instrumenting your services, you can capture the entire lifecycle of a request as it traverses through various components, services, and databases. This skill helps developers move beyond simple logs to visualize complex call chains, identify latency bottlenecks, and debug error propagation in distributed systems. It acts as a guide and automation assistant to help you set up infrastructure, configure collectors, and integrate OpenTelemetry SDKs into your existing codebase.

Installation

To integrate this skill into your environment, run the following command in your terminal: clawhub install openclaw/skills/skills/anton-abyzov/sw-distributed-tracing Once installed, the agent will have access to deployment templates for Kubernetes and Docker Compose, as well as language-specific instrumentation snippets for Python, Go, and Node.js.

Use Cases

This skill is indispensable when you are faced with 'distributed monolith' syndrome where requests seem to disappear or hang without clear logs. Key scenarios include:

• Debugging inter-service latency: Identify which specific service in a request chain is causing delays.
• Bottleneck Analysis: Pinpoint database query overhead or slow external API calls.
• Service Mapping: Generate dynamic dependency graphs to visualize how your microservices communicate.
• Error Triage: Trace exactly where a request failed and inspect span attributes to see why an error occurred.

Example Prompts

1. "OpenClaw, please generate a Docker Compose configuration for a Jaeger all-in-one instance and guide me on how to connect my Flask service to it using OpenTelemetry."
2. "I'm experiencing intermittent 500 errors in my user-service. How can I use the distributed-tracing skill to instrument the request flow and isolate the failing span?"
3. "Help me identify which downstream service is adding the most latency to my /api/orders endpoint based on the Jaeger trace data."

Tips & Limitations

• Tip: Always use unique Trace IDs and ensure your headers (like B3 or W3C TraceContext) are propagated across service boundaries to maintain consistent trace continuity.
• Tip: Avoid capturing PII in span attributes or tags; sanitize your data before sending it to the collector.
• Limitation: Distributed tracing introduces a small performance overhead; in high-traffic production systems, consider using head-based or tail-based sampling to reduce the volume of spans processed.