Lab: Python APM and log-trace correlation

What you will learn in this TP

• Instrument with ddtrace
• Read the service page
• Correlate logs and traces

Introduction #

What you will learn in this section

• Instrument with ddtrace
• Read the service page
• Correlate logs and traces

Context. Developers want to go from an error log to the full trace of the request, and vice versa.

Prerequisites: Lab 1 completed, Python 3  ·  Estimated time: 45 min

Linked traces and logs.

Hands-on #

Guided steps, with their solutions.

1. 1. Write the API

   Why? A slow route that sometimes fails produces interesting traces.

   Create app.py.

2. 2. Run with ddtrace

   Why? ddtrace-run instruments Flask and injects the IDs into the logs.

   Install ddtrace and run the API with unified tagging.

3. 3. Generate traffic

   Why? You need requests, including errors, to populate the service page.

   Send 300 requests.

4. 4. Read the service page

   Why? RED metrics and deployments appear automatically.

   Open APM > Services > paiement-api. Note the error rate and the p95 latency, then open an error trace.

5. 5. Correlate

   Why? The injected IDs link the two signals.

   Send app.log to Datadog (agent log configuration on this file, service:paiement-api), then go from a trace to its logs and from an error log to its trace.

Difficulty level : ●●●○○ (3/5)