Lab: Instrument a Python application with OpenTelemetry

What you will learn in this TP

• Auto-instrument an API
• Add a manual span
• Find the trace

The application #

What you will learn in this section

• Auto-instrument an API
• Add a manual span
• Find the trace

Context. A small Python API does not emit any traces yet. You instrument it with OpenTelemetry without rewriting its logic, then add a business span to measure a specific step (discount calculation).

Prerequisites: Previous lab (Tempo), Python 3  ·  Estimated time: 40 min

An API that calls a second API.

1. 1. Create the application

   Why? A minimal application is enough to practise; what matters is seeing how instrumentation is added without changing the routes.

   Write a Flask application with a /commande route that waits 100 ms.

2. 2. Auto-instrument

   Why? Auto-instrumentation automatically creates spans for HTTP requests: it is the fastest way to get initial visibility.

   Install the OpenTelemetry packages and run the application with the agent.

3. 3. Add a manual span

   Why? Manual spans provide the business context that auto-instrumentation cannot guess (discount, customer type).

   Add a calcul_remise span with a remise.pourcent attribute.

4. 4. Find the trace

   Why? Filtering by name and duration shows how to isolate a slow operation among thousands of traces.

   Search for traces whose calcul_remise span lasts more than 50 ms.

Difficulty level : ●●○○○ (2/5)