Local ELK Observability Test Stack with a Simple FastAPI Web Application

This repository demonstrates how to use the Elastic Stack (Elasticsearch, Kibana, and APM Server) to observe a basic web application using OpenTelemetry automatic instrumentation.

It uses EDOT (Elastic Distribution of OpenTelemetry) to collect and export telemetry data to the Elastic Stack.

The web application is a simple FastAPI service that returns a JSON message when accessed at:

curl http://localhost:8000/

The ELK stack is provisioned via Docker Compose.

Logs, traces, and custom metrics (e.g., the endpoint_call_count metric) are captured and can be viewed in the Kibana UI.

Getting Started

• Start the stack: docker-compose up
• Create the Kibana service token: docker exec -it elasticsearch elasticsearch-service-tokens create elastic/kibana kibana-system
• Copy the generated token and set it as the value of the ELASTICSEARCH_SERVICEACCOUNTTOKEN environment variable in the kibana service section of docker-compose.yml.
• Restart Kibana: docker compose up -d --force-recreate kibana
• Access ElasticSearch at: http://localhost:5601
  • Login with username: elastic, Password: password
• Install the APM integration with:
  • host = apm-server:8200
  • URL: http://apm-server:8200
• Hit the app to create some data: curl http://localhost:8000/
• Data should now be present in ElasticSearch