example

flow example

In this example a demo application is run the computes random Fibonacci numbers. The application is traced using OpenTelemetry with a flow SpanProcessor. The metrics produced with by flow are collected with an OpenTelemetry collector and ultimately shipped to a Prometheus instance.

Prerequisites

This demo expects docker-compose is installed and working.

Start the demo

Start all services.

docker-compose up --detach

This should do a few things.

1. Build the demo Fibonacci application and package it as a Docker image.
2. Start the Fibonacci application.
3. Start an OpenTelemetry Collector instance. This collector will forward traces from the application to a Jaeger instance and metrics to a Prometheus instance.
4. Start a Jaeger instance to capture traces from the collector.
5. Start a Prometheus instance to scrape metrics from the collector.

Verify

Fibonacci Application

First, make sure the Fibonacci application is reporting metrics.

$ curl -s http://localhost:41820/metrics | grep 'spans_total'

# HELP spans_total The total number of processed spans
# TYPE spans_total counter
spans_total{state="ended"} 762
spans_total{state="started"} 762

The total number of spans will be different, but you should see something like above.

Jaeger Traces

Next, make sure the traces are making it to the Jaeger instance. Navigate through the Jaeger UI you are running locally to ensure traces from the application are there.

Prometheus Metrics

Finally, verify the flow SpanProcessor is working. Go to the Prometheus targets page and make sure all targets are up.

From here you can go to the Prometheus graph page and visualize that data being scraped.

For example, run the following query to see the exported span rate.

rate(spans_total[1m])

Stop and Clean Up

Stop all services.

docker-compose down

If you are all done with the demo you can remove everything as well.

docker-compose rm