The primary use case for fake_exporter is to evaluate prometheus in a
synthetic environment. To do so a Flask application is provided that
will export a /metrics endpoint for Prometheus to scrape.
To run a single instance of fake_exporter install Flask and run
./fake_exporter.py. The metrics will be available with e.g. curl -s http://localhost:5000/metrics.
The endpoint accepts some arguments, notably metrics and seed
respectively to fix the number of metrics returned and the random seed
if needed.
To simulate many targets in an easy way, it is possible to run
fake_exporter with any WSGI compatible application server, a sample
configuration for uwsgi is provided that binds multiple ports on
localhost. Generate it with make (you'll need jinja2 installed).
Once fake_exporter.py is available on multiple ports, Prometheus
should be instructed about the targets. A sample configuration is
provided in prometheus.yml with the respective targets file in
fake_exporter.yml. Note that if the latter file is modified
Prometheus will pick up the changes automatically.