prometheus-hyper

Helper library to export prometheus metrics using tokio and hyper. It's intended to help writing prometheus exporters without the need to setup and maintain a http (no https) webserver. If the program also uses a http server for other purposes this package is probably not the best way and prometheus should be used directly.

This crate is similar to prometheus_exporter. If you are not in a tokio environment you might choose this one. Keep in mind there is an vulnerability for tiny-http < 0.8: https://rustsec.org/advisories/RUSTSEC-2020-0031.html

Use this crate when:

• you write an async app that wants to export metrics
• want to interact directly with prometheus.
• don't want to care about the webserver to host the /metrics endpoint

Don't use this crate when:

• you write a rest-api, in that case you should prob just implement the /metrics endpoint with your rest-framework.
• you want to avoid tokio.
• you are writing a library, in that case use plain prometheus and let the binary application decide which HTTP server to use.

This crate optimizes for typical metrics scraping, e.g. a scrape interval by few prometheus instances (usually 1) with a usual interval (e.g. 1s). It's optimized for a low foot-print, rather than being able to serve 100.000 metrics requests per sec.

Usage

Add this to your Cargo.toml:

[dependencies]
prometheus-hyper = "0.1"

The crate exports a single struct Server with a single fn run: Pass your registry to the server, and provide it with a shutdown_future. The server will shut down, once the future completes.

tokio::spawn(Server::run(
    Arc::clone(&registry),
    SocketAddr::from(([0; 4], 8080)),
    shutdown_future,
));

This will start the exporter and bind the http server to 0.0.0.0:8080. After that you can just update the metrics as you used to.

See the documentation and the examples for more information on how to use this crate.