Getting Started

Verified against the README at commit fc561264 (3.13.0-rc.0). Commands assume a Unix shell with Docker, or Go and Node if you build from source.

Prerequisites

• Docker, for the quickest trial run, or
• Go (version in go.mod or greater), Node (version in web/ui/.nvmrc or greater), and npm 10+ to build from source (README:70-72).

Install

The fastest path is the official container image (README:60-61):

docker run --name prometheus -d -p 127.0.0.1:9090:9090 prom/prometheus

To build from source instead (README:76-78, 101):

git clone https://github.com/prometheus/prometheus.git
cd prometheus
make build

make build compiles the prometheus and promtool binaries with the web assets embedded, so the binary runs from anywhere (README:97-101).

A first working setup

The container above starts with a default config. To run your own, point Prometheus at a config file that scrapes itself.

1. Write a minimal prometheus.yml:

   global:
     scrape_interval: 15s
   
   scrape_configs:
     - job_name: prometheus
       static_configs:
         - targets: ["localhost:9090"]

2. Start the server with that config (README:85-86):

   ./prometheus --config.file=prometheus.yml

When built with go install rather than make build, run the binary from the root of the cloned repository so it can find web assets under web/ui/static (README:88-92).

Verify it works

Open the web UI at http://localhost:9090/ (README:64). Then confirm Prometheus is scraping itself: run the query up in the expression browser, or hit the API directly.

curl 'http://localhost:9090/api/v1/query?query=up'

A healthy server returns JSON with "status":"success" and an up series whose value is 1 for the prometheus job.

Where to go next

• The official first steps guide covers adding more targets and writing queries.
• For high availability, long-term retention, and scaling beyond a single node, see the Adoption & Ecosystem page and the project documentation at prometheus.io rather than re-documenting it here.