Grafana 13.0.0

Grafana draws graphs from data it queries elsewhere. It stores no metrics itself, which is the thing to understand before configuring it.

Import rather than build

The single most useful piece of advice. Grafana has a large library of community dashboards, and dashboard 1860 for node_exporter covers essentially everything about a host.

Import it, use it, and build your own only for the handful of things specific to your setup. People lose entire evenings assembling panels that already existed.

Data sources

Prometheus is the common pairing. Grafana also queries InfluxDB, Loki for logs, PostgreSQL, MySQL, Elasticsearch, and many others, and a single dashboard can mix them.

Running it

services:
  grafana:
    image: grafana/grafana:latest
    restart: unless-stopped
    ports:
      - "127.0.0.1:3000:3000"
    volumes:
      - grafana-data:/var/lib/grafana
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=changeme
      - GF_USERS_ALLOW_SIGN_UP=false

volumes:
  grafana-data:

Bind to 127.0.0.1 and put a reverse proxy in front. Our compose generator produces files like this.

Persist /var/lib/grafana. Dashboards, users, and data source configuration live there, and losing it means rebuilding all of it.

Alerting

Grafana has its own alerting engine, which overlaps with Prometheus Alertmanager. Pick one rather than running both, since duplicate alerts from two systems is a fast route to ignoring all of them.

Grafana alerting is easier to configure through the interface; Alertmanager is better at routing, grouping, and silencing.

Where it fits

Grafana is the visualization layer and needs something underneath it. For a single machine where you want detail immediately, Netdata gives you dashboards with no separate stack at all.

Our monitoring guide covers when each is the right answer.

Verify Your Download

Release archives on the download page are published with SHA-256 checksums.