Monitor your service with Grafana & Prometheus

Prometheus collects and stores metrics (CPU, RAM, traffic), Grafana renders them in dashboards. The pair is the self-hosted observability standard.

Cause / The problem

Without monitoring, you discover outages when they hit: a full disk, saturated RAM, climbing latency. You must capture metrics continuously and visualize them to get ahead.

Solution

1. Run Prometheus with this prometheus.yml:

   global:
     scrape_interval: 15s
   scrape_configs:
     - job_name: "node"
       static_configs:
         - targets: ["localhost:9100"]

2. Start the stack with Docker:

   docker run -d -p 9090:9090 -v ./prometheus.yml:/etc/prometheus/prometheus.yml prom/prometheus

3. Add node_exporter for system metrics (CPU, RAM, disk):

   docker run -d -p 9100:9100 --net=host prom/node-exporter

4. Start Grafana:

   docker run -d -p 3000:3000 grafana/grafana

   Default login: admin / admin (change it immediately).
5. Add Prometheus as a data source in Grafana: Connections → Data sources → Add → type Prometheus, URL http://localhost:9090, then Save & test.
6. Import a dashboard: Dashboards → New → Import → enter ID 1860 (Node Exporter Full) and pick your Prometheus source.
7. Set up an alert: in Alerting, create a rule (e.g. node_memory_MemAvailable_bytes / node_memory_MemTotal_bytes < 0.1 for a RAM < 10% alert), then wire it to a contact point (email, webhook).

Put Grafana and Prometheus behind an HTTPS reverse proxy (see the dedicated article), and never expose them without authentication.