13.4 Primary path — Prometheus + Node Exporter + Grafana
This course's primary monitoring path for labs:
- Node Exporter — exposes host metrics on a local HTTP endpoint.
- Prometheus — scrapes that endpoint on an interval and stores time series.
- Grafana — graphs those series and can fire alert rules (depending on your setup).
CloudWatch (brief mention): on AWS you can use the CloudWatch agent and CloudWatch dashboards/alarms for the same ideas (CPU, disk, alarms). Prefer CloudWatch when your team standardises on AWS-native ops. For depth on AWS console services, use the AWS Cloud course and official AWS docs — we do not invent free-tier limits or prices here. This chapter sticks to Prometheus + Grafana so every student can practise the same open stack on a laptop or one lab VM.
Node Exporter (:9100 typical)
│ scrape
▼
Prometheus
│ query
▼
Grafana dashboards / alerts
Ports here are published ports: Prometheus talks to Node Exporter on a logical port (commonly 9100 in docs); Grafana serves its UI on another (often 3000 in local labs). Publish only what your lab needs; do not expose admin UIs to the whole internet without auth.
Versions and installs
Package names and default ports can change. Follow current Prometheus / Node Exporter / Grafana install docs for your OS. Commands below are sketches — adapt paths and compose files to the version you install.