Prometheus turns exported time-series metrics into a queryable view of system behavior and evaluates rules against that data. This hands-on course builds a small local monitoring pipeline from binary installation and scraping through PromQL, alert rules, and Alertmanager routing.
You will scrape Prometheus itself and Linux host metrics from Node Exporter, query memory and CPU data, load an InstanceDown rule, and send a test alert to Alertmanager. The final challenge independently configures both scrape jobs and a custom HighCpuLoad rule, then verifies it through the Rules API.
What You Will Learn
- Download, configure, start, and verify a local Prometheus server
- Run Node Exporter and confirm its metrics endpoint on port 9100
- Add static scrape jobs and inspect target health through the Prometheus API
- Query raw memory metrics, CPU rates, and a calculated memory percentage with PromQL
- Create an alert rule file and load it from
prometheus.yml - Connect Prometheus to Alertmanager and configure a local webhook receiver
- Submit a test alert and inspect the resulting Alertmanager state
- Build and verify a custom CPU-load rule from the command line
Who This Course Is For
This intermediate course is for Linux administrators, DevOps learners, and operations engineers who want a practical first workflow for collecting metrics, querying them, and defining alerts with Prometheus.
Prerequisites: Comfort with Linux shell commands, YAML, background processes, curl, ports, and basic CPU and memory concepts. Prior Prometheus or PromQL experience is not required.
Learning environment: A single Linux lab machine running downloaded Prometheus, Node Exporter, Alertmanager, and a local webhook endpoint. Components are started as lab processes rather than installed as persistent production services.
Frequently Asked Questions
Will I monitor multiple remote servers?
No. Prometheus scrapes itself and one Node Exporter on localhost. Static multi-target configuration, service discovery, federation, and distributed storage are outside the course.
How advanced are the PromQL exercises?
They cover a raw memory metric, rate() over a CPU counter, and a simple memory-use calculation. Aggregation, label matching, joins, recording rules, histograms, and query optimization are not taught.
Will alerts reach email, Slack, or another external service?
No. Alertmanager routes to a local webhook, and a test alert is submitted manually. The final challenge confirms that HighCpuLoad is loaded; it does not generate sustained CPU load or verify an external notification.
