Ravindra BagaleCourses & study guides

Chapter 13: Monitoring Basics

13.13 Lab — disk metric path and a triggered alert

By Last updated:

Lab

Build the primary monitoring path for a badge API lab host and prove one disk alert.

  1. Run Node Exporter; confirm /metrics responds on your chosen port.
  2. Run Prometheus with a scrape job for that exporter; confirm target UP.
  3. Run Grafana; add Prometheus data source; create a small dashboard (CPU, memory, disk %).
  4. Define one disk-used alert with a sustained duration.
  5. Trigger safely with a temporary fill file on a lab path; confirm alert fires.
  6. Delete the fill file; confirm alert resolves and disk recovers.
  7. Write a five-line runbook in your notes (not committed secrets).
  8. Optional: skim CloudWatch docs and write three bullets on how the same disk story would look there — no need to implement both.

Practice task

Explain golden signals to a project manager in six lines, mapping latency/traffic/errors/saturation to badge API peak traffic and a full disk.

Thodkyaat sangaycha tar

  • Monitor so you see trouble before customers do — bike dashboard before the highway stop.
  • Golden signals: latency, traffic, errors, saturation; host metrics explain saturation.
  • Primary path: Node Exporter → Prometheus → Grafana; CloudWatch is the AWS-native alternate (link AWS course for depth).
  • Prefer few clear panels and one trustworthy alert over noisy walls.
  • Trigger disk alerts safely on lab disks and always clean up; ports are published ports — do not expose admin UIs publicly without care.
  • No invented AWS prices or free-tier claims — stop idle labs.

Samajla ka? Nasel tar lab (13.13) scrape UP → dashboard → disk alert fire/resolve paryant sodu naka. Aata pudhe jaauya logging aani light observability kade.