Lab trackHelm Labs

Helm hands-on labs

Twenty-two labs that take one real four-service application from a folder of YAML to a tested, versioned chart in a registry — templating, releases, upgrades and rollbacks, schemas, tests and hooks on a live Kubernetes cluster.

  • 22 labs
  • 15.1 h
  • Intermediate
  • Free
Start with the first lab

Free with a free account — it runs in your browser, nothing to install.

  • A real environment in your browser
  • You run the commands yourself
  • Your result is checked automatically

Labs in this track

22 labsFree

Before Helm

3 labs

Charts and Templating

6 labs

Build the Real Chart

5 labs

Operate and Trust the Chart

5 labs

Package, Share, Compose

3 labs
  1. Package It and Push It40 min
  2. Charts Inside Charts45 min
  3. Ship It60 min

About this track

A lab-only path that starts without Helm, so every feature it has afterwards is an answer to something you have already lived through.

Show more

Every lab is a real Kubernetes cluster in your browser, and the same application runs through the whole track: a store with a front end, two APIs and a message queue. First you deploy it from raw YAML, copy it into a second environment by hand and watch the copies drift. Then you learn what a chart and a release are, where values come from and the template language that turns them into objects. In the middle of the track you build the store's chart yourself, one piece at a time, until the shop opens in your browser. The last two sections are what owning a chart is like: upgrades that put themselves back, changes reviewed before they land, values refused at the door, tests and hooks, then packaging, an OCI registry, subcharts and a capstone release to production.

There are no video lectures and no reading-only lessons. Each lab states the task, checks what you actually did on the cluster, and shows a worked solution when you ask for one.

What you'll learn

  • Deploy a multi-service application from raw manifests, and see exactly what Helm replaces
  • Install, upgrade and inspect releases, and read the history Helm keeps in the cluster
  • Write a chart from nothing: Chart.yaml, values, templates, .helmignore, lint and template
  • Override values from files and the command line, and know which source wins
  • Use the template language: range, quote, toYaml with nindent, tpl, named helpers and labels
  • Toggle objects, generate one object per key of a map, and avoid Go's idea of false
  • Render Secrets and ConfigMaps, and roll Pods on a config change with a checksum
  • Set probes and resources per service, and express environments as values files
  • Put the application behind an Ingress and write a NOTES.txt that tells the truth
  • Upgrade with --atomic, roll back, diff before applying, and catch bad values with a server-side dry run
  • Refuse bad values with values.schema.json, unit-test the templates, and gate upgrades with hooks
  • Package a chart, push it to an OCI registry, compose subcharts and library charts, and ship a release

Prerequisites

The Kubernetes hands-on labs, or the same ground from experience: comfortable with kubectl get, describe, logs and apply, Deployments, Services and namespaces.

Nothing to install. The cluster, the terminal and an editor are in your browser.

Who it's for

DevOps and platform engineers who install other people's charts and want to write their own, developers who ship to Kubernetes, and anyone preparing to own the deployment of a real application.

FAQ

Do I need the Kubernetes labs first? Not formally, but every lab assumes kubectl. If Deployments and Services are new to you, start there; they are free.

How many times can I launch a lab? Ten launches per lab, one at a time.

Is it the same application all the way through? Yes. The store you deploy by hand in the first lab is the one you package and ship in the last.

Do I have to go in order? The labs build on each other, so in order is best, but each one prepares its own starting point and stands on its own.