Most Argo CD tutorials hand you a cluster someone else built, a sample app, one argocd app sync, and a green checkmark. Then you try it at work and the checkmark turns out to have taught you nothing. You do not know why the repo-server rendered that manifest, what a sync wave orders, how an RBAC rule locks a teammate out of the wrong project, or what happens to your running pods when you delete an app.
This book takes the other route. You start from empty servers and finish with an automated GitOps pipeline you built yourself. Nothing is pre-baked.
The "hard way" is not difficulty for its own sake. It is doing the work yourself.
What you will build
You begin with bare servers and end with a pipeline that deploys on every Git push. Along the way you will:
- Provision infra with Terraform, bootstrap a cluster, and deploy an app onto it.
- Trace a change from a Git commit through Argo's reconciliation loop to a running resource, and learn what each component does and what breaks when it fails.
- Install Argo, choosing deliberately between the full, core, and custom installs, then log in from the CLI and add more clusters.
- Create your first Application from the CLI, then redefine the same setup as declarative manifests you keep in Git.
- Configure sync policy on purpose: automated sync, self-healing, pruning, retries and backoff, namespace creation, and more.
- Run your own code around a sync with resource hooks and sync phases, order resources with sync waves, and learn when to use hooks and when not to.
- Scope a teammate to exactly one project with RBAC, then verify the grant.
- Package your apps as a Helm chart and have Argo manage it as a versioned release.
- Generate many Apps from one definition with AppSets.
- Build a multi-branch pipeline that gives every Git branch its own live environment.
- And more!
This book assumes you can read a Deployment, run kubectl, and open a pull request without looking it up.
Every code listing (+ full-color diagram, and screenshot) lives in the companion kit.
