Skip to content
Tutorials Center
My learning

Topics

Lesson 1 of 37 Β· Getting Started

What Is Argo CD? Declarative GitOps for Kubernetes

Learn what Argo CD is, what problem it solves, how it compares with push-based CI/CD deployments, and which words (Application, sync, drift) you will meet throughout this course.

Argo CD is a Kubernetes controller that keeps a cluster in the state described by files in a Git repository. You commit YAML (or Helm charts, or Kustomize overlays) to Git; Argo CD notices, compares the files with what is running, and changes the cluster until the two match. It is a graduated project of the Cloud Native Computing Foundation and one of the most widely used GitOps tools.

This course teaches Argo CD and the ideas behind it: GitOps, declarative configuration, reconciliation and safe delivery. You will learn the concepts, the exact YAML and the commands to use.

Argo CD needs a Kubernetes cluster, and this site cannot run one. The YAML and shell commands in these lessons are shown read-only. Every runnable example is a clearly labelled JavaScript stand-in: a small model of one idea (a diff, a status calculation, a sync plan) that you can edit and run in the browser. The stand-ins are teaching models, not Argo CD itself.

The problem Argo CD solves

A traditional pipeline builds an image and then pushes a change to the cluster: the CI job runs kubectl apply or helm upgrade with credentials for production. That works, but it has weak spots.

  • The CI system holds powerful cluster credentials.
  • Nobody knows what is really running unless they ask the cluster. If someone edits a Deployment by hand, Git and reality quietly diverge.
  • Recreating a cluster means re-running old pipelines in the right order and hoping.

Argo CD reverses the direction. It runs inside the cluster and pulls the desired state from Git. Git becomes the single source of truth, and the controller continuously reports whether reality agrees.

flowchart LR
  Dev["Developer"] -->|"pull request"| Git[("Git repository: desired state")]
  Git -->|"Argo CD pulls and renders"| Argo["Argo CD in the cluster"]
  Argo -->|"compares and applies"| Cluster["Kubernetes resources: live state"]
  Cluster -->|"status and health"| Argo

Changes flow from Git to the cluster, never from a person's laptop straight to it.

Key words you will meet

Desired vs live

Desired state What Git says
Live state What runs now
Diff The gap between them

Argo CD exists to close this gap.

1 / 3
  • Application: a custom resource that connects one source in Git to one destination in a cluster.
  • Sync: applying the desired state to the cluster. OutOfSync means live differs from desired.
  • Drift: live state that changed without a Git commit, for example a manual kubectl edit.
  • Health: whether the resources actually work (a Deployment can be in sync but crash-looping).

A model of desired versus live

The first stand-in compares two small objects and lists the differences, the way Argo CD's diff view highlights them. Edit live and run it again.

JS Stand-in: desired vs live diff

Press Run (or Ctrl + Enter) to see the console output here.

No output.

Sync status is not health

Sync compares configuration; health inspects behavior. This second stand-in combines them the way the Argo CD UI does.

JS Stand-in: sync plus health

Press Run (or Ctrl + Enter) to see the console output here.

No output.

The third line is the important one: the manifests match Git exactly, yet no pod runs. Argo CD shows both facts separately so you can tell a deployment problem from a configuration problem.

What Argo CD is not

  • Not a CI tool. It does not build images or run tests. Your CI builds and pushes an image, then updates Git (often a one-line tag change); Argo CD delivers it.
  • Not a package manager. It renders Helm and Kustomize but does not replace them.
  • Not Argo Workflows. The Argo family also has Argo Workflows (job pipelines), Argo Rollouts (canary and blue-green) and Argo Events. Each is a separate project.

Why teams adopt it

  1. Audit trail: every production change is a Git commit with an author, a review and a revert button.
  2. Drift visibility: the controller tells you when the cluster no longer matches Git, and can heal it automatically.
  3. Disaster recovery: a new cluster plus the same Git repository gives you the same applications.
  4. Safer credentials: only the cluster-side controller needs write access; CI needs none.
  5. Multi-cluster: one Argo CD can manage many clusters.

Check your understanding

Quick check Β· 4 questions

of 4 answered You scored / 4 Β· perfect!

  1. 1. In the pull model used by Argo CD, where does the deployment logic run?

    Argo CD runs in the cluster and pulls from Git, so CI does not need cluster credentials.
  2. 2. An app's manifests match Git exactly but its pods keep crashing. What does Argo CD report?

    Sync status compares configuration with Git; health inspects whether the resources work. They are independent.
  3. 3. Which task is NOT done by Argo CD?

    Argo CD is a delivery tool. Building images is the job of CI.
  4. 4. What is drift?

    Drift means the cluster no longer matches Git, for example after a manual kubectl edit.

Summary

  • Argo CD is a pull-based GitOps controller that makes a cluster match a Git repository.
  • Core vocabulary: Application, AppProject, sync, drift, health.
  • Sync status (matches Git?) and health status (works?) are independent.
  • It delivers; it does not build. Pair it with CI.
  • Gains: audit trail, drift detection, recovery and credential isolation.

Next: Argo CD Architecture