Skip to content
All posts
AKS GitOps & CI/CDMay 14, 20261 min read

Flux on AKS in Twenty Minutes

GitOps sounds like a platform project. Getting your first Flux reconciliation running on AKS is closer to a coffee break, and it changes how you think about deploys immediately.

People defer GitOps because it sounds like a quarter-long platform initiative. The full platform might be. But your first working reconciliation on AKS is a coffee break, and doing it once teaches the model better than any diagram.

The shape, start to finish:

  1. Get Flux into the cluster. On AKS you have two clean options: enable the managed GitOps extension (Flux, lifecycle handled by Azure) so you don't operate the reconciler yourself, or run flux bootstrap with the CLI for full control. For a first taste, the managed extension is the least friction.
  2. Point it at a repo. Create a Git repository with a couple of plain manifests, a Deployment and a Service for some trivial app. This repo is now the desired state.
  3. Declare the source and a reconciliation. Tell Flux which repo and path to watch. That's the whole wiring: a source, and a thing that says "make the cluster match this path."
  4. Watch it converge. Flux pulls the repo and applies it. Your app appears, not because a pipeline pushed it, but because the cluster pulled it.

Now do the thing that makes it click: edit a manifest in Git, bump the replica count, and commit. Within the reconcile interval, the cluster changes to match. You didn't run kubectl. You changed the source of truth, and the cluster followed. Then try the reverse: kubectl scale the deployment by hand and watch Flux quietly revert it back to what Git says. That snap-back is GitOps. Drift doesn't survive.

That's the entire mental shift, earned in twenty minutes: the cluster follows the repo, continuously. Everything else in GitOps (environments, promotions, secrets, multi-cluster) is elaboration on that one loop. Start with the loop. Build the platform later.

Want this distilled for your stack? Ask the assistant for the key takeaways or related reading.

Keep reading

Related notes

AKS GitOps & CI/CD1 min read

Kustomize Over Helm for Cluster Config? Sometimes.

Helm became the default for everything, including jobs it's bad at. For your own cluster configuration, Kustomize is often the calmer choice, and knowing when is the whole skill.

AzureKubernetes
May 28, 2026Read

The Clarity Brief

Every other Tuesday · unsubscribe anytime

A short, high-signal briefing on architecture, AI, observability, and engineering leadership, written to make hard things clear.

Topics you care about

We respect your inbox. Your email is used only to send the newsletter and is never sold or shared.

Sathpal-OS