← Documentation

Installation

Kaja has two parts: the dashboard (the website you log into) and a small program called the agent that runs inside your Kubernetes cluster. The agent is what lets Kaja see your cluster and manage things from one place. This page explains how to get the agent onto your cluster.

You don’t install from a generic “Kaja repo”

Unlike many tools, you don’t add a public Helm repo and type your cluster name and server URL yourself. Instead:

  1. You create the cluster in the Kaja console (name + slug).
  2. Kaja gives you a custom install command that already includes your cluster id and a secret token.
  3. You run that command once on the cluster you want to connect.

So the right order is: create in Kaja first, then run the command it gives you.

What the install command does

The command uses Helm to install the agent in a namespace called kaja. It passes your cluster id and token so the agent knows which cluster it is and can securely talk to Kaja. After that, the agent keeps running and Kaja shows the cluster as connected (or “Active”) when it’s healthy.

Two ways to run it

On the cluster setup page in Kaja you’ll see:

  • From a release tarball — Downloads the agent chart from GitHub and installs it. Good if you prefer a fixed release.
  • From OCI — Installs from a container registry (ghcr.io). Same result, different source.

Copy the command that matches your preference and run it with kubectl pointing at your cluster. You can run the same command again later to upgrade; it’s safe to re-run.

Check that it worked

After running the command, you can check that the agent is running:

kubectl get pods -n kaja

You should see the agent pod(s). Once they’re running, Kaja will show the cluster as connected.

If you don’t use Helm

If your environment doesn’t allow Helm, you can still install the agent using the same chart: export the manifests from the chart (e.g. with helm template) and apply them with kubectl apply. You’ll need to set the cluster id and token in the same way the Helm command does (via the chart’s values or env). The setup page in Kaja is the place to get the correct cluster id and token.

Next steps

  • Connecting clusters — Add more clusters; each gets its own command from the console.
  • Quick start — Full walkthrough from sign-in to a connected cluster.