Deploy the heapster helm chart on Azure Kubernetes Service (AKS)

Question

Pulumi · Accepted Answer

In order to deploy the Heapster Helm chart on an Azure Kubernetes Service (AKS) cluster using Pulumi, you'll need to perform a few steps:

1. **Set up an AKS cluster**: Before deploying any applications, you'll need a Kubernetes cluster. With Pulumi, you can define and create an AKS cluster programmatically.

2. **Install Heapster using Helm**: Once you have your Kubernetes cluster, you can deploy Heapster using Helm. Pulumi provides resources for working with Helm charts directly.

Heapster is a deprecated monitoring tool, superseded by Metrics Server in Kubernetes, so this guide assumes you have specific reasons for installing Heapster. If you're looking to set up a monitoring solution for a new cluster, consider using Metrics Server or other modern alternatives such as Prometheus.

Below is a Pulumi program in TypeScript which sets up an AKS cluster and deploys the Heapster Helm chart to it. Please note that you should have Pulumi CLI installed and be logged into your Azure account for the code to work.

First, we'll begin by creating the AKS cluster:

```typescript
import * as azure from "@pulumi/azure-native";
import * as kubernetes from "@pulumi/kubernetes";

// Create an AKS cluster.
const cluster = new azure.containerservice.ManagedCluster("aksCluster", {
    resourceGroupName: azureResourceGroupName, // Replace with your Azure resource group name
    agentPoolProfiles: [{
        count: 1,
        maxPods: 110,
        mode: "System",
        name: "agentpool",
        nodeLabels: {},
        osDiskSizeGB: 30,
        osType: "Linux",
        vmSize: "Standard_DS2_v2",
    }],
    dnsPrefix: "aks-heapster", // Replace with your own DNS prefix
    enableRBAC: true,
    kubernetesVersion: "1.20.7", // Use an appropriate version
    linuxProfile: {
        adminUsername: "aksuser", // Replace with the desired admin username
        ssh: {
            publicKeys: [{
                keyData: "ssh-rsa YOUR_PUBLIC_KEY", // Replace with your SSH public key
            }],
        },
    },
    nodeResourceGroup: "aksHeapsterNode", // Replace with the desired node resource group name
    servicePrincipalProfile: {
        clientId: "YOUR_CLIENT_ID", // Replace with your service principal's client ID
        secret: "YOUR_SECRET", // Replace with your service principal's secret
    },
});

// Export the cluster's kubeconfig.
export const kubeconfig = cluster.kubeConfigRaw;
```

This code sets up an AKS cluster with a single node pool and exports the kubeconfig, which will be used to connect to the Kubernetes cluster.

Next, we will deploy Heapster using the Helm chart:

```typescript
// Helm chart version and name
const heapsterChart = "heapster";
const heapsterVersion = "1.5.4"; // Use the appropriate chart version

// Create a Kubernetes provider instance that uses our AKS kubeconfig.
const k8sProvider = new kubernetes.Provider("k8sProvider", {
    kubeconfig: cluster.kubeConfigRaw,
});

// Deploy Heapster using the Helm chart.
const heapster = new kubernetes.helm.v3.Chart("heapster", {
    chart: heapsterChart,
    version: heapsterVersion,
    fetchOpts: {
        repo: "https://kubernetes-charts.storage.googleapis.com/", // Heapster chart is located within the deprecated repo
    },
}, { provider: k8sProvider });
```

In the above code block, we define the name and version of the Heapster Helm chart we wish to install, create a Pulumi Kubernetes provider that uses the AKS kubeconfig, and then deploy Heapster with the `Chart` resource.

If the Heapster chart is not available in the default repositories, you might need to add a custom repository URL in the `fetchOpts.repo` field.

Lastly, you should be aware of the current status of Heapster. Since Heapster is deprecated, its Helm chart may no longer be maintained or available in the default chart repositories. If Heapster is specifically required, you might need to obtain the chart from a different source or use an archived version.

Always review the code and customize according to your requirements, like selecting appropriate VM sizes, Kubernetes versions, or replacing deprecated Helm chart repositories with their updated locations.