Deploy the kubeflow-kfserving-inference helm chart on Rancher

Question

Pulumi · Accepted Answer

To deploy the Kubeflow KFServing inference Helm chart on Rancher, you would typically follow these steps:

1. Set up Rancher and a Kubernetes cluster where the Helm chart will be deployed.
2. Add the Helm chart repository that contains the Kubeflow KFServing charts.
3. Use Pulumi to programmatically execute the Helm chart on your Kubernetes cluster managed by Rancher.

Below is a program in TypeScript using Pulumi to deploy a Helm chart on Rancher. In Pulumi, the `@pulumi/rancher2` package provides a way to interact with Rancher 2.x. However, deploying Helm charts within a Rancher-managed cluster typically involves interacting with Kubernetes directly rather than through Rancher's resources.

This program will create a `HelmRelease` resource which represents the deployment of a Helm chart. For that, you can use the `@pulumi/kubernetes` package, which is the Pulumi provider for Kubernetes.

First, you would install the required Pulumi packages by running:

```bash
npm install @pulumi/pulumi @pulumi/kubernetes
```

Then, here is the TypeScript program which defines the deployment:

```typescript
import * as pulumi from "@pulumi/pulumi";
import * as k8s from "@pulumi/kubernetes";

// Configuration for the Kubeflow KFServing chart.
const kfservingChart = new k8s.helm.v3.Chart("kfserving", {
    // Replace 'CHART_REPO' with the actual chart repository URL and 'CHART_VERSION' with the desired version.
    chart: "kubeflow-kfserving-inference",
    version: "CHART_VERSION",
    fetchOpts: {
        repo: "CHART_REPO_URL",
    },
    // Define any values we want to override. This would be similar to
    // the `values.yaml` file when installing Helm charts manually.
    values: {
        // Your custom values here.
    },
});

// Export the name of the chart
export const chartName = kfservingChart.metadata.apply(metadata => metadata.name);
```

Replace `CHART_REPO_URL` with the URL of the Helm chart repository where the Kubeflow KFServing chart is hosted, and `CHART_VERSION` with the version number of the chart you wish to deploy.

Make sure you are authenticated to the Kubernetes cluster which is managed by Rancher, and have set up your `kubeconfig` file correctly so that Pulumi can interact with your cluster.

If you need to perform any specific configurations to integrate with Rancher-managed Kubernetes clusters, such as setting up namespaces or RBAC, you should include that in your Pulumi program before defining the `HelmRelease`.

After you've set up the program, you would run it using the Pulumi CLI. Executing `pulumi up` would initiate the deployment process, which Pulumi would perform according to the configuration set in the program.

Keep in mind that the exact values and configurations may need to be tailored to the specifics of your Rancher setup and the requirements of the Kubeflow KFServing Helm chart you are deploying.

Remember to review and customize the Helm chart values according to the official Kubeflow KFServing documentation or the README file usually provided within the Helm chart package for proper configuration options.