Install Helm on the Management Node¶
GPU Manager requires Helm version 3.14 or newer to install the AI Platform that is delivered as Helm Charts. Follow the steps below to install helm on the management host (mvai-mgmt) and the GPU worker node (mvai-nvgpu01). These steps mirror the Official Helm Installation Guide.
Helm Prerequisites¶
- Ubuntu 22.04 LTS installed on both mvai-mgmt and mvai-nvgpu01
- A running Kubernetes cluster
- Cluster admin permissions
kubectlinstalled and configured to access your cluster
1. Install Helm¶
Option 1: Use the Helm Installer Script¶
Helm now has an installer script that will automatically grab the latest version of Helm and install it locally.
- Fetch the script, then execute it locally.
- It's well documented, so you can read through it and understand what it is doing before running it.
Option 2: Install Helm via Console¶
$ curl -fsSL -o get_helm.sh https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3
$ chmod 700 get_helm.sh
$ ./get_helm.sh
Alternatively, you can run curl https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash to install helm with one command.
2. Add and update the Helm Stable Chart repository¶
Add the official Helm stable chart repository using:
Update Helm's chart repository to ensure you have the latest chart information:
To test your Helm installation, you can search for available charts:
This command should return a list of available charts from the stable repository.