Skip to content

Upgrade GPU Cluster Manager

Warning

To upgrade an existing GPU Cluster Manager installation from 0.4.0 to 0.5.0, you must be using Helm version 3.18 or newer. This version is required to support the --reset-then-reuse-values flag, which is essential for preserving your existing configuration during the upgrade process.

Prerequisites

If you don't have Helm 3.18, you can install it by running the following command:

curl https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash

Upgrade Command

Use the following command to upgrade your installation. This command will update the chart to version 0.5.0-rc1 while keeping your custom configurations intact.

Bash

/usr/local/bin/helm upgrade mvai oci://ghcr.io/memverge/charts/mvai --namespace cattle-system --version 0.5.0-rc1 --wait --timeout 2h0m0s --debug --reset-then-reuse-values

Here is a breakdown of the key flags:

  • --version 0.5.0-rc1: Specifies the new version of the chart you're upgrading to.
  • --wait: Helm will wait for all resources to be in a ready state before marking the upgrade as successful.
  • --timeout 2h0m0s: Sets a maximum timeout of 2 hours for the upgrade process.
  • --reset-then-reuse-values: This flag ensures that all chart values are reset to the new defaults, and then any customizations from your previous installation are reapplied. This preserves your specific changes while updating the default settings.