Upgrade Platform Applications

Prerequisites

Before you begin, you must:

Set the WORKSPACE_NAMESPACE environment variable to the name of the workspace’s namespace where the cluster is attached:
```
export WORKSPACE_NAMESPACE=<workspace_namespace>
```
CODE

Set the WORKSPACE_NAME environment variable to the name of the workspace where the cluster is attached:
```
export WORKSPACE_NAME=<workspace_name>
```
CODE

Upgrade Platform Applications from the CLI

The DKP upgrade process deploys and upgrades Platform applications as a bundle for each cluster or workspace. For the Management Cluster or workspace, DKP upgrade handles all Platform applications; no other steps are necessary to upgrade the Platform application bundle. However, for managed or attached clusters or workspaces, you MUST manually upgrade the Platform applications bundle with the following command.

If you are upgrading your Platform applications as part of the DKP upgrade, upgrade your Platform applications on any additional Workspaces before proceeding with the Konvoy upgrade. Some applications in the previous release are not compatible with the Kubernetes version of this release, and upgrading Kubernetes is part of the DKP Konvoy upgrade process.

Prerequisites for upgrading clusters with NVIDIA GPU Nodes

NOTE: The information in this subsection is applicable only to clusters with GPU nodes. If your cluster is not running GPU nodes, then skip these steps and proceed to the Run DKP upgrade command subsection.

If your attached or managed cluster is running GPU nodes and has deployed the nvidia Platform Application, you must run additional steps prior to upgrading the workspace. In DKP 2.4, the nvidia application is replaced by nvidia-gpu-operator during upgrade, and the configuration for the new application must be created prior to the upgrade to ensure that the new application is deployed with the proper configuration, or it may not deploy successfully.

NOTE: The following steps enable the application for all clusters in the workspace and add configuration overrides at the workspace level, which apply to all clusters in the workspace. If only a subset of clusters in the workspace are running GPU nodes, or some clusters require different configurations, refer to Customize an Application per Cluster on how to customize the application with different configuration overrides per cluster.

Create the ConfigMap with the necessary workspace-level configuration overrides to set the correct Toolkit version. For example, if you’re using Centos 7.9 or RHEL 7.9 as the base operating system for your GPU enabled nodes, set the toolkit.version parameter:
```
cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: ConfigMap
metadata:
  namespace: ${WORKSPACE_NAMESPACE}
  name: nvidia-gpu-operator-ws-overrides
data:
  values.yaml: |
    toolkit:
      version: v1.10.0-centos7
EOF
```
CODE

Create an AppDeployment named nvidia-gpu-operator in the workspace namespace:

cat <<EOF | kubectl apply -f -
apiVersion: apps.kommander.d2iq.io/v1alpha3
kind: AppDeployment
metadata:
  name: nvidia-gpu-operator
  namespace: ${WORKSPACE_NAMESPACE}
spec:
  appRef:
    kind: ClusterApp
  configOverrides:
    name: nvidia-gpu-operator-ws-overrides
EOF

CODE

You may now proceed to the workspace upgrade steps.

Execute the DKP Upgrade Command

Use this command to upgrade all platform applications in the given workspace and its projects to the same version as the platform applications running on the management cluster:

dkp upgrade workspace ${WORKSPACE_NAME}

CODE

An output similar to this appears:

✓ Ensuring HelmReleases are upgraded on clusters in namespace "${WORKSPACE_NAME}"
...

CODE

If the upgrade fails or times out, retry the command with higher verbosity to get more information on the upgrade process:

dkp upgrade workspace ${WORKSPACE_NAME} -v 4

CODE

If you find any HelmReleases in a “broken” release state such as “exhausted” or “another rollback/release in progress”, you can trigger a reconciliation of the HelmRelease using the following commands:

kubectl -n ${WORKSPACE_NAMESPACE} patch helmrelease <HELMRELEASE_NAME> --type='json' -p='[{"op": "replace", "path": "/spec/suspend", "value": true}]'
kubectl -n 4{WORKSPACE_NAMESPACE} patch helmrelease <HELMRELEASE_NAME> --type='json' -p='[{"op": "replace", "path": "/spec/suspend", "value": false}]'

CODE