#Install Alauda GPU Management

#Prerequisites

• Cluster administrator access to the target workload cluster where Alauda GPU Management will be installed.
• ACP v4.0-v4.3.
• NVIDIA GPU worker nodes, x86_64 architecture.
• Alauda Build of Node Feature Discovery installed on the target workload cluster.
• Driver mode selected before creating the ClusterPolicy instance.
• Access to the Alauda GPU Management package from Customer Portal.

See Versions and Components and Compatibility for supported combinations, and Prerequisites for the full node checklist.

#Prepare packages

Alauda GPU Management is a controller-based (OLM) operator. After it is installed, the ClusterPolicy instance controls which components are reconciled on the GPU nodes. The driver, container toolkit, device plugin, GPU Feature Discovery, and DCGM-Exporter — together with a driver manager that sequences driver rollout — are included in the product package. The SUSE Linux Micro precompiled driver tuple validated for this release also ships in the package; other kernel tuples and the upstream mutable-OS driver are prepared separately (see Configure the driver mode).

#Choose a driver mode

Choose the driver mode before creating the ClusterPolicy instance. This is the one decision that changes what you prepare and how the driver is delivered.

In preinstalled-driver mode, the operator does not create the driver DaemonSet. The host driver must already be installed and healthy on each GPU node before you enable the device plugin, runtime integration, or metrics.

The mechanics of each mode — image preparation, air-gap handling, and the driver spec fields — are documented once in Configure the driver mode. Complete that page for your mode before creating the ClusterPolicy.

#Installation flow

#1. Prepare the driver according to the selected mode

Complete Configure the driver mode for the mode chosen above. This includes importing the precompiled tuple for your node kernel, or mirroring the upstream driver image, or confirming the host driver is present.

#2. Allow driver and operand images in the cluster image policy

If ACP enforces image whitelist policies and the operator manages the driver, add every required image tag to an ImageWhiteList in the target workload cluster. Create the whitelist in the cluster that runs the GPU nodes and the driver DaemonSet.

Use one exact rule for each supported image tag; do not replace the tag with a broad wildcard. Apply the rules before creating or updating the ClusterPolicy instance, and recheck them after a product upgrade.

Skip this step in preinstalled-driver mode unless another enabled component image is blocked by the cluster image policy.

#3. Upload packages

Download and upload these packages from Customer Portal:

• Alauda GPU Management operator package.
• Alauda Build of Node Feature Discovery cluster plugin package.

For the package upload workflow, see Upload Packages.

#4. Install Node Feature Discovery

Install Alauda Build of Node Feature Discovery from Administrator > Marketplace > Cluster Plugins.

NFD is the shared node discovery layer for the cluster. It provides OS, kernel, architecture, and PCI hardware labels that the GPU installation depends on. Alauda GPU Management does not deploy NFD itself — it relies on the labels NFD publishes to detect GPU nodes — so NFD must be installed as a separate prerequisite. Install it once on the target workload cluster before creating the ClusterPolicy instance; without it, GPU nodes are not detected and no GPU components are reconciled.

#5. Install Alauda GPU Management

1. Go to Administrator > Marketplace > OperatorHub.
2. Switch to the target workload cluster and open Alauda GPU Management.
3. Click Install.
4. Keep the default namespace unless your deployment plan requires a different one. Commands in this guide use the default; if you select another namespace, replace it accordingly.
5. Select Manual upgrade strategy for production environments.
6. Wait until the operator subscription succeeds and the controller pods are running.

#6. Create a ClusterPolicy instance

1. Open Installed Operators > Alauda GPU Management.
2. Open the ClusterPolicy tab.
3. Click Create ClusterPolicy.
4. Configure the components that should be managed by the operator. The default form is set for the precompiled driver flow validated for this release.

Common component switches:

For the precompiled-driver mode, set the driver repository and revisioned version on ClusterPolicy.spec.driver as described in Configure the driver mode; the operator resolves the kernel-matched image. The default form does not auto-deploy a separate driver CR that might not match your node kernel.

#Next step

After the instance is created, continue with Verification.