Skip to content

Commit

Permalink
Change AKS to use nodepool cmdline arg
Browse files Browse the repository at this point in the history
Signed-off-by: Mike McKiernan <[email protected]>
  • Loading branch information
mikemckiernan committed Feb 13, 2024
1 parent 1ede270 commit 98e83c4
Show file tree
Hide file tree
Showing 2 changed files with 19 additions and 7 deletions.
5 changes: 5 additions & 0 deletions .codespell_exclude_lines.txt
Original file line number Diff line number Diff line change
Expand Up @@ -12,3 +12,8 @@ After you start your Azure AKS cluster, you are ready to install the NVIDIA GPU
GPU Operator with Azure AKS <microsoft-aks.rst>
* Added support for running the Operator with Microsoft Azure Kubernetes Service (AKS).
You must use an AKS image with a preinstalled NVIDIA GPU driver and a preinstalled
Create AKS Cluster with a Node Pool to Skip GPU Driver installation
command-line argument to the ``az aks nodepool add`` command.
$ az aks nodepool add -g <rg-name> -n gpunodes --cluster-name <cluster-name> \
`Skip GPU driver installation (preview) https://learn.microsoft.com/en-us/azure/aks/gpu-cluster?source=recommendations&tabs=add-ubuntu-gpu-node-pool#skip-gpu-driver-installation-preview`__
After you start your Azure AKS cluster with an image that includes a preinstalled NVIDIA GPU Driver
21 changes: 14 additions & 7 deletions gpu-operator/microsoft-aks.rst
Original file line number Diff line number Diff line change
Expand Up @@ -30,26 +30,33 @@ NVIDIA GPU Operator with Azure Kubernetes Service
Approaches for Working with Azure AKS
*************************************

Create AKS Cluster with Node Pool Tags to Prevent Driver installation
=====================================================================
Create AKS Cluster with a Node Pool to Skip GPU Driver installation
===================================================================

When you create an AKS cluster, you can specify the ``--nodepool-tags`` argument to the Azure CLI
command to customize the nodes.
If you specify ``--nodepool-tags SkipGPUDriverInstall=true``, AKS does not install
Azure Kubernetes Service has a preview feature that enables a ``--skip-gpu-driver-install``
command-line argument to the ``az aks nodepool add`` command.
This argument prevents installing
the NVIDIA GPU Driver in the stock Ubuntu operating system.

This approach enables you to take advantage of the lifecycle management
that the NVIDIA GPU Operator provides for managing your cluster.

.. code-block:: console
:caption: Sample Cluster Create Command
:caption: Sample Node Pool Add Command
$ az aks create -g <rg-name> -n <cluster-name> ... --nodepool-tags SkipGPUDriverInstall=true
$ az aks nodepool add -g <rg-name> -n gpunodes --cluster-name <cluster-name> \
--node-count <n> \
--skip-gpu-driver-install \
...
When you follow this approach, you can install the Operator without any special
considerations or arguments.
Refer to :ref:`Install NVIDIA GPU Operator`.

For more information about this preview feature, see
`Skip GPU driver installation (preview) https://learn.microsoft.com/en-us/azure/aks/gpu-cluster?source=recommendations&tabs=add-ubuntu-gpu-node-pool#skip-gpu-driver-installation-preview`__
in the Azure Kubernetes Service documentation.


Default AKS configuration without the GPU Operator
==================================================
Expand Down

0 comments on commit 98e83c4

Please sign in to comment.