Introduction

Red Hat OpenShift on DGX User Guide is provided as a companion document to the official Red Hat OpenShift documentation. It provides additional information for installing and configuring OpenShift 4 with Red Hat CoreOS on clusters incorporating DGX worker nodes.

Red Hat OpenShift is an Enterprise-grade container-management solution based on Kubernetes for automating deployment, scaling, and management of containerized applications. It is developed and supported by Red Hat and includes additional security features and tooling for managing complex infrastructures on-premises as well as in hybrid cloud installations.

Red Hat OpenShift 4 is a major release upgrade from version 3 incorporating many technologies from the acquisition of CoreOS. It follows a new paradigm where systems are always reimaged with the latest version with only minimal provisioning. At its core are the immutable Red Hat CoreOS (RHCOS) system images based on Red Hat Enterprise Linux 8. All additional software, drivers, and configuration are ephemeral and provided through kubernetes primitives, such as containers, deployments, and operators. This includes the NVIDIA GPU operator for supporting NVIDIA GPUs and the NVIDIA Network Operator for the ConnectX network interfaces.

While OpenShift 4 still supports Red Hat Enterprise Linux 7 and 8 on the worker nodes, customers are advised to move to the newer Red Hat CoreOS deployments for improved supportability. Refer to the corresponding installation instructions for Red Hat Enterprise Linux on DGX and the OpenShift documentation when you are not planning to use Red Hat CoreOS on DGX.

This user guide provides additional information for installing and configuring OpenShift 4 with Red Hat CoreOS on clusters incorporating DGX worker nodes. It should be seen as a companion document to the official Red Hat OpenShift documentation. The following chapters describe additional configuration steps and best practices that are specific to NVIDIA DGX^TM systems. Refer to the OpenShift Container Platform Documentation for generic information about OpenShift and installation instructions.

Customer Support

Customer support for running OpenShift on DGX systems is provided by Red Hat for OpenShift and NVIDIA for the DGX platform and drivers. For CoreOS and OpenShift support, visit the Red Hat Enterprise Support website: https://www.redhat.com/en/services/support. For DGX hardware, firmware / drivers, or NGC application issues, visit the NVIDIA Enterprise Support website : https://www.nvidia.com/en-us/support/enterprise/

Additional Documentation

Refer to the following documents for additional Information: