XML

Word

Printable

Epic Goal

We have a fully supported way to use NVIDIA GPUs on MicroShift, but for rpm{} based deployments. We expect the majority of edge deployments to be ostree based (aka: RHEL/Edge).
The goal of this epic is to provide a fully supported way on how to embed the necessary drivers, modules, device plugins etc. into an ostree based image.
The focus is for X86 first, but we should keep in mind that the same approach should work on ARM in the long term.

Why is this important?

Machine Vision at the Edge is a key use case. We have several customers on the MicroShift early access program implementing this

An admin follows the provided instructions (e.g. adding the necessary packages, manifests, containers etc. to the image builder blueprint). The resulting image can be installed on an edge device with an NVIDIA GPU, and the gpu workload works without any further interventions.

CI - MUST be running successfully with tests automated
Release Technical Enablement - Provide necessary release enablement details and documents.
...

There are some manual post-installation steps in the RPM instructions (e.g. disabling noveau). How do integrate these into ostree based? Post-Installation kickstart might not be a good idea, as not all installation methods might use kickstart (e.g. simplified installer). We might need to create a new package (e.g. microshift-nividia-gpu) which adds the necessary systemd scripting.

CI - CI is running, tests are automated and merged.
Release Enablement <link to Feature Enablement Presentation>
DEV - Upstream code and tests merged: <link to meaningful PR or GitHub Issue>
DEV - Upstream documentation merged: <link to meaningful PR or GitHub Issue>
DEV - Downstream build attached to advisory: <link to errata>
QE - Test plans in Polarion: <link or reference to Polarion>
QE - Automated tests merged: <link or reference to automated tests>
DOC - Downstream documentation merged: <link to meaningful PR>