[Feature]: Add support for OpenShift/OCP guidance in DCM partitioning doc #428
Description
Suggestion Description
DCM partitioning doc lacks OpenShift (OCP) procedure and examples
Doc/page: Applying Partition Profiles (https://instinct.docs.amd.com/projects/gpu-operator/en/latest/dcm/applying-partition-profiles.html)
Problems:
- The procedure reads like vanilla Kubernetes and omits OpenShift-specific details needed to succeed on OCP.
- Examples reference kube-amd-gpu, but on OpenShift the operator is commonly deployed in openshift-amd-gpu.
- Taint/toleration guidance is incomplete for OpenShift (NoExecute taint can evict critical pods).
- The doc provides (or implies) tolerations guidance suitable for simple Kubernetes setups, but doesn’t cover OpenShift’s reality, where essential cluster DaemonSets/Deployments may land on GPU nodes. Applying a NoExecute taint can evict important components unless they tolerate it indeed.
Impact: OCP users follow the doc and get stuck or apply changes in the wrong namespace / wrong assumptions about cluster components.
Operating System
No response
GPU
No response
ROCm Component
No response