rhoai-kueue-workload-management

name	rhoai-kueue-workload-management
metadata	{"author":"rhoai3-demo","version":"1.0.0","platform-family":"rhoai","platform-baseline":"repo","ocp-baseline":"repo","skill-group":"RHOAI Platform"}
description	Use when documenting, reviewing, or rebuilding Red Hat OpenShift AI workload management with Kueue: Red Hat build of Kueue Operator integration, DataScienceCluster kueue managementState, migration from embedded Kueue, namespace queue enforcement with kueue.openshift.io/managed=true, workload queue labels, dashboard enablement with disableKueue=false, Kueue-enabled hardware profile restrictions, default ClusterQueue and LocalQueue behavior, and documented Kueue troubleshooting signals. Do NOT use for initial distributed workload component installation or Ray/Kubeflow Training Operator installation; use rhoai-distributed-workloads. Do NOT use for distributed workload ResourceFlavor, ClusterQueue, LocalQueue, RDMA, or Ray administrator troubleshooting; use rhoai-distributed-workload-operations. Do NOT use for user workflows that run Ray, PyTorchJob, or TrainJob workloads; use rhoai-distributed-workload-workflows. Do NOT use for GPU node scaling or live cluster changes without the OpenShift safety guard.

RHOAI Kueue Workload Management

Use this skill to manage RHOAI integration with the Red Hat build of Kueue for the active baseline in docs/PLATFORM_BASELINE.md.

Source Grounding

Read references/source-capture.md before using product behavior details. Official Red Hat documentation is product authority. This skill adapts the official Kueue workload-management chapter to this repo's GitOps review model.

Scope

This skill covers:

• RHOAI integration with the Red Hat build of Kueue Operator
• DataScienceCluster.spec.components.kueue.managementState
• migration from deprecated embedded Kueue to Red Hat build of Kueue Operator
• queue enforcement on namespaces labeled kueue.openshift.io/managed=true
• workload kueue.x-k8s.io/queue-name label requirements
• dashboard enablement through OdhDashboardConfig
• Kueue-enabled hardware profile restrictions
• default ClusterQueue and namespace LocalQueue behavior
• documented Kueue troubleshooting patterns

This skill does not cover distributed workload quota resource authoring, ResourceFlavor, ClusterQueue, or LocalQueue schema beyond the RHOAI integration points. Use rhoai-distributed-workload-operations for the official RHOAI distributed workload quota chapter and use Red Hat build of Kueue documentation only as supplemental object detail. Use ocp-ai-workloads for OCP Red Hat build of Kueue Operator installation, Kueue API behavior, Leader Worker Set Operator, and JobSet Operator platform guidance.

Use rhoai-distributed-workload-workflows for data scientist workflows that submit Ray, PyTorchJob, TrainJob, fine-tuning, checkpointing, and monitoring workloads after queue policy is in place.

Demo Policy

For this repo:

• Use Unmanaged Kueue in DataScienceCluster so RHOAI integrates with the Red Hat build of Kueue Operator.
• Do not use deprecated embedded Kueue Managed mode.
• Do not install embedded Kueue and the Red Hat build of Kueue Operator on the same cluster.
• Enable Kueue in the dashboard only after the Red Hat build of Kueue Operator, cert-manager, and quota resources are ready.
• Label existing demo namespaces explicitly with kueue.openshift.io/managed=true when they should enforce queue usage.
• Ensure dashboard-created workloads use hardware profiles with a local queue.
• Do not use node selectors or tolerations in Kueue-enabled hardware profiles; use resource flavors and queues for placement.
• Keep queue and quota objects GitOps-managed when the active implementation introduces them.

Workflow

1. Confirm the active baseline in docs/PLATFORM_BASELINE.md.
2. Confirm installation prerequisites with rhoai-distributed-workloads.
3. Read references/official-doc-extraction.md.
4. Decide whether the task is:
   • activating Kueue integration
   • enabling Kueue in the dashboard
   • enabling queue enforcement for projects
   • creating Kueue-enabled hardware profile guidance
   • troubleshooting queue or webhook failures
   • migrating from embedded Kueue
5. Use examples/kueue-workload-management-patterns.md for review patterns.
6. Validate with references/validation-checklist.md.

References

• references/source-capture.md
• references/official-doc-extraction.md
• references/validation-checklist.md
• examples/kueue-workload-management-patterns.md