| name | rhoai-distributed-workload-operations |
| 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 distributed workload operations from the Administer documentation: Kueue ResourceFlavor, ClusterQueue, and LocalQueue quota resources for distributed workloads; NVIDIA GPU quota examples; RDMA setup boundaries; and administrator troubleshooting for RayCluster, RayJob, PyTorchJob, KubeRay, CodeFlare, and Kueue Workload status. Do NOT use for initial distributed workload component installation; use rhoai-distributed-workloads. Do NOT use for RHOAI/Kueue dashboard integration, embedded Kueue migration, or namespace enforcement setup; use rhoai-kueue-workload-management. Do NOT use for GPU enablement; use rhoai-nvidia-gpu-accelerators. Do NOT use for user workflows that run Ray, PyTorchJob, or TrainJob workloads; use rhoai-distributed-workload-workflows. For live cluster diagnostics, pair with env-troubleshoot and the OpenShift safety guard.
|
RHOAI Distributed Workload Operations
Use this skill to manage distributed workload quotas, RDMA readiness, and
administrator troubleshooting on the active product baseline in
docs/PLATFORM_BASELINE.md.
Source Grounding
Read references/source-capture.md before using product configuration details.
Official Red Hat documentation is product authority. This skill adapts the
official Administer chapter to this repo's GitOps and NVIDIA-first demo model.
Demo Policy
For this repo:
- Use NVIDIA GPU resources by default with
nvidia.com/gpu.
- Do not create AMD GPU queue examples unless the user explicitly asks for AMD
coverage.
- Do not use shared cohorts for OpenShift AI 3.4 distributed workload queues.
- Treat RDMA as optional advanced infrastructure. Do not enable it unless the
active environment has supported NVIDIA GPUs, compatible accelerated
networking, and an explicit implementation need.
- Keep Kueue quota resources GitOps-managed once the active implementation
introduces them.
- Use
ocp-ai-workloads for OCP Kueue API behavior, Leader Worker Set
Operator, JobSet Operator, and platform queue/quota references before
treating them as RHOAI-specific guidance.
- Keep user-facing Ray, PyTorchJob, TrainJob, fine-tuning, checkpointing,
monitoring, and troubleshooting workflows in
rhoai-distributed-workload-workflows.
Workflow
- Confirm the active baseline in
docs/PLATFORM_BASELINE.md.
- Confirm distributed workload components were installed with
rhoai-distributed-workloads.
- Confirm RHOAI/Kueue integration and namespace enforcement with
rhoai-kueue-workload-management.
- Read
references/official-doc-extraction.md.
- For quota work, define or review:
ResourceFlavor
ClusterQueue
- project
LocalQueue
- requested CPU, memory, and GPU quotas
- For RDMA work, verify the environment supports the official prerequisites
before authoring MachineConfig, NVIDIA Network Operator, or PyTorchJob
changes.
- For troubleshooting, inspect
Workload.status.conditions.message,
RayCluster.status.conditions.message, KubeRay pod health, and pod events.
- Validate with
references/validation-checklist.md.
References
references/source-capture.md
references/official-doc-extraction.md
references/validation-checklist.md
examples/distributed-workload-operations-patterns.md