| name | ocp-machine-configuration |
| metadata | {"author":"rhoai3-demo","version":"1.0.0","platform-family":"ocp","platform-baseline":"repo","ocp-baseline":"repo","skill-group":"OpenShift Platform"} |
| description | Use when documenting, reviewing, or rebuilding OpenShift Container Platform machine configuration guidance from official OCP documentation: Machine Config Operator, MachineConfig, MachineConfigPool, MachineConfiguration, KubeletConfig, ContainerRuntimeConfig, PinnedImageSet, node disruption policies, boot image management, rendered machine config pruning, image mode for OpenShift, MachineOSConfig, MachineOSBuild, Machine Config Daemon metrics, systemd, CRI-O, kubelet, kernel, NetworkManager, and RHCOS host configuration. Do NOT use for MachineSet or cloud machine lifecycle; use ocp-machine-management. Do NOT use for pod scheduling and node operations; use ocp-nodes. Do NOT run live MCO operations without the env safety guard and explicit approval.
|
OCP Machine Configuration
Use this skill to ground OpenShift Machine Config Operator (MCO), RHCOS host
configuration, and machine config pool rollout guidance in the official
OpenShift Container Platform Machine configuration guide for the active
baseline in docs/PLATFORM_BASELINE.md.
Source Grounding
Read references/source-capture.md before using product behavior. Official Red
Hat documentation is product authority. This skill covers node OS and runtime
configuration after machines exist; it does not replace ocp-machine-management
for MachineSet and cloud machine lifecycle, ocp-nodes for scheduling and node
operations, or RHOAI skills for accelerator and OpenShift AI component
configuration.
Demo Posture
For this AWS-hosted RHOAI demo:
- Treat MCO,
MachineConfig, MachineConfiguration, KubeletConfig,
ContainerRuntimeConfig, PinnedImageSet, MachineOSConfig, and MCP changes
as high-risk platform changes because they can drain or reboot nodes, reload
CRI-O, affect upgrades, or block cluster operators.
- Prefer the default OpenShift host configuration unless the demo has a clear,
official-doc-backed need for a node OS or runtime change.
- Do not modify control plane machine config pools unless the user explicitly
requests a planned control plane operation and
ocp-etcd review is included.
- Keep GPU node behavior aligned with
rhoai-nvidia-gpu-accelerators; do not
use ad hoc MachineConfig tuning as a substitute for NVIDIA GPU Operator,
Node Feature Discovery, or supported RHOAI hardware profile configuration.
- Use GitOps only after verifying API shape and rollout impact. Do not invent
Ignition, Butane, CRI-O, kubelet, systemd, or kernel settings.
Machine Configuration Model
Use the official docs to frame:
- Machine Config Operator: manages RHCOS host configuration and applies
changes to nodes through machine config pools.
- MachineConfig: defines host-level configuration applied to a target role
or pool, including files, systemd units, kernel arguments, extensions, and
related RHCOS settings.
- MachineConfigPool: groups nodes and tracks rendered machine config
rollout, update, degraded, and paused state.
- MachineConfiguration: singleton operator configuration in the
openshift-machine-config-operator namespace, including node disruption
policy configuration.
- KubeletConfig and ContainerRuntimeConfig: supported custom resources
for kubelet and CRI-O settings that generate machine configs.
- PinnedImageSet: an official OCP resource for documented cases where
nodes must preload selected images and avoid garbage collection removing
them. Do not use it as a general repeatability mechanism for this demo.
- Image mode for OpenShift: builds or applies custom layered RHCOS images
with
MachineOSConfig and MachineOSBuild when the documented limitations
and registry requirements are acceptable.
- Machine Config Daemon metrics and logs: indicate node-level update,
drain, pivot, reboot, kubelet, and degraded-state problems.
Workflow
- Confirm the active OpenShift baseline in
docs/PLATFORM_BASELINE.md.
- Read
references/official-doc-extraction.md.
- Identify whether the task concerns:
MachineConfig files, systemd, kernel arguments, extensions, firmware, or
host access configuration
- node disruption policies through
MachineConfiguration
- kubelet settings through
KubeletConfig
- CRI-O settings through
ContainerRuntimeConfig
PinnedImageSet image preloading
- boot image management or rendered machine config pruning
- image mode with
MachineOSConfig or MachineOSBuild
- Machine Config Daemon metrics, logs, drift, pool status, or degraded state
- For GitOps manifests, verify the target pool, labels, CRD fields, and
disruption behavior before writing resources.
- For live operations, use the repo environment guard and pair this skill with
env-manage-resources, env-troubleshoot, or env-deploy-and-evaluate.
- Validate the output with
references/validation-checklist.md.
Related Skills
- Use
ocp-machine-management for Machine API, MachineSet, autoscaling, and
cloud machine lifecycle.
- Use
ocp-nodes for labels, taints, drain, schedulability, node health, and
pod placement.
- Use
ocp-etcd before any control plane or cluster-wide change that could
affect quorum, upgrades, or cluster operator stability.
- Use
ocp-node-feature-discovery for specialized hardware labels.
- Use
rhoai-nvidia-gpu-accelerators for GPU workers, accelerator profiles,
NVIDIA GPU Operator handoff, and RHOAI GPU scheduling intent.
- Use
project-gitops-authoring and project-manifest-review for manifests.
References
references/source-capture.md
references/official-doc-extraction.md
references/validation-checklist.md
examples/machine-config-review-patterns.md