gke-compute-classes

Name: gke-compute-classes
Availability: InStock
Author: gke-labs

Configure GKE ComputeClasses for node pool auto-creation and workload scheduling.

42|32|Updated May 7, 2026

Search Tags:#kubernetes #infrastructure #cost-optimization #autoscaling #gke #computeclass

Ready to run

npx skills add https://github.com/gke-labs/kube-agents --skill gke-compute-classes-gke-labs

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: gke-compute-classes
Source: https://github.com/gke-labs/kube-agents/tree/main/agents/platform/skills/gke-compute-classes
Command: npx skills add https://github.com/gke-labs/kube-agents --skill gke-compute-classes-gke-labs

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) and assets (resource) components.

What problem does it solve?

This Skill addresses the complexity of managing GKE node provisioning, helping you avoid resource stockouts, optimize costs through intelligent fallback, and resolve scheduling failures for specialized workloads.

Core Features & Use Cases

Intelligent Provisioning: Automates the configuration of GKE ComputeClasses to handle Spot VM fallbacks, GPU/TPU targeting, and machine family selection.
Cost & Performance Optimization: Provides guardrails for aligning infrastructure with Committed Use Discounts (CUDs) and managing active migration to cost-effective nodes.
Use Case: When your AI/ML training job fails due to GPU unavailability, use this Skill to design a robust ComputeClass that automatically falls back from reserved capacity to on-demand and spot instances across multiple zones.

Quick Start

Use the gke-compute-classes skill to generate a ComputeClass YAML template for a high-performance N4 machine family with spot instance fallback and proper GPU tolerations.

Frequently Asked Questions about gke-compute-classes

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I configure GKE ComputeClasses for automatic Spot VM fallback?▼

GKE ComputeClasses automate Spot VM fallback by defining intent-driven provisioning rules that transition workloads to on-demand or spot instances across multiple zones when reserved capacity is unavailable.

What is a GKE ComputeClass and when do I need it for workload scheduling?▼

A GKE ComputeClass manages node pool auto-creation and workload scheduling. You need it when targeting specific hardware like GPUs or isolating stateful workloads to prevent resource stockouts.

Can I use GKE ComputeClasses to align node provisioning with Committed Use Discounts?▼

Yes, GKE ComputeClasses provide guardrails to align infrastructure provisioning with Committed Use Discounts (CUDs), enabling active migration to cost-effective nodes while enforcing security and resource constraints.

How do I resolve GKE scheduling failures for AI/ML training jobs requiring GPUs?▼

Resolve GKE GPU scheduling failures by designing a robust ComputeClass that targets accelerators and automatically falls back from reserved capacity to on-demand and spot instances across multiple zones during unavailability.

Does GKE ComputeClass support high-performance N4 machine families with spot fallback?▼

Yes, you can generate a ComputeClass YAML template to provision high-performance N4 machine families with spot instance fallback, applying proper GPU tolerations and managing stateful workload isolation.

Why do my GKE workloads experience resource stockouts during node auto-creation?▼

Resource stockouts during GKE node auto-creation often occur due to unmanaged provisioning; ComputeClasses solve this by applying intelligent fallback logic and machine family selection to optimize availability and costs.