kueue

Automate Kubernetes workload admission and scheduling with queue-based quota management.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/tylertitsworth/skills --skill kueue
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kueue
Source: https://github.com/tylertitsworth/skills/tree/main/kueue
Command: npx skills add https://github.com/tylertitsworth/skills --skill kueue

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Manage and schedule Kubernetes workloads (batches, serving, and ML jobs) across clusters and teams, ensuring fair resource sharing and controlled admission.

Core Features & Use Cases

  • Kubernetes-native queueing for quotas and preemption
  • Cohorts, borrowing, and multi-tenant configurations for shared clusters
  • Submitting and admitting a variety of workload types (Job, RayJob, PyTorchJob, Deployments, etc.)

Quick Start

Create ClusterQueues, LocalQueues, and ResourceFlavors; then submit a workload labeled for a queue.

Frequently Asked Questions about kueue

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

FAQPage Schema
How do I manage Kubernetes quotas for multi-tenant workloads?

Kubernetes batch workload admission is controlled via queue-based quota management. You create ClusterQueues, LocalQueues, and ResourceFlavors, then label submitted jobs to match a queue for controlled admission and scheduling.

Can I use Kubernetes scheduling for both batch jobs and ML training?

Kubernetes scheduling handles batch jobs and ML training through queue-based admission. It supports submitting and admitting various workload types including standard Jobs, RayJobs, PyTorchJobs, and Deployments via configured LocalQueues.

Do I need custom resource definitions to run Kubernetes job queuing?

Custom resource definitions are required to run Kubernetes job queuing. You need proper CRDs installed, a running Kueue controller, and configured quotas, ResourceFlavors, and admission checks to operate the scheduling system.

What is the best way to share GPU resources across teams in Kubernetes?

Sharing GPU resources across teams in Kubernetes is managed through cohorts and flavor-based resource assignment. This enables fair resource sharing, borrowing, and multi-tenant configurations across shared clusters using ClusterQueues.

How does preemption work in Kubernetes workload scheduling?

Preemption in Kubernetes workload scheduling operates through queue-based quota management. When resources are constrained, ClusterQueues can preempt lower-priority workloads to admit new jobs, ensuring controlled admission and fair resource sharing across tenants.