deploy-sunk-on-eks

Automate SUNK deployment on Amazon EKS with cost-optimized defaults.

7|Updated Jul 9, 2026
One-click install
npx skills add https://github.com/coreweave/sunk-anywhere --skill deploy-sunk-on-eks
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deploy-sunk-on-eks
Source: https://github.com/coreweave/sunk-anywhere/tree/main/skills/eks/deploy-sunk-on-eks
Command: npx skills add https://github.com/coreweave/sunk-anywhere --skill deploy-sunk-on-eks

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Automates the deployment of SUNK (Slurm on Kubernetes) on Amazon Elastic Kubernetes Service (EKS), focusing on a sub-$50/day budget. Streamlines cluster creation, configurations, and resource setup for cost-sensitive environments.

Core Features & Use Cases

  • Cluster Setup: Fully automated setup of an EKS cluster with CPU and GPU nodegroups.
  • Optimized for Cost: Defaults to cost-effective configurations; provides explicit opt-out options for optional features to keep costs low.
  • GPU Integration: Option to include GPU compute nodegroups with GPU health checks and monitoring.
  • Monitoring: Integrated monitoring using Kubernetes-native and CloudWatch dashboards for observability.
  • Use Case: A researcher at a university with a limited budget can deploy a cost-effective, high-performance computing environment for running complex simulations.

Quick Start

Run the "deploy-sunk-on-eks" skill and follow the prompts for a cost-effective SUNK deployment on EKS.

Frequently Asked Questions about deploy-sunk-on-eks

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

FAQPage Schema
How do I deploy Slurm on Amazon EKS while keeping costs low?

You can deploy SUNK (Slurm on Kubernetes) on Amazon EKS by automating cluster creation, OIDC+IRSA setup, shared storage configuration, and Helm installation with default settings optimized for minimal spend.

What is SUNK and how does it work for high-performance computing on Kubernetes?

SUNK integrates Slurm job scheduling into Kubernetes clusters, providing a cost-effective high-performance computing environment by combining containerized CPU and GPU nodegroups with shared storage.

Can I run GPU compute nodegroups on Amazon EKS with this Slurm deployment?

Yes, you can include GPU compute nodegroups on Amazon EKS with built-in GPU health checks and monitoring dashboards to ensure high-performance computing workloads run reliably and cost-effectively.

What's the best way to set up a budget-conscious Kubernetes cluster for complex simulations?

Setting up a budget-conscious Kubernetes cluster for complex simulations involves using cost-effective default configurations with explicit opt-out flags for optional features, keeping daily costs under $50.

Do I need to configure shared storage and monitoring separately for Slurm on EKS?

No, shared storage configuration and integrated monitoring using Kubernetes-native and CloudWatch dashboards are handled automatically during the SUNK Helm installation and cluster bootstrap process.

What are the limitations of deploying a sub-$50/day Slurm cluster on Amazon EKS?

The main limitation of a sub-$50/day Slurm cluster on Amazon EKS is that optional features are disabled by default to keep costs low, requiring explicit opt-out flags to manage budget constraints.