OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Infrastructure & Serving

Kubernetes Gpu Guide

Kubernetes Gpu Guide is an open-source repository of setup instructions, shell scripts and Kubernetes YAML files for creating a GPU cluster across multiple Ubuntu 16.04 bare-metal servers.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Kubernetes Gpu Guide — OpenViST project profile in AI Infrastructure & Serving.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when planning a self-managed Kubernetes GPU cluster on the Ubuntu 16.04 environment described by the guide, and after checking that its system and network assumptions match your setup.

Who it is for

  • Researchers and hobbyists setting up their own GPU cluster.
  • Operators who can administer Ubuntu servers and want Kubernetes infrastructure for deep-learning training.

Documented uses

  • Initialize a master node and join worker nodes to a Kubernetes cluster.
  • Use a multi-server GPU cluster as infrastructure for more extensive deep-learning training.
Before you choose

Requirements & limits

The guide is publicly available on GitHub and is identified as MIT-licensed.

  • The instructions describe Ubuntu 16.04 bare-metal servers and may not transfer to other operating systems.
  • The listed prerequisites include SSH with a sudo user and internet access on each node, plus specified UDP and TCP ports.
  • The README warns that its Kubernetes guidance may become outdated.
  • Disabling ufw appears in the example setup constraints, but the guide explicitly says this is not recommended.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Initialize a Kubernetes GPU cluster on Ubuntu 16.04

Use the documented fast-track scripts to initialize a master and join worker nodes to it.

Before you start

  • Master and worker servers running Ubuntu 16.04, with SSH access using a sudo user and internet access.
  • The guide lists ufw deactivated for both node types, while warning that this is not recommended.
  • For the master, the guide lists UDP and TCP ports 6443, 443, and 8080; ports 30000–32767 are listed only if applications need them. For workers, it lists UDP and TCP ports 6443 and 443.
  • The corresponding initialization scripts copied to the master and worker nodes.

Steps

  1. On the master, make the initialization script executable and run it with the master IP: chmod +x init-master.sh then sudo ./init-master.sh .

  2. Remember the token produced by master initialization.

  3. On each worker, make its initialization script executable and run it with the master token and master IP plus port, usually 6443: chmod +x init-worker.sh then sudo ./init-worker.sh :.

Expected outcome

The documented process initializes the master and runs worker initialization with the master's token and address.

Guide limitations

  • The setup instructions describe the author's experience on Ubuntu 16.04 and may not transfer to other operating systems.
  • The guide warns that its Kubernetes instructions may become obsolete or change.
  • The guide labels disabling ufw as not recommended.
  • The available instructions do not document a verification step or GPU training execution.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
6 November 2026
Original website title
Langhalsdino/Kubernetes-GPU-Guide · Official homepage title · retained in its source language
How we review sourcesSuggest a correction