How does GPU passthrough work in these clusters?

The skill automates the configuration of the NVIDIA device plugin, allowing your Kubernetes pods to access host GPU resources for machine learning and compute tasks.

How do I update an existing cluster configuration?

Running the config command on an existing instance will recreate the cluster with your new settings, though you should export important resources first as data is not preserved.

Can I run multiple Kubernetes clusters simultaneously?

Yes, the k3d skill supports multi-instance management, allowing you to create and run several clusters on different ports using the --instance flag.

Can k3d pods communicate with other Bazzite AI services?

Yes, clusters are automatically joined to the bazzite-ai network, enabling DNS-based service discovery for tools like Ollama, JupyterLab, and ComfyUI.

Do I need to install kubectl separately?

The skill provides an integrated shell command to run kubectl within the cluster context, though you can also use a local kubectl by referencing the generated kubeconfig.

k3d Kubernetes Manager

Name: k3d Kubernetes Manager
Author: atrawog

byatrawog

0•

デプロイメントとDevOps

Manages lightweight Kubernetes clusters using k3d and Podman with integrated GPU support and service discovery.

The k3d skill provides a comprehensive interface for managing k3s Kubernetes clusters within the Bazzite AI ecosystem. It allows developers to create, configure, and scale containerized Kubernetes environments directly on Podman, featuring seamless NVIDIA GPU passthrough for machine learning workloads and DNS-based service discovery to connect with other AI tools like Ollama and JupyterLab. This skill is ideal for local development, testing k8s-based applications, and running GPU-accelerated microservices without the overhead of a full cloud environment.

主な機能

01Automated service discovery within the bazzite-ai network

02Lightweight k3s cluster orchestration via Podman containers

030 GitHub stars

04Integrated shell for direct kubectl command execution

05NVIDIA GPU passthrough for hardware-accelerated ML workloads

06Multi-instance support for running parallel development environments

ユースケース

01Deploying GPU-accelerated machine learning models in a cluster environment

02Local Kubernetes development and testing of k8s-based applications

03Prototyping microservices that require inter-pod communication with AI tools

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add atrawog/bazzite-ai-plugins k3d

For use in Claude.ai and ChatGPT

Download Skill