Loading Search...
Crater
Deployment

NVIDIA GPU Operator Deployment Guide

Enable GPU support in Kubernetes clusters using NVIDIA GPU Operator and integrate with Crater.

Overview

The NVIDIA GPU Operator automates the deployment and management of all necessary components to support GPUs in Kubernetes clusters.

In Crater, it provides:

  • GPU driver installation
  • NVIDIA container runtime setup
  • dcgm-exporter for GPU monitoring (used by Prometheus stack)
  • Smooth integration with Crater's job scheduling and GPU metrics display

Crater requires GPU Operator to ensure GPU jobs are correctly scheduled and monitored.

Installation

We recommend installing GPU Operator via Helm with Crater’s preconfigured values.

📦 Helm values: deployments/gpu-operator/values.yaml
📖 Detailed guide: deployments/gpu-operator/README.md

Edit on GitHub

CloudNativePG Deployment Guide

Integrate PostgreSQL into Crater using the CloudNativePG Operator with OpenEBS storage.

Harbor Deployment Guide

Crater relies on an internal Harbor registry to support image building, storage, and secure distribution within the cluster.