Hacklab

Self-hosted Kubernetes CDE for ML with GPU support, including JupyterHub, Monitoring, and a codespace.

Directory Structure

.
├── jupyterhub
│   ├── flake.nix
│   ├── secrets.nix
│   └── values
│       ├── configmap.yaml
│       ├── deployment.yaml
│       └── services.yaml
├── monitoring
│   ├── flake.lock
│   ├── flake.nix
│   ├── README.md
│   ├── secrets.nix
│   └── values
│       ├── alertmanager
│       ├── grafana
│       └── prometheus
└── README.md

JupyterHub

• Contains configurations for deploying JupyterHub on Kubernetes.
  • flake.nix: Nix flake that fetches utils & scoped secrets.
  • secrets.nix: Where your secrets templating definiton goes.
  • values: Kubernetes YAML files for JupyterHub.
    • configmap.yaml
    • deployment.yaml
    • services.yaml

Monitoring

• Contains configurations for monitoring stack including Prometheus and Grafana.
  • flake.nix: Nix flake that fetches utils & scoped secrets.
  • README.md: Documentation for setting up and managing the monitoring stack.
  • secrets.nix: Where your secrets templating definiton goes.
  • values: Kubernetes YAML files for monitoring components.
    • alertmanager
    • grafana
    • prometheus

Codespace

• Contains configurations for deploying Coder.com as a codespace.
  • flake.nix: Nix flake that fetches utils & scoped secrets.
  • secrets.nix: Where your secrets templating definiton goes.
  • values: Kubernetes YAML files for Coder.com.
    • configmap.yaml
    • deployment.yaml
    • services.yaml

Getting Started

Prerequisites

• Kubernetes cluster with GPU support
• Nix package manager

Setup

1. Clone the repository:

  git clone https://github.com/heshdotcc/hacklab.git && cd $_