JupyterHub 5 on Ubuntu 24.04

JupyterHub 5 on Ubuntu 24.04 by cloudimg

JupyterHub 5.4.6 on Ubuntu 24.04 LTS (Noble Numbat), purpose-built for Microsoft Azure and maintained by cloudimg. JupyterHub is the de facto open-source multi-user platform for hosting Jupyter notebooks at organisations: data science teams, classrooms, research labs, and engineering groups all use it to share a single VM among many users while keeping each user's notebooks, files, and kernels isolated.

Why Choose cloudimg?

• 24/7 Expert Support with guaranteed 24-hour response and 1-hour critical-issue average. Contact support@cloudimg.co.uk
• Production-Ready from Launch — security-patched, validated, and pre-configured before publication
• Azure-Native Integration — Azure Linux Agent, cloud-init, Gen2 Hyper-V boot
• Per-VM Admin Credential Generation — `jupyter` user password rotated uniquely on first boot, written to `/etc/cloudimg-credentials.txt` (mode 0600). No two virtual machines ever share an admin password.
• Websocket-Aware Reverse Proxy — nginx is pre-configured with `proxy_http_version 1.1`, `Upgrade`/`Connection` headers, and `proxy_buffering off` so kernel comms work end-to-end without manual tuning

What is Included

• JupyterHub 5.4.6 installed into a Python 3.12 venv at `/opt/jupyterhub` (no system pip pollution)
• JupyterLab + classic Notebook UIs available; default landing is JupyterLab via `c.Spawner.default_url = '/lab'`
• configurable-http-proxy (Node.js) — JupyterHub's required routing component
• PAMAuthenticator using Linux system users — log in with the `jupyter` user and the per-VM password
• nginx reverse proxy on port 80 → 127.0.0.1:8000 with websocket Upgrade for kernel channels
• jupyterhub-5-firstboot.service systemd one-shot rotating the `jupyter` password and writing credentials
• jupyterhub.service systemd unit auto-starting on boot
• Ubuntu 24.04 LTS base with latest security patches at build time
• Azure Linux Agent for SSH key injection and seamless cloud integration

Use Cases

• Single-VM multi-user JupyterLab for small data-science teams
• Educational notebook server (classroom labs, training, workshops)
• Engineering shared analysis environment with persistent kernels and files
• Research-lab compute server for data scientists who don't want a Kubernetes deployment
• Replacement for shared SaaS notebook services where data residency matters

Technical Specifications

• Operating System: Ubuntu 24.04 LTS (Noble Numbat)
• JupyterHub Version: 5.4.6
• Python Runtime: Python 3.12 (system) + venv at /opt/jupyterhub
• Authenticator: PAMAuthenticator (Linux system users)
• Default User: azureuser (sudo, OS); jupyter (JupyterHub admin)
• Web UI Port: 80 (nginx) → 8000 (JupyterHub)
• Service Management: systemd (jupyterhub.service, jupyterhub-5-firstboot.service)
• Recommended Size: Standard_B2s for development; Standard_D4s_v3 or larger for multiple concurrent users
• VM Generation: Hyper-V Gen2 with UEFI boot

Notes on Hardening for Multi-User Production

The default PAMAuthenticator authorises only the `jupyter` user. To add more users, create them as Linux system users (`useradd -m -s /bin/bash <name>`, set a password via `chpasswd`) and add their usernames to `c.Authenticator.allowed_users` in `/etc/jupyterhub/jupyterhub_config.py`, then `systemctl restart jupyterhub.service`. For external authentication (GitHub, Google, OIDC), install the `oauthenticator` package and switch `c.JupyterHub.authenticator_class`.

Support

cloudimg provides 24/7/365 expert technical support. Contact support@cloudimg.co.uk or visit www.cloudimg.co.uk.