Massed Compute

Jupyter notebooks are essential for data science, research, and interactive Python development. This guide shows you how to deploy a secure, production-ready JupyterLab server on Ubuntu 22.04 with systemd service management and token-based authentication accessible from any browser.

Massed Compute VM Pricing

Step-by-Step Deployment

ssh username@YOUR_VM_IP

sudo apt-get update
sudo apt-get install -y pipx
pipx ensurepath

# --include-deps exposes the jupyter wrapper script
pipx install jupyterlab --include-deps

source ~/.bashrc

jupyter lab --generate-config
TOKEN=$(openssl rand -hex 24)
echo "Your Jupyter token: $TOKEN"

cat <<EOF >> ~/.jupyter/jupyter_lab_config.py
c.ServerApp.ip = "0.0.0.0"
c.ServerApp.port = 8888
c.ServerApp.open_browser = False
c.ServerApp.token = "YOUR_TOKEN"
c.ServerApp.allow_root = False
EOF

sudo tee /etc/systemd/system/jupyter.service > /dev/null <<EOF
[Unit]
Description=JupyterLab
After=network.target

[Service]
Type=simple
User=USERNAME
ExecStart=/home/USERNAME/.local/bin/jupyter lab --config /home/USERNAME/.jupyter/jupyter_lab_config.py
Restart=on-failure

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl daemon-reload
sudo systemctl enable --now jupyter

sudo systemctl status jupyter

# Check service is active
sudo systemctl is-active jupyter

# Verify port 8888 is listening
sudo ss -ltnp | grep 8888

curl -sS -o /dev/null -w "%{http_code}\n" http://YOUR_VM_IP:8888/lab

http://YOUR_VM_IP:8888/?token=YOUR_TOKEN

Troubleshooting

pipx Not Found

If pipx isn’t available in the apt repository, fall back to pip:

pip install --user jupyterlab

Update the systemd ExecStart path to use ~/.local/bin/jupyter instead.

Service Fails to Start

Check the service logs for detailed error information:

sudo journalctl -u jupyter -n 50

Common issues include wrong config file paths or incorrect user settings in the service file.

Token Authentication Issues

If the browser rejects your token, regenerate both the config and token together:

TOKEN=$(openssl rand -hex 24)
echo "c.ServerApp.token = \"$TOKEN\"" >> ~/.jupyter/jupyter_lab_config.py
sudo systemctl restart jupyter

Port 8888 Blocked

Massed Compute instances may filter inbound ports by default. Check your instance settings to open port 8888 for external access.

Skip All of This: Deploy with an AI Agent

This entire guide exists as a tested, machine-readable recipe in the Massed Compute MCP (tested June 2, 2026). Connect an MCP-compatible AI agent with this config:

{
  "mcpServers": {
    "massed-compute": {
      "type": "http",
      "url": "https://vm.massedcompute.com/api/mcp",
      "headers": { "Authorization": "Bearer MC_TOKEN" }
    }
  }
}

Then say:

The agent will match your request against the recipe catalog, provision a VM with the right specs, run the setup and verification steps above, and report back with your server URL and access token. If any step fails, it stops immediately and shows you the exact error.

Quick Setup Reference

For experienced users, here’s the condensed command sequence:

# Install and configure
sudo apt-get update && sudo apt-get install -y pipx
pipx ensurepath && source ~/.bashrc
pipx install jupyterlab --include-deps
jupyter lab --generate-config
TOKEN=$(openssl rand -hex 24)

# Configure server
cat <<EOF >> ~/.jupyter/jupyter_lab_config.py
c.ServerApp.ip = "0.0.0.0"
c.ServerApp.port = 8888
c.ServerApp.open_browser = False
c.ServerApp.token = "$TOKEN"
c.ServerApp.allow_root = False
EOF

# Create and start service
sudo tee /etc/systemd/system/jupyter.service > /dev/null <<EOF
[Unit]
Description=JupyterLab
After=network.target

[Service]
Type=simple
User=$USER
ExecStart=$HOME/.local/bin/jupyter lab --config $HOME/.jupyter/jupyter_lab_config.py
Restart=on-failure

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl daemon-reload
sudo systemctl enable --now jupyter
echo "Access: http://$(curl -s ifconfig.me):8888/?token=$TOKEN"

Frequently Asked Questions