Python, pip and virtual environments on an Ubuntu 24.04 VPS

Ubuntu 24.04 ships Python 3.12, but installing packages with plain pip now fails with an "externally-managed-environment" error. The fix is to isolate each project in a virtual environment. This guide shows how to prepare your VPS Linux Infrawire, manage dependencies properly and keep an application running as a systemd service.

📋 Prerequisites

• A VPS running Ubuntu 24.04 or Debian 12 with root SSH access (IP and root password are in your customer area)
• A Python project or a simple script to test with
• Basic command-line knowledge

1. Install Python 3, pip and venv

Python 3 is usually preinstalled, but pip and the venv module are separate packages.

Bash
1apt update
2apt install -y python3 python3-pip python3-venv python3-dev build-essential
3python3 --version
4pip3 --version

2. Understand the PEP 668 error

Running pip install requests as root on Ubuntu 24.04 or Debian 12 prints:

TEXT
error: externally-managed-environment

This protection (PEP 668) stops pip from overwriting files owned by apt and breaking system tools. Do not bypass it with --break-system-packages. Use one of the three clean options below.

3. Create a virtual environment

A virtual environment is a private copy of Python and pip for one project.

Bash
1mkdir -p /opt/myapp && cd /opt/myapp
2python3 -m venv venv
3source venv/bin/activate
4pip install --upgrade pip
5pip install requests flask gunicorn

The prompt now starts with (venv). Leave the environment with deactivate.

4. Manage dependencies with requirements.txt

Freeze the exact versions so the project can be rebuilt identically on another server.

Bash
1pip freeze > requirements.txt
2# On a fresh machine
3python3 -m venv venv
4source venv/bin/activate
5pip install -r requirements.txt

5. Install command-line tools with pipx

For standalone tools (httpie, yt-dlp, black...), pipx gives each one its own hidden virtual environment and exposes the command globally.

Bash
1apt install -y pipx
2pipx ensurepath
3pipx install httpie

Open a new shell afterwards so the updated PATH is loaded. For a library that has a Debian package, apt install python3-requests is also a valid choice.

6. Run the application as a systemd service

Create a dedicated unprivileged user, then a unit that starts the interpreter from the virtual environment (here Gunicorn serving a Flask object named app in app.py).

Bash
1useradd --system --home /opt/myapp --shell /usr/sbin/nologin myapp
2chown -R myapp:myapp /opt/myapp

Create /etc/systemd/system/myapp.service:

INI
1[Unit]
2Description=My Python application
3After=network.target
4
5[Service]
6User=myapp
7WorkingDirectory=/opt/myapp
8ExecStart=/opt/myapp/venv/bin/gunicorn --bind 127.0.0.1:8000 app:app
9Restart=on-failure
10RestartSec=5
11
12[Install]
13WantedBy=multi-user.target

Bash
1systemctl daemon-reload
2systemctl enable --now myapp

Put a reverse proxy such as nginx in front of port 8000 to expose it on the Internet.

✅ Verify

Bash
1which python   # must point inside /opt/myapp/venv when activated
2pip list
3systemctl status myapp --no-pager
4journalctl -u myapp -n 50 --no-pager
5curl -I http://127.0.0.1:8000

🔧 Troubleshooting

The virtual environment was not created successfully because ensurepip is not available : the venv package is missing. Install it with apt install python3-venv (or python3.12-venv) and run the command again.

status=203/EXEC in systemd : the path in ExecStart is wrong or the file is not executable by the service user. Check it with ls -l /opt/myapp/venv/bin/gunicorn and the ownership set earlier.

ModuleNotFoundError only under systemd : the service uses the system Python instead of the virtual environment. Always call /opt/myapp/venv/bin/python or the binary inside venv/bin in ExecStart.

pip: command not found : use python3 -m pip or activate the virtual environment first.

---

🎉 Your Python projects are now isolated, reproducible and running as a reliable service!