Use a Docker Container as a PyCharm Remote Interpreter

What

This project updates a technique and set of scripts originally posted by Gabe Rives-Corbett.

A simple set of scripts and Dockerfiles, the project allows you to use a Docker container as a remote Python interpreter in the PyCharm integrated development environment (IDE).

Note: While you can probably use this project with other IDEs that support remote Python interpreters, we have only verified it with PyCharm.

This project was developed using Mac OS with Docker for Mac installed.

Why

A Python project should fully isolate its compilers and dependencies from those of other projects, to avoid confusion and mistakes.

What we can avoid by using a Docker container as a PyCharm remote interpreter:

• Any individual project's dependencies interfering with other projects' dependencies.
• Installation of multiple versions of Python in a project's development environment.
• Installation of Python compilers or packages that could interfere with the Host OS Python version.

Other Choices

There are other ways to isolate dependencies and compilers. The simpliest is to make sure the Python path in your project's development environment contains the desired compiler. The python utility virtualenv is a good choice, but it uses Python compilers installed on the Host OS. ValGrind can also be used to create remote Python interpreters.

Prerequisites

Set up SSH access

Pycharm will use ssh to connect to the Docker container, via sshd. The setup script and Dockerfile look for an ssh key file in your /.ssh directory named docker-dev.

• Create the docker_dev key:

ssh-keygen -t rsa -f ~/.ssh/docker_dev -N ""

• Configure your .ssh/config to use the docker_dev key:

Host localhost
    Hostname localhost
    User root
    IdentityFile ~/.ssh/docker_dev

Create a container

The Dockerfile in this repo uses the Python 3.6 DockerHub image as its base. You can edit the Dockerfile to use the base image that best suits your needs.

To create a container, run the make-python-dev.sh script. Pass in arguments to name the Docker image (e.g., proj-one) and the port for the ssh connection (e.g., 5822).

The example command below creates a Docker container named proj-one-5822.

make-python-dev.sh proj-one 5822 Dockerfiles/Dockerfile

You can now use your docker_dev key to ssh into the container:

ssh localhost -p 5822

Note: If the above command doesn't work, there may be issues with the ssh configuration; if this is the case, try the command below:

ssh -i ~/.ssh/docker_dev root@localhost -p 5822

Configure PyCharm

To add this container as the project interpreter:

1. Open the Preferences window.
2. Click on the Project settings.
3. Select Project Interpreter.
4. One the far right side of the project Interpreter line, click the "..." button to open the menu.
5. Click "add remote".

In the next window that pops up:

1. Select "SSH credentials".
2. Enter "localhost" for the hostname.
3. Use the port number you assigned to the Docker container (5822 in the example above).
4. The username is "root".
5. Auth Type is "private key".
6. Set the private key file to ~/.ssh/docker-dev.
7. Leave the passphrass blank.
8. Set the path to the python interpreter to /usr/local/bin/python.

That should be it! PyCharm will upload some support scripts after you save the new configuration. To test your configuration, open the Python Console in PyCharm (bottom Left tab); it should display the Python version used in the Docker container. Now, when you run or debug scripts in PyCharm, it will use your Docker container.

Notes

• The remote interpreter script mounts the path to the home directory of the host OS in the Docker container. The assumption is that your Python project is somewhere in the home directory.
• You'll need to install your project's dependencies in the container using pip or a setup.py script.
• The base docker image for python 3 actually has two versions of python:
  • /usr/bin/python is version 2.7 of Python.
  • /usr/local/bin/python is version 3 of Python.
• The install scripts use apt-get to install sshd. If you change the base docker image, be sure to update the install commands in the script accordingly.