Remote container templates with VScode

Templates for some docker configurations with vscode remote containers.

Our typical use case is to launch a container running on a remote host while using vscode locally to edit the source code in the host machine. The templates are configured to mount a remote local folder in the remote host machine.

A list of the templates:

• ./docker_base
  • A basic remote container template
  • Base image: python:3.9.7-slim
• ./docker_glfw
  • A graphics library support (OpenGL + GLFW)
  • Base image: python:3.9.7-slim
  • Display variable is set to host machine's display
• ./docker_pytorch)
  • A GPU-accelerated pytorch environment for a host mahine with CUDA-nabled NVIDIA GPUs.
  • Base image: nvcr.io/nvidia/pytorch:21.10-py3 from NVIDIA GPU Cloud
• ./docker_ximea
  • A Ximea camera-supported image.
  • Base image: python:3.9.7-slim
  • This image downloads Ximea Linux Software Package at the build time.
  • The container currently mounts the entire /dev/bus/usb of the host machine

Requirement

• Visual Studio Code
• Visual Studio Code extensions:
  • Remote - Containers
• Docker Desktop 2.0+
  • (For Windows, WSL2 backend)

How to use

Step 1: Choose a template project

• Choose one of the template projects and open the folder in the vscode editor.

Step 2: Configure a remote host

Create a setting file, .vscode/settings.json, in the template project and specify your remote machine:

Example:

{
    "docker.host": "ssh://[email protected]:22"
}

Step 3: Configure .env file's variables

We make a remote host's folder accessible from the container.

Open .devcontainer/.env, and configure variables to be compatible with your remote machine's variables:

Example:

CONTAINER_USER="username"
CONTAINER_WORKSPACE="/workspace"
HOST_WORKSPACE="/your/host/machine/workspace:"

With this setting, the pipeline will perform the following:

• creates a local user username in the image
• mounts /your/host/machine/workspace in the remote host machine to /workspace in the container when launching.

NOTE: The current hard-code requirement in .env is due to the limitation in the vscode remote container extension where docker-compose is called in the local machine environment instead of inside the remote host machine.

Step 4: Build and Open the container

In the vscode editor, press F1 and choose Remote-Containers: Rebuild and Reopen in Container