Docker Tutorial: Setting up Conda and PyTorch in Containers

Docker is a powerful platform for containerization that allows developers to package applications and their dependencies into portable, lightweight containers. These containers ensure consistent environments across different systems, eliminating issues like "it works on my machine" by encapsulating the application, libraries, and configurations.

Docker simplifies deployment, enhances reproducibility, and isolates environments, making it ideal for development, testing, and production. By combining Docker with Conda, a package and environment manager, you can create a reproducible Python environment with specific libraries like PyTorch, which can be shared with others, ensuring they can run the same setup without compatibility issues.

This is meant to be a quick start tutotiral, for further explanations and features see: https://www.digitalocean.com/community/tutorials/how-to-install-and-use-docker-on-ubuntu-20-04

Docker Installation

Linux Installation

Ubuntu/Debian

# Update package index
sudo apt-get update

# Install required packages
sudo apt-get install apt-transport-https ca-certificates curl gnupg lsb-release

# Add Docker's official GPG key
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo gpg --dearmor -o /usr/share/keyrings/docker-archive-keyring.gpg

# Set up the stable repository
echo "deb [arch=amd64 signed-by=/usr/share/keyrings/docker-archive-keyring.gpg] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable" | sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

# Install Docker Engine
sudo apt-get update
sudo apt-get install docker-ce docker-ce-cli containerd.io

# Add your user to the docker group to run Docker without sudo
sudo usermod -aG docker $USER

# Start and enable Docker service
sudo systemctl start docker
sudo systemctl enable docker

Windows Installation

1. Download Docker Desktop for Windows from the official Docker website: https://www.docker.com/products/docker-desktop
2. Run the installer and follow the installation wizard
3. Enable WSL 2 (Windows Subsystem for Linux) if prompted - this is recommended for better performance
4. Restart your computer when installation is complete
5. Launch Docker Desktop from the Start menu
6. Verify installation by opening PowerShell or Command Prompt and running:

   docker --version

Creating the Dockerfile

Create a new file named Dockerfile (no extension) in your project directory and add the following content:

# Use an official Ubuntu base image for compatibility and familiarity
FROM ubuntu:20.04

# Set environment variables to avoid interactive prompts during installation
ENV DEBIAN_FRONTEND=noninteractive

# Update package lists and install essential tools
RUN apt-get update && apt-get install -y \
    wget \
    curl \
    git \
    && rm -rf /var/lib/apt/lists/*

# Download and install Miniconda
RUN wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /tmp/miniconda.sh \
    && bash /tmp/miniconda.sh -b -p /opt/conda \
    && rm /tmp/miniconda.sh

# Add Conda to PATH for easy access
ENV PATH=/opt/conda/bin:$PATH

# Initialize Conda for bash
RUN /opt/conda/bin/conda init bash

# Create a Conda environment named 'pytorch_env' with Python 3.8
RUN conda create -n pytorch_env python=3.8 -y

# Activate the Conda environment and install PyTorch
RUN /bin/bash -c "source /opt/conda/bin/activate pytorch_env && \
    conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch -y"

# Set the working directory
WORKDIR /workspace

# Set the default command to open a bash shell in the Conda environment
CMD ["/bin/bash", "-c", "source /opt/conda/bin/activate pytorch_env && bash"]

Building and Running the Container

Build the Docker Image

# Navigate to the directory containing your Dockerfile
cd /path/to/your/dockerfile

# Build the Docker image (replace 'pytorch-conda' with your preferred image name)
docker build -t pytorch-conda .

Run the Container

# Run the container interactively
docker run -it pytorch-conda

# Run with volume mounting to share files between host and container
docker run -it -v $(pwd):/workspace pytorch-conda

# Run with GPU support (if you have NVIDIA GPU and nvidia-docker installed)
docker run --gpus all -it -v $(pwd):/workspace pytorch-conda

Distributing the Docker Environment

Method 1: Export/Import Docker Image

For offline distribution or when Docker Hub isn't suitable:

Export Image

# Save the image to a tar file
docker save -o pytorch-conda.tar pytorch-conda:latest

Import Image

# Load the image from tar file
docker load -i pytorch-conda.tar

# Run the container
docker run -it pytorch-conda:latest

Method 2: Share the Dockerfile

The simplest method is to share just the Dockerfile:

1. Create a repository (GitHub, GitLab, etc.) containing your Dockerfile
2. Add a README with build and run instructions
3. Others can clone and build the image themselves:

   git clone https://github.com/yourusername/pytorch-docker.git
   cd pytorch-docker
   docker build -t pytorch-conda .
   docker run -it pytorch-conda

Verification

To verify that PyTorch is correctly installed in your container:

# Run Python in the container and test PyTorch
docker run -it pytorch-conda python -c "import torch; print(f'PyTorch version: {torch.__version__}'); print(f'CUDA available: {torch.cuda.is_available()}')"

Further Reading

As mentioned at the start, this is meant to be a quick start tutotiral, for further explanations and features see: https://www.digitalocean.com/community/tutorials/how-to-install-and-use-docker-on-ubuntu-20-04