Assignment 0 Introduction to Docker

Please feel free to skip this section if you are working on the MathLab machines @mathlab.utsc.utoronto.ca as they already have all the necessary software components installed.

Introduction

Docker runs our applications in a lightweight container by using the image that has all the necessary software components installed. Think of it as conda or virtualenv (both are used to as an containerized Python environment), but at a broader scope for almost all software libraries. Compared with the virtual machines, Docker has negligible performance overhead, making itself an ideal solution for software development.

Instructions

Docker

  • Download and Install Docker: Windows, Mac, Ubuntu

  • Build the Docker image:

    • An image is an isolated environment that includes all necessary software components.

    # Build the docker image from file Dockerfile and tag (-t) it with name `cscd70:2023S`.
cd Assignment0-Introduction_to_Docker
docker build . -t cscd70:2023S

  • Create a Docker container:

    • A container is an instantiation of an image. Multiple containers can be created out of the same image.

    # Create a container for the first assignment.
cd Assignment1-Introduction_to_LLVM/FunctionInfo
docker run -it -v $(pwd):/mnt --rm --name cscd70_a1 cscd70:2023S

    • -t: Allocate a pseudo terminal; -i: Connect to STDIN (i.e., keyboard). -it must be used jointly for an interactive bash shell.

    • -v: Mount the current working directory (i.e., FunctionInfo) to /mnt in the container. Any changes made to the mounted directory outside the container will be reflected in the container as well (and vice versa). The mounted directory must be in absolute path.

    • --rm: [Optional] Automatically clean the container when we exit.

    • --name: [Optional] Name of the container

    • The image name (i.e., cscd70:2023S) should always come last.

  • Run the experiments:

    cd /mnt
mkdir build && cd build
cmake .. && make

Docker-Compose

Docker-Compose is a simple wrapper on top of the docker commands. There is a docker-compose file provided in the root folder of this repository. We can do the same things as we have previously described, but much simpler:

# Install docker-compose via python-pip.
pip3 install docker-compose

docker-compose build dev     # Build the image.
docker-compose run --rm dev  # Create a container.