README.md

Docker container creation

This tutorial covers the basic concepts of how to create your own Docker container and get it running for your runai project.

Steps:

1. Create a folder: Within your home directory, create a folder to store your Dockerfile and related files for container creation.

2. Create a Dockerfile: Use your preferred editor to create a Dockerfile specifying the desired software available in your container. Here's a recommended approach:

   • Start by pulling an existing container (e.g., Nvidia TensorFlow container).
   • Install any supplementary programs required.

   Basic Template:

   FROM nvcr.io/nvidia/tensorflow:22.11-tf2-py3
   
   ARG USER_ID
   ARG GROUP_ID
   ARG USER
   
   RUN addgroup --gid $GROUP_ID $USER
   RUN adduser --disabled-password --gecos '' --uid $USER_ID --gid $GROUP_ID $USER
   WORKDIR /nfs/home/$USER
   EXPOSE 8888
   
   # (Optional): If you have additional files the Dockerfile needs to read, place them in the same folder. Before 
   # using them in a command, explicitly copy them into the current directory e.g. arequirements.txt containing packages 
   # to install using pip.
   
   COPY requirements.txt .
   RUN pip3 install -r requirements.txt

Tip

If you are using multiple consecutive RUN statements in your dockerfile, you can chain them together using && This reduces both the memory size and build time of the subsequent dockerfile.

E.g., for the previous together file:

   RUN addgroup --gid $GROUP_ID $USER && adduser --disabled-password --gecos '' --uid $USER_ID --gid $GROUP_ID $USER

3. Build and push the image: Create a bash script containing the following commands to build and push your image to the registry (assuming the script is in the same directory as your Dockerfile named Dockerfile):

   #!/bin/bash
   
   # Create a tag or name for the image
   docker_tag="your_image_name:tag"
   
   # Build the image using your Dockerfile and arguments
   docker build . -f Dockerfile \
     --tag ${docker_tag} --network=host \
     --build-arg USER_ID=$(id -u) --build-arg GROUP_ID=$(id -g) --build-arg USER=${USER}
   
   docker push ${docker_tag}

   Run this bash script to build and push the image.

4. Verify image creation: Use docker images to confirm successful image creation.

5. Run a job with your custom image: You can now use your custom image just like a standard container by specifying its name in the --image argument of the runai command.