docker

code
analysis
Author

Vincenzo Palazeti

Published

December 2, 2023

Two stage Dockerfile. Runner image is slim with only virtual environment.

FROM ubuntu:20.04 AS builder-image

# avoid stuck build due to user prompt
ARG DEBIAN_FRONTEND=noninteractive

RUN apt-get update && \
    apt-get install --no-install-recommends -y python3.9 python3.9-dev python3.9-venv python3-pip python3-wheel build-essential && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

# create and activate virtual environment
# using final folder name to avoid path issues with packages
RUN python3.9 -m venv /home/myuser/venv
ENV PATH="/home/myuser/venv/bin:$PATH"

# install requirements
COPY app/requirements.txt .
RUN pip3 install --no-cache-dir wheel
RUN pip3 install --no-cache-dir -r requirements.txt

FROM ubuntu:20.04 AS runner-image
RUN apt-get update && \
    apt-get install --no-install-recommends -y python3.9 python3-venv && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

RUN useradd --create-home myuser
COPY --from=builder-image /home/myuser/venv /home/myuser/venv

USER myuser
RUN mkdir /home/myuser/code
WORKDIR /home/myuser/code
COPY app .

# make sure all messages always reach console
ENV PYTHONUNBUFFERED=1

# activate virtual environment
ENV VIRTUAL_ENV=/home/myuser/venv
ENV PATH="/home/myuser/venv/bin:$PATH"

I’ve been told installing python manually is not worth the effort & nudged to use the official docker python images.

This is good advice, because I have struggled with C package installs (specifically for LGBM).

Create a container & attach a volume to the image. This command opens the container in interactive mode, mounts the /app directory as a volumne, and links the port 8080.

This is useful because changes to application, or whatever you are working on, are reflected inside of the docker container.

I have been using this with streamlit, but I believe it should work with jupyter. Linking to jupyter through a docker container is a pain, so maybe not. I’ll have to check.

docker run -it --rm -v $(pwd)/app:/home/myuser/code -p 8080:8080 img_name

If your environment requires variables you can pass them through with --env-file

docker run --env-file .env-local

With the format:

DS_BUCKET=XXXX
AWS_ACCESS_KEY_ID_DEV=XXXX
AWS_SECRET_ACCESS_KEY_DEV=XXXX