Why isn't my .Dockerignore file ignoring files?

9.2k views Asked by At

When I build the container and I check the files that should have been ignored, most of them haven't been ignored.

This is my folder structure.

Root/

  data/

  project/
    __pycache__/
    media/
    static/

  app/
    __pycache__/
    migrations/
    templates/

  .dockerignore
  .gitignore
  .env
  docker-compose.yml
  Dockerfile
  requirements.txt
  manage.py

Let's say i want to ignore the __pycache__ & data(data will be created with the docker-compose up command, when creating the container) folders and the .gitignore & .env files.

I will ignore these with the next .dockerignore file

.git
.gitignore

.docker

*/__pycache__/
**/__pycache__/

.env/
.venv/
venv/

data/

The final result is that only the git & .env files have been ignored. The data folder hasn't been ignored but it's not accesible from the container. And the __pycache__ folders haven't been ignored either.

Here are the docker files.

docker-compose.yml

version: "3.8"

services: 

  app: 
    build: .
    volumes: 
      - .:/django-app
    ports: 
      - 8000:8000
    command: /bin/bash -c "sleep 7; python manage.py migrate; python manage.py runserver 0.0.0.0:8000"
    container_name: app-container
    depends_on: 
      - db

  db:
    image: postgres
    volumes:
      - ./data:/var/lib/postgresql/data
    environment: 
      - POSTGRES_DB=${DB_NAME}
      - POSTGRES_USER=${DB_USER}
      - POSTGRES_PASSWORD=${DB_PASSWORD}
    container_name: postgres_db_container

Dockerfile

FROM python:3.9-slim-buster

ENV PYTHONUNBUFFERED=1

WORKDIR /django-app
EXPOSE 8000

COPY requirements.txt requirements.txt

RUN apt-get update \
    && adduser --disabled-password --no-create-home userapp \
    && apt-get -y install libpq-dev \
    && apt-get -y install apt-file \
    && apt-get -y install python3-dev build-essential \
    && pip install -r requirements.txt

USER userapp
2

There are 2 answers

3
David Maze On BEST ANSWER

You're actually injecting your source code using volumes:, not during the image build, and this doesn't honor .dockerignore.

Running a Docker application like this happens in two phases:

  1. You build a reusable image that contains the application runtime, any OS and language-specific library dependencies, and the application code; then
  2. You run a container based on that image.

The .dockerignore file is only considered during the first build phase. In your setup, you don't actually COPY anything in the image beyond the requirements.txt file. Instead, you use volumes: to inject parts of the host system into the container. This happens during the second phase, and ignores .dockerignore.

The approach I'd recommend for this is to skip the volumes:, and instead COPY the required source code in the Dockerfile. You should also generally indicate the default CMD the container will run in the Dockerfile, rather than requiring it it the docker-compose.yml or docker run command.

FROM python:3.9-slim-buster

# Do the OS-level setup _first_ so that it's not repeated
# if Python dependencies change
RUN apt-get update && apt-get install -y ...

WORKDIR /django-app

# Then install Python dependencies
COPY requirements.txt .
RUN pip install -r requirements.txt

# Then copy in the rest of the application
# NOTE: this _does_ honor .dockerignore
COPY . .

# And explain how to run it
ENV PYTHONUNBUFFERED=1
EXPOSE 8000
USER userapp
# consider splitting this into an ENTRYPOINT that waits for the
# the database, runs migrations, and then `exec "$@"` to run the CMD
CMD sleep 7; python manage.py migrate; python manage.py runserver 0.0.0.0:8000

This means, in the docker-compose.yml setup, you don't need volumes:; the application code is already inside the image you built.

version: "3.8"
services:
  app: 
    build: .
    ports: 
      - 8000:8000
    depends_on: 
      - db
    # environment: [PGHOST=db]
    # no volumes: or container_name:

  db:
    image: postgres
    volumes: # do keep for persistent database data
      - ./data:/var/lib/postgresql/data
    environment: 
      - POSTGRES_DB=${DB_NAME}
      - POSTGRES_USER=${DB_USER}
      - POSTGRES_PASSWORD=${DB_PASSWORD}
    # ports: ['5433:5432']

This approach also means you need to docker-compose build a new image when your application changes. This is normal in Docker.

For day-to-day development, a useful approach here can be to run all of the non-application dependencies in Docker, but the application itself outside a container.

# Start the database but not the application
docker-compose up -d db

# Create a virtual environment and set it up
python3 -m venv venv
. venv/bin/activate
pip install -r requirements.txt

# Set environment variables to point at the Docker database
export PGHOST=localhost PGPORT=5433

# Run the application locally
./manage.py runserver

Doing this requires making the database visible from outside Docker (via ports:), and making the database location configurable (probably via environment variables, set in Compose with environment:).

0
Mikaelblomkvistsson On

That's not actually your case, but in general an additional cause of ".dockerignore not ignoring" is that it applies the filters to whole paths relative to the context dir, not just basenames, so the pattern:

__pycache__
*.pyc

applies only to the docker context's root directory, not to any of subdirectories.

In order to make it recursive, change it to:

**/__pycache__
**/*.pyc