Why docker-compose don't let me create a volume? - docker

Question

I am writing this request today because I will like to create my first Docker container. I watched a lot of tutorials, and there I come across a problem that I cannot solve, I must have missed a piece of information.
My program is quite basic, I would like to create a volume so as not to lose the information retrieved each time the container is launched.
Here is my docker-compose
version: '3.3'
services:
homework-logger:
build: .
ports:
- '54321:1235'
volumes:
- ./app:/app
image: 'cinabre/homework-logger:latest'
networks:
- homeworks
networks:
homeworks:
name: homeworks-logger
and here is my DockerFile
FROM debian:9
WORKDIR /app
RUN apt-get update -yq && apt-get install wget curl gnupg git apt-utils -yq && apt-get clean -y
RUN apt-get install python3 python3-pip -y
RUN git clone http://192.168.5.137:3300/Cinabre/Homework-Logger /app
VOLUME /app
RUN ls /app
RUN python3 -m pip install bottle beaker bottle-cork requests
CMD ["python3", "main.py"]
I did an "LS" in the container to see if the / app folder was empty: it is not
Any ideas?
thanks in advance !

Answer 1

Volumes are there to hold your application data, not its code. You don't usually need the Dockerfile VOLUME directive and you should generally avoid it unless you understand exactly what it does.
In terms of workflow, it's commonplace to include the Dockerfile and similar Docker-related files in the source repository yourself. Don't run git clone in the Dockerfile. (Credential management is hard; building a non-default branch can be tricky; layer caching means Docker won't re-pull the branch if it's changed.)
For a straightforward application, you should be able to use a near-boilerplate Dockerfile:
FROM python:3.9 # unless you have a strong need to hand-install it
WORKDIR /app
# Install packages first. Unless requirements.txt changes, Docker
# layer caching won't repeat this step. Do not list out individual
# packages in the Dockerfile; list them in Python-standard setup.py
# or Pipfile.
COPY requirements.txt .
# ...in the "system" Python space, not a virtual environment.
RUN pip3 install -r requirements.txt
# Copy the rest of the application in.
COPY . .
# Set the default command to run the container, and other metadata.
EXPOSE 1235
CMD ["python3", "main.py"]
In your application code you need to know where to store the data. You might put this in an environment variable:
import os
DATA_DIR = os.environ.get('DATA_DIR', '.')
with open(f"${DATA_DIR}/output.txt", "w") as f:
...
Then in your docker-compose.yml file, you can specify an alternate data directory and mount that into your container. Do not mount a volume over the /app directory containing your application's source code.
version: '3.8'
services:
homework-logger:
build: .
image: 'cinabre/homework-logger:latest' # names the built image
ports:
- '54321:1235'
environment:
- DATA_DIR=/data # (consider putting this in the Dockerfile)
volumes:
- homework-data:/data # (could bind-mount `./data:/data` instead)
# Use the automatic `networks: [default]`
volumes:
homework-data: