Reputation: 3847
When I run docker-compose up
, why does a new image get created? How can I tell docker-compose to run with the image I just created? What am I doing wrong here?
When I run the command docker build . -t prod_srdc_web
from a clean system, using docker system prune -a
, everything is built fine. When I run docker image ls
after building the image I see the following:
REPOSITORY TAG IMAGE ID CREATED SIZE
prod_srdc_web latest 923538608c06 5 seconds ago 1.1GB
<none> <none> 04b24dbf7c93 About a minute ago 4.23MB
python 3.7 42d620af35be 9 days ago 918MB
alpine 3.7 6d1ef012b567 4 months ago 4.21MB
After I run docker-compose up
everything gets built again and the container launches. Now docker image ps
shows an identical image with a new name, srdc_django_web
:
REPOSITORY TAG IMAGE ID CREATED SIZE
srdc_django_web latest 7e75a841e0fd About a minute ago 1.1GB
prod_srdc_web latest 923538608c06 2 minutes ago 1.1GB
<none> <none> 04b24dbf7c93 4 minutes ago 4.23MB
python 3.7 42d620af35be 9 days ago 918MB
alpine 3.7 6d1ef012b567 4 months ago 4.21MB
Why does docker-compose up
build a new image? Ideally I'd like to use the image I just built and tagged, prod_srdc_web
.
Dockerfile:
FROM alpine:3.7
RUN apk upgrade --no-cache build-base gcc python=3.7 python3-dev postgresql-dev bash git
ENV LIBRARY_PATH=lib:/usr/lib
FROM python:3.7
ENV PYTHONUNBUFFERED 1
RUN mkdir -p /code && mkdir -p /var/www/website.com/static/
WORKDIR /code
# Install dependencies
COPY requirements.txt /code/
RUN pip3 install --no-cache-dir --upgrade pip
RUN pip3 install -r requirements.txt
RUN pip3 install gunicorn
# Copy project
COPY . /code
# Copy entrypoint.prod.sh
COPY ./entrypoint.prod.sh /
EXPOSE 4500
# Run entrypoint.prod.sh
RUN [ "chmod", "+x", "./entrypoint.prod.sh" ]
ENTRYPOINT [ "./entrypoint.prod.sh" ]
docker-compose.yml
version: '3'
services:
django_web:
build: .
volumes:
- .:/srdc
ports:
- "4500:4500"
container_name: srdc_c_django_web
entrypoint.prod.sh
#!/usr/bin/env bash
python manage.py makemigrations
python manage.py migrate --noinput
python manage.py collectstatic --noinput
gunicorn srdc.wsgi:application -w 3 -b 0.0.0.0:4500
It looks like the repository name srdc_django_web
is created by concatenating the directory name (srdc
) with the service name in the docker-compose.yml file (django_web
). If I eventually add Nginx I want to use docker-compose up to run the entire container prod_srdc_web
, not have it build each service fresh.
FYI my versions are:
Docker version 18.09.2, build 6247962
docker-compose version 1.23.2, build 1110ad01
Upvotes: 2
Views: 3694
Reputation: 263666
Your compose file does not specify an image name. Therefore, even if you build the image in advance manually, docker-compose does not have any mapping from the build you want to run and the image you've already created. To give an image name, specify:
version: '3'
services:
django_web:
build: .
image: prod_srdc_web:latest
volumes:
- .:/srdc
ports:
- "4500:4500"
container_name: srdc_c_django_web
Docker will use the image cache to avoid recreating the same image layers that it already has. However, since you have the following:
COPY . /code
Any file in the build context (the current directory, assuming you do not ignore files with a .dockerignore
) that changes will result in a new image being created. That can include file ownership and permissions, and files like the docker-compose.yml if they are in this directory are also included.
Upvotes: 5