So I am deploying django, postgress and nginx containers via docker-compose and I have an issue that I can't seem to figure out.
In order to resolve the following error in my Django app, I knew I just had to run a Django migration.
docker@postgres ERROR: relation "accounts_myprofile" does not exist
In an attempt to run migrations, I tried:
docker-compose run web python manage.py makemigrations
docker-compose run web python manage.py migrate
which returned the following:
Migrations for 'accounts':
accounts/migrations/0001_initial.py:
- Create model Entry
- Create model MyProfile
Running migrations:
No migrations to apply.
I was only able to successfully migrate from within the Django container, example:
docker exec -i -t 6dc97c6a305c /bin/bash
python manage.py makemigrations
python manage.py migrate
Although I have resolved the issue, I still don't understand why running the migrate via docker-compose run does not actually migrate anything. I'm hoping someone can maybe point me in the right direction on this.
Also, I don't know if this is a related issue or not, but when I run those docker-compose run web commands, they seem to be creating new containers that won't shutdown unless I manually stop them, docker-compose stop doesn't remove them.
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
a7bb3c7106d1 accounts_web "python manage.py che" 4 hours ago Restarting (0) 41 minutes ago 8000/tcp accounts_web_run_62
ee19ca6cdf49 accounts_web "python manage.py mig" 4 hours ago Restarting (0) 43 minutes ago 8000/tcp accounts_web_run_60
2d87ee35de3a accounts_web "python manage.py mak" 4 hours ago Restarting (0) 43 minutes ago 8000/tcp accounts_web_run_59
1c6143c13097 accounts_web "python manage.py mig" 4 hours ago Restarting (1) 44 minutes ago 8000/tcp accounts_web_run_58
6dc97c6a305c b1cb7debb103 "python manage.py run" 3 days ago Up 4 hours 8000/tcp accounts_web_1
Note: Docker-compose stop will properly stop the container at the bottom (as it should), but the other container that were created by docker-compose run web python manage.py migrate, will need to be manually stopped.
my docker-compose
web:
restart: always
build: ./web
expose:
- "8000"
links:
- postgres:postgres
volumes:
- /usr/src/app
- /usr/src/app/static
env_file: .env
environment:
DEBUG: 'true'
command: python manage.py runserver 0.0.0.0:8000
postgres:
restart: always
image: kartoza/postgis:9.4-2.1
ports:
- "5432:5432"
volumes:
- pgdata:/var/lib/postgresql/data/
You have already noticed the problem. When you use docker-compose run, a new container is created.
When you ran the first command (makemigrations), a new container was created, makemigrations ran, and the migration files were written to the (new) container's filesystem.
When you ran the second command (migrate), another new container was created. The migration ran, but it had nothing to do. That's because the migration files were not available - they were written in a different container than this new one.
You can solve this in a couple of ways.
First, you can do what you already did, but use docker-compose exec instead of run.
docker-compose exec web python manage.py makemigrations
docker-compose exec web python manage.py migrate
exec will use the already-running container, rather than creating new containers.
Another option is to use an entrypoint script and run the migration there, before the server is started. This is the way to go if you'd prefer things to be more automatic.
Dockerfile:
COPY entrypoint.sh /entrypoint.sh
RUN chmod +x /entrypoint.sh
entrypoint.sh:
#!/bin/sh
python manage.py makemigrations
python manage.py migrate
exec "$@"
docker-compose.yml (under 'web'):
entrypoint: /entrypoint.sh
In this scenario, when the container starts, the entrypoint script will run, handle your migration, then hand off to the command (which in this case is Django runserver).
As you noticed, the new containers stay running. That is normally unexpected, because you overrode the command with one that should exit (rather than stay running). However, in docker-compose.yml, you specified restart: always. So they will run the migration commands over and over, restarting each time the command exits.
Dan Lowe gave a very nice answer, but the entrypoint script was not working for me. The problem is that some "makemigrations" expect your input, for instance "yes"/"no".
You can complement Dan Lowe answer with:
python manage.py makemigrations --noinput
instead of
python manage.py makemigrations
(This works at least for simple "yes"/"no" questions)
This awnser is a complement to Dan Lowe and Rexcirus responses.
To work well during de CodeBuild e Fargate I did some changes:
Dockefile:
COPY ./docker/entrypoint.sh /usr/local/bin/
COPY ./docker/entrypoint.sh /${projectName}/
# backwards compat
RUN ln -s usr/local/bin/entrypoint.sh /
ENTRYPOINT ["entrypoint.sh"]
CMD ["entrypoint.sh"]
./docker/entrypoint.sh
#!/bin/sh
python manage.py makemigrations --noinput
python manage.py migrate
python manage.py runserver 0.0.0.0:8000
Now everything is running OK.