Docker Compose, Django: role "_" does not exist - django

Question

Context
I am trying to run my Django application and Postgres database in a docker development environment using docker-compose (it's my first time using Docker).
I want to use my application with a custom role and database both named teddycrepineau (as opposed to using the default postgres user and db).
Goal
My goal is to deploy a web app powered on the front end by react and the backend by django restapi, the whole running in a docker.
System/Version
python: 3.7
django: 2.1
OS: Mac OS High Sierra
What error am I getting
When running docker-compose up with my custom role and db, I am getting the following error django.db.utils.OperationalError: FATAL: role "teddycrepineau" does not exist. When running the same command with the default role and db postgres Django is able to start normally.
My understanding was that running docker-compose up would create the role and db passed as environment variable.
What I have tried so far
I read multiple threat on this site, GitHub, and docker:
tried to delete my container and rebuilt it with formatting as suggested here
Went through this GitHub issue
Tried to move my environment variable from .env file the environment inside my docker-compose.yml file and rebuild my container
Files
docker-compose.yml
version: '3'
volumes:
postgres_data: {}
services:
postgres:
image: postgres
volumes:
- postgres_data:/var/lib/postgresql/data
env_file: .env
ports:
- "5432"
django:
build:
context: teddycrepineau-backend
dockerfile: teddycrepineau-root/Dockerfile
command: ./teddycrepineau-backend/teddycrepineau-root/start.sh
env_file: .env
volumes:
- .:/teddycrepineau-backend
ports:
- "8000:8000"
depends_on:
- postgres
Dockerfile
FROM python:3.7
ENV PYTHONUNBUFFERED 1
WORKDIR /teddycrepineau-backend/
ADD ./teddycrepineau-root/requirements.txt /teddycrepineau-backend/
RUN pip install -r requirements.txt
ADD . /teddycrepineau-backend/
RUN chmod +x ./teddycrepineau-root/start.sh
start.sh
#!/usr/bin/env bash
python3 ./teddycrepineau-backend/teddycrepineau-root/manage.py runserver
.env
POSTGRES_PASSWORD=
POSTGRES_USER=teddycrepineau
POSTGRES_DB=teddycrepineau
EDIT
My file structure is as follow
root
|___ teddycrepineau-backend
|___ teddycrepineau-root
|___ teddycrepineau
|___ Dockerfile
|___ manage.py
|___ start.sh
|___ teddycrepineau-frontend
|___ React-App
|___ .env
|___ docker-compose.yml
When I move my docker-compose.yml file inside my backend folder, it starts as expected (though I am not able to access my site when going to 127.0.0.1:8000 but that is mostly a different issue) with custom user and db. When I put my docker-compose.yml file to my root folder, I get the error django.db.utils.OperationalError: FATAL: role "teddycrepineau" does not exist

Answer 1

This happens because your pgsql db was launched without any envs. The pgsql docker image only uses the envs the first time you created the container, after that it won't recreate DB and users.
The solution is to remove the pgsql volume so next time you docker-compose up you will have a fresh db with envs read. Simple way to do it is docker-compose down -v

Answer 2

Change your env order like this.
POSTGRES_DB=teddycrepineau
POSTGRES_USER=teddycrepineau
POSTGRES_PASSWORD=
I find it at this issue. I hope it works.

Answer 3

when you run the
sudo docker-compose exec web python manage.py migrate
yes of course you will receive
"django.db.utils.OperationalError: FATAL: role "user" does not exist
first you need to put
sudo docker-compose down -v
sudo docker system prune
check container, they should be deleted
sudo docker ps -a
then check images
sudo docker image ls
don`t forget delete images
sudo docker image rm 3e57319a7a3a
run to the project folder and then check out
python manage.py migrate
if it didn`t works put the
python manage.py migrate —run-syncdb
and
sudo docker-compose up -d --build
sudo docker-compose exec web python manage.py collectstatic --no-input
sudo docker-compose exec web python manage.py makemigrations
sudo docker-compose exec web python manage.py migrate auth
sudo docker-compose exec web python manage.py migrate --run-syncdb

Answer 4

I encountered the issue due to a mismatch between the $POSTGRES_DB and $POSTGRES_USER variables. By default, psql will attempt to set the database to the same name as the user logging in, so when there is a mismatch between the variables it fails with an error along the lines of psql:
FATAL: database "root" does not exist
I had to edit the init script that I was writing to include the -d "$POSTGRES_DB" option like so:
#!/bin/bash
set -e
psql -v ON_ERROR_STOP=1 --username "$POSTGRES_USER" -d "$POSTGRES_DB" <<-EOSQL
CREATE USER docker;
CREATE DATABASE docker;
GRANT ALL PRIVILEGES ON DATABASE docker TO docker;
EOSQL