Locally, the Celery instance accepts the settings. In production in Docker it does not accept

Question

Project structure:

proj:
  my_app:
    __init__.py
    tasks.py
  settings:
    __init__.py
    base_settings.py
    settings.py
    prod_settings.py
    celery_config.py
    celery.py

---

tasks.py

from settings import celery_app

@celery_app.task(bind=True, max_retries=5, default_retry_delay=3,name='my_app.tasks.send_mail')
def send_email_task(self, message, subject, recipient_list):
    send_mail_func()

---

celery_config.py

from os import getenv as env

broker_url = env('CELERY_BROKER_URL', 'not_value')
result_backend = env('CELERY_RESULT_BACKEND', 'not_value')
timezone = env('CELERY_TIMEZONE', 'UTC')
task_serializer = 'json'
accept_content = ['json']
result_serializer = 'json'
enable_utc = False
broker_connection_retry_on_startup = True

celery.py:

from celery import Celery

celery_app = Celery('app_celery')
celery_app.config_from_object('settings.celery_config')
celery_app.autodiscover_tasks()

docker-compose:

rabbitmq:
    restart: always
    env_file:
      - .env
    volumes:
      - ~/.docker-conf/rabbitmq/data/:/var/lib/rabbitmq/
      - ~/.docker-conf/rabbitmq/log/:/var/log/rabbitmq
    healthcheck:
      test: [ "CMD-SHELL", "rabbitmqctl status" ]
      interval: 30s
      timeout: 10s
      retries: 5
    command: >
      bash -c "
        rabbitmq-plugins enable rabbitmq_management &&
        rabbitmq-server &
        sleep 10 &&
        rabbitmqctl await_startup &&
        rabbitmqctl add_vhost my_vhost &&
        rabbitmqctl set_user_tags ${RABBITMQ_DEFAULT_USER} administrator &&
        rabbitmqctl set_permissions -p my_vhost ${RABBITMQ_DEFAULT_USER} '.*' '.*' '.*' &&
        wait
      "

  backend:
    restart: always
    volumes:
      - static:/static/
      - media:/media/
      - /home/user/logs/gunicorn:/var/log/gunicorn
      - /home/user/logs/django:/var/log/django
    env_file:
      - .env
    command: bash -c "
      export DJANGO_SETTINGS_MODULE=settings.prod_settings
      && python manage.py migrate --settings=settings.prod_settings
      && python manage.py collectstatic --noinput --settings=settings.prod_settings
      && python manage.py create_admin ${ADMINER_ADMIN_EMAIL} ${ADMINER_ADMIN_PASSWORD} --settings=settings.prod_settings
      && gunicorn settings.wsgi -b 0.0.0.0:8080 --workers 2 --access-logfile /var/log/gunicorn/access.log --error-logfile /var/log/gunicorn/error.log
      "
    depends_on:
      pg:
        condition: service_healthy


  celery_worker:
    restart: always
    env_file:
      - .env
    volumes:
      - /home/user/logs/django:/var/log/django
    command: bash -c "
      sleep 10 && \
      export DJANGO_SETTINGS_MODULE=settings.prod_settings
      && celery -A settings worker --loglevel=DEBUG --concurrency=2 --pool=gevent --hostname=worker1
      "
    depends_on:
      rabbitmq:
        condition: service_healthy

  celery_beat:
    restart: always
    env_file:
      - .env
    volumes:
      - /home/user/logs/django:/var/log/django
    command: bash -c "
      export DJANGO_SETTINGS_MODULE=settings.prod_settings
      && celery -A settings beat --loglevel=INFO
      "
    depends_on:
      rabbitmq:
        condition: service_healthy

---

the problem is that the Celery worker is running in the local environment. But in the product environment, tasks do not go away even in RabbitMQ, they are always with the PENDING status.

I connect to a docker container with a backend and watch celery_app. But for some reason it has empty broker_url = None and backend = None. It seems that the settings are not applied to the Celery instance and therefore the tasks do not go to RabbitMQ.

I tried to submit settings in capital letters, add the CELERY_ prefix, and submit settings via .update() - nothing helps. I sincerely don't understand what the problem might be. Thanks!