celery django setup

what is it?

how celery works under the hood when paired with django

django ORM access in celery

when celery initialises a new task, django.setup is called

celery/celery/fixups/django.py

class DjangoWorkerFixup:
    _db_recycles = 0
 
    def __init__(self, app: "Celery") -> None:
        self.app = app
        self.db_reuse_max = self.app.conf.get('CELERY_DB_REUSE_MAX', None)
        self._db = cast("DjangoDBModule", import_module('django.db'))
        self._cache = import_module('django.core.cache')
        self._settings = symbol_by_name('django.conf:settings')
 
        self.interface_errors = (
            symbol_by_name('django.db.utils.InterfaceError'),
        )
        self.DatabaseError = symbol_by_name('django.db:DatabaseError')
 
    def django_setup(self) -> None:
        import django
        django.setup()
 
    def validate_models(self) -> None:
        from django.core.checks import run_checks
        self.django_setup()
        if not os.environ.get('CELERY_SKIP_CHECKS'):
            run_checks()

django.setup is used to create a standalone django app, which provides access to the django ORM and application registry inside each worker

logging configuration

by default, Celery “hijacks” and modifies the root logger

celery/celery/app/log.py

def setup_logging_subsystem(....)
 
	root = logging.getLogger()
 
	if self.app.conf.worker_hijack_root_logger:
		root.handlers = []
		get_logger('celery').handlers = []
		get_logger('celery.task').handlers = []
		get_logger('celery.redirected').handlers = []
 
	# Configure root logger
	self._configure_logger(
		root, logfile, loglevel, format, colorize, **kwargs
	)
 
	# Configure the multiprocessing logger
	self._configure_logger(
		get_multiprocessing_logger(),
		logfile, loglevel if MP_LOG else logging.ERROR,
		format, colorize, **kwargs
	)

worker_hijack_root_logger defaults to True, which means celery clears the root handlers and resets the root log level