Easy Configuration For Celery App Using a Yaml File

celery-yaml is a library to inject a --yaml option to the celery worker command in order to inject its configuration.

It also handle help to configurate this application for Pyramid application.

Usage

celery -A my_application.module_containing_my.app worker --yaml development.yaml ...

This will configure the application my_application containing an application app in a submodule module_containing_my.

The celery app must register the --yaml using the add_yaml_option on the app instance, this way:

from celery import Celery
from celery_yaml import add_yaml_option

app = Celery()
add_yaml_option(app)

Yaml format

celery:
  broker_url: "amqp://guest:guest@localhost:5672//"
  result_backend: "rpc://"
  imports:
    - my_application.tasks
  # see all settings in the celery docs:
  # https://docs.celeryproject.org/en/stable/userguide/configuration.html

logging:
  version: 1
  # dictConfig format
  # https://docs.python.org/3/library/logging.config.html#logging-config-dictschema

NOTE

To avoid password in the configuration file:

• Every variables can be substitute with an environment variable. The yaml file will replace variable using envsub.

  celery:
    broker_url: "${CELERY_BROKER_URL-amqp://guest:guest@localhost:5672//}"
  

• DEPRECATED: The broker_url is always overriden by the environment variable CELERY_BROKER_URL without any declaration.

Using celery beat

Celery beat can schedule tasks using interval or crontab.

The schedule parameter can be an int, a float or a mapping. If it is a mapping, it will be cast to a crontab, the mapping will be passed as acrontab parameters.

celery:
  beat_schedule:
    beat_second:
      task: beat_it
      schedule: 10 # example of interval schedule, every 10 seconds

    beat_crontab:
      task: beat_it
      schedule: # example of crontab
        hour: 9
        minute: 0
        day_of_week: 1

Using Celery in a Pyramid App.

The extras "pyramid" must be added to install the extras depencencies.

With PEP 621 (uv, pdm, ...)

[dependency-groups]
celery = ["celery >=5.2.7,<6", "celery-yaml[pyramid]>=2.0.0,<3"]

[project.entry-points."paste.app_factory"]
main = "pyramid_helloworld:main"

[project.entry-points."celery_yaml.app"]
main = "pyramid_helloworld.backend:app"

With poetry

[tool.poetry.dependencies]
celery-yaml = { version = "^2.0.0", extras = ["pyramid"] }

[tool.poetry.plugins."paste.app_factory"]
main = "pyramid_app.wsgi:main"

[tool.poetry.plugins."celery_yaml.app"]
main = "pyramid_app.backend:app"

[tool.poetry.plugins."plaster.loader_factory"]
"file+yaml" = "plaster_yaml:Loader"

the paste.app_factory is used by Pyramid itself to build the WSGI sergivice but we add a plaster.loader_factory to configure the usage of a yaml file instead of an ini file to configure it.

Then the celery_yaml.app is used by celery-yaml as an entrypoint to the celery app.

Then, in the configuration file,

celery: &celery
  result_backend: "rpc://"
  imports:
    - pyramid_app.tasks

app:
  "use": "egg:pyramid_app"
  "pyramid.includes": ["celery_yaml"]
  "celery":
    <<: *celery
    "use": "egg:pyramid_app"

foo: "bar"

More configuration

if the celery app as a method on_yaml_loaded then the function is called with the data and the filepath in parameter. It may be used to get mode configuration available readed from the yaml file.

Example

from celery import Celery as CeleryBase


class Celery(CeleryBase):

    def on_yaml_loaded(self, data: dict[str, Any], config_path: str):
        data['foo']  # you can access to the value of foo here

This is particullary usefull for depenency injection purpose.

Environment substitution

celery:
  broker_url: ${CELERY_BROKER_URL}
  result_backend: ${CELERY_BACKEND_RESULT}
  imports:
    - pyramid_app.tasks

database_dsn: ${DATABASE_DSN}

The CELERY_BROKER_URL and CELERY_BACKEND_RESULT environment variable are interpreted by celery, so can be outside your yaml file, but, it is not True for all the variable, using celery-yaml, you can still inject configuration in the yaml that are substituted in the configuration during the yaml load.

This is particullary usefull to keep your configuration structured but keep secrets in environment variable, for security reason. And also to build docker container that can be easily customizable.

See full example in the examples directory:

https://github.com/mardiros/celery-yaml/tree/master/examples/pyramid-app