IN DEVELOPMENT

celery-sql-beat-reloader

A celery schedule that will sync with a db for the latest schedule of tasks.

Features

• Pulls schedule from a database instead of code.
• Every schedule is timezone aware.

Getting Started

Prerequisites

• Python 3
• celery >= 5.2
• sqlalchemy

Installing

Install from PyPi:

$ pip install celery-sql-beat-reloader

Install from source by cloning this repository:

$ git clone git@github.com:sir-wiggles/celery-sql-beat-reloader.git
$ cd celery-sql-beat-reloader
$ python setup.py install

Usage

celery_sql_beat_reloader uses sqlalchemy to manage database connections so you should in theory be able to use this on any flavor of SQL database.

Schema

CREATE TABLE celery_beat_schedule (
    id SERIAL PRIMARY KEY,
    created_date TIMESTAMPTZ DEFAULT NOW(),
    updated_date TIMESTAMPTZ DEFAULT NOW(),
    last_scheduled_date TIMESTAMPTZ,
    active BOOLEAN DEFAULT TRUE,

    name TEXT NOT NULL UNIQUE,
    task TEXT NOT NULL,

    args   JSONB DEFAULT '[]',
    kwargs JSONB DEFAULT '{}',

    type          TEXT NOT NULL DEFAULT 'cron',
    seconds       INTEGER DEFAULT 60,
    minute        TEXT DEFAULT '*',
    hour          TEXT DEFAULT '*',
    day_of_week   TEXT DEFAULT '*',
    day_of_month  TEXT DEFAULT '*',
    month_of_year TEXT DEFAULT '*',
    timezone      TEXT DEFAULT 'America/Los_Angeles'
);

NOTE: Table is not auto generated.

Column Descriptions

• active - indicates if the schedule is active or not.
  • active=false, the schedule will be ignored at startup or removed at the next sync interval.
  • active=true, the schedule will be include at startup or added at the next sync interval.
• name - is a unique name for the schedule.
• task - is the import path of the task e.g. test_app.beat1, test_app.beat2.
• type - is the type of the schedule and supports two types: cron and periodic.
  • cron - is a crontab with timezone awareness.
  • periodic - is a schedule with a frequency of seconds from the seconds column.
• last_scheduled_date - is when the beat was last sent to be processed.

NOTE: last_scheduled_date does not mean the task was successful or even if the task was processed. It just means the scheduler sent a request at that time.

Config

In your celery app.conf you'll need to set beat_reloader_dburl to the url of your database.

# test_app.py
import celery

app = celery.Celery(
    "test",
    broker_url="redis://localhost:6379",
    timezone="UTC",
    beat_reloader_dburl="postgresql+psycopg2://user:pass@127.0.0.1:5432/db",
)


@app.task
def beat1():
    print("beat1")


@app.task
def beat2():
    print("beat2")

Running

celery --app test_app beat -S celery_sql_beat_reloader:Reloader --max-interval=300 --loglevel=INFO

Specifying the --max-interval argument will allow you to control how frequently the schedule syncs with the database

Acknowledgments

• celery-sqlalchemy-scheduler
• django-celery-beat
• celerybeatredis
• celery