กำหนดและเรียกใช้งาน

A Celery task is just a Python function marked so workers can run it later. You decorate it once and it becomes enqueueable.

Mark a function with the Celery task decorator. Now it has extra methods like delay that schedule it onto the queue.

Calling add(2, 3) runs it right now in your process. To run it on a worker instead, you call its delay method.

The delay shortcut puts the job on the broker and returns immediately. The work happens later on a worker.

Need more control like a countdown or queue name? Use apply_async, which takes args plus scheduling options.

Both delay and apply_async return an AsyncResult. It carries the task id you use to look up the result later.

Task arguments travel through the broker, so they must be serializable: use plain data like ints, strings, lists, and dicts.

Do not send a database row to a task. Send its id instead, then re-fetch the object inside the task with a fresh session.

A view validates input, enqueues the job, and returns a fast response. The user is not kept waiting for the slow part.

Give tasks descriptive names like send_welcome_email. Clear names make logs and the broker dashboard far easier to read. 🔍

A good task does one clear job. Small, focused tasks are easier to retry, monitor, and reason about when something breaks.

You want a function to run on a worker.

Decorate a function as a task, call delay or apply_async to queue it, pass serializable args, and enqueue from thin views. ✅