使用芹菜作为模型层

Question

The requirement

I received an unusual requirement: must use celery for communication with the database layer.

Application is using Flask and MongoDB on the backend and Angular on the frontend.

The sponsor believes that I can fire a task, let one of the workers move data in/out of Mongo and get the results back.

My research so far

As far as I know, if I call a celery task synchronously (using __call__ or apply ), the task is executed locally instead of being sent to the database; if I call it asynchronously, I must block on the Flask side and keep polling the task queue until I get results or an error back.

Another approach would be modifying the REST API to be make asynchronous calls (return the task ID immediately and keep polling for results on the client side. Personally I abhor the idea, this looks silly for tasks in the millisecond range like basic CRUD operations.

Question

Is there an elegant way to wrap task.delay in order to make it look like a synchronous call and still have the task executed by one of the workers?

Answer 1

Yes, there is - don't know how I missed AsyncResult.get in the docs:

AsyncResult.get(timeout=None, propagate=True, interval=0.5)

Wait until task is ready, and return its result.

So you just call the task like this:

task.delay(args).get(timeout=1)