I have this Celery task:

@app.task
def do_something(with_this):# instantiate a class from a third party libraryinstance = SomeClass()# this class uses callbacks to send progress info about# the status and progress of what we're doingdef progress_callback(data):# this status will change to 'finished' later# but the return value that I want as the task result won't be returned# so this is where I should mark the task as done manuallyif data['status'] == 'working':# I create a custom state for this taskdo_something.update_state(state = 'PROGRESS',meta = data['progress'])# adding the callback to the instanceinstance.add_callback(progress_callback)# use the instance to do what I want# this functions returns a value that I don't want as the task result# so leaving this function without a return statement will make it Noneinstance.do_job(with_this)

How can I mark a task as done manually ?

In this case the function reaches the end without any return statement so the task.result I get is None, I want to set the data passed to the callback function as the result and mark the task as done.

I tried using:

app.backend.mark_as_done(do_something.request.id, data)

It's successfully setting the state and the result of the task but later the result is set to the return value of the function which is here None.

I finally found the solution which is storing the task state and result then ignoring the task by raising an Ignore exception, for example:

from celery.exceptions import Ignore@app.task
def do_something(with_this):# store the state and result manually# the SUCCESS state is set by this methodapp.backend.mark_as_done(do_something.request.id,the_data_to_store)# we can also use update_state which calls# backend.store_result just like mark_as_done# but we have to set the state in this casedo_something.update_state(state = celery.states.SUCCESS,meta = the_data_to_store)# ignore the task so no other state is recorded# like what was happening with my function in the question# the task will still be acknowledgedraise Ignore()

This is helpful when you cannot return the data that you want to store as the result.