Calling async_result.get() from within a celery task

Here is a fragment that hushes the warning if you know what you are doing is safe:

from celery.result import allow_join_result

with allow_join_result():
    result.get()

source

There is the solution of chaining of course, except it's impossible to chain remote tasks. The only way to call remote tasks is using app.send_task, which returns an AsyncResult, and that I can't chain as I need the task function itself..

No, it is possible to chain remote tasks. I've just tried it in a project of mine and it works. I suggest you try it with a trivial test task first to make sure you got the basics down before moving to something more complex. I've created these tasks:

@app.task
def foo(arg):
    return arg + 1

@app.task
def bar(arg):
    return "I barred " + str(arg)

The two tasks are held in a module named app.tasks. (It is part of a Django project.)

Then I wrote a command that does:

import celery
print (celery.signature("app.tasks.foo", args=(1, )) |
       celery.signature("app.tasks.bar")).delay().get()

And I got on the screen:

I barred 2

If you want your task to be synchronous, you can use ready() to hold a loop:

import time

while not result_from_remote.ready():
    time.sleep(5)

return result_from_remote.get()