I want to be sure that a task is killed at a certain time if it is still running. The context is an overloaded worker, where tasks are not picked up straight away. Image this "busy" worker (concurrency=1 to simulate business): uv run celery --app src.tasks.example worker --concurrency 1 --queues celery --loglevel info --hostname default I am going to run this task, which just sleeps for the number of seconds given in parameter: @app.task() def naptime(seconds: float) -> None: sleep(seconds) I submit this task that way: naptime.apply_async(args=(3,)) task2 = naptime.apply_async(args=(3,)) start = time() result = task2.get() total = time() - start I expect total to be around 6 (2 tasks of 3 seconds, which are executed sequentially) and it is indeed close enough. Now, I want to make sure that task2 is killed 5 seconds after submission time if it runs. task2 = naptime.apply_async(args=(3,), time_limit=5) => does not work as I want, because it would kill the task 5 seconds after it is picked up The task is done by then, but later than 5 seconds after submission. task2 = naptime.apply_async(args=(3,), time_limit=5, expires=5) => does not work as I want, the task does not have time to expire. Note: in reality, I do not know how long the task will last, so I can not compute an expires (task of S seconds that needs to be done T seconds after submission could have an expire of (T-S). I know T but not S.) How can I make sure that the task is killed at a specific time? Continue reading...
