Threading in Django is not working in production

Question

I have a function in my Django views.py that looks like this.

def process(request):
form = ProcessForm(request.POST, request.FILES)
if form.is_valid():
    instance = form.save(commit=False)
    instance.requested_by = request.user
    instance.save()
    t = threading.Thread(target=utils.background_match, args=(instance,), kwargs={})
    t.setDaemon(True)
    t.start()
    return HttpResponseRedirect(reverse('mart:processing'))

Here, I'm trying to call a function 'background_match' in a separate thread when ProcessForm is submitted. Since this thread takes some time to complete, I redirect the user to another page named 'mart:processing'.

The problem I am facing is that it all works fine in my local machine but doesn't work on production server which is an AWS EC2 instance. The thread doesn't start at all. There's a for loop inside the background_match function which doesn't move forward.

However, if I refresh (CTRL + R) the 'mart:processing' page, it does move by 1 or 2 iterations. So, for a complete loop consisting of 1000 iterations to run, I need to refresh the page 1000 times. If after, say, 100 iterations I don't refresh the page it gets stuck at that point and doesn't move to the 101st iteration. Please help!

Answer 1

Wrong architecture. Django and other web apps should be spawning threads like this. The correct way is to create an async task using a task queue. The most popular task queue for django happens to be Celery .

The mart:processing page should then check the async result to determine if the task has been completed. A rough sketch is as follows.

from celery.result import AsynResult
from myapp.tasks import my_task

...
if form.is_valid():
    ...
    task_id = my_task()
    request.session['task_id']=task_id
    return HttpResponseRedirect(reverse('mart:processing'))
    ...

On the subsequent page

task_id = request.session.get('task_id')
if task_id:
    task = AsyncResult(task_id)