CODEWITHSUNDEEP
pythonflaskwsgigunicorn
Dmiters
Dmiters

Reputation: 2031

How do I start background threads when launching a Flask app through gunicorn?

I'm launching my Flask app directly through the interpreter.

if __name__ == '__main__':
    t = Thread(target=run_schedule)
    t.start()
    context = ('cert.pem', 'key.pem')
    app.run(host='0.0.0.0',port=8080,debug=False,ssl_context=context)

The run_schedule function loops forever, unblocking once in a while to do a task.

Apparently I'm supposed to use a wsgi server like gunicorn when I'm not debugging anymore, but it doesn't call the module through main(), so the thread doesn't start.

Putting them outside that block won't work because then the thread would get started if any other code imports the module!

@before_first_request is almost what I need, but it requires me to poke the server with a request first. Not ideal.

What's the recommended way to do it? (Or do background threads go against wsgi philosophy?)

Upvotes: 1

Views: 3481

Answers (2)

David42
David42

Reputation: 103

You can do this if you want to. And you may want to because Cron jobs can be fragile adn Celery may be overkill.

Create a file inside your app folder, say production.py which looks like this:

from . import app, start_background
start_background()

Now invoke Gunicorn like this:

gunicorn --workers=1 --threads=4 app.production:app

As long as you set workers to 1, you should get only one background task.

Upvotes: 0

Nils Werner
Nils Werner

Reputation: 36765

You shouldn't spawn background threads in your server application. For example a WSGI server might spawn several server apps and then you have several background threads. Instead look into cronjobs or job queues like Celery.

Upvotes: 1

Related Questions