from socketify import WSGI import os import multiprocessing import logging from hello_socketify.hello.wsgi import application _is_travis = os.environ.get('TRAVIS') == 'true' workers = int(multiprocessing.cpu_count()) if _is_travis: workers = 2 def run_app(): WSGI(application).listen(8080, lambda config: logging.info(f"Listening on port http://localhost:{config.port} now\n")).run() def create_fork(): n = os.fork() # n greater than 0 means parent process if not n > 0: run_app() # fork limiting the cpu count - 1 for i in range(1, multiprocessing.cpu_count()): create_fork() run_app() # run app on the main process too :)