@maestrofjp This was always the issue with gunicorn sync that made it a bad idea on Heroku. To ramp up concurrency you had to add processes.
