When I specify only the broker_read_url in celeryconfig.py, for some reason celery falls back to the default amqp://localhost:5672
This doesn't make sense to me. Why would a worker need to write to the broker? Looking at the source code, I can't see anything besides the purge option which would require it.
Is there a way around this? I've tried putting in a dummy url, and that doesn't work.
edit: More details about my setup. I've got a rmq shovel set up, which I'm writing to, and a rmq instance I'm reading from. In my producer app, I want to configure only broker_read_url, and in the consuming worker, I want to only configure broker_write_url.
Since I can't do this, my assumption is that celery must be using it for something, but I can't really tell what looking through the code.