HireFire makes my Django app loose jobs in queue, using Redis and Celery for tasks

Question

I'm sure I'm not the only person using HireFire to manage worker dynos.

My workers send emails using Celery tasks and Redis.

When HireFire is off, everything works great. I set the Dynos for Celery and CeleryBeat to 1 and the tasks all succeed.

When I turn HireFire on, jobs are getting lost in the Queue.

I found this article on the Ruby library to handle exceptions caused by dyno restarts:

https://devcenter.heroku.com/articles/queuing-ruby-resque#job-termination

Does anyone know of a similar Python version, or another source of information?

https://github.com/jezdez/hirefire/

This is the Python library for HireFire on Github. I've looked through it, but I'm not having any moments of clarity.

https://github.com/AnthonyBRoberts/NNS

This is my repo on Github. There's no settings files in the repo (for obvious reasons). Here's the relevant settings:

HIREFIRE_PROCS = ["procs.CelerydProc", "procs.CelerybeatProc"]

MIDDLEWARE_CLASSES = (
    'hirefire.contrib.django.middleware.HireFireMiddleware',
    ...
)

INSTALLED_APPS = (
...
'django_ses',
'djcelery',
...
)

BROKER_URL = os.getenv('REDISTOGO_URL','redis://localhost:6379')

import djcelery
djcelery.setup_loader()

Any thoughts are welcome, and let me know what additional information would be useful.

Thanks,

Anthony