Python has support for GUIDs (or UUIDs) that would let us track each request coming in, and let us debug issues associated with each request. The chief problem in this would be finding some way of storing request IDs, and relevant parameters associated with each one. I propose doing this in RabbitMQ, and building out some celery workers, but this could complicate our stack quite a bit.