I suggest digging through search for prior discussions about rate limiting.
I’m not profound in this topic but to my understanding rate limiting is applied in two stages: first nginx does rate-limiting (not knowing about trust levels) and then the rails app does another layer of protection.
If you wish to test the hypothesis that you encounter rate limits within the nginx layer, you may comment out the following line in your app.yml:
- templates/web.ratelimited.template.yml