rate limit tracker scaling #77
Description
We have a good starting point, but as we and others start to use the rate limit tracker component, we should keep an eye on how its scaling performance and stability. Some things to look for:
- execution time: Track how this scales with the number of installations. If it scales anything even close to linearly, we should definitely split it up into a queue-based system.
- rate limiting the rate limit tracker: While the
/rate_limitAPI does not appear to itself be rate limited, we should monitor for any failures that would indicate that the installation token API is being throttled. I can't find any docs that mention a rate limit for that API, but I assume there is one. That would be worth asking GitHub about. This will be important both for Apps with many installations and if we want to increase the frequency of rate limit checks.