To protect our endpoint from being overwhelmed, we have rate limits in place. For query/mutation, there is a 5000 limit per 5 minute interval.
Every response header will have 3 rate-limit headers:
Ratelimit-Limit — The maximum number of api calls in this interval (currently always 5000).
Ratelimit-Remaining — The number of api calls you have left in this interval.
Ratelimit-Reset — A Unix epoch timestamp of when the current interval ends.
Any request beyond the limit would receive an HTTP response code of 429 Too Many Requests.
Client could listen up to 50 endpoints per web socket connection. Attempting to listen to more would return an error.
Each endpoint has a buffer size of 30. If client are too slow to read, the particular endpoint would be closed.
Each IP could establish up to 100 web socket connections.