How exactly does nginx limit_req burst "delay" requests?

Question

Nginx

Let's say we have nginx with the given limits:
limit_req_zone $binary_remote_addr zone=one:10m rate=30r/s;
limit_req zone=one burst=10;
Suppose 31 requests come to this nginx in some second, one request is accordingly postponed for the next second, the next second 31 requests come again, the one that was postponed will be executed in priority and guaranteed?

Answer 1

As far as I remember from my experience, extra requests are not delayed, but rejected with the 500th error.

Answer 2

Burst is supposed to do more than rate. If there is a rate of 30 and a burst of 40, then the 31st request will be delayed for the next second, and the 41st will be rejected with a 503 status.