Hi,

I'm GET-ing the page/html/{title} endpoint at https://en.wikipedia.org/api/rest_v1/ for information extraction. I'm trying to nail down a polite request rate, and to determine whether the current rate limit is likely to change soon.

- the doc at https://en.wikipedia.org/api/rest_v1/ pegs the rate limit at 200 req/s;
- on #wikimedia-services, +gwicke noted that the varnish cache's rate limit is much lower -- around 100 req/s; but
- in practice, I get 429's whenever I exceed 70 req/s for more than a few minutes.

Pchelolo suggested additional debug logs on 429's might help get to the bottom of this lower-than-expected rate limit.

What kind of debugging info can I provide from my end? Any chance I'll be able to hit the 200 req/s mark in the next few months?

Thanks,

Shahin