At what level is the in-memory rate limiting meant to perform at?

[ulule]

At what level is the in-memory rate limiting meant to perform at?

[angela-flores-wdc]

Hmm, the simplified version I created for the gist is not experiencing the same problem. I will close this issue then because the rate limiter is behaving as intended, it must be some other factor in my code.

[novln]

Hello,

The known limitation is about performance (see #86), but the counter shouldn't have any issue on incrementing itself.

Could you provide a simple gist that reproduces this issue so I could test it locally and find out what's going on.

Thank you in advance.

[justincampbell]

I just noticed a similar issue. When testing the memory store through HTTP handlers, the memory store appears to behave correctly with 1 key (IP address), but introducing a 2nd key causes odd behavior where a limit of 2 req/s allows ~3 requests for the first key and 2 for the 2nd key, and a limit of 3 req/s allows anywhere between 40 and 100 req/s through per key. I'm attempting to write a reproduction now.

[novln]

Ok, please let me know 🙂