Home / server / Questions / 518634
Accepted
Gajus
Asked: 2013-06-27 00:05:16 +0800 CST

What is the definition of a burst in Nginx limit_req setting?

  • 772

What is the difference between this:

limit_req_zone $binary_remote_addr zone=flood:10m rate=30r/s;
limit_req zone=flood nodelay

and:

limit_req_zone $binary_remote_addr zone=flood:10m rate=30r/s;
limit_req zone=flood burst=5 nodelay

Documentation http://wiki.nginx.org/HttpLimitReqModule.

nginx

3 Answers

  1. As it says about burst on http://nginx.org/en/docs/http/ngx_http_limit_req_module.html

    Excessive requests are delayed until their number exceeds the maximum burst size

    This means in your case the first 5 requests are 'ignored' and only requests 6 and up are counted. So on request 35 the limit of 30 is reached. Then, on request 36 within the same 1 second coming from the same origin, nginx will send a 503 response instead of treating the request any further.

    • 15

  2. Burst means it will be delayed for next second, if there is excess 30 request/second then nginx send 503 error code.

    Burst works just like a queue

    • 6
  3. Best Answer

    In your case:

    allow no more than 30 requests per second at an average, with bursts not exceeding 5 requests.

    Reference: http://nginx.org/en/docs/http/ngx_http_limit_req_module.html

    • 1