I had following nginx configuration:
worker_processes 1;
error_log logs/error.log;
events {
worker_connections 1024;
}
http {
limit_req_zone $request_uri zone=by_uri_6000:10m rate=6000r/s;
server {
listen 80;
server_name "openresty.com";
location = /limit-6000 {
limit_req zone=by_uri_6000 nodelay;
return 200 "ok"
}
}
}
But it is not working at all when testing using wrk:
wrk -c100 -t10 -d5s http://localhost/limit-6000
Running 5s test @ http://localhost/
10 threads and 100 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 5.57ms 1.65ms 23.47ms 81.17%
Req/Sec 1.81k 495.91 9.27k 99.60%
90400 requests in 5.10s, 17.93MB read
Requests/sec: 17713.29
Transfer/sec: 3.51MB
But, if I change my configuration to something like below (I was using openresty in fact):
location = /limit-6000 {
limit_req zone=by_uri_6000 nodelay;
default_type 'application/json';
content_by_lua_block {
ngx.say('{"code": 0, "msg": "成功"}')
}
it works then, what could be the reason, I didn't see any explanation in official document.
nginx return
directive works on rewrite phase, it immediately returns the result.
I believe ngx_http_limit_req_module works on access phase. So with return directive ngx_http_limit_req_module doesn't have any chances to be in game.