How to set/estimate proper value for nginx's "limit_req_zone" and "limit_req"? I have nginx as reverse-proxy for Django app. During loading for example admin panel (where Django request a lot of css/js files) I receiving more than ten 503's errors.
Current configuration:
limit_req_zone $binary_remote_addr zone=perip:5m rate=20r/s;
limit_req zone=perip burst=5 nodelay;
Is it a good practice, to set burst parameter for example to 20?
Is there maybe another way to handle it?