limit_zone and custom variable

Sergej Kurakin sergej.kurakin at
Thu Feb 11 22:50:59 MSK 2010


I have a question about limit_zone and custom variable. I'm analysing
situation with strange 503 errors messages for users that do first
request to server with NGINX 0.6.x As I see from NGINX source error
503 showed in only one situation - when client reaches connection

In NGINX configuration file in http section limit_zone with custom
variable defined in next way:

http {
limit_zone ten $my_remote_addr 1m;
[servres section]

and $my_remote_addr is not defined anywhere in http section.

In server section $my_remote_addr defined for any user in one way,
except for some IP range (some proxy servers) - you can see it above:

server {
set $my_remote_addr $binary_remote_addr;

if ($remote_addr ~ "^(XXX\.XXX\.XXX|YYY\.YYY\.YYY)\.") {
    set $my_remote_addr $binary_remote_addr$remote_port;

limit_conn ten 11;

Will this configuration impact NGINX's limit functionality in non-
predictable behaviour? I can't understand how limits work with this
settings from source code :-( Maybe someone know answer for that

Should we change it and before setting "limit_zone ten $my_remote_addr
1m;" use "set $my_remote_addr $binary_remote_addr$remote_port;" for
any IP?

More information about the nginx mailing list