Nginx post ONE request to one more upstream_addr

Server NGINX
#1

hi, everyone,

nginx version is 0.8.53,
configure (http definition) is :

http{
upstream my_loads {
server 10.10.1.12;
server 10.10.1.13;
}

log_format combined_addsvr '$remote_addr - $remote_user

[$time_local] ’
'"$request" $status $body_bytes_sent ’
'"$http_referer" “$http_user_agent”
$upstream_addr
$upstream_cache_status ';
access_log logs/access.log combined_addsvr;

server {
    listen 80;
    server_name mytest.com;
    client_max_body_size 100m;
    proxy_buffering off;
    proxy_read_timeout 120;
    proxy_send_timeout 120;
    client_body_timeout 120;
    send_timeout  120;

    location / {
        proxy_pass http://my_loads/;
        proxy_set_header Host $host;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    }
}

===========================================

there is a strange line in log file: (sorry to use xxx replace of true
url)
192.168.1.1 - - [14/Jun/2011:20:53:24 +0800] “POST /1.php HTTP/1.1” 200
3729558 “http://mytest.com/xxx” “Mozilla/4.0 (compatible; MSIE 7.0;)”
10.10.1.12:80, 10.10.1.13:80 -

when nginx deals with 1.php, it sends POST to all upstream servers!

the next is normal log:
192.168.1.1 - - [14/Jun/2011:20:53:44 +0800] “POST /1.php HTTP/1.1” 302
2 "
http://mytest.com/xxx" “Mozilla/4.0 (compatible; MSIE 7.0;)”
10.10.1.12:80 -

Could anyone tell me if it is a bug or just any configure improperly?
thanks.

#2

Hello!

On Wed, Jun 15, 2011 at 02:38:08PM +0800, MaLi wrote:

    proxy_buffering off;
}

192.168.1.1 - - [14/Jun/2011:20:53:44 +0800] “POST /1.php HTTP/1.1” 302 2 "
http://mytest.com/xxx" “Mozilla/4.0 (compatible; MSIE 7.0;)” 10.10.1.12:80 -

Could anyone tell me if it is a bug or just any configure improperly?
thanks.

nginx will retry request on next upstream server if first one
failed, see here:

http://wiki.nginx.org/HttpProxyModule#proxy_next_upstream

Maxim D.

#3

Thanks Dounin,

actually, both 10.10.1.12 and 10.10.1.13 received the POST items with
1.php
(so repeat ran the script) ,
and why both servers being recorded as Upstream_addr in the log file?

2011/6/15 Maxim D. [email protected]

#4

oh, I see,

Maybe the POST data is too big, then first upstream server cannot
response
in time. I forgot to point one fact , that is one server maybe only
receive
part of POST data.

thanks again, Dounin.

2011/6/15 Maxim D. [email protected]

#5

Hello!

On Wed, Jun 15, 2011 at 04:03:30PM +0800, MaLi wrote:

Thanks Dounin,

actually, both 10.10.1.12 and 10.10.1.13 received the POST items with 1.php
(so repeat ran the script) ,

Most likely it’s failed to respond in time, and that’s why nginx
retried request. Don’t hesitate to look into error_log, it should
has details.

and why both servers being recorded as Upstream_addr in the log file?

Variable $upstream_addr should contain addresses of all upstream
servers tried, this is how it works.

Maxim D.

’

    proxy_buffering off;
}

the next is normal log:

http://wiki.nginx.org/HttpProxyModule#proxy_next_upstream

Maxim D.

nginx mailing list
[email protected]
http://nginx.org/mailman/listinfo/nginx