# Strange error: nginx does not flush the buffer back to the client

**URL:** <https://www.ruby-forum.com/t/strange-error-nginx-does-not-flush-the-buffer-back-to-the-client/161179>\
**Category:** NGINX\
**Created:** [February 24, 2009, 6:44am UTC](https://www.ruby-forum.com/t/strange-error-nginx-does-not-flush-the-buffer-back-to-the-client/161179 "2009-02-24T06:44:12Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Peter\_Langhans](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/54ee81/32.png) [@Peter\_Langhans](https://www.ruby-forum.com/u/Peter_Langhans)\
**Post date:** [February 24, 2009, 6:44am UTC](https://www.ruby-forum.com/t/strange-error-nginx-does-not-flush-the-buffer-back-to-the-client/161179/1 "2009-02-24T06:44:12Z")

</div>

Hi,

I’m completely new to nginx - I’ve just started using it a day ago, but  
I’ve  
managed to set up a reverse proxy szenario with nginx as a front end and  
3  
apache servers (2.0.63) as the backend servers.

\*The strange thing that is happening is: sometimes the pages are not  
displayed to the user, but it just freezes, i.e. the page is not  
loading,  
BUT when I kill the nginx process (the only connection to the outer  
world)  
all files from the current request (that was hanging) are served  
immediately. It seems like the buffer is not written to the open  
connection  
and killing nginx causes nginx to release the connection and thus the  
data  
on it?!?!

Can anyone help solving the Gordic know as per why the data is not send,  
but  
the request hangs? Is nginx waiting for something before sending the  
request  
out? If so why and more importantly: can I turn this behavior off? \*

The apache servers are listening on the following addresses:

- 192.168.0.1:80 → web1
- 192.168.0.2:80 → web2
- 192.168.0.3:80 → web3

My public IP address is 69.64.147.249 and nginx is listening on web1 on  
port  
80. I have to match 4 different domains to consider:

- [example.com](http://example.com)
- [app.example.com](http://app.example.com)
- \*.app.example.com (should be distributed between web1, web2, web3)
- [static.example.com](http://static.example.com) (for static file serving)

I have defined 2 upstreams

- [proxy1.example.com](http://proxy1.example.com): for dealing with requests to [example.com](http://example.com) and  
[app.example.com](http://app.example.com) on only web1
- [proxy2.example.com](http://proxy2.example.com): for dealing with requests to \*.example.com on  
web1,  
web2, web3

\*\*  
The described problem occurs in versions nginx-0.7.38 and nginx-0.6.35.

Thanks so much for your help!

Bests,  
Peter

This is my configuration:

##########

user nginx nginx;

worker\_processes 1;

events {  
worker\_connections 1024;  
}

http {  
include mime.types;  
default\_type text/plain;

```
sendfile on;

keepalive_timeout 65;

server_name_in_redirect off;

gzip on;
gzip_min_length 1100;
gzip_types text/plain text/html text/css

```

application/x-javascript text/xml application/xml application/xml+rss  
text/javascript image/x-icon image/png;

```
upstream proxy1.example.com {
    server 192.168.0.1:80;
}

upstream proxy2.example.com {
    server 192.168.0.1:80;
    server 192.168.0.2:80;
    server 192.168.0.3:80;
}

server {
    listen 69.64.147.249;
    server_name www.example.com;
    rewrite ^(.*) http://example.com$1 permanent;
}

server {
    listen 69.64.147.249;
    server_name example.com app.example.com;;
    location / {
        access_log off;
        proxy_pass http://proxy1.example.com;

        proxy_set_header Host $host;
        proxy_set_header X-Real-IP 

```

$remote\_addr;  
proxy\_set\_header X-Forwarded-For  
$proxy\_add\_x\_forwarded\_for;  
client\_max\_body\_size 10m;  
client\_body\_buffer\_size 128k;  
proxy\_connect\_timeout 90;  
proxy\_send\_timeout 90;  
proxy\_read\_timeout 90;  
proxy\_buffer\_size 4k;  
proxy\_buffers 4 32k;  
proxy\_busy\_buffers\_size 64k;  
proxy\_temp\_file\_write\_size 64k;  
}  
}

```
server {
    listen 69.64.147.249;
    server_name *.app.example.com;
    location / {
        access_log off;
        proxy_pass http://proxy2.example.com;

        gzip off;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP 

```

$remote\_addr;  
proxy\_set\_header X-Forwarded-For  
$proxy\_add\_x\_forwarded\_for;  
client\_max\_body\_size 10m;  
client\_body\_buffer\_size 128k;  
proxy\_connect\_timeout 90;  
proxy\_send\_timeout 90;  
proxy\_read\_timeout 90;

```
        proxy_buffering on;
        proxy_buffer_size 4k;
        proxy_buffers 4 32k;
        proxy_busy_buffers_size 64k;
        proxy_temp_file_write_size 64k;
    }
}

server {
    listen 69.64.147.249;
    server_name static.example.com;
    access_log logs/access.www.example.com.log;

    # deny access to .htaccess files, if Apache's document root
    # concurs with nginx's one
    location ~ /\.ht {
        deny all;
    }
}

```

}

##########

---

<div class="post-metadata">

**Author:** ![Peter\_Langhans](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/p/54ee81/32.png) [@Peter\_Langhans](https://www.ruby-forum.com/u/Peter_Langhans)\
**Post date:** [February 24, 2009, 9:19pm UTC](https://www.ruby-forum.com/t/strange-error-nginx-does-not-flush-the-buffer-back-to-the-client/161179/2 "2009-02-24T21:19:37Z")

</div>

I would be sure to check the error logs (for nginx, apache, and maybe  
the  
system too), because this does not seem at all like normal behavior. If  
you  
find something in the error logs, you should provide that. If you  
can’t,  
you might try more hardcore debugging and provide an strace or  
something.

On Mon, Feb 23, 2009 at 9:24 PM, Peter Langhans \<