Hi,

first, the setup: I have 2 identical maschines running latest CentOS 6.6
and uWSGI 2.0.10, both machines had nginx-1.4.7-1.el6.ngx.x86_64
installed. I saw almost identical values for $request_time in the Nginx
log files of both machines.

We are using Nginx to serve some static files and pass requests to an
uWSGI application.

The relevant configuration is as follows:

upstream backend {
server unix:///tmp/backend_blue.socket max_fails=0;
server unix:///tmp/backend_blue.socket max_fails=0;
}

location / {
proxy_next_upstream error timeout http_502;
proxy_set_header Host $host;
include uwsgi_params;
uwsgi_param HTTP_X_REQUEST_ID $uuid;
uwsgi_param HTTPS on;
uwsgi_buffering off;
uwsgi_pass backend;
}

location /static {
alias /home/webadmin/blue/static;
index index.php;
if ( !-e $request_filename) {
rewrite ^(/static.*)/test__(.*)$ $1/$2 redirect;
}
}

After upgrading Nginx to one of the following stable releases I've
discovered that the values for $request_time have significantly
increased (almost twice as high) but only for upstream requests.
There are no differences in $request_time for static requests.

Affected packages:
nginx-1.6.0-1.el6.ngx.x86_64
nginx-1.6.1-1.el6.ngx.x86_64
nginx-1.6.2-1.el6.ngx.x86_64
nginx-1.6.3-1.el6.ngx.x86_64
nginx-1.8.0-1.el6.ngx.x86_64

No further changes were made, only an update of nginx RPM. uWSGI itself
reports unchanged times to generate responses.

Any idea?

_______________________________________________
nginx mailing list
nginx@nginx.org
http://mailman.nginx.org/mailman/listinfo/nginx