99.999% my config works - then default_server is used

[Michal Kralik]

Hi,

We're facing super strange issues with nginx 1.10.3 (CentOS 7,
3.10.0-327.36.3.el7.x86_64). Our config works ok, we get a lot of traffic,
but every day a couple of requests (5-10) don't get processes correctly by
a server directive with the defined server_name, but rather by a server
directive with listen's "default_server". Here's the config
https://paste.ngx.cc/29345124359e0447

Most of the time a request to 01.app.example.com gets processed correctly,
but every day a couple of requests get processed by the server directive
from file conf.d/01-default.conf. There's nothing in the nginx system logs
around the time when this issue happens.

The access.log from the default_server shows this
(/data/logs/nginx/access.log):
1.1.1.1 - - [06/Apr/2017:12:20:57 +0000] "01.app.example.com" "GET
/robots.txt HTTP/1.1" 302 161 "-" "Mozilla/5.0 (compatible; Googlebot/2.1; +
http://www.google.com/bot.html)" "-"

It's clearly a redirect processed by the default_server, but the request
should be rather processed by server directive in conf.d/02-apps.conf
because the host header clearly states "01.app.example.com".

Any ideas what's going on?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nginx.org/pipermail/nginx/attachments/20170406/497fae47/attachment.html>