check https://nginx.org/en/docs/http/ngx_http_sub_module.html

alrighty, let me read on that, thanks

Ребята подскажите я могу параметр proxy_read_timeout на весь сервер поставить? то есть в глобальный конфиг?

подскажите, установлен модуль rtmp и транслирует картинку нормально в rtmp://ip:port
можно ли nginx настроить чтобы ещё в rtsp транслировал?

Hi, may I clear a doubt here regarding an odd behavior in Nginx?
(forgive me for writing in english, I can't speak Russian)

Do you run it in docker?
Then it's a docker's behaviour, not nginx's one :)

Nope, it's running on a VMWare instance with Ubuntu 20.04, x86_64

Share some details then

According to the official documentation, access_log can be used in the contexts http, server, location, if in location, and limit_except.

I have 2 virtualhosts. One is a catch-all virtualhost with por 80 declared as default_server and the other is an actual website. The idea is that when someone tries to access by IP address, the request will be processed by the default virtualhost.

However, what is happening is that if I declare access_log within a location block in my default virtualhost, not only the request isn't logged, but passes to the actual website virtualhost.

If the access_log is declared at the server block instead, everything works as expected.

I have no Idea why is this happening. This is with Nginx 1.18, on Ubuntu 20.04 x86_64

Please share the whole configuration.
I guess, there's just a typo in config.

via pastebin

it's rather short:

It works as expected this way:

server {
  listen 80 default_server;

  access_log /var/log/nginx/catchall.log custom;

  location / {
    return 404;
  }
}

But not this way:

server {
  listen 80 default_server;

  location / {
    access_log /var/log/nginx/catchall.log custom;
    return 404;
  }
}

server {
  listen 44480 default_server;

  location / {
    access_log /tmp/nginx_test/default-access.log main ;
    return 404;
  }
}

server {
  listen 44480 ;
  server_name lala ;

  access_log /tmp/nginx_test/vhost-access.log main ;

  location / {
    return 200 "OK";
  }
}

Works as expected.

Not in my case. Could it be a result of my custom log format?

Yep. That's why I've asked for a configuration.

I just removed the custom format and same thing happens.

I'm done with guessing here. It's clear that the request is matched with other vhost somehow.

That's what puzzles me. It should never match that because it's a named virtualhost.

I just found what it is. I had an include in the default virtualhost with a few custom error_page instructions.