Nginx: Unknown log format error

I have a proxy set up that is taking UDP traffic to load balance internal DNS. It is routing the traffic properly to my backend DNS servers, but I cannot seem to get the log formatting to work correctly.

The error I'm receiving is:

nginx[32103]: nginx: [emerg] unknown log format "dns" in /etc/nginx/nginx.conf:23

I have the log format in the http block, and define it in the stream-->server block. I tried to set the log_format within the stream-->server block, but none of the variables would work.

    user www-data;
    worker_processes auto;
    pid /run/nginx.pid;
    include /etc/nginx/modules-enabled/*.conf;
    
    events {
            worker_connections 768;
            # multi_accept on;
    }
    stream {
    
      upstream dns_servers {
        least_conn;
        server 192.168.3.222:53 fail_timeout=10s;
        server 192.168.70.80:53 fail_timeout=10s;
        server 192.168.70.81:53 fail_timeout=10s;
      }
      server { 
        listen 53 udp;
        proxy_pass dns_servers;
        proxy_responses 1;
        proxy_timeout 1s;
        access_log /var/log/nginx/access.log dns;
    
      }
    }
    
    http {
    
            sendfile on;
            tcp_nopush on;
            tcp_nodelay on;
            keepalive_timeout 65;
            types_hash_max_size 2048;
    
            include /etc/nginx/mime.types;
            default_type application/octet-stream;
    
            ssl_protocols TLSv1 TLSv1.1 TLSv1.2; # Dropping SSLv3, ref: POODLE
            ssl_prefer_server_ciphers on; 
    
            log_format dns '[$time_local] | $remote_addr | $remote_user | $server_name $host to: $upstream_addr | '
                               '"$request" | $status | upstream_response_time $upstream_response_time msec '
                               '$msec | request_time $request_time';
    
            access_log /var/log/nginx/access.log dns;
            error_log /var/log/nginx/error.log;
}

I tried to place log_format in the stream block, but I start getting unknown <variable name> variable.

user www-data;
worker_processes auto;
pid /run/nginx.pid;
include /etc/nginx/modules-enabled/*.conf;

events {
        worker_connections 768;
        # multi_accept on;
}
stream {
 log_format dns '[$time_local] | $remote_addr | $remote_user | $server_name $host to: $upstream_addr | '
                           '"$request" | $status | upstream_response_time $upstream_response_time msec '
                           '$msec | request_time $request_time';

  upstream dns_servers {
    least_conn;
    server 192.168.3.222:53 fail_timeout=10s;
    server 192.168.70.80:53 fail_timeout=10s;
    server 192.168.70.81:53 fail_timeout=10s;
  }
  server { 
    listen 53 udp;
    proxy_pass dns_servers;
    proxy_responses 1;
    proxy_timeout 1s;
    access_log /var/log/nginx/access.log dns;

  }
}

This will not be a web server at all. It will only serve as a proxy for DNS traffic. How do I properly get the logging to function correctly? This is a default nginx.conf file except for the stream block.


Solution 1:

A log_format for use by stream servers must be declared in the stream block, rather than the http block. These directives are independent of each other. See the documentation for the stream log_format.