2.5. Configuring NGINX as an HTTP load balancer

Procedure

1. Edit the /etc/nginx/nginx.conf file and add the following settings:

   http {
       upstream backend {
           least_conn;
           server server1.example.com;
           server server2.example.com;
           server server3.example.com backup;
       }
   
       server {
           location / {
               proxy_pass http://backend;
           }
       }
   }

   The least_conn directive in the host group named backend defines that NGINX sends requests to server1.example.com or server2.example.com, depending on which host has the least number of active connections. NGINX uses server3.example.com only as a backup in case that the other two hosts are not available.

   With the proxy_pass directive set to http://backend, NGINX acts as a reverse proxy and uses the backend host group to distribute requests based on the settings of this group.

   Instead of the least_conn load balancing method, you can specify:

   • No method to use round robin and distribute requests evenly across servers.
   • ip_hash to send requests from one client address to the same server based on a hash calculated from the first three octets of the IPv4 address or the whole IPv6 address of the client.
   • hash to determine the server based on a user-defined key, which can be a string, a variable, or a combination of both. The consistent parameter configures that NGINX distributes requests across all servers based on the user-defined hashed key value.
   • random to send requests to a randomly selected server.

2. Restart the nginx service:

   # systemctl restart nginx