// Tutorial //

How To Set Up Nginx Load Balancing with SSL Termination

Published on July 17, 2014
Default avatar
By Jesin A
Developer and author at DigitalOcean.
How To Set Up Nginx Load Balancing with SSL Termination

Introduction

This article shows you how to set up Nginx load balancing with SSL termination with just one SSL certificate on the load balancer. This will reduce your SSL management overhead, since the OpenSSL updates and the keys and certificates can now be managed from the load balancer itself.

About SSL Termination

Nginx can be configured as a load balancer to distribute incoming traffic around several backend servers. SSL termination is the process that occurs on the load balancer which handles the SSL encryption/decryption so that traffic between the load balancer and backend servers is in HTTP. The backends must be secured by restricting access to the load balancer’s IP, which is explained later in this article.

SSL Termination Diagram

Prerequisites

In this tutorial the commands must be run as the root user or as a user with sudo privileges. You can see how to set that up in the Users Tutorial.

The following guides can be used as reference:

A LAMP server is not required, but we’ll be using it as an example in this tutorial.

Setup

This tutorial makes use of the following 3 droplets:

Droplet 1 (Frontend)

  • Image: Ubuntu 14.04
  • Hostname: loadbalancer
  • Private IP: 10.130.227.33

Droplet 2 (Backend)

  • Image: Ubuntu 14.04
  • Hostname: web1
  • Private IP: 10.130.227.11

Droplet 3 (Backend)

  • Image: Ubuntu 14.04
  • Hostname: web2
  • Private IP: 10.130.227.22

Domain name - example.com

All these Droplets must have private networking enabled.

Update and upgrade the software on all three servers:

apt-get update && apt-get upgrade -y

Reboot each server to apply the upgrades. This is important, since OpenSSL needs to be on its latest version to be secure.

We will be setting up a new Nginx virtual host for the domain name with the upstream module load balancing the backends.

Prior to setting up Nginx loadbalancing, you should have Nginx installed on your VPS. You can install it quickly with apt-get:

apt-get install nginx

On the two backend servers, update your repositories and install Apache:

apt-get install apache2

Install PHP on both backend servers:

apt-get install php5 libapache2-mod-php5 php5-mcrypt

For more information, see this article.

Generate Keys And Create An SSL Certificate

In this section, you will run through the steps needed to create an SSL certificate. This article explains in detail about SSL certificates on Nginx.

Create the SSL certificate directory and switch to it.

mkdir -p /etc/nginx/ssl/example.com
cd /etc/nginx/ssl/example.com

Create a private key:

openssl genrsa -des3 -out server.key 2048

Remove its passphrase:

openssl rsa -in server.key -out server.key

Create a CSR (Certificate Signing Request):

openssl req -new -key server.key -out server.csr

Use this CSR to obtain a valid certificate from a certificate authority or generate a self-signed certificate with the following command.

openssl x509 -req -days 365 -in server.csr -signkey server.key -out server.crt

Once this is done this directory will contain the following files:

  • server.key - The private key
  • ca-certs.pem - A collection of your CA’s root and intermediate certificates. Only present if you obtained a valid certificate from a CA.
  • server.crt - The SSL certificate for your domain name

Virtual Host File And Upstream Module

Create a virtual hosts file inside the Nginx directory

nano /etc/nginx/sites-available/example.com

Add the upstream module containing the private IP addresses of the backend servers

upstream mywebapp1 {
    server 10.130.227.11;
    server 10.130.227.22;
}

Begin the server block after this line. This block contains the domain name, references to the upstream servers, and headers that should be passed to the backend.

server {
    listen 80;
    server_name example.com www.example.com;
	
    location / {
        proxy_pass http://mywebapp1;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

The proxy_set_header directive is used to pass vital information about the request to the upstream servers.

Save this file and create a symbolic link to the sites-enabled directory.

ln -s /etc/nginx/sites-available/example.com /etc/nginx/sites-enabled/example.com

Perform a configuration test to check for errors.

service nginx configtest

If no errors are displayed, reload the nginx service.

service nginx reload

Load balancing has now been configured for HTTP.

Enable SSL

Add the following directives to the virtual hosts file (/etc/nginx/sites-available/example.com) inside the server {} block. These lines will be shown in context in the next example.

listen 443 ssl;
ssl on;
ssl_certificate         /etc/nginx/ssl/example.com/server.crt;
ssl_certificate_key     /etc/nginx/ssl/example.com/server.key;
ssl_trusted_certificate /etc/nginx/ssl/example.com/ca-certs.pem;

Ignore the ssl_trusted_certificate directive if you are using self-signed certificates. Now the server block should look like this:

server {
    listen 80;
    listen 443 ssl;
    server_name example.com www.example.com;

    ssl on;
    ssl_certificate         /etc/nginx/ssl/example.com/server.crt;
    ssl_certificate_key     /etc/nginx/ssl/example.com/server.key;
    ssl_trusted_certificate /etc/nginx/ssl/example.com/ca-certs.pem;

    location / {
        proxy_pass http://mywebapp1;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Check for configuration errors and reload the Nginx service.

service nginx configtest && service nginx reload

Securing The Backend Servers

Currently, the website hosted on the backend servers can be directly accessed by anyone who knows its public IP address. This can be prevented by configuring the web servers on the backends to listen only on the private interface. The steps to do this in Apache are as follows.

Edit the ports.conf file.

nano /etc/apache2/ports.conf

Find the following line:

Listen 80

Replace it with the backend server’s own private IP address:

Listen 10.130.227.22:80

Do this on all the backend servers and restart Apache.

service apache2 restart

The next step is to restrict HTTP access to the load balancer’s private IP. The following firewall rule achieves this.

iptables -I INPUT -m state --state NEW -p tcp --dport 80 ! -s 10.130.227.33 -j DROP

Replace the example with the load balancer’s private IP address and execute this rule on all the backend servers.

Testing The Setup

Create a PHP file on all the backend servers (web1 and web2 in this example). This is for testing and can be removed once the setup is complete.

nano /var/www/html/test.php

It should print the accessed domain name, the IP address of the server, the user’s IP address, and the accessed port.

<?php
    header( 'Content-Type: text/plain' );
    echo 'Host: ' . $_SERVER['HTTP_HOST'] . "\n";
    echo 'Remote Address: ' . $_SERVER['REMOTE_ADDR'] . "\n";
    echo 'X-Forwarded-For: ' . $_SERVER['HTTP_X_FORWARDED_FOR'] . "\n";
    echo 'X-Forwarded-Proto: ' . $_SERVER['HTTP_X_FORWARDED_PROTO'] . "\n";
    echo 'Server Address: ' . $_SERVER['SERVER_ADDR'] . "\n";
    echo 'Server Port: ' . $_SERVER['SERVER_PORT'] . "\n\n";
?>

Access this file several times with your browser or using curl. Use curl -k on self-signed certificate setups to make curl ignore SSL errors.

curl https://example.com/test.php https://example.com/test.php https://example.com/test.php

The output will be similar to the following.

   Host: example.com
   Remote Address: 10.130.245.116
   X-Forwarded-For: 117.193.105.174
   X-Forwarded-Proto: https
   Server Address: 10.130.227.11
   Server Port: 80

   Host: example.com
   Remote Address: 10.130.245.116
   X-Forwarded-For: 117.193.105.174
   X-Forwarded-Proto: https
   Server Address: 10.130.227.22
   Server Port: 80

   Host: example.com
   Remote Address: 10.130.245.116
   X-Forwarded-For: 117.193.105.174
   X-Forwarded-Proto: https
   Server Address: 10.130.227.11
   Server Port: 80

Note that the Server Address changes on each request, indicating that a different server is responding to each request.

Hardening SSL Configuration

This section explains configuring SSL according to best practices to eliminate vulnerabilities with older ciphers and protocols. Individual lines are shown in this section and the complete configuration file is shown in the last section of this tutorial.

Enabling SSL session cache improves the performance of HTTPS websites. The following directives must be placed after ssl_trusted_certificate. They enable shared caching of size 20MB with a cache lifetime of 10 minutes.

ssl_session_cache shared:SSL:20m;
ssl_session_timeout 10m;

Specify the protocols and ciphers to be used in the SSL connection. Here we have omitted SSLv2 and disabled insecure ciphers like MD5 and DSS.

ssl_prefer_server_ciphers       on;
ssl_protocols                   TLSv1 TLSv1.1 TLSv1.2;
ssl_ciphers                     ECDH+AESGCM:DH+AESGCM:ECDH+AES256:DH+AES256:ECDH+AES128:DH+AES:ECDH+3DES:DH+3DES:RSA+AESGCM:RSA+AES:RSA+3DES:!aNULL:!MD5:!DSS;

Strict Transport Security instructs all supporting web browsers to use only HTTPS. Enable it with the add_header directive.

add_header Strict-Transport-Security "max-age=31536000";

Check for configuration errors and reload the Nginx service.

service nginx configtest && service nginx reload

Complete Configuration

After configuring and hardening SSL termination, the complete configuration file will look like this:

/etc/nginx/sites-available/example.com

upstream mywebapp1 {
    server 10.130.227.11;
    server 10.130.227.22;
}

server {
    listen 80;
    listen 443 ssl;
    server_name example.com www.emxaple.com;

    ssl on;
    ssl_certificate         /etc/nginx/ssl/example.com/server.crt;
    ssl_certificate_key     /etc/nginx/ssl/example.com/server.key;
    ssl_trusted_certificate /etc/nginx/ssl/example.com/ca-certs.pem;
	
    ssl_session_cache shared:SSL:20m;
    ssl_session_timeout 10m;
	
    ssl_prefer_server_ciphers       on;
    ssl_protocols                   TLSv1 TLSv1.1 TLSv1.2;
    ssl_ciphers                     ECDH+AESGCM:DH+AESGCM:ECDH+AES256:DH+AES256:ECDH+AES128:DH+AES:ECDH+3DES:DH+3DES:RSA+AESGCM:RSA+AES:RSA+3DES:!aNULL:!MD5:!DSS;
	
    add_header Strict-Transport-Security "max-age=31536000";

    location / {
        proxy_pass http://mywebapp1;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Do a SSL Server Test and this setup should get an A+ grade. Run the curl test again to check if everything is working properly.

curl https://example.com/test.php https://example.com/test.php https://example.com/test.php

Further Reading

To learn more about load-balancing algorithms read this article.


Want to learn more? Join the DigitalOcean Community!

Join our DigitalOcean community of over a million developers for free! Get help and share knowledge in our Questions & Answers section, find tutorials and tools that will help you grow as a developer and scale your project or business, and subscribe to topics of interest.

Sign up
About the authors
Default avatar
Jesin A

author

Developer and author at DigitalOcean.

Default avatar
Developer and author at DigitalOcean.

Still looking for an answer?

Was this helpful?
10 Comments

If you have a app behind the SSL that need a stick or persistent session (Magento for example), I strongly reccomend that you put “ip_hash” algorithm instead the regular “round robin”, this way your clients will not be redirected on every request.

Source: http://nginx.org/en/docs/http/load_balancing.html#nginx_load_balancing_with_ip_hash

Overall, great tutorial !!! However, it would be better to add some info about which plan to choose, and how many users/requests can handle load-balancer/web-servers approximately.

How does this setup disallows network sniffing from within the DO internal network itself ? It doesn’t

Excellent. I had to remove “ssl on;”, otherwise I get the message “The plain HTTP request was sent to HTTPS port”: https://stackoverflow.com/questions/8768946/dealing-with-nginx-400-the-plain-http-request-was-sent-to-https-port-error

Hello i want Redirect all invalid https domain ( havent SSL), redirect to http in nginx SSL termination. I do not want to do this for all sites Separately for each one, I am a general solution for all requests dont have https but open with https and redirect to http Automatically. i have nginx for SSL termination. Thanks

Hi,

Thanks for the tutorial, it helped me a lot.

However, the following command didn’t work for me (with the current ubuntu 16.04 droplet): iptables -I INPUT -m state --state NEW -p tcp --dport 80 ! -s LB_IP_ADDRESS -j DROP

This one however seemed to be working: iptables -I INPUT \! --src LB_IP_ADDRESS -m tcp -p tcp --dport 80 -j DROP

I am still wondering though, I am using Nginx on the backend servers, and couldn’t get the listen directive to work on only the private IP address of the LB. It gives me a bind() error every time. If anyone knows how to fix this that would be great. I haven’t seen any tutorials about nginx load balancing covering the security regarding private network to allow only connection coming via the LB.

If your ISP works for KGB, FSB, CIA or Mosad this might not work (it’s easy to fake 10.130.227.33 for an ISP): iptables -I INPUT -m state --state NEW -p tcp --dport 80 ! -s 10.130.227.33 -j DROP

Would be great to see how to setup a secure channel. What are the options:

  1. VPN (might behave like tamagotchi);
  2. Https between Nginx LB and webapp servers ( can we just use an IP like proxy_pass http://10.130.227.11 then? what would happen to redirects ?);
  3. Any other idea ?

Friend how i setup this using easyengine? is possible? can you help me!?

I do not understand what to put in the mywebapp1 field…

Good tutorial…and it would have it saved me hours and hours of frustration if you said that the 3 droplets **have to be on the same datacenter **because the private networking doesn’t working across datacenters.

Besides that - great piece.