Highly Available Web Infrastructure

Tech Stack

• Ubuntu 22.04 LTS
• Vagrant
• VirtualBox
• NGINX
• Keepalived (VRRP)
• BIND9 DNS
• Bash

Project Structure

.
├── Vagrantfile
├── provision/
│   ├── dns.sh
│   └── web.sh
├── config/
│   ├── bind/
│   │   ├── db.test.com
│   │   ├── named.conf.local
│   │   └── named.conf.options
│   └── nginx/
│       ├── master.conf
│       ├── backup.conf
│       ├── web1.conf
│       ├── web2.conf
│       ├── web1-index.html
│       └── web2-index.html
├── screenshots/
│   ├── dig_via_the_dns_server.png
│   ├── resloving_the_master.png
│   ├── ping.png
│   └── after_halting_web1.png
└── README.md

Flow

Client → BIND9 DNS → Virtual IP → Web1 (MASTER)
                              └→ Web2 (BACKUP)

1. Start All Machines

vagrant up

2. Test DNS Resolution

dig @192.168.56.10 test.com

3. Send a Request to the Master

curl --resolve test.com:80:$(dig +short @192.168.56.10 test.com) http://test.com

4. Stop the Master to Trigger Failover

vagrant halt web1

5. Send the Request Again

curl --resolve test.com:80:$(dig +short @192.168.56.10 test.com) http://test.com

The same virtual IP now serves the response from web2.

Results

Machine	Role	IP Address
dns1	BIND9 DNS server	192.168.56.10
web1	NGINX master	192.168.56.11
web2	NGINX backup	192.168.56.12
Virtual IP	Keepalived service address	192.168.56.20

• test.com and www.test.com resolve to the virtual IP 192.168.56.20.
• NGINX is served by web1 as the master node.
• Traffic automatically fails over to web2 when web1 is unavailable.

DNS Lookup

Master Server

Connectivity Test

Automatic Failover