Blue/Green Deployment for FastAPI with Nginx and Ansible

This project implements zero-downtime deployments for FastAPI applications using Ansible and Nginx with a blue/green deployment strategy.

Prerequisites

• Vagrant 2.2+
• Ansible 2.9+
• VirtualBox 6.0+
• Python 3.6+

Infrastructure Overview

• Load Balancer: lb.example.com (Nginx)
• Blue Environment: blue.example.com
• Green Environment: green.example.com

File Structure

.
├── playbooks/
│   ├── deploy.yml            # Deploys application
│   ├── setup_loadbalancer.yml # Configures Nginx
│   ├── switch.yml           # Switches traffic
│   └── cleanup.yml          # Cleans old deployments
├── templates/
│   ├── fastapi.service.j2    # Systemd service template
│   └── loadbalancer.conf.j2  # Nginx config template
├── inventory/
│   └── production           # Inventory file
└── app/
    ├── blue/                # Blue deployment files
    └── green/               # Green deployment files

Deployment Workflow

1. Initial Setup

# Create all VMs
vagrant up

# Configure load balancer
ansible-playbook -i inventory/production playbooks/setup_loadbalancer.yml

2. First Deployment (Blue)

ansible-playbook -i inventory/production playbooks/deploy.yml \
  -e "target_group=blue deployment_color=blue"

3. Deploy New Version (Green)

ansible-playbook -i inventory/production playbooks/deploy.yml \
  -e "target_group=green deployment_color=green"

4. Test Green Environment

# Directly access green environment
curl http://green.example.com:8000

# Or through load balancer with test header
curl -H "X-Deployment: green" http://lb.example.com

5. Switch Traffic to Green

ansible-playbook -i inventory/production playbooks/switch.yml \
  -e "deployment_color=green"

6. (Optional) Clean Up Blue

ansible-playbook -i inventory/production playbooks/cleanup.yml \
  -e "old_deployment_group=blue old_deployment_color=blue"

Rollback Procedure

# Switch back to blue
ansible-playbook -i inventory/production playbooks/switch.yml \
  -e "deployment_color=blue"

# Clean up green if needed
ansible-playbook -i inventory/production playbooks/cleanup.yml \
  -e "old_deployment_group=green old_deployment_color=green"

Maintenance Commands

Check current traffic distribution:

vagrant ssh lb
sudo tail -f /var/log/nginx/access.log | awk '{print $6}' | sort | uniq -c

Validate Nginx config:

vagrant ssh lb
sudo nginx -t

View service status:

# On blue/green nodes
systemctl status fastapi-blue
systemctl status fastapi-green

Customization

Environment Variables

Set in inventory/production:

[all:vars]
app_port=8000
domain=example.com

Weighted Traffic Distribution

For canary releases:

ansible-playbook -i inventory/production playbooks/switch.yml \
  -e "deployment_color=green blue_weight=90 green_weight=10"

ansible-playbook -i inventory/production playbooks/switch.yml \
  -e "deployment_color=green blue_weight=50 green_weight=50"

ansible-playbook -i inventory/production playbooks/switch.yml \
  -e "deployment_color=green blue_weight=10 green_weight=90"

ansible-playbook -i inventory/production playbooks/switch.yml \
  -e "deployment_color=green blue_weight=0 green_weight=100"

Best Practices

1. Always test new deployments before switching traffic
2. Monitor both environments during cutover
3. Keep previous deployment until new one is verified stable
4. Use low DNS TTL if using DNS-based switching
5. Implement health checks in your application

Troubleshooting

Nginx fails to reload:

• Check syntax: sudo nginx -t
• Verify ports are open
• Ensure DNS resolution works

Service fails to start:

• Check logs: journalctl -u fastapi-blue
• Verify Python dependencies
• Check port availability

This README provides:
1. Clear step-by-step deployment instructions
2. Rollback procedures
3. Maintenance commands
4. Customization options
5. Troubleshooting tips
6. Best practices

The workflow ensures zero-downtime deployments with proper testing and rollback capabilities.