ClusterFlow

Distributed Multi-Node Backend Architecture using FastAPI, NGINX, and MongoDB

Distributed Architecture

Architecture Overview

                          Nginx (Load Balancer)
                               |
                          192.168.0.179:80
                               |
                _________________|_______________
               |                                 |
          FastAPI 1                         FastAPI 2
      192.168.0.120:8000              192.168.0.11:8001
               |                                 |
               |_______________|_________________|
                               |
                          MongoDB
                    192.168.0.179:27017
                    Database: "ems"
                    Collection: "meters"

Components

1. Nginx (Reverse Proxy & Load Balancer)

• Host: 192.168.0.179 (Windows Machine)
• Port: 80
• Role: Routes incoming HTTP requests to FastAPI instances
• Configuration: C:\nginx-1.29.3\conf\nginx.conf
• Status: Running

2. FastAPI Backend Servers

• FastAPI 1: 192.168.0.120:8000
• FastAPI 2: 192.168.0.11:8001
• Role: Process meter data, handle API requests
• Database Connection: MongoDB at 192.168.0.179:27017

3. MongoDB

• Host: 192.168.0.179
• Port: 27017
• Database: ems
• Collection: meters
• Role: Stores meter data (voltage, current, power, frequency, etc.)

---

API Endpoints

Home Route

GET /
Response: {"server": "hostname", "message": "EMS Backend Running"}

Insert Meter Data

POST /meters
Body: {
    "meter_id": "M001",
    "plant": "Plant A",
    "voltage": 230.5,
    "current": 15.2,
    "power": 3500.5,
    "frequency": 50.0,
    "power_factor": 0.95,
    "timestamp": "2026-05-15T14:30:00"
}
Response: {"status": "saved", "meter_id": "M001", "server": "hostname"}

Get All Meter Data

GET /meters
Response: [list of all meter records]

Get Data by Meter ID

GET /meters/{meter_id}
Example: GET /meters/M001
Response: [list of records for M001]

Get Latest Data for Meter

GET /latest/{meter_id}
Example: GET /latest/M001
Response: {latest meter record}

WebSocket (Real-time Data Stream)

WebSocket /ws
Continuously sends latest meter data every second

---

Access URLs

From Local Machine

http://localhost/
http://192.168.0.179/

From Another Device on Same LAN

http://192.168.0.179/

---

Setup Instructions

Prerequisites

• Python 3.8+
• MongoDB running at 192.168.0.179:27017
• Nginx running at 192.168.0.179:80

1. Install Python Dependencies

cd C:\Users\Admin\Desktop\testing
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

2. Install Required Packages

pip install fastapi uvicorn motor pydantic

3. Run FastAPI Server (on each backend machine)

On 192.168.0.120:

cd C:\Users\Admin\Desktop\testing
python main.py

Server starts at: http://192.168.0.120:8000

On 192.168.0.11:

cd C:\Users\Admin\Desktop\testing
python main.py

Server starts at: http://192.168.0.11:8001

4. Verify MongoDB Connection

# Test connection to MongoDB
mongosh --host 192.168.0.179 --port 27017

5. Start Nginx

cd C:\nginx-1.29.3
.\nginx.exe

---

Nginx Configuration

Located at: C:\nginx-1.29.3\conf\nginx.conf

Key Settings:

• Upstream cluster with two FastAPI servers
• Load balancing (round-robin by default)
• Keepalive connections: 16
• Timeout settings:
  • Connect: 3s
  • Read: 10s
  • Send: 10s
• Failover: If one server fails (5xx errors, timeout), request is sent to next upstream

---

Testing

1. Test Nginx Proxy

curl http://192.168.0.179/

2. Test Direct FastAPI

curl http://192.168.0.120:8000/
curl http://192.168.0.11:8001/

3. Insert Test Data

$body = @{
    meter_id = "M001"
    plant = "Plant A"
    voltage = 230.5
    current = 15.2
    power = 3500.5
    frequency = 50.0
    power_factor = 0.95
    timestamp = Get-Date -Format "o"
} | ConvertTo-Json

Invoke-WebRequest -Uri "http://192.168.0.179/meters" `
    -Method POST `
    -ContentType "application/json" `
    -Body $body

4. Retrieve Data

curl http://192.168.0.179/meters/M001

---

Troubleshooting

Nginx returns 502 Bad Gateway

• Cause: FastAPI servers not reachable
• Fix:
  • Verify FastAPI is running on 192.168.0.120:8000 and 192.168.0.11:8001
  • Check network connectivity: Test-NetConnection 192.168.0.120 -Port 8000
  • Restart nginx: C:\nginx-1.29.3\nginx.exe -s reload

MongoDB Connection Error

• Cause: MongoDB not running or wrong IP
• Fix:
  • Verify MongoDB is running: mongosh --host 192.168.0.179
  • Check connection string in main.py: mongodb://192.168.0.179:27017

Firewall Blocking Port 80

• Cause: Windows Firewall blocking inbound traffic
• Fix:

  New-NetFirewallRule -DisplayName "Allow Nginx Port 80" `
      -Direction Inbound -LocalPort 80 -Protocol TCP -Action Allow

Cannot Access from Remote Device

• Cause: Device not on same network or firewall blocks it
• Fix:
  • Verify device is on 192.168.0.x network
  • Check Windows Firewall allows port 80
  • Run: Test-NetConnection 192.168.0.179 -Port 80 from remote device

---

Performance Notes

• Load Balancing: Nginx distributes requests equally between both FastAPI servers
• Failover: If one server goes down, traffic is automatically routed to the other
• Keepalive: Reduces connection overhead (16 concurrent connections per upstream)
• MongoDB: Single instance (consider replica set for production HA)

---

Production Recommendations

1. Enable HTTPS: Add SSL/TLS certificates to Nginx
2. Health Checks: Add nginx health check directives
3. MongoDB: Deploy as replica set for high availability
4. Logging: Enable detailed request logging in Nginx and FastAPI
5. Monitoring: Add monitoring for Nginx, FastAPI, and MongoDB
6. Auto-restart: Configure Windows Services or use process managers for auto-restart

---

Files

• Nginx Config: C:\nginx-1.29.3\conf\nginx.conf
• FastAPI App: C:\Users\Admin\Desktop\testing\main.py
• Requirements: C:\Users\Admin\Desktop\testing\requirements.txt
• MongoDB: External at 192.168.0.179:27017

---

Contact & Support

For issues or questions about this setup, verify:

1. All services are running
2. Network connectivity between components
3. MongoDB is accessible
4. Nginx is not reporting errors in C:\nginx-1.29.3\logs\error.log