Skip to content

Getting Started

Prerequisites

  • Raspberry Pi 5 with fresh 64-bit Raspberry Pi OS (Bookworm)
  • NVMe SSD connected via HAT
  • 7× Raspberry Pi 3B with SD cards
  • 1× Raspberry Pi 4 with IMX500 AI Camera
  • TP-Link TL-SG108E Gigabit switch
  • All nodes connected via ethernet

Step 1 — Flash Pi 5 OS

Flash Raspberry Pi OS 64-bit (Bookworm) to NVMe SSD using Raspberry Pi Imager:

  • Hostname: raspberrypi
  • Username: pi
  • Password: 1234
  • Enable SSH: ✅

Step 2 — Clone Repository

git clone https://github.com/Sapkale26/Cloud_Computing_Project.git ~/cluster
cd ~/cluster

Step 3 — Bootstrap Pi 5

chmod +x scripts/bootstrap-pi5.sh
./scripts/bootstrap-pi5.sh

This installs all dependencies, configures dnsmasq, sets up NVMe, and installs k3s.

Step 4 — Setup SSH Keys

./scripts/setup-ssh-keys.sh

This distributes SSH keys from Pi 5 to all Pi 3s and Pi 4.

Step 5 — Configure Worker Nodes

cd infrastructure/ansible
ansible-playbook -i inventory.yml playbooks/setup-all.yml

This installs OpenMPI, OpenCV, node_exporter on all workers.

Step 6 — Start Everything

./scripts/start-cluster.sh

Step 7 — Start Pi 4 Camera

ssh pi@192.168.50.98
source yolo-env/bin/activate
python sender.py

Access Services

Service URL Credentials
Dashboard http://192.168.50.1:30080 —
Grafana http://192.168.50.1:3000 admin/admin
Prometheus http://192.168.50.1:9090 —
MinIO Console http://192.168.50.1:9001 admin/admin123
Backend API http://192.168.50.1:5000 —

Health Check

./scripts/health-check.sh