Monitor Raspberry Pi Performance

Introduction

Keeping tabs on your Raspberry Pi's health is essential for stability and performance, especially in remote deployments. This guide covers essential monitoring tools—from built-in commands to full stacks like Prometheus and Grafana—so you can track CPU, memory, temperature, and disk usage in real-time.

Prerequisites

• Raspberry Pi 3B+ or newer
• SSH access or terminal
• Optional: Docker for advanced monitoring stacks

Step 1 — Quick System Checks with Built-in Tools

Start with commands already on your Pi. These give you instant insight without additional setup:

# Check CPU temperature
vcgencmd measure_temp

# Display memory usage
free -h

# Check disk usage
df -h

# Check system uptime and load average
uptime

Run these commands to establish a baseline. Temperature above 70°C indicates throttling risk on Pi 4/5.

Step 2 — Real-Time Process Monitoring with htop

Install htop for interactive process viewing:

sudo apt install htop -y
htop

Navigate with arrow keys, press T to sort by memory, P for CPU. Press q to exit. htop shows per-core usage clearly—useful for spotting single-threaded bottlenecks.

Step 3 — Install Netdata for Web Dashboard

Netdata offers a lightweight (~100MB RAM), real-time web UI. Install with one command:

wget -O /tmp/netdata-kickstart.sh https://get.netdata.cloud/kickstart.sh
sh /tmp/netdata-kickstart.sh --stable-channel --disable-telemetry

Access the dashboard at http://your-pi-ip:19999. Netdata tracks CPU, memory, disk I/O, network, and temperature automatically. No configuration needed.

Step 4 — Advanced Stack: Prometheus + Node Exporter + Grafana

For persistent historical data, set up Prometheus:

# Create monitoring directory
mkdir -p ~/monitoring && cd ~/monitoring

# Create docker-compose.yml
cat > docker-compose.yml <<'EOF'
version: '3.8'
services:
  node-exporter:
    image: arm32v6/node-exporter:latest
    container_name: node-exporter
    restart: always
    ports:
      - "9100:9100"
    volumes:
      - /:/rootfs:ro
      - /sys:/sys:ro
      - /proc:/proc:ro
    command:
      - '--path.rootfs=/rootfs'
      - '--path.sysfs=/sys'
      - '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'

  prometheus:
    image: arm32v6/prometheus:latest
    container_name: prometheus
    restart: always
    ports:
      - "9090:9090"
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml
      - prometheus_data:/prometheus
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
      - '--storage.tsdb.retention.time=30d'

  grafana:
    image: grafana/grafana:latest
    container_name: grafana
    restart: always
    ports:
      - "3000:3000"
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=admin
    volumes:
      - grafana_data:/var/lib/grafana

volumes:
  prometheus_data:
  grafana_data:
EOF

# Create prometheus.yml
cat > prometheus.yml <<'EOF'
global:
  scrape_interval: 15s
  evaluation_interval: 15s

scrape_configs:
  - job_name: 'node'
    static_configs:
      - targets: ['localhost:9100']
EOF

# Start services
docker-compose up -d

Access Grafana at http://your-pi-ip:3000 (default: admin/admin). Add Prometheus data source at http://localhost:9090, then import dashboard ID 1860 for Node Exporter metrics.

Step 5 — Custom Telegram Alert Script

Alert when temperature exceeds 70°C:

cat > ~/temp-alert.sh <<'EOF'
#!/bin/bash
TELEGRAM_TOKEN="your-bot-token"
TELEGRAM_CHAT_ID="your-chat-id"
THRESHOLD=70

while true; do
  TEMP=$(vcgencmd measure_temp | grep -oP '\d+\.\d+')
  TEMP_INT=${TEMP%.*}
  
  if [ "$TEMP_INT" -gt "$THRESHOLD" ]; then
    curl -s -X POST "https://api.telegram.org/bot$TELEGRAM_TOKEN/sendMessage" \
      -d "chat_id=$TELEGRAM_CHAT_ID" \
      -d "text=Alert: Pi temperature is ${TEMP}°C"
  fi
  
  sleep 300  # Check every 5 minutes
done
EOF

chmod +x ~/temp-alert.sh
nohup ~/temp-alert.sh > /tmp/temp-alert.log 2>&1 &

Replace your-bot-token and your-chat-id with actual values from BotFather and your Telegram account.

Monitoring Tools Comparison

Tool	Memory	CPU	Real-Time	Historical	Web UI
Built-in (vcgencmd)	<5MB	minimal	yes	no	no
htop	~10MB	minimal	yes	no	no
Netdata	100MB	2-3%	yes	24h default	yes
Prometheus	150-200MB	3-5%	yes	30+ days	yes (Grafana)

Troubleshooting

High memory usage despite low workload Netdata and Prometheus grow over time. Reduce retention: edit docker-compose and lower storage.tsdb.retention.time to 7d or restart services to clear old data.

Temperature alerts firing too often Increase THRESHOLD in the alert script or extend the sleep interval to 600 seconds (10 minutes) for less noise.

Grafana dashboard not updating Verify Node Exporter is running: curl http://localhost:9100/metrics | head -20. Restart Prometheus if metrics are stale.

Port already in use If port 9090 is occupied, change Prometheus port: edit docker-compose to "9091:9090" and update Grafana data source accordingly.

Summary

Use built-in tools for quick health checks, Netdata for a lightweight dashboard, and Prometheus/Grafana for production monitoring with alerting. Most homelabs thrive on Netdata alone—only reach for the full stack if you need historical analysis or multi-Pi federation.