Skip to main content

LED Control Demo - Grafana Monitoring

This is a Grafana monitoring system for visualizing LED ON/OFF tick events of TIMPANI vs Normal LED controllers.

Architecture​

┌─────────────────────────────────────────────────────────────┐
│                    Guest Node                              │
│                                                            │
│  ┌──────────────────┐       ┌──────────────────┐           │
│  │ led-timpani-ctrl │       │ led-normal-ctrl  │           │
│  │   (C + TIMPANI)  │       │     (Python)     │           │
│  │   Port: 9101     │       │   Port: 9102     │           │
│  └────────┬─────────┘       └────────┬─────────┘           │
│           │                          │                     │
│           │   Prometheus Metrics     │                     │
│           └──────────┬───────────────┘                     │
│                      │                                   │
│           ┌──────────▼───────────┐                        │
│           │     Prometheus       │                        │
│           │     Port: 9090       │                        │
│           └──────────┬───────────┘                        │
│                      │                                   │
│           ┌──────────▼───────────┐                        │
│           │      Grafana         │                        │
│           │     Port: 3000       │                        │
│           └──────────────────────┘                        │
└─────────────────────────────────────────────────────────────┘

Collected Metrics​

MetricDescriptionUnit
led_signal_countTotal LED toggle count (source for ON/OFF ticks)count
led_interval_msActual LED toggle intervalms
led_stateCurrent LED state (0=OFF, 1=ON)-

Usage​

1. Start Monitoring System​

./start.sh

2. Access Grafana​

  • URL: http://localhost:3000
  • Default account: admin / admin
  • Dashboard: "LED Control Demo - TIMPANI vs Normal"

3. Start LED Controllers​

Both LED controller containers must be running:

# led-timpani-ctrl (Port 9101 must be exposed)
# led-normal-ctrl (Port 9102 must be exposed)

4. Load Testing​

In the demo, CPU load is toggled by pressing the joystick button on the master: the kuksa-bridge runs stress-ng directly on the guest host (see ../kuksa-bridge/README.md).

To reproduce the load manually on the guest:

stress-ng --cpu 0 --cpu-method all --timeout 30s

Dashboard Panel Description​

LED ON/OFF Tick Timeline​

  • Real-time comparison of ON/OFF toggle ticks between TIMPANI and Normal controllers
  • Tick events are derived from led_signal_count increments
  • Spikes (1) indicate a toggle event occurred in that short window

LED Toggle Interval​

  • Shows actual interval compared to expected 500ms
  • TIMPANI maintains 500ms even under load
  • Normal interval increases under load

Current Stats​

  • Shows current tick, signal count, and LED state
  • Includes 1-minute average toggle rate

Troubleshooting​

Prometheus Not Collecting Metrics​

# Check each controller's metrics endpoint
curl http://localhost:9101/metrics  # timpani
curl http://localhost:9102/metrics  # normal

No Data in Grafana​

  1. Check Prometheus data source connection
  2. Check time range (recommend Last 2 minutes)
  3. Enable auto-refresh (500ms)

Stop Monitoring​

./stop.sh

Notes​

  • network_mode: host allows containers to access the host network directly
  • Prometheus scrape interval: 100ms (for finer ON/OFF tick capture)
  • Grafana refresh: 200ms~500ms (for real-time monitoring)