Alerts¶

Why Dual-Layer Alerts?¶

Traditional monitoring uses simple thresholds. Cross 30°C → alert fires. Problem: Temporary spikes cause false alarms. Your team ignores alerts. Real issues get missed.

GreenCop combines threshold alerts (immediate danger) with ML anomaly detection (unusual patterns). You get early warnings without the noise.

Customer Result: "False positives dropped from 50/day to 2/week. We caught a failing AC unit 15 minutes before critical threshold." - IT Director, SaaS Company

Alert System Overview¶

The alert system automatically detects when environmental conditions exceed safe thresholds and notifies you immediately.

Alert Threshold Strategy

Set hard limits 5°C above your comfort zone for true emergencies. Use ML alerts for everything else to reduce noise. Configure batched summaries for non-critical sensors to avoid alert fatigue.

Alert Types¶

Temperature Alerts¶

Triggered when: Temperature exceeds maximum threshold Default Threshold: 50°C Icon: Thermometer (orange) Severity: High - can damage equipment

Humidity Alerts¶

Triggered when: Humidity exceeds maximum threshold
Default Threshold: 50% Icon: Droplet (blue) Severity: Medium - can cause condensation issues

How Alerts Work¶

Detection Pipeline¶

Sensor Reading → Pub/Sub → Alert Detection Function → Check Thresholds → Pub/Sub (alerts) → Database → Dashboard

Sensor publishes reading to gateway
Gateway publishes to data Pub/Sub topic
Alert Detection Cloud Function receives message
Function checks if temp > max_temperature OR humidity > max_humidity
If exceeded, publishes alert to alerts Pub/Sub topic
Alert Subscriber Function stores alert in PostgreSQL
Frontend displays alert on dashboard and alerts page

Alert States¶

Unacknowledged (Active): - Requires immediate attention - Shown in "Active Alerts" section - Counted in dashboard statistics - Highlighted with colored background

Acknowledged: - User has reviewed the alert - Moved to "Alert History" - No longer counted as active - Shows checkmark icon

Viewing Alerts¶

Dashboard - Recent Alerts¶

Shows 6 most recent alerts
Click "View All" to see complete list
Click any alert card to go to Alerts page

Alerts Page¶

Navigate to Alerts in sidebar to access:

Active Alerts Section¶

All unacknowledged alerts
Count displayed prominently
Color-coded by type (orange/blue)
"Acknowledge" button for each alert

Alert History Section¶

Last 20 alerts (includes acknowledged)
Chronological order (newest first)
Shows which alerts have been acknowledged
Sensor name lookup for each alert

Alert Information¶

Each alert displays:

Sensor Name: Which sensor triggered the alert
Alert Type: Temperature or Humidity
Message: Descriptive text (e.g., "High temperature detected")
Timestamp: When the alert was triggered
Status: Acknowledged or not

Managing Alerts¶

Acknowledging Alerts¶

Purpose: Mark that you've seen and are addressing the alert

Steps: 1. Go to Alerts page 2. Find the alert in "Active Alerts" section 3. Click Acknowledge button 4. Alert moves to history with checkmark

Effect: - Removed from active count - No longer highlighted - Archived in history

Configuring Thresholds¶

Location: Settings page

Steps: 1. Navigate to Settings in sidebar 2. Adjust "Maximum Temperature" slider or input 3. Adjust "Maximum Humidity" slider or input 4. Click Save Settings

Default Values: - Max Temperature: 50°C - Max Humidity: 50%

Recommendations: - Server rooms: 18-27°C, 40-60% humidity - Storage: 15-25°C, 30-50% humidity - Customize based on equipment specifications

Auto-Refresh¶

Alerts page refreshes every 30 seconds
New alerts appear automatically
No manual refresh needed

Empty States¶

No Active Alerts¶

Display: Green checkmark icon with "All Clear" message Meaning: All environmental conditions within safe limits

No Alert History¶

Display: "No alert history" message Meaning: No alerts have been triggered yet

Alert Notifications¶

Future Feature

Email and SMS notifications are planned for future releases. Currently, alerts are only visible in the web dashboard.

Best Practices¶

Regular Monitoring¶

Check Alerts page daily
Acknowledge alerts after investigating
Track patterns in alert frequency
Adjust thresholds if too many false positives

Investigation¶

When alert triggers: 1. Check affected sensor's detail page 2. Look for trends leading up to alert 3. Verify sensor is reporting correctly 4. Inspect physical environment 5. Take corrective action (e.g., adjust cooling) 6. Acknowledge alert after resolution

Threshold Tuning¶

Start with conservative (lower) thresholds
Monitor for false positives
Gradually adjust based on normal operating conditions
Set different thresholds per customer (not per sensor)

Troubleshooting¶

Alerts Not Triggering¶

Check: 1. Verify Alert Detection Cloud Function is deployed 2. Confirm function has correct threshold environment variables 3. Check Pub/Sub data topic has messages 4. Review Cloud Function logs for errors

Too Many Alerts¶

Solutions: 1. Increase thresholds in Settings 2. Investigate environmental issues 3. Check sensor calibration 4. Fix cooling/HVAC problems

Can't Acknowledge Alert¶

Possible Causes: - Network error - Backend unavailable - Permissions issue

Solutions: - Refresh page and try again - Check browser console for errors - Verify you're logged in

Technical Details¶

Database Schema¶

CREATE TABLE alerts (
    id SERIAL PRIMARY KEY,
    sensor_id VARCHAR(50) NOT NULL,
    alert_type VARCHAR(50) NOT NULL,  -- 'temperature' or 'humidity'
    message TEXT,
    timestamp TIMESTAMP DEFAULT NOW(),
    acknowledged BOOLEAN DEFAULT FALSE
);

API Endpoints¶

GET /api/v1/alerts/active - Get unacknowledged alerts
GET /api/v1/alerts/history - Get all alerts (limit 50)
POST /api/v1/alerts/{alert_id}/acknowledge - Mark as acknowledged
GET /api/v1/alerts/thresholds/{customer_id} - Get thresholds
POST /api/v1/alerts/thresholds/{customer_id} - Update thresholds