Alerting¶
Overview¶
Alerting notifies operators when sensor readings fall outside acceptable ranges or when a sensor stops reporting entirely. The alerting stack sits on top of the Monitoring stack (Prometheus → Alertmanager).
Alert Rules¶
Alert rules are defined in Prometheus and fire when conditions persist beyond a configurable threshold duration (to avoid noise from transient spikes).
| Alert | Condition | Severity |
|---|---|---|
| pH out of range | pH < 5.0 or pH > 7.0 for > 5 min | critical |
| EC out of range | EC outside target band for > 15 min | warning |
| Water level low | float sensor triggered | critical |
| Water flow stopped | flow rate = 0 while pump relay is on | critical |
| Air temp high | air temp > 35 °C for > 5 min | warning |
| Sensor missing | no reading received for > 2× polling interval | warning |
| VPD out of range | VPD outside target band for > 10 min | warning |
Notification Channels¶
| Channel | Use case |
|---|---|
| All alerts (Alertmanager SMTP) | |
| TBD (e.g. Telegram, Pushover) | Critical alerts on mobile |
Silencing and Inhibition¶
- Silence alerts during planned maintenance windows
- Inhibit downstream alerts when a sensor-missing alert is already firing (avoids alert storms)