Link Stability
The link-stability monitor watches network links and interfaces for flapping and outages, and opens a single graduated ticket per troubled link. It's separate from device ping monitoring and the internet probe — this one answers "a link keeps going down".
What's monitored
- Gateway WAN — always monitored.
- Every other link — monitored when the device has the 🔔 monitor toggle on (Devices → Onboarded → per-device toggle). Turning the toggle off stops watching that device immediately (any open episode is closed).
Data sources, merged automatically: the UniFi controller (gateway WAN health + per-port status), SNMP interface status (for devices with SNMP), and the device's online/offline status as a fallback.
The graduated ticket lifecycle
flowchart LR
STABLE["Stable baseline"] -->|"first down/up transition"| P2["P2 — Link flap<br/>(kept open)"]
P2 -->|"3+ flaps in 30 min"| P1R["P1 — Recurring link flap"]
P2 -->|"down 10+ min"| P1O["P1 — Link outage"]
P1R -->|"no events for 30 min"| CLOSE["Auto-closed + summary"]
P1O -->|"no events for 30 min"| CLOSE
| Stage | Trigger | Result |
|---|---|---|
| Flap | A link transitions down→up (or up→down) from a known-good baseline | P2 ticket opens, kept open |
| Recurrence | 3+ flaps within the 30-minute window | Same ticket escalates to P1 |
| Outage | The link stays down longer than 10 minutes | Same ticket escalates to P1 |
| Stable | No further events for 30 minutes | Ticket auto-closes with a summary note |
The thresholds are configurable by an admin (defaults: 30-min window, 3 flaps, 10-min down, 30-min stable).
WAN and internet outages
The WAN flap ticket is the internet-outage ticket — when the internet probe confirms a real outage, it promotes the open WAN flap ticket to P1 instead of opening a duplicate "Internet connectivity down" ticket. Recovery closes it.
Persistence
Link episodes are persisted, so a container restart resumes an in-flight episode instead of losing it or double-alerting.
Who sees what
- Tickets are system-generated (
source=auto) and visible to all roles. - The 🔔 per-device toggle is set on the Devices page (operator/admin).
- Email alerts (down / P1 / recovery / stable) go to the alert recipients, best-effort (SMTP required).