Link Health¶
The Link health tab on a gateway detail page lists the history of the gateway's link outages as time intervals. It answers "when did the link drop, how long did each outage last, and who detected it" — questions the live presence badge cannot answer.
What each row means¶
- Down at — the start of the outage.
- Recovered at — the end; an Ongoing tag means the outage is still open and its duration is still growing.
- Duration — recovered minus down, rendered compactly (
45s,12m 30s,2h 5m). - Source — who detected the outage:
- Lease detector — the platform detected it from an expired gateway lease (heartbeats stopped or the transport dropped).
- Edge watchdog — the gateway's local watchdog reported the episode over
v1/gateway/linkhealth. When an edge report matches a lease episode, the row is corrected with the edge ground truth and shown as edge-sourced; a micro-flap the lease never saw appears as an edge-only row. - Reason — the recorded reason text.
- Alarm — long outages link to the gateway's Alarms tab; alarm-free sub-lease micro-flaps show none. An edge flap burst is represented by a
GATEWAY_LINK_FLAPPINGalarm; see the Alarms help.
Filtering¶
- Use the Source dropdown to show only lease-detected or edge-reported episodes.
- Use the time-range picker to narrow to a specific window.
- Pages load server-side; an open outage stays on the first page until it ends and the list is refetched.
Relation to other tabs¶
- The presence badge and the Devices page show only the current online/offline state.
- The Alarms tab tracks alarm lifecycle (open/ack/clear); an outage can exist without an alarm, and one burst can drive one alarm across multiple episodes.
- Alarm-free micro-flaps are visible only here and, aggregated, in the tenant's daily outage digest.