Added
- A webhook can now wait before reporting a device offline. Set a few minutes on the webhook and that target is told only if the device is still down when the time has passed — a link that flaps and returns no longer wakes anyone at three in the morning. It is set per webhook, so a dashboard can keep firing immediately while a phone waits five minutes. Existing webhooks are unchanged: the delay starts at 0, which sends on the first failed poll exactly as before.
- If a target was never told the device went down, it is not told it came back either — otherwise the quiet night ends with a "back online" message about an outage you never heard about.
- The wait survives a restart of the manager: it is a recorded time the device went down, not a countdown held in memory, so nothing is lost if mikr restarts while an alert is waiting.
Fixed
- A BGP peer that went down disappeared from its own history instead of being recorded as down. The graph drew a straight line across the outage and reported no downtime. Peers are now taken from the router's BGP connection list, so a peer with no session is recorded as down for as long as it stays down.
- A router whose peers were all down looked like a router without BGP. The Routing tab vanished from its device page and history collection slowed to every 30 minutes — during the outage. Both now go by the configured peers, not the live sessions.
- A peer you disabled yourself is no longer counted as an outage. Those samples are drawn grey and counted separately from real downtime.
- The graph now names a peer the way the router's own log does, so the two can be compared.
Notes
- Graphs for peers whose session name differed from the connection name start a new series. Older samples are still shown until the new one has data, and expire with the usual 90-day history retention.