Access points behind NAT, cameras, door readers, handsets and cloud instances cannot be pinged from outside. For every device imported through a vendor integration, the vendor's own status in this poll is the monitor — with the same alerts, tickets in your IT Service Management (ITSM) system, and uptime reports as a pinged device.
monitor_source = vendor_api).A monitor that stays green while a device is gone is worse than no monitor. Many vendor APIs quietly omit offline devices, or keep answering statistics for a device that has left the controller. SmartTile reads presence from this poll only, never refreshes that reading from any other code path, and hands the result to the same alert and ticket engine that watches pinged devices.
Import through a vendor connector and the device is monitored from that connector's own polls. Imports land on the company's plan (never below Monitor) with auto-ticket on and Device Health armed when the company is entitled.
UniFi: a device missing from the list for 5 minutes is offline. Aruba and Cisco FMC: missing for 3 consecutive periods. A single transient omission changes nothing.
When the Meraki org or the UniFi / cloud credential stops answering for 3 periods, devices that were UP turn UNKNOWN (grey) and the NOC is e-mailed — never left green.
UniFi's uptime counter going backwards is recorded as DOWN → UP with its real duration; 3 reboots or drops in 60 minutes open a reboot-loop ticket.
The vendor's DOWN is the confirmation — no ping required. After the ticket gate, one ticket per location in Rev.io and the company's connected ITSM system; the vendor's UP confirms recovery and sends the restored e-mail.
Per device, Decide DOWN using: ping only, the vendor API only, or both — DOWN only when both agree, and whichever still has a fresh reading decides if the other goes blind.
API monitors write the same state history as ping, so outages and reboots reach the Uptime modal, availability math and reports.
Every 5 minutes a second pass looks for devices that are truly down but were missed by the main 60-second cycle, and puts them on the ticket path.
| Setting | Default |
|---|---|
| UniFi / Meraki status check | Every 60 s |
| Omission → offline | UniFi 5 min · Aruba / FMC 3 periods |
| Controller / org blind → UNKNOWN | 3 periods (Meraki ~15 min) |
| Reboot-loop ticket | 3 in 60 min |
| GDMS handsets | Own notifier; engine inclusion off |
| The moment | Vendor console alone | With SmartTile |
|---|---|---|
| An AP drops off the controller | It silently disappears from the list. | 5 min absent → offline → DOWN in SmartTile, state history written, ticket after the gate. |
| An AP reboots five times overnight | Shows online in the morning. | Each reboot recorded with its duration; the third in an hour opens a reboot-loop ticket; uptime drops accordingly. |
| The API key is revoked | Every device keeps its last status. | After 3 failed periods devices go UNKNOWN and the NOC is e-mailed. |
| The device comes back | Someone closes the ticket later. | The vendor's UP is the recovery: note, restored e-mail, close per the company's policy. |
| Monitoring source | Where status comes from |
|---|---|
| meraki_api · ubiquiti_api | Meraki device availability; UniFi controller / Integration API list and stats |
| ubiquiti_protect_api · ubiquiti_access_api | Cameras, door hubs and readers from the UniFi console |
| pfsense_api · fmc_api · aruba_api | The firewall / manager / Central list, with omission detection on FMC and Aruba |
| verkada_api · epa_api | Verkada camera status; Grandstream GDMS handset registration |
| aws_api · digitalocean_api | CloudAutomate instance state (DOWN after 2 consecutive bad polls) |
Only list presence refreshes last-seen; statistics, enrichment and identity writes never touch the field the offline sweep reads.
A poll SmartTile could not make never marks a device absent; it marks it UNKNOWN.
Each company's vendor credentials import and monitor its own devices; status is shown only to that company's users.
API-sourced alerting, strict down detection, reboot detection, status mode and import defaults each have a platform switch.
| Objection | Answer |
|---|---|
| “The vendor already alerts us.” | Into its own console. SmartTile puts the outage on the same ticket, report and uptime figure as the rest of the site. |
| “A controller blip will flood us.” | Omission is debounced, a blind controller goes UNKNOWN (no ticket), and down devices at a site share one ticket. |