One GDMS OpenAPI connection brings every Grandstream phone and ATA the account manages into SmartTile — tied to the right location, confirmed before it is called down, one site outage kept to one event — with a guarded, call-aware remote reboot.
A desk phone that drops off its SIP server is usually reported by the person who could not make a call. GDMS already knows — but it also reports the last known account status for a phone that is offline, so a naive integration shows a dead phone as registered. SmartTile's EndpointAutomate reads GDMS the careful way: raw online/offline decides, stale account status is never trusted, and nothing is called down until it has been seen down poll after poll.
Name, model, firmware, serial, private and public IP, GDMS site and live status for every device in the account, refreshed every poll. Each phone's SIP extension and SIP server are read once and kept.
Extension prefix map, then location-name match, then device name, then GDMS site name; optionally auto-creates "GDMS – <PREFIX>" locations, else a default location.
Offline → Unreachable, some accounts unregistered → Unregistered. A fault needs 3 confirming polls on the legacy 5-min scale (≈15 min at 60 s), a restore 2 (≈10 min). No transitions in the first 5 min after a restart.
When a configurable share of a location's handsets (and at least 3) fault together, SmartTile raises one site event and marks the rest Suppressed — each suppression written to the audit trail.
Each handset seen online at least once becomes a VoIP Phone device (monitored by the API, never pinged), with up/down, state history and a 7-day uptime % from the GDMS polls.
Reboot from the handset panel as a GDMS task — refused while the phone is on a call, during a site event, or twice in 24 h; logged with who asked.
https://www.gdms.cloud), API Client ID, Secret Key, GDMS username and password.| Setting | What it controls |
|---|---|
| Engine | Polling on/off platform-wide — super admin; env kill switch wins |
| Confirm polls | Fault confirmations, 1–10 (default 3) |
| Site-flood stickiness | A muted phone still counts as faulted (default on) |
| NOC notifications | PHONE DOWN / restored e-mails to the NOC inbox, max 10 per cycle |
| Alert engine | Handsets also evaluated for ITSM tickets — off by default |
/oapi/oauth/tokenPassword grant with the API ID + Secret; token cached until expiry/oapi/v1.0.0/device/listEvery poll: the full device list, 100 per page/oapi/v1.0.0/device/account/infoOnce per new phone: SIP extension(s) and SIP server/oapi/v1.0.0/device/account/statusBefore a reboot: is any line on a call?/oapi/v1.0.0/task/addOnly when someone clicks Reboot (task type 1)access_token, timestamp and a SHA-256 signature built from the token, client ID, secret and timestamp (plus the body hash) — the scheme GDMS documents. An expired token is dropped and retried once.Device: GRP2615 · ext 1204 · Location "Head Office"
Seen by SmartTile: GDMS status offline, account status ignored as stale; candidate Unreachable held pending
Then: confirmed DOWN, device record and state history updated, and one PHONE DOWN e-mail to the NOC inbox with MAC, model, IP, extension and SIP server
Secret Key and GDMS password are encrypted at rest; the API only ever says whether each is stored.
Every connection belongs to one company; lists, handset detail, events and reboots are scoped to the caller's companies. A connection's organization and default location must belong to that company.
The EndpointAutomate API answers admins and super admins only; the engine and flood switches are super admin only.
No reboot mid-call, while suppressed, or more than once per 24 h; unknown call state aborts unless a super admin forces it. Every attempt is recorded.
Transitions keep the device's confirmed status and state history in step, so reports never show 100 % for a phone that was down (switch on by default).
Every site event lists how many handsets it muted; muted phones return to evaluation when the site recovers.
| Question | Answer |
|---|---|
| Will an unplugged phone open a ticket? | Not by default — handsets notify the NOC inbox. Switching on the alert-engine setting evaluates them like any API-monitored device for the ITSM ticket path. |
| Our GDMS account serves several sites? | Every handset is placed at a location by extension prefix, name or GDMS site; unmatched ones fall to the default location. |