Monitors

A Monitor is a scheduled rule that evaluates device metrics. On each tick it pulls the configured metrics over a lookback window, computes any expressions you defined, and checks your alert conditions. When a condition trips, the monitor fires the integrations attached to that alert.

Each monitor rolls up to two independent states: a health state and a data state.

Monitors and alerts

Alerts aren't separate objects. A monitor holds a list of alert configurations, and each evaluation emits alert events inline. The alerts you see on a device, site, or fleet are read-only views of those events, filtered to the ones currently tripped. An alert id uses the alt_ prefix.

You can see all the monitors in your organization and filter them. Creating a monitor walks you through picking a target, setting a schedule, choosing metrics, optionally defining variables, and adding alerts. From a monitor you can edit it, enable or disable it, view per-device health and charts over a time range, edit its membership, and delete it.

Configuration

Target

Pick what the monitor watches.

Target type	Watches
`SINGLE_THING`	One device
`ALL_THINGS_OF_TYPE_IN_PLACE`	Every device of a type within a site or fleet
`ALL_THINGS_OF_TYPE`	Every device of a type, org-wide
`EXPLICIT_LIST`	A hand-picked list of devices of one type
`ALL_THINGS_AT_PLACE`	Every device at a site or fleet (any type)

Plus a monitor name (required, up to 80 characters).

Schedule

Field	Options
Frequency	`1m`, `2m`, `5m`, `10m`, `12m`, `15m`, `30m`, `1h`, `2h`, `3h`, `6h`, `12h`, `1d`
Lookback period	Same set

Lookback is capped relative to frequency, so faster monitors look back over a shorter window.

Metrics

Add one or more metrics to pull each tick. For each metric, choose how to handle gaps:

Missing-data behavior	Effect
`NO_INFILL`	Leave gaps as missing
`INFILL_ZERO`	Fill gaps with zero
`INFILL_LAST_VALUE`	Carry the last known value forward

Variables (optional)

Named expressions computed from your metrics, reusable in alert conditions. Each has a unique name and an expression (up to 2000 characters). Expressions are validated live, with a badge showing the inferred type.

Alerts

Conditions are written as expressions, not picked from operator dropdowns. Aggregations like average, min, and max are written in the expression text (for example avgSignal < 30).

Field	Type	Notes
Alert name	string	Required, unique, up to 80 chars
Condition	expression	Required; must evaluate to a boolean
Message	string	Required, up to 500 chars
Threshold percent	0–100	Share of devices that must trip; `0` or empty means any device
Severity	`0`–`9`	Required, unique per alert
Integrations	multi-select	Outbound integrations to fire (webhook, email, SMS, phone)
Per-device alert	toggle	One payload listing every affected device; webhook only, bypasses the threshold
Contributed state	`HEALTHY` / `WARNING` / `UNHEALTHY`	The health state this alert sets when tripped (default `WARNING`)

States

Health state

State	Meaning
`HEALTHY`	No alert condition tripped
`WARNING`	A warning-level condition tripped
`UNHEALTHY`	An unhealthy-level condition tripped

Data state

State	Meaning
`COMPLETE_DATA`	All expected data was present
`MISSING_DATA`	Some expected data was missing
`NO_DATA`	No data arrived

Enabled, disabled, auto-disabled

A monitor is Enabled or Disabled via a toggle.

The platform auto-disables a monitor when its target stops being valid: a metric is removed from the type, the thing type or metric definition is deleted, or the place is deleted. A banner shows the reason and time. Re-enable it with the toggle once the target is restored.

Alert events

Each tick records whether the condition tripped. Alert events carry the contributed state (HEALTHY / WARNING / UNHEALTHY) and the severity (0–9). The detail charts show tripped versus not-tripped over time.

API

Monitor management endpoints are coming soon to the v2 API. For now, manage monitors in the console. A Place's monitor count is available through GET /v2/sites/{placeId}/monitor/count and GET /v2/fleets/{placeId}/monitor/count.