Update Strategy

The update strategy block controls how Gridone synchronizes its data with the actual, physical device. All fields are optional and omitting the block entirely uses the following defaults.

update_strategy:
  polling_interval: 10s
  read_timeout: 10s

Fields

Field	Alias	Type	Default	Description
polling_interval	polling	duration or integer	10 (seconds)	How often attributes are read from the device. Must be positive.
read_timeout	timeout	duration or integer or null	10 (seconds)	Maximum time to wait for a device response. Must be between 1 and 60 seconds, or null to disable.
polling_groups	—	map of name to duration or integer	{}	Named polling groups, each with its own interval. See Polling groups.

Duration format

polling_interval and read_timeout accept either a plain integer (seconds) or a duration string:

Unit	Accepted forms
Seconds	s, sec, second, seconds
Minutes	m, min
Hours	h
Days	d

Examples: 30s, 1min, 2h, 90 (= 90 seconds).

Disabling polling

Set polling: disable to stop the system from polling the device. The device remains reachable for on-demand reads and writes, but no background polling occurs. This can be especially useful for push transport based devices, like mqtt, knx, or lorawan, where devices spontaneously publish their attribute updates and don't require polling.

update_strategy:
  polling: disable

polling: disable switches off the default polling only. Attributes assigned to a polling group keep polling on that group's schedule: assigning an attribute to a named group is an explicit opt-in. This is how a push device gets a single "trigger" attribute polled — a request the device answers with a burst of pushes carrying every value — while its siblings stay listen-only:

update_strategy:
  polling: disable
  polling_groups:
    full_refresh: 1min

attributes:
  - name: firmware_version
    polling_group: full_refresh   # the only polled attribute
    read:
      topic: data/${device_id}
      request: { topic: ${device_id}, message: READ_ALL }
    ...
  - name: temperature                # listen-only, fed by the burst
    read:
      topic: data/${device_id}
    ...

Polling groups

Some devices expose many attributes that don't all need to be read at the same rate — a temperature reading might be worth polling every 10 seconds, while a rarely-changing configuration value only needs an hourly check. polling_groups declares named groups with their own interval, and each attribute is assigned to one via its polling_group field:

update_strategy:
  polling_groups:
    core: 10s
    realtime_other: 1min
    config: 1h

attributes:
  - name: temperature
    polling_group: core
    ...
  - name: fan_speed
    polling_group: realtime_other
    ...
  - name: temperature_setpoint_min
    polling_group: config
    ...

Each group polls on its own schedule, and all attributes in a group are read together in a single batch request per sweep. Attributes with no polling_group fall back to the driver's polling_interval instead, and stop polling altogether under polling: disable — named groups do not (see Disabling polling). Every polling_group referenced by an attribute must be declared in polling_groups — an undeclared reference is rejected when the driver is loaded.

Silence detection for push devices

Silence detection for push devices is now configured under the health check block via expected_push_interval.

Examples

Custom interval and timeoutPolling disabled

update_strategy:
  polling_interval: 1min
  read_timeout: 5s

update_strategy:
  polling: disable