experimental.window() function

The experimental.window() function groups records based on a time value. New columns are added to uniquely identify each window. Those columns are added to the group key of the output tables. Input tables must have _start, _stop, and _time columns.

A single input record will be placed into zero or more output tables, depending on the specific windowing function.

By default the start boundary of a window will align with the Unix epoch (zero time) modified by the offset of the location option.

import "experimental"

experimental.window(
    every: 5m,
    period: 5m,
    offset: 12h,
    location: "UTC",
    createEmpty: false,
)

Parameters

every

(Required) Duration of time between windows. Defaults to period value.

period

Duration of the window. Period is the length of each interval. It can be negative, indicating the start and stop boundaries are reversed. Defaults to every value.

offset

Offset is the duration by which to shift the window boundaries. It can be negative, indicating that the offset goes backwards in time. Defaults to 0, which will align window end boundaries with the every duration.

location

Location used to determine timezone. Default is the location option.

Flux uses the timezone database (commonly referred to as “tz” or “zoneinfo”) provided by the operating system.

createEmpty

Specifies whether empty tables should be created. Defaults to false.

tables

Input data. Default is piped-forward data (<-).

Examples

Window data into 10 minute intervals

import "experimental"

from(bucket:"example-bucket")
  |> range(start: -12h)
  |> experimental.window(every: 10m)
  // ...

Window by calendar month

import "experimental"

from(bucket:"example-bucket")
  |> range(start: -1y)
  |> experimental.window(every: 1mo)
  // ...