puzzled time with aggregateWindow() function

[ccbellflower]

Raw data are inserted every 30s. so when the interval param, every, is set in minutes, every interval should have value, the time of which .But that's not.

• When every is setted to "7m", data is retrieved from 00:03:59, which should be "00:08:59"

• When every is setted to "8m", data is retrieved from 00:07:59

[ccbellflower]

code referred to as below: from(bucket: "bucket") |> range(start: 2024-05-29T16:01:00.000Z, stop: 2024-05-29T16:30:00.000Z) |> filter(fn: (r) => r["_measurement"] == "measurement") |> filter(fn: (r) => r["_field"] == "pi01") |> aggregateWindow(every: 8m, fn: last, createEmpty: false, timeSrc: "_time", // offset: 1m ) |> yield(name: "last")

[sanderson]

@ccbellflower Window boundaries originate at the Unix epoch and are calculated based on that starting point. So window boundaries will be started every 7-8 minutes starting at 1970-01-01T00:00:00Z. I'm curious to see what your actual window boundaries are in your query. Run the following to get the start boundary of each window:

from(bucket: "bucket")
    |> range(start: 2024-05-29T16:01:00.000Z, stop: 2024-05-29T16:30:00.000Z)
    |> filter(fn: (r) => r["_measurement"] == "measurement")
    |> filter(fn: (r) => r["_field"] == "pi01")
    |> window(every: 8m, createEmpty: true)
    |> unique(column: "_start")

[ccbellflower]

@sanderson could you please check data in this link, because i have failed to post screenshoots on this comment. And i dont think wind boundaries based on starting point referring to this link

[sanderson]

@ccbellflower Thanks for the link to the community post. When I query for the actual window boundaries that are being used in your query (limited to the time range you defined), I get the following:

7m windows

_start	_stop
2024-05-29T16:01:00Z	2024-05-29T16:04:00Z
2024-05-29T16:04:00Z	2024-05-29T16:11:00Z
2024-05-29T16:11:00Z	2024-05-29T16:18:00Z
2024-05-29T16:18:00Z	2024-05-29T16:25:00Z
2024-05-29T16:25:00Z	2024-05-29T16:30:00Z

8m windows

_start	_stop
2024-05-29T16:01:00Z	2024-05-29T16:08:00Z
2024-05-29T16:08:00Z	2024-05-29T16:16:00Z
2024-05-29T16:16:00Z	2024-05-29T16:24:00Z
2024-05-29T16:24:00Z	2024-05-29T16:30:00Z

The last function you're using in aggregateWindow() is a selector function that doesn't change the timestamp of the row that is returned for each window. The timestamp in your results is the original timestamp associated with the last row in each window. To modify the timestamp for each window, you need to use an aggregate function or define a custom function that replaces the timestamp of the last row with the boundary of the window.

[ccbellflower]

@sanderson thanks for your reply. But what I am describing is that the time returned is not the original timestamp associated with the last row in each wind. data are inserted into every 30s, so there is one piece of data at least every minute. So when every is 7m and timeSrc is _time, 2024-05-30 00:06:59GMT+8 is the actual data rather than 2024-05-30 00:03:59GMT+8.

[github-actions[bot]]

This issue has had no recent activity and will be closed soon.

[ccbellflower]

这是来自QQ邮箱的假期自动回复邮件。您好，您的邮件我已收到，会尽快进行处理，谢谢。                                    陈艺