Task - daily min/max

InfluxDB 2 Tasks
1

Dear all,

I’d like to downsample MIN/MAX data from a sensor (outside temperature).

  • The source bucket is called “openhab” and is written every 5mins.
  • The target bucket for MIN-values is called “persistent_MIN” and should contain one datapoint per day.
  • The target bucket for MAX-values is called “persistent_MAX” and should contain one datapoint per day.
  • A day should be defined as 00:00-24:00 (so not 12:00-12:00 e.g.).
  • TZ is CET

In order to downsample my data, I used the data explorer and came up with the following query that I saved as a task;

option v = {
    timeRangeStart: -1d,
    timeRangeStop: now(),
}

option task = {
    name: "KM200_Sensor_Aussentemp_MIN",
    cron: "0 0 * * *",
}

from(bucket: "openhab")
    |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
    |> filter(fn: (r) => r["_measurement"] == "KM200_Sensor_Aussentemp")
    |> filter(fn: (r) => r["_field"] == "value")
    |> aggregateWindow(every: 24h, fn: min, createEmpty: false)
    |> yield(name: "min")
    |> to(bucket: "persistent_MIN", org: "dc1")

As you can see, that task should be executed once a day via cron (0 0 * * *).

The reason I’m posting here is that, for example if I run that query in the data explorer for source-bucket, it would show a MIN-value of 3.4°C and a MAX-value of 11.1°C for 05.11.2021.

If I do the same for the downsample-buckets, the graph shows those values with a _time of 2021-11-06 01:00:00 GMT+1 (so basically “+1 day”).

Could you please tell me what’s needed so that my downsampled data would show up “correctly” for the actual day they were retrieved from?

Many thanks & KR

2

Hello @Influx_Compensator,
so you’re saying that the output of


from(bucket: "openhab")
    |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
    |> filter(fn: (r) => r["_measurement"] == "KM200_Sensor_Aussentemp")
    |> filter(fn: (r) => r["_field"] == "value")
    |> aggregateWindow(every: 24h, fn: min, createEmpty: false)
    |> yield(name: "min")

Is at 05.11.2021?
Can you verify that please?
My guess is that aggregateWindow() puts the timestamp at the end of the window period.
Give this a whirl:

from(bucket: "openhab")
    |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
    |> filter(fn: (r) => r["_measurement"] == "KM200_Sensor_Aussentemp")
    |> filter(fn: (r) => r["_field"] == "value")
    |> aggregateWindow(every: 24h, fn: min, createEmpty: false)
    |> yield(name:"raw")
    |> timeShift(duration: -1d, columns: ["_time"])
    |> yield(name:"final")

yield() functions are there to help you visualize different points of your query

3

Hi there,

Sorry for not replying yet!

yes - I can confirm this works well until I put it in a task, then it would show the data for “date+1d” at 00:30.

I can also confirm that this query shows 2 curves, “raw” and “final”, while “final” has the correct date. Many thanks for that - I’ve put it in a task and added

|> yield(name:"final")
|> to(bucket: "persistent_MIN", org: "dc1")

at the bottom.

Anyhow, can you tell me why - eventhough the CRON is set to 00:00 - the result of the query would show the data-points at 01:00 (and not at 00:00)? I have several systems reporting to influxDB and if let’s say data is being sent at 14:21, influxDB also shows a data-point at 14:21 (so I’m trying to say there shouldn’t be a TZ-conflict).

Many thanks!

4

Helo @Influx_Compensator,
The Flux team had to make an executive decision about whether to put the data points at the start or after a window period duration. I don’t think you should encounter a problem.
However, this might be helpful to you: