Trying to get both min and max as part of window aggregation into a single value list

Fluxlang
1

I have data returning a single series, so I have _value and _time. Now I need to aggregate by aggregate window. I am trying to find all outliers by aggregating using both min and max (so I want the min and max value for the aggregation). This will allow me to find spikes and drops when looking at graphs through time. I do not want multiple columns - one for min and one for max, I want a single column (so one line in grafana), making sure in each aggregation window I pick one min and one max and never miss the min and max values.

Example with min:

from(bucket: "aerolab")
  |> range(start: v.timeRangeStart, stop:v.timeRangeStop)
  |> filter(fn: (r) =>
      r._measurement == "nsDevice" and
      r._field == "UsedBytes"
  )
  |> aggregateWindow(every: v.windowPeriod, fn: min)
2

Hello @iptable,
The following post could be helpful

3

Thank you. I got to this, which works (for whoever might be looking for/at this):

data = () => from(bucket: "aerolab")
  |> range(start: v.timeRangeStart, stop:v.timeRangeStop)
  |> filter(fn: (r) =>
    r._measurement == "nsDevice" and
    r._field == "ReadBytesPerSecond" and 
    r.nodeIdent =~ /${nodeIdent:regex}/
  )

min = data()
  |> aggregateWindow(every: v.windowPeriod, fn: min)

max = data()
  |> aggregateWindow(every: v.windowPeriod, fn: max)

union(tables: [min, max])

The nodeIdent is for a multi-select filter in in grafana (dropdown variables). Now my problem is that, while this works, it produces 2x the number of datapoints that I should be producing (since I am picking 2 datapoints per window instead of one).

So a followup question, which would solve this for me, is there a way to make the aggregateWindow 2x the variable?

I tried the below, but that throws an error:

...
|> aggregateWindow(every: v.windowPeriod*2, fn: min)
...
|> aggregateWindow(every: v.windowPeriod*2, fn: max)
...

Side note for others who might find this thread, the reason data = is a function and not a result is because it works much faster this way. Something to do with pipeline optimisation.