Bar chart data panel, last timing group (aggregation to day, week or month) shows twice with different data

Grafana Bar Chart Panel
1

Hi friends,

I’m using Grafana 10.11 and it is getting the data from InfluxBD 2.0 version (flux)

I’m gathering energy consumption data and the aggregating it to days, weeks and months.

But data shown in different formats shows the last aggregation time (day, week or month) typically twice as long as the aggregation for the last period is ongoing.

Like here day (18.9)

image
image1159×459 15.3 KB

Three separate queries

from(bucket: "IELEC2")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "Usage_kW")
  |> aggregateWindow(every: 24h, fn: sum, createEmpty: false)


from(bucket: "IELEC2")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "Price")
  |> aggregateWindow(every: 24h, fn: sum, createEmpty: false)

from(bucket: "IELEC2")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "Nordpool_price")
  |> filter(fn: (r) => r["_field"] == "Nordpool_Price_kW")
  |> aggregateWindow(every: 24h, fn: sum, createEmpty: false)

and here on Months (09-23)

image
image1153×455 22.3 KB

bucket1 = from(bucket: "IELEC2")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "Usage_kW")
  |> aggregateWindow(every: 1mo, fn: sum, createEmpty: false)

bucket2 = from(bucket: "IELEC_CT_CLAMPS_HIST")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT_CLAMPS")
  |> filter(fn: (r) => r["Name"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "CT4_Usage_kW")
  |> aggregateWindow(every: 1mo, fn: sum, createEmpty: false)

    union(tables: [bucket1, bucket2])

bucket1 = from(bucket: "IELEC2")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "Price")
  |> aggregateWindow(every: 1mo, fn: sum, createEmpty: false)

  bucket2 = from(bucket: "IELEC_CT_CLAMPS_HIST_PRICE")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT_CLAMPS_1")
  |> filter(fn: (r) => r["Name"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "Price")
  |> aggregateWindow(every: 1mo, fn: sum, createEmpty: false)

      union(tables: [bucket1, bucket2])

bucket1 = from(bucket: "IELEC2")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "Nordpool_price")
  |> filter(fn: (r) => r["_field"] == "Nordpool_Price_kW")
  |> aggregateWindow(every: 1mo, fn: sum, createEmpty: false)

  bucket2 = from(bucket: "IELEC_CT_CLAMPS_HIST")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "CT_CLAMPS")
  |> filter(fn: (r) => r["Name"] == "CT4")
  |> filter(fn: (r) => r["_field"] == "Nordpool_Price_kW")
  |> aggregateWindow(every: 1mo, fn: sum, createEmpty: false)


      union(tables: [bucket1, bucket2])

I use transformations on the last panel (as it reads two data set from old historical data and combines it with new (slightly different naming rules). No transforms on the 1st panel on daily aggregates.

image
image2179×1019 173 KB

The numbers get “correct” once I move to next aggregation interval ,like day/week or month, but during that period I see them twice. Haven’t really figured how the data is divided or how I could only see the only one value; aggregated data up to that point from start of the aggregation period.

Any help or pointers where to find answers is appreciated. Thank you !

2

Answering my own question. Seems that the Flux query requires additional item in the aggregation window.

timeSrc:“_start”

So total line should look like this

|> aggregateWindow(every: 1h, fn: sum, createEmpty: false, timeSrc:"_start")

This corrects the data. Works on other aggregation times (months weeks etc)