Bug in using influxdata cloud v2 Data Explorer

InfluxDB 2
, ,
1

Hi everyone,

I’m a new user who just start to use influxDB cloud v2. And I have trouble with using the Data Explorer. I’m not sure whether it’s a bug or I’m not using it correctly.

Here is how I reproduce it:

  1. create a bucket named test

  2. insert 1 point into that bucket by following commands:
    curl --request POST “https://us-west-2-1.aws.cloud2.influxdata.com/api/v2/write?org=XXXX&bucket=test&precision=s” --header “Authorization: Token XXXX” --data-raw ‘my_measure height=100 1603749600’

    I expect I insert 1 data point at unix time stamp 1603749600 at precision second. So that is 2020-10-26 and at 10pm UTC time (google any unix time converter to check it).

  3. view the data in data explorer.

    So here is the problem, when I view data in past 7 days, the point shows in the graph says 2020-10-26 and at 10:30pm UTC time (check my uploaded image)

    Then I change the timerange to view data in past 30 days, the point shows in the graph says the point shows in the graph says 2020-10-26 and at 11:00pm UTC time

Could anyone helps me to figure out what’s the problem here?
Thanks a lot

2
22427×1340 140 KB

2

@H_F - Very interesting result here! I’m able to reproduce what you are seeing.

I’m struggling to decide if I think this is a bug or not, but it is certainly very confusing. Because you are querying on just one point, this is sort of a special case. The Data Explorer is running a query that includes aggregateWindow with a mean function. Naturally the mean is not impacting the result here, but the window size will change timestamps to the center of the window. The window size changes with 7 days and 30 days setting. If your point used a timestamp that isn’t on the UTC hour, this behavior would be even more obvious.

For my own dashboards, I display the dynamic variable v.windowPeriod which is what is changing when you adjust the time range. This variable is what is fed into the windowAggregate function. I do wish that these automatic variables were displayed by default and that the explorer indicated clearly that mean is the default too.

I encourage you to add to this Issue ticket or create a new one: you can pretty much copy your message into the ticket. v.windowPeriod is not consistently calculated · Issue #18907 · influxdata/influxdb · GitHub

Let us know your thoughts.

If you switch to the Script Editor view, you can see the Flux query that the explorer is running. For me, it is this

from(bucket: "bucket")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "my_measure")
  |> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
  |> yield(name: "mean")
3

Thanks for you reply. Another thing I observe is this doesn’t just happen when I insert and query only 1 point, it shows the same symptom when I insert and query multiple daily points.

I’ll open a new github issue to see anyone can help on this.

4

Yes of course - it will window aggregate and average even if there are multiple points.

Please link the ticket here after you’ve made it!

5

So i asked the issue there. So apparently it’s a by design things, since they use [open, close) boundaries. I need to manually remove the window function to make dashboard correct.

1 Like