I use Time Series Visual Builder and make a nice graph calculation throughput.
It works fine when I use 1s resolution and 15 minutes time span.
When I change time span to lets say 6 hours kibana complains that there are to many buckets.
The counters I use are stored in ES once per second.
To make the graph smooth I would like to calculate the average over a couple of samples.
(Average method does not work, I need to use sum)
I would like to divide the sum of my counters with the time interval used when I set interval to "auto".
Is there a variable I can use calculation aggregation?
I can hardcode interval to 10s and the divide my formula with 10.
That works for short time periods.
but that does not work when I change time span. I think I need a variable to solve this.
Is there a variable keeping track of time interval?
Something like this:
Here is an exampe:
First I have Interval set to auto and I show last 15 minutes. Then the sum of all samples are 10 times to high.
Seems like interval is set to 10 secs in this case
When I show last hour I get the following result:
30 times higher than correct value.
Then I specify Interval = 1s. This gives correct result:
So in the first case the variable I am searching for should have the value 10 and in the second case it should be 30.
Then I could always calculate the correct value in the formula.