Kibana - Incorrect Count # with Bucket Aggregation "Terms"

Hi,

I'd like to create a Visualisation for our Sales team, to easily show what the top 3 sold items are.

I have come across a strange behaviour when using a simple Visualisation where I would like to count the top 3 Terms from all the documents in my index. Doesn't matter which Visualisation I choose, its always the same result.

My index has 999 documents, so I leave the Y-Axis aggregation as: "Count"
-> The result shows correctly 999 results

I then create a Bucket with an Aggregation of "Terms" and pick the Field "Listing Title" and select "Descending, Size: 3"
-> The result shows the 3 Items but the Aggregated number is incorrect. Its shows 12 instead of 14.
I checked the source CSV and the Top sold items appears indeed 14 times. There are no typos/whitespaces which could mess with the Visualisation.

When I change the Bucket settings from: "Descending, Size: 3" to: "Descending, Size: 9999", I get the correct Amount of 14 displayed. But then I also get all the other Items we are not interested in, displayed.

Am I missing something here or is this an intended behaviour? I have also played around wth other Bucket types/Aggregations, but I never get the result that I'd like. Any ideas what I might be doing wrong here?

Hello,

This sounds related to the terms returned per shard. Might be worth reading through this discuss issue which includes some helpful links on how to account for and resolve issues related to smaller queries and their interplay with shard size:

Regards,
Aaron

Excellent. That has helped pointing me in the right direction.
I've created a default Index template with a default shard size and replica of 1 and now I am getting the correct results!

Thanks for that, much appreciated!

1 Like