I'm doing a "terms aggregation" + "top-bucket" subaggregation on each bucket. And I have several indexes: index1, index2, index3, index4 etc.
And I want to get x=20 "buckets" on the top aggregation.
Can I(how?) modify the aggregation functioning, so I can do:
aggregate on index1 if buckets < 20: aggregate on index2 + merge if buckets <20: aggregate on index3 + merge if buckets < 20: aggregate on index4 + merge etc
I understand I can do this on the client, with custom merging logic, but maybe it's faster/better/possible? to do it in the server ?