I was exploring TEXT_MATCH functionality with pinot-0.7.0/0.6.0 and had configured one of the columns for it. Is there any configuration for the refresh time interval for the index - https://docs.pinot.apache.org/basics/indexing/text-search-support After enabling indexing (with

index type: Text

and

encoding type: RAW

) on the column and doing TEXT_MATCH, I was first getting an empty result, but after some time, I was getting the result. So, what is the initial delay for such a column to be searchable? Is any settings/configuration (e.g num of docs, indexed size, etc) for the same?

Hi Ronak, The refresh threshold isn't yet configurable. We can make it though. However, there was a bug I recently fixed. It was causing the lag. Just realized it's in my branch. Will create a PR

So by design/indexing there won’t be any significant lag?

@Sidd @Kishore G @Xiang Fu On similar lines, I was also exploring JSON index support with apache-pinot-0.7.0. In that context, I have few questions. • is JSON index supported on multi-value column? • Is there any plan in extending text_match kind of support on json value space? As shown in below example as Q1 doing term query on

message

field of json document.

Table = json_doc_example
Example data with 3 rows:
id_col, json_doc_col
"1234", {"message": "too many retries"}
"1235", {"message": "error sql timeout"}
"1236", {"message": "too large filter"}

Q1: select * from json_doc_example JSON_MATCH(json_doc_col, 'message=''too''') : empty result -> can this return in future -> [1234, 1236] // or is it recommended to use both text index + json index for such use case on that column?
Q2 select * from json_doc_example JSON_MATCH(json_doc_col, 'message=''too lareg filter''') : returns -> [1236]

Text is not supported for strings within json as of now. We can add that since we have the primitives.. please file an issue

👍 sounds good, I will file one.