Filtering out empty multi-value values

Hello. I don’t quite understand the results of my queries :slight_smile:

So, I have a datasource, where one of the dimensions (“jvmId”) is multi-valued. Some rows contain an empty array as the value, not null. I want to execute a groupBy query (group by this very same jvmId) to return only those rows, where this dimension does contain some non-null value. When I try to add

  "type": "not",
  "field": {
    "type": "selector",
    "dimension": "jvmId",
    "value": null


to my filters, I still get rows with “jvmId = null”:

“event”: {

        "count": 69106,

        "jvmId": null,

        "success": 61402



What am I doing wrong?

A, found more information. not-null filter fails on segments that were created by Kafka Indexing spec, but works correctly on segments created by index_hadoop spec. Seems like a bug in Kafka indexing spec?