Inefficient handling of 'limit'

[caveadventure]

A query like select day,count(*) from index where x = query('FreeBSD') group by day limit 200 generates an Elasticsearch query like this:

{
  "from" : 0,
  "size" : 200,
  "query" : {
    "bool" : {
      "must" : {
        "query_string" : {
          "query" : "FreeBSD"
        }
      }
    }
  },
  "_source" : {
    "includes" : [ "day", "COUNT" ],
    "excludes" : [ ]
  },
  "fields" : "day",
  "aggregations" : {
    "day" : {
      "terms" : {
        "field" : "day",
        "size" : 200
      },
      "aggregations" : {
        "COUNT(*)" : {
          "value_count" : {
            "field" : "_index"
          }
        }
      }
    }
  }
}

Testing shows that the first top-level "size": 200 is unnecessary and results in much slower queries. It should probably be set to "size": 0 since the hits aren't needed anyways when doing aggregations.

[eliranmoyal]

Sure, I'll fix it.

[eliranmoyal]

this is fixed on the new versions
thanks for mention this
you can also specify specific sizes for next terms agg like this:

SELECT COUNT(*) FROM account GROUP BY gender, terms('alias'='ageAgg','field'='age','size'=3)