Understanding Percolator query extraction


(Evgeniy Fateev) #1

Hi, all!

Fighting with performance of the percolate query.

The first thing I need to understand is the conditions when terms can be extracted from the query during the indexing phase.

What I need to understand is whether terms will be extracted from the bool query in its filter part?

For example, giving this mapping:

curl "localhost:9200/queries/_mapping?pretty"
{
  "queries": {
    "mappings": {
      "_doc": {
        "dynamic": "false",
        "properties": {
          "location": {
            "type": "geo_point"
          },
          "query": {
            "type": "percolator"
          },
          "status": {
            "type": "keyword"
          }
        }
      }
    }
  }
}

And this query:

curl -X PUT "localhost:9200/queries/_doc/1" -d ' {
  "query": {
    "bool": {
      "filter": [{
          "term": {
            "status": "active"
          }
        },
        {
          "geo_distance": {
            "distance": "15km",
            "location": {
              "lat": 55.8899,
              "lon": 37.5926
            }
          }
        }
      ],
      "must": [],
      "must_not": []
    }
  }
}
'

Will {"term": {"status": "active"}} be extracted or I need to manually index it this way:

curl -X PUT "localhost:9200/queries/_doc/1" -d ' {
  "status": "active",
  "query": {
    "bool": {
      "filter": [
        {
          "geo_distance": {
            "distance": "15km",
            "location": {
              "lat": 55.8899,
              "lon": 37.5926
            }
          }
        }
      ],
      "must": [],
      "must_not": []
    }
  }
}
'

and modify my percolate query from this:

{
  "query": {
    "bool": {
      "filter": [
        {
          "percolate": {
            "field": "query",
            "document": {
              "title": "Some document title",
              "status": "active"
            }
          }
        }
      ]
    }
  }
}

to this:

{
  "query": {
    "bool": {
      "filter": [
        {
          "term": {
            "status": "active"
          }
        },
        {
          "percolate": {
            "field": "query",
            "document": {
              "title": "Some document title",
              "status": "active"
            }
          }
        }
      ]
    }
  }
}

Or may be I need to move filter part of the bool query to its must part?

Massive thank you in advance!


(Evgeniy Fateev) #2

After some IDE debugging I realized that terms are extracted in both bool query parts: must and filter. Extracted information is stored in the document (indexed query) field.

Not sure how this information is used in the percolate query.


(system) closed #3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.