ElasticSearch _Type with geolocation

Question

I have set up an elastic Search index which includes different _type mapping for every country.

So there is a mapping for "us" "au" "uk" etc.

Each mapping includes a location mapping of type "geo_point"

prior to adding different _types

My query sort would look like:

"sort" : [
    {

            "_geo_distance" : {
            "postcode.location" : [' . $mylocation_long . ',' . $mylocation_lat . '], 
            "order" : "asc",
            "unit" : "km"
        }
    }
],

with adding _types to the data and mapping this no longer works, instead I specify it like:

 "sort" : [
    {

            "_geo_distance" : {
            "$country.location" : [' . $mylocation_long . ',' . $mylocation_lat . '], 
            "order" : "asc",
            "unit" : "km"
        }
    }
],

this works fine.

However there are times when queries need to be done beyond a single country. So setting it to "us.location" isn't correct, and wont work.

In this case, how do I make this sorting work, when I don't know the country and I need to sort it by a mapped location.

Or is it a case of this can not be done and all docs must have the same _type in order for this to work?

Answer 1

Sorry if I am missing something obvious, but why cannot you just sort on "location". It seems to work just fine:

curl -XDELETE localhost:9200/test-idx/ && echo 
curl -XPUT localhost:9200/test-idx/ -d '
{
    "settings":{
        "number_of_shards":1,
        "number_of_replicas":0
    },
    "mappings": {
        "us": {
            "properties": {
                "location": {
                  "type": "geo_point"
                }        
            }
        },
        "uk": {
            "properties": {
                "location": {
                  "type": "geo_point"
                }        
            }
        },
        "au": {
            "properties": {
                "location": {
                  "type": "geo_point"
                }        
            }
        }
    }
}' && echo 
curl -XPUT localhost:9200/test-idx/us/1 -d '
{
    "location": "42.3606402,-71.0674569"
}
' && echo 
curl -XPUT localhost:9200/test-idx/uk/2 -d '
{
    "location": "51.5286416,-0.1017943"
}
' && echo 
curl -XPUT localhost:9200/test-idx/au/3 -d '
{
    "location": "-33.8471226,151.0594183"
}
' && echo 

curl -XPOST localhost:9200/test-idx/_refresh && echo 
curl "localhost:9200/test-idx/_search?pretty" -d '{
    "query": {
        "match_all": {}
    },
    "sort" : [
        {
                "_geo_distance" : {
                "location" : "52.3712989,4.8937347", 
                "order" : "asc",
                "unit" : "km"
            }
        }
    ]
}' && echo

output:

{"ok":true,"acknowledged":true}
{"ok":true,"acknowledged":true}
{"ok":true,"_index":"test-idx","_type":"us","_id":"1","_version":1}
{"ok":true,"_index":"test-idx","_type":"uk","_id":"2","_version":1}
{"ok":true,"_index":"test-idx","_type":"au","_id":"3","_version":1}
{"ok":true,"_shards":{"total":1,"successful":1,"failed":0}}
{
  "took" : 3,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "failed" : 0
  },
  "hits" : {
    "total" : 3,
    "max_score" : null,
    "hits" : [ {
      "_index" : "test-idx",
      "_type" : "uk",
      "_id" : "2",
      "_score" : null, "_source" : {"location": "51.5286416,-0.1017943"},
      "sort" : [ 355.2735714686373 ]
    }, {
      "_index" : "test-idx",
      "_type" : "us",
      "_id" : "1",
      "_score" : null, "_source" : {"location": "42.3606402,-71.0674569"},
      "sort" : [ 5563.606078215864 ]
    }, {
      "_index" : "test-idx",
      "_type" : "au",
      "_id" : "3",
      "_score" : null, "_source" : {"location": "-33.8471226,151.0594183"},
      "sort" : [ 16650.926847312003 ]
    } ]
  }
}

Answer 2

What happens when you point the working query at /index/_search instead of /index/type/_search ?