3

I have a collection called search2 with about 20000 documents like this:

    {
        "loc": {
        "type": "Polygon",
        "coordinates": [
            [
            [
                43.78526674007639,
                11.14739998758569
            ],
            [
                43.78526674007639,
                11.183372851822439
            ],
            [
                43.79443488391605,
                11.183372851822439
            ],
            [
                43.79443488391605,
                11.264311796355125
            ],
            [
                43.812771171595415,
                11.264311796355125
            ],
            [
                43.83110745927479,
                11.264311796355125
            ],
            [
                43.83110745927479,
                11.273305012414314
            ],
            [
                43.849443746954144,
                11.273305012414314
            ],
            [
                43.858611890793824,
                11.273305012414314
            ],
            [
                43.858611890793824,
                11.264311796355125
            ],
            [
                43.8769481784732,
                11.264311796355125
            ],
            [
                43.8769481784732,
                11.246325364236752
            ],
            [
                43.88611632231286,
                11.246325364236752
            ],
            [
                43.88611632231286,
                11.237332148177565
            ],
            [
                43.895284466152546,
                11.237332148177565
            ],
            [
                43.895284466152546,
                11.228338932118376
            ],
            [
                43.904452609992234,
                11.228338932118376
            ],
            [
                43.904452609992234,
                11.165386419704065
            ],
            [
                43.895284466152546,
                11.165386419704065
            ],
            [
                43.895284466152546,
                11.156393203644878
            ],
            [
                43.88611632231286,
                11.156393203644878
            ],
            [
                43.8769481784732,
                11.156393203644878
            ],
            [
                43.858611890793824,
                11.156393203644878
            ],
            [
                43.849443746954144,
                11.156393203644878
            ],
            [
                43.849443746954144,
                11.165386419704065
            ],
            [
                43.83110745927479,
                11.165386419704065
            ],
            [
                43.83110745927479,
                11.156393203644878
            ],
            [
                43.812771171595415,
                11.156393203644878
            ],
            [
                43.812771171595415,
                11.14739998758569
            ],
            [
                43.79443488391605,
                11.14739998758569
            ],
            [
                43.78526674007639,
                11.14739998758569
            ]
            ]
        ]
        },
        "docId": 1,
        "docVote": 0,
        "title": "title-1",
        "_id": {
        "$oid": "5248725d2dd5622510000001"
        }
    }

I define an index with this command:

    db.search2.ensureIndex({"docVote": 1,"loc":"2dsphere"});

On the collection there are only this index and the default index on "_id" field.

When i execute the following query i expect "nscannedObjects" to be = 10 :

    db.search2.find({
        loc: {
        $geoIntersects: {
            $geometry: {
            type: "Polygon",
            coordinates: [
                [
                    [43.7269795, 11.1540365],
                    [43.8329368, 11.1540365],
                    [43.8329368, 11.3310908],
                    [43.7269795, 11.3310908],
                    [43.7269795, 11.1540365]
                ]
            ]
            }
        }
        }
    }, {
        "docVote": 1,
        _id: 0
    }).sort({
        "docVote": 1
    }).limit(10).hint({
        "docVote": 1,
        "loc": "2dsphere"
    }).explain()

But this is the result:

    {
    "cursor" : "S2Cursor",
    "isMultiKey" : true,
    "n" : 10,
    "nscannedObjects" : 44283,
    "nscanned" : 648117,
    "nscannedObjectsAllPlans" : 44283,
    "nscannedAllPlans" : 648117,
    "scanAndOrder" : true,
    "indexOnly" : false,
    "nYields" : 13,
    "nChunkSkips" : 0,
    "millis" : 12632,
    "indexBounds" : {

    },
    "nscanned" : 648117,
    "matchTested" : NumberLong(46642),
    "geoTested" : NumberLong(46642),
    "cellsInCover" : NumberLong(8),
    "server" : "*********"
    }

If i remove sort from the query i obtain this:

    {
        "cursor" : "S2Cursor",
        "isMultiKey" : true,
        "n" : 10,
        "nscannedObjects" : 10,
        "nscanned" : 25,
        "nscannedObjectsAllPlans" : 10,
        "nscannedAllPlans" : 25,
        "scanAndOrder" : false,
        "indexOnly" : false,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "millis" : 3,
        "indexBounds" : {

        },
        "nscanned" : 25,
        "matchTested" : NumberLong(10),
        "geoTested" : NumberLong(10),
        "cellsInCover" : NumberLong(8),
        "server" : "******"
    }

So, why index is not used to sort results? From this documentation: http://docs.mongodb.org/manual/applications/geospatial-indexes/ http://docs.mongodb.org/manual/tutorial/sort-results-with-indexes/ I understand that MongoDb support "A compound index with scalar index fields (i.e. ascending or descending) as a prefix or suffix of the 2dsphere index field" and that "If the sort document is a subset of a compound index and starts from the beginning of the index, MongoDB can use the index to both retrieve and sort the query results."

What am I missing?

Thanks in advance

MarcoBiagi
  • 163
  • 7
  • db.searchtest2path.ensureIndex({"docVote": 1,"loc":"2dsphere"}); Collection on the sample and on the index statement are not the same. Probably a typo, i would be better to fix on your question :) – fgakk Sep 30 '13 at 13:17

2 Answers2

0

My thoughts are it is because you're defining docVote as the first field in the compound index, yet you dont use the docVote in your query.

My suggestion is to try db.search2.ensureIndex({"loc":"2dsphere"}); to see whether the query uses the index. And then try db.search2.ensureIndex({"loc":"2dsphere", "docVote": 1});

Bertie
  • 17,277
  • 45
  • 129
  • 182
-1

Remove the index on "loc" and try only with the "docVote" index.

RevH
  • 58
  • 2
  • 8
  • Hi Rev,thanks you for your answear. I've try it and it doesn't work because in this way 2dsphere index is not used. – MarcoBiagi Oct 01 '13 at 07:51