MongoDB indexes and the $or operator

nlyn picture nlyn · Jul 29, 2011 · Viewed 9.3k times · Source

Sorry if this has been asked before. I couldn't find a definitive answer. Can I query on a mongodb index if my query contains the $or operator? My query looks something like this:

    // find everything in the collection
    $cursor = $collection->find(array(
   '$or' => array( 
        array('album_id' => array(
            '$in' => $this->my_album_ids
            ),
        'type' => array(
            '$in' => array('like','comment')
            )
         ),
       array(
    'user_id' => (int)session_item('user_id'),
        'type' => 'message',
        'reply' => 'no'
         )
       ),
        'timestamp' => array('$gt' => (int)$since)))->sort(array('timestamp'=>-1))->skip($start)->limit($amount);  

The Example is in PHP, but I guess this is applicable to any language.

Update:

The following are my indexes, but the above query does not use them. It looks right to me though.

    $collection->ensureIndex(array(
        'album_id' => 1,
        'type' => 1,
        'timestamp' => -1,
    ));

    $collection->ensureIndex(array(
        'user_id' => 1,
        'type' => 1,
        'reply' => 1,
        'timestamp' => -1,
    ));

Here is my explain()

Array
(
    [cursor] => BasicCursor
    [nscanned] => 12
    [nscannedObjects] => 12
    [n] => 6
    [scanAndOrder] => 1
    [millis] => 0
    [nYields] => 0
    [nChunkSkips] => 0
    [isMultiKey] => 
    [indexOnly] => 
    [indexBounds] => Array
        (
        )

    [allPlans] => Array
        (
            [0] => Array
                (
                    [cursor] => BasicCursor
                    [indexBounds] => Array
                        (
                        )

                )

        )

    [oldPlan] => Array
        (
            [cursor] => BasicCursor
            [indexBounds] => Array
                (
                )

        )

)

Answer

Remon van Vliet picture Remon van Vliet · Jul 29, 2011

Yes, an $or query will use indexes as appropriate. For example :

> db.test.ensureIndex({a:1})
> db.test.ensureIndex({b:1})
> db.test.find({$or:[{a:1}, {b:2}]}).explain()
{
        "clauses" : [
                {
                        "cursor" : "BtreeCursor a_1",
                        "nscanned" : 0,
                        "nscannedObjects" : 0,
                        "n" : 0,
                        "millis" : 0,
                        "nYields" : 0,
                        "nChunkSkips" : 0,
                        "isMultiKey" : false,
                        "indexOnly" : false,
                        "indexBounds" : {
                                "a" : [
                                        [
                                                1,
                                                1
                                        ]
                                ]
                        }
                },
                {
                        "cursor" : "BtreeCursor b_1",
                        "nscanned" : 0,
                        "nscannedObjects" : 0,
                        "n" : 0,
                        "millis" : 1,
                        "nYields" : 0,
                        "nChunkSkips" : 0,
                        "isMultiKey" : false,
                        "indexOnly" : false,
                        "indexBounds" : {
                                "b" : [
                                        [
                                                2,
                                                2
                                        ]
                                ]
                        }
                }
        ],
        "nscanned" : 0,
        "nscannedObjects" : 0,
        "n" : 0,
        "millis" : 1
}