We're having an issue where a FindAndModify in our code, in some of our environments/servers, decides to use an index that fails horribly (i.e. is sparse, when we're querying for nulls), and would like a way to force it to use a correct index. We're open to using an Update instead, if it would mean a faster turn-around on this issue.

Our current example:

SECONDARY> db.version()
2.0.3

SECONDARY> db.ourcollection.getIndexes()
[
{
"v" : 1,
"key" :

{ "_id" : 1 }

,
"ns" : "ourdb.ourcollection",
"name" : "id"
},
{
"v" : 1,
"key" :

{ "StartedOn" : -1 }

,
"ns" : "ourdb.ourcollection",
"name" : "StartedOn_-1",
"background" : true,
"sparse" : true
}
]

SECONDARY> db.ourcollection.find(

{ "_id" : new BinData(3,"L9UXPwbYdEiSQ68DG7A1iQ=="), "StartedOn" : null }

).explain()
{
"cursor" : "BtreeCursor StartedOn_-1",
"nscanned" : 30,
"nscannedObjects" : 30,
"n" : 0,
"millis" : 0,
"nYields" : 0,
"nChunkSkips" : 0,
"isMultiKey" : false,
"indexOnly" : false,
"indexBounds" :

{ "StartedOn" : [ [ null, null ] ] }

}

SECONDARY> db.ourcollection.find(

{ "_id" : new BinData(3,"L9UXPwbYdEiSQ68DG7A1iQ==") }

,

{ _id: 1, StartedOn: 1 }

).toArray()
[

{ _id: BinData(3,"L9UXPwbYdEiSQ68DG7A1iQ==") }

]

SECONDARY> db.ourcollection.find(

{ "_id" : new BinData(3,"L9UXPwbYdEiSQ68DG7A1iQ=="), "StartedOn" : null }

).toArray()
[ ]