I have run into a case where querying against a sparse index on a nested field puts the DB in a bad state that affects later queries on the same collection. I originally discovered this problem using the ruby driver, but was able to reproduce the problem completely from the mongo console (see below). I inserted comments, prefixed with #. The mongo_bug db did not exist prior to this example.

> use mongo_bug
switched to db mongo_bug
> db.bug_collection.ensureIndex(

{"nested.key2": 1}

,

{sparse: true}

)
> db.bug_collection.insert(

{key: 'a'}

)
> db.bug_collection.insert(

{key: 'b'}

)
> db.bug_collection.insert(

{key: 'c'}

)
> db.bug_collection.find({})

{ "_id" : ObjectId("4f0e7daf9be3426d0ba2e7b8"), "key" : "a" } { "_id" : ObjectId("4f0e7db59be3426d0ba2e7b9"), "key" : "b" } { "_id" : ObjectId("4f0e7dba9be3426d0ba2e7ba"), "key" : "c" }

> db.bug_collection.find({key: {$in: ['a', 'b', 'c']}, "nested.key2": {$ne: 'd'}})

1. No results returned (expected, because of the sparse index?)
   > db.bug_collection.find({key: {$in: ['a', 'b', 'c']}})
2. No results returned (NOT expected)
   > db.bug_collection.dropIndex("nested.key2_1") { "nIndexesWas" : 2, "ok" : 1 }

   > db.bug_collection.find({key: {$in: ['a', 'b', 'c']}})

   { "_id" : ObjectId("4f0e7daf9be3426d0ba2e7b8"), "key" : "a" } { "_id" : ObjectId("4f0e7db59be3426d0ba2e7b9"), "key" : "b" } { "_id" : ObjectId("4f0e7dba9be3426d0ba2e7ba"), "key" : "c" }
3. Query now works, after the index has been dropped.

The order of the queries also matters:

> db.bug_collection.dropIndex("nested.key2_1")

{ "nIndexesWas" : 2, "ok" : 1 }

> db.bug_collection.ensureIndex(

{"nested.key2": 1}

,

{sparse: true}

)
> db.bug_collection.find({key: {$in: ['a', 'b', 'c']}})

{ "_id" : ObjectId("4f0e811f89fc95d0a1a0356b"), "key" : "a" } { "_id" : ObjectId("4f0e812389fc95d0a1a0356c"), "key" : "b" } { "_id" : ObjectId("4f0e812589fc95d0a1a0356d"), "key" : "c" }

1. Results returned (expected)
   > db.bug_collection.find({key: {$in: ['a', 'b', 'c']}, "nested.key2": {$ne: 'd'}}) { "_id" : ObjectId("4f0e811f89fc95d0a1a0356b"), "key" : "a" } { "_id" : ObjectId("4f0e812389fc95d0a1a0356c"), "key" : "b" } { "_id" : ObjectId("4f0e812589fc95d0a1a0356d"), "key" : "c" }
2. Results returned (this same query returns nothing in first example)
   > db.bug_collection.dropIndex("nested.key2_1") { "nIndexesWas" : 2, "ok" : 1 }

   > db.bug_collection.ensureIndex(

   {"nested.key2": 1}

   ,

   {sparse: true}

   )
   > db.bug_collection.find({key: {$in: ['a', 'b', 'c']}, "nested.key2": {$ne: 'd'}})

3. No results returned (as in first example)
   > db.bug_collection.find({key: {$in: ['a', 'b', 'c']}})
4. No results returned (as in first example)

It's as if the queries are considered equivalent and being cached.