How to query MongoDB to test if an item exists?

Starting Mongo 2.6, count has a limit optional parameter, which makes it a viable alternative to find whether a document exists or not:

db.collection.count({}, { limit: 1 })
// returns 1 if exists and 0 otherwise

or with a filtering query:

db.collection.count({/* criteria */}, { limit: 1 })

Limiting the number of matching occurrences makes the collection scan stop whenever a match is found instead of going through the whole collection.

---

Starting Mongo 4.0.3, since count() is considered deprecated we can use countDocuments instead:

db.collection.countDocuments({}, { limit: 1 })

or with a filtering query:

db.collection.countDocuments({/* criteria */}, { limit: 1 })

I dont believe that there is a straight way of checking the existence of the item by its value. But you could do that by just retrieving only id (with field selection)

db.your_collection.find({..criteria..}, {"_id" : 1});

It is significantly faster to use find() + limit() because findOne() will always read + return the document if it exists. find() just returns a cursor (or not) and only reads the data if you iterate through the cursor.

db.collection.find({_id: "myId"}, {_id: 1}).limit(1)

(instead of db.collection.findOne({_id: "myId"}, {_id: 1})).

Look at more details: Checking if a document exists – MongoDB slow findOne vs find

Since you don't need the count, you should make sure the query will return after it found the first match. Since count performance is not ideal, that is rather important. The following query should accomplish that:

db.Collection.find({ /* criteria */}).limit(1).size();

Note that find().count() by default does not honor the limit clause and might hence return unexpected results (and will try to find all matches). size() or count(true) will honor the limit flag.

If you want to go to extremes, you should make sure that your query uses covered indexes. Covered indexes only access the index, but they require that the field you query on is indexed. In general, that should do it because a count() obviously does not return any fields. Still, covered indexes sometimes need rather verbose cursors:

db.values.find({"value" : 3553}, {"_id": 0, "value" : 1}).limit(1).explain();

{
  // ...
  "cursor" : "BtreeCursor value_1",
  "indexOnly" : true,  // covered!
}

Unfortunately, count() does not offer explain(), so whether it's worth it or not is hard to say. As usual, measurement is a better companion than theory, but theory can at least save you from the bigger problems.