Unique Indexes in MongoDB

By default, MongoDB happily lets two documents share the same value in any field — including things that logically should be unique, like an email address. A unique index enforces that rule at the database level.

Creating a Unique Index

db.collection.createIndex({ field: 1 }, { unique: true })

db.student.createIndex({ "email": 1 }, { "unique": true })

If duplicate values already exist in the collection, index creation fails:

MongoServerError: Index build failed ... E11000 duplicate key error
collection: rahultechnology.student index: email_1
dup key: { email: "rahul@gmail.com" }

You have to remove or fix the duplicates first, then create the index. Once it exists, any insert or update that would create a duplicate is rejected immediately:

db.student.insertMany([{ roll: 6, name: "rohan", age: 24, email: "rahul@gmail.com" }])
// MongoBulkWriteError: E11000 duplicate key error ... dup key: { email: "rahul@gmail.com" }

Unique Index on an Embedded Field

Indexes work on nested fields too, using dot notation — this also speeds up queries that filter on the embedded field.

db.student.createIndex({ "marks.hindi": 1 })

Once created, a query on marks.hindi uses IXSCAN instead of a full collection scan (see the Indexing Basics lesson for how to verify this with explain()).

Creating Multiple Indexes at Once: createIndexes()

db.student1.createIndexes([
  { roll: 1, email: 1 }
])
// [ 'roll_1_email_1' ]

Common Errors & Troubleshooting

• Index build failing on existing duplicates — you must find and resolve the duplicate values first (e.g. with an aggregation $group + $match count > 1, covered in the Aggregation section) before createIndex(..., {unique:true}) will succeed.
• Multiple documents missing the unique field all conflicting with each other — MongoDB treats a missing field as null, and by default a unique index only allows one document with a null (or missing) value for that field — insert a second one and you'll get a duplicate-key error on null.
• Forgetting a unique index applies per-field, not per-document combination — a single-field unique index only guarantees that one field is unique; use a compound unique index ({fieldA:1, fieldB:1}, {unique:true}) if the combination of two fields together needs to be unique.

Key Takeaways

• MongoDB's default _id index is unique — this is the same mechanism applied manually to any field you choose.
• Existing duplicates must be cleaned up before a unique index can be created.
• createIndexes() (plural) lets you define several indexes in a single call.