db.collection.bulkWrite()

Write Operations

db.collection.bulkWrite( [
   { insertOne : { "document" : <document> } }
] )

db.collection.bulkWrite( [
   { updateOne :
      {
         "filter": <document>,
         "update": <document or pipeline>,
         "upsert": <boolean>,
         "collation": <document>,
         "arrayFilters": [ <filterdocument1>, ... ],
         "hint": <document|string>
      }
   }
] )

db.collection.bulkWrite( [
   { updateMany :
      {
         "filter" : <document>,
         "update" : <document or pipeline>,
         "upsert" : <boolean>,
         "collation": <document>,
         "arrayFilters": [ <filterdocument1>, ... ],
         "hint": <document|string>
      }
   }
] )

db.collection.bulkWrite([
   { replaceOne :
      {
         "filter" : <document>,
         "replacement" : <document>,
         "upsert" : <boolean>,
         "collation": <document>,
         "hint": <document|string>
      }
   }
] )

db.collection.bulkWrite([
   { deleteOne : {
      "filter" : <document>,
      "collation" : <document>                   // Available starting in 3.4
   } }
] )

db.collection.bulkWrite([
   { deleteMany: {
      "filter" : <document>,
      "collation" : <document>                    // Available starting in 3.4
   } }
] )

_id Field

If the document does not specify an _id field, then mongod adds the _id field and assign a unique ObjectId() for the document before inserting or upserting it. Most drivers create an ObjectId and insert the _id field, but the mongod will create and populate the _id if the driver or application does not.

If the document contains an _id field, the _id value must be unique within the collection to avoid duplicate key error.

Update or replace operations cannot specify an _id value that differs from the original document.

Execution of Operations

The ordered parameter specifies whether bulkWrite() will execute operations in order or not. By default, operations are executed in order.

The following code represents a bulkWrite() with five operations.

db.collection.bulkWrite(
   [
      { insertOne : <document> },
      { updateOne : <document> },
      { updateMany : <document> },
      { replaceOne : <document> },
      { deleteOne : <document> },
      { deleteMany : <document> }
   ]
)

In the default ordered : true state, each operation will be executed in order, from the first operation insertOne to the last operation deleteMany.

If ordered is set to false, operations may be reordered by mongod to increase performance. Applications should not depend on order of operation execution.

The following code represents an unordered bulkWrite() with six operations:

db.collection.bulkWrite(
   [
      { insertOne : <document> },
      { updateOne : <document> },
      { updateMany : <document> },
      { replaceOne : <document> },
      { deleteOne : <document> },
      { deleteMany : <document> }
   ],
   { ordered : false }
)

With ordered : false, the results of the operation may vary. For example, the deleteOne or deleteMany may remove more or fewer documents depending on whether the run before or after the insertOne, updateOne, updateMany, or replaceOne operations.

The number of operations in each group cannot exceed the value of the maxWriteBatchSize of the database. The default value of maxWriteBatchSize is 100,000. This value is shown in the hello.maxWriteBatchSize field.

This limit prevents issues with oversized error messages. If a group exceeds this limit, the client driver divides the group into smaller groups with counts less than or equal to the value of the limit. For example, with the maxWriteBatchSize value of 100,000, if the queue consists of 200,000 operations, the driver creates 2 groups, each with 100,000 operations.

If the error report for a single batch grows too large, MongoDB truncates all remaining error messages to the empty string. If there are at least two error messages with total size greater than 1MB, they are trucated.

The sizes and grouping mechanics are internal performance details and are subject to change in future versions.

Executing an ordered list of operations on a sharded collection will generally be slower than executing an unordered list since with an ordered list, each operation must wait for the previous operation to finish.

Capped Collections

bulkWrite() write operations have restrictions when used on a capped collection.

updateOne and updateMany throw a WriteError if the update criteria increases the size of the document being modified.

replaceOne throws a WriteError if the replacement document has a larger size than the original document.

deleteOne and deleteMany throw a WriteError if used on a capped collection.

Time Series Collections

bulkWrite() write operations have restrictions when used on a time series collection. Only insertOne can be used on time series collections. All other operations will return a WriteError.

Error Handling

db.collection.bulkWrite() throws a BulkWriteError exception on errors. See Error Handling inside Transactions.

Excluding write concern errors, ordered operations stop after an error, while unordered operations continue to process any remaining write operations in the queue, unless when run inside a transaction. See Error Handling inside Transactions.

Write concern errors are displayed in the writeConcernErrors field, while all other errors are displayed in the writeErrors field. If an error is encountered, the number of successful write operations are displayed instead of the inserted _id values. Ordered operations display the single error encountered while unordered operations display each error in an array.

Transactions

db.collection.bulkWrite() can be used inside distributed transactions.

Ordered Bulk Write Example

It is important that you understand bulkWrite() operation ordering and error handling. By default, bulkWrite() runs an ordered list of operations:

• Operations run serially.

• If an operation has an error, that operation and any following operations are not run.

• Operations listed before the error operation are completed.

The bulkWrite() examples use the pizzas collection:

db.pizzas.insertMany( [
   { _id: 0, type: "pepperoni", size: "small", price: 4 },
   { _id: 1, type: "cheese", size: "medium", price: 7 },
   { _id: 2, type: "vegan", size: "large", price: 8 }
] )

The following bulkWrite() example runs these operations on the pizzas collection:

• Adds two documents using insertOne.

• Updates a document using updateOne.

• Deletes a document using deleteOne.

• Replaces a document using replaceOne.

try {
   db.pizzas.bulkWrite( [
      { insertOne: { document: { _id: 3, type: "beef", size: "medium", price: 6 } } },
      { insertOne: { document: { _id: 4, type: "sausage", size: "large", price: 10 } } },
      { updateOne: {
         filter: { type: "cheese" },
         update: { $set: { price: 8 } }
      } },
      { deleteOne: { filter: { type: "pepperoni"} } },
      { replaceOne: {
         filter: { type: "vegan" },
         replacement: { type: "tofu", size: "small", price: 4 }
      } }
   ] )
} catch( error ) {
   print( error )
}

Example output, which includes a summary of the completed operations:

{
   acknowledged: true,
   insertedCount: 2,
   insertedIds: { '0': 3, '1': 4 },
   matchedCount: 2,
   modifiedCount: 2,
   deletedCount: 1,
   upsertedCount: 0,
   upsertedIds: {}
}

If the collection already contained a document with an _id of 4 before running the previous bulkWrite() example, the following duplicate key exception is returned for the second insertOne operation:

writeErrors: [
   WriteError {
      err: {
         index: 1,
         code: 11000,
         errmsg: 'E11000 duplicate key error collection: test.pizzas index: _id_ dup key: { _id: 4 }',
         op: { _id: 4, type: 'sausage', size: 'large', price: 10 }
      }
   }
],
result: BulkWriteResult {
   result: {
      ok: 1,
      writeErrors: [
         WriteError {
            err: {
               index: 1,
               code: 11000,
               errmsg: 'E11000 duplicate key error collection: test.pizzas index: _id_ dup key: { _id: 4 }',
               op: { _id: 4, type: 'sausage', size: 'large', price: 10 }
            }
         }
      ],
      writeConcernErrors: [],
      insertedIds: [ { index: 0, _id: 3 }, { index: 1, _id: 4 } ],
      nInserted: 1,
      nUpserted: 0,
      nMatched: 0,
      nModified: 0,
      nRemoved: 0,
      upserted: []
   }
}

Because the bulkWrite() example is ordered, only the first insertOne operation is completed.

To complete all operations that do not have errors, run bulkWrite() with ordered set to false. For an example, see the following section.

Unordered Bulk Write Example

To specify an unordered bulkWrite(), set ordered to false.

In an unordered bulkWrite() list of operations:

• Operations can run in parallel (not guaranteed). For details. See Ordered vs Unordered Operations.

• Operations with errors are not completed.

• All operations without errors are completed.

Continuing the pizzas collection example, drop and recreate the collection:

db.pizzas.insertMany( [
   { _id: 0, type: "pepperoni", size: "small", price: 4 },
   { _id: 1, type: "cheese", size: "medium", price: 7 },
   { _id: 2, type: "vegan", size: "large", price: 8 }
] )

In the following example:

• bulkWrite() runs unordered operations on the pizzas collection.

• The second insertOne operation has the same _id as the first insertOne, which causes a duplicate key error.

try {
   db.pizzas.bulkWrite( [
      { insertOne: { document: { _id: 3, type: "beef", size: "medium", price: 6 } } },
      { insertOne: { document: { _id: 3, type: "sausage", size: "large", price: 10 } } },
      { updateOne: {
         filter: { type: "cheese" },
         update: { $set: { price: 8 } }
      } },
      { deleteOne: { filter: { type: "pepperoni"} } },
      { replaceOne: {
         filter: { type: "vegan" },
         replacement: { type: "tofu", size: "small", price: 4 }
      } }
   ],
   { ordered: false } )
} catch( error ) {
   print( error )
}

Example output, which includes the duplicate key error and a summary of the completed operations:

writeErrors: [
   WriteError {
      err: {
         index: 1,
         code: 11000,
         errmsg: 'E11000 duplicate key error collection: test.pizzas index: _id_ dup key: { _id: 3 }',
         op: { _id: 3, type: 'sausage', size: 'large', price: 10 }
      }
   }
],
result: BulkWriteResult {
   result: {
      ok: 1,
      writeErrors: [
         WriteError {
            err: {
               index: 1,
               code: 11000,
               errmsg: 'E11000 duplicate key error collection: test.pizzas index: _id_ dup key: { _id: 3 }',
               op: { _id: 3, type: 'sausage', size: 'large', price: 10 }
            }
         }
      ],
      writeConcernErrors: [],
      insertedIds: [ { index: 0, _id: 3 }, { index: 1, _id: 3 } ],
      nInserted: 1,
      nUpserted: 0,
      nMatched: 2,
      nModified: 2,
      nRemoved: 1,
      upserted: []
   }
}

The second insertOne operation fails because of the duplicate key error. In an unordered bulkWrite(), any operation without an error is completed.

Bulk Write with Write Concern Example

Continuing the pizzas collection example, drop and recreate the collection:

db.pizzas.insertMany( [
   { _id: 0, type: "pepperoni", size: "small", price: 4 },
   { _id: 1, type: "cheese", size: "medium", price: 7 },
   { _id: 2, type: "vegan", size: "large", price: 8 }
] )

The following bulkWrite() example runs operations on the pizzas collection and sets a "majority" write concern with a 100 millisecond timeout:

try {
   db.pizzas.bulkWrite( [
      { updateMany: {
         filter: { size: "medium" },
         update: { $inc: { price: 0.1 } }
      } },
      { updateMany: {
         filter: { size: "small" },
         update: { $inc: { price: -0.25 } }
      } },
      { deleteMany: { filter: { size: "large" } } },
      { insertOne: {
         document: { _id: 4, type: "sausage", size: "small", price: 12 }
      } } ],
      { writeConcern: { w: "majority", wtimeout: 100 } }
   )
} catch( error ) {
   print( error )
}

If the time for the majority of replica set members to acknowledge the operations exceeds wtimeout, the example returns a write concern error and a summary of completed operations:

result: BulkWriteResult {
   result: {
      ok: 1,
      writeErrors: [],
      writeConcernErrors: [
         WriteConcernError {
            err: {
               code: 64,
               codeName: 'WriteConcernTimeout',
               errmsg: 'waiting for replication timed out',
               errInfo: { wtimeout: true, writeConcern: [Object] }
            }
         }
      ],
      insertedIds: [ { index: 3, _id: 4 } ],
      nInserted: 0,
      nUpserted: 0,
      nMatched: 2,
      nModified: 2,
      nRemoved: 0,
      upserted: [],
      opTime: { ts: Timestamp({ t: 1660329086, i: 2 }), t: Long("1") }
   }
}