CRUD Operations

CRUD operations are those which deal with creating, reading, updating, and deleting documents.

Key-value Pair Notation

Key-value pairs appear in many different contexts in the MongoDB Ruby driver, and there are some quirks of syntax with regard to how they can be notated which depend on which version of Ruby you’re using.

When constructing a document, the following syntax is acceptable and correct for Ruby version 1.9 and later:

document = { name: "Harriet", age: 36 }

If you’re using Ruby version 2.2 or greater, you can optionally enclose your keys in quotes.

document = { "name": "Harriet", "age": 36 }

If you need to use any MongoDB operator which begins with $, such as $set, $gte, or $near, you must enclose it in quotes. If you’re using Ruby version 2.2 or greater, you can notate it as follows:

collection.update_one({ name: "Harriet" }, { "$set": { age: 42 } })

If you’re using an earlier version of Ruby, use the hashrocket symbol:

collection.update_one({ name: "Harriet" }, { "$set" => { age: 42 } })

Quoted strings and hashrockets for key-value pairs will work with any version of Ruby:

collection.update_one({ "name" => "Harriet" }, { "$set" => { age: 42 } })

Creating Documents

To insert documents into a collection, select a collection on the client and call insert_one or insert_many.

Insert operations return a Mongo::Operation::Result object which gives you information about the insert itself.

On MongoDB 2.6 and later, if the insert fails, an exception is raised, because write commands are used.

On MongoDB 2.4, an exception is only raised if the insert fails and the write concern is 1 or higher.

client =[ '' ], :database => 'music')

result = client[:artists].insert_one( { :name => 'FKA Twigs' } )
result.n # returns 1, because 1 document was inserted.

result = client[:artists].insert_many([
  { :name => 'Flying Lotus' },
  { :name => 'Aphex Twin' }
result.inserted_count # returns 2, because 2 documents were inserted.

Specify a Decimal128 number

New in version 3.4.

Decimal128 is a BSON datatype that employs 128-bit decimal-based floating-point values capable of emulating decimal rounding with exact precision. This functionality is intended for applications that handle monetary data, such as financial and tax computations.

The following example inserts a value of type Decimal128 into the price field of a collection named inventory:

client =[ '' ], :database => 'test')

price ="428.79")
client[:inventory].insert_one({ "_id" => 1,
                                "item" => "26 inch monitor",
                                "price" => price })

The above operation produces the following document:

{ "_id" : 1, "item" : "26 inch monitor", "price" : NumberDecimal("428.79") }

You can also create a Decimal128 object from a Ruby BigDecimal object, or with Decimal128.from_string().

big_decimal =, 5)
price =
# => BSON::Decimal128('428.79')

price = BSON::Decimal128.from_string("428.79")
# => BSON::Decimal128('428.79')


The Ruby driver provides a fluent interface for queries using the find method on the collection. Various options are available to the find method.

The query is lazily executed against the server only when iterating the results - at that point the query is dispatched and a Mongo::Cursor is returned.

To find all documents for a given filter, call find with the query:

client =[ '' ], :database => 'music')

client[:artists].find(:name => 'Flying Lotus').each do |document|
  #=> Yields a BSON::Document.

To query nested documents, specify the keys in nested order using dot notation.

client =[ '' ], :database => 'music')
client[:artists].find("records.releaseYear": 2008).each do |document|
  #=> Yields a BSON::Document.

Query Options

To add options to a query, chain the appropriate methods after the find method. Note that the underlying object, the Mongo::Collection::View, is immutable and a new object will be returned after each method call.

client =[ '' ], :database => 'music')

documents = client[:artists].find(:name => 'Flying Lotus').skip(10).limit(10)
documents.each do |document|
  #=> Yields a BSON::Document.

The following is a full list of the available options that can be added when querying and their corresponding methods as examples.

Option Description
allow_partial_results For use with sharded clusters. If a shard is down, allows the query to return results from the shards that are up, potentially only getting a portion of the results.
batch_size(Integer) Specifies the size of each batch of documents the cursor will return on each GETMORE operation.
comment(String) Adds a comment to the query.
hint(Hash) Provides the query with an index hint to use.
limit(Integer) Limits the number of returned documents to the provided value.
max_scan(Integer) Sets the maximum number of documents to scan if a full collection scan would be performed. Deprecated as of MongoDB server version 4.0.
no_cursor_timeout MongoDB automatically closes inactive cursors after a period of 10 minutes. Call this for cursors to remain open indefinitely on the server.

Specifies the fields to include or exclude from the results.

client[:artists].find.projection(:name => 1)

Changes the read preference for this query only.

client[:artists] => :secondary_preferred)
show_disk_loc(Boolean) Tells the results to also include the location of the documents on disk.
skip(Integer) Skip the provided number of documents in the results.
snapshot Execute the query in snapshot mode. Deprecated as of MongoDB server version 4.0.

Specifies sort criteria for the query.

client[:artists].find.sort(:name => -1)

Additional Query Operations

Get the total number of documents an operation returns.
client =[ '' ], :database => 'music')

client[:artists].find(:name => 'Flying Lotus').count
Filters out documents with duplicate values. Equivalent to the SQL distinct clause.
client =[ '' ], :database => 'music')

client[:artists].find.distinct(:name )

Tailable Cursors

For capped collections you may use a tailable cursor that remains open after the client exhausts the results in the initial cursor. The following code example shows how a tailable cursor might be used:

client =[ '' ], :database => 'music')
client[:artists, capped: true, size: 512].create

result = client[:artists].insert_many([
  { :name => 'Flying Lotus' },
  { :name => 'Aphex Twin' }

enum = client[:artists].find({}, cursor_type: :tailable_await).to_enum

while true
  doc =
  # do something

Read Preference

Read preference determines the candidate replica set members to which a query or command can be sent. They consist of a mode specified as a symbol, an array of hashes known as tag_sets, and two timing options: local_threshold and server_selection_timeout.

Defines the upper limit in seconds of the latency window between the nearest server and suitable servers to which an operation may be sent. The default is 15 milliseconds, or 0.015 seconds.
Defines how long to block for server selection before throwing an exception. The default is 30,000 milliseconds, or 30 seconds.

For more information on the algorithm used to select a server, please refer to the Server Selection documentation, available on GitHub.

Read preference can be set as an option on the client or passed an option when a command is run on a database:

# Set read preference on a client, used for all operations
client =[ '' ],
                           read: { mode: :secondary,
                                   tag_sets: [ { 'dc' => 'nyc' } ]
                                  } )

# Set read preference for a given command
client.database.command( { collstats: 'test' }, read: { mode: secondary,
                                                     tag_sets: [ { 'dc' => 'nyc' } ] } )

Read preference can also be set for specific operations on a collection using the with method:

client =[ '' ], :database => 'music')
artists = client[:artists]
artists.with(:read => { :mode => :primary_preferred }).find.to_a


There are five possible read preference modes. They are :primary, :secondary, :primary_preferred, :secondary_preferred, :nearest. Please see the read preference documentation in the MongoDB Manual for an explanation of the modes and tag sets.

Tag sets

The tag_sets parameter is an ordered list of tag sets used to restrict the eligibility of servers for selection, such as for data center awareness.

A read preference tag set (T) matches a server tag set (S) – or equivalently a server tag set (S) matches a read preference tag set (T) — if T is a subset of S.

For example, the read preference tag set { dc: 'ny', rack: 2 } matches a secondary server with tag set { dc: 'ny', rack: 2, size: 'large' }.

A tag set that is an empty document matches any server, because the empty tag set is a subset of any tag set. This means the default tag_sets parameter [{}] matches all servers.


Updating documents is possible by executing a single or multiple update, or by using the $findAndModify command.


client =[ '' ], :database => 'music')
artists = client[:artists]

result = artists.find(:name => 'Goldie').update_one("$inc" => { :plays =>  1 } )
result.n # Returns 1.

result = artists.update_one( { :name => 'Goldie' }, { "$inc" => { :plays =>  1 } } )
result.n # Returns 1.


result = artists.find(:label => 'Hospital').update_many( "$inc" => { :plays =>  1 } )
result.modified_count # Returns the number of documents that were updated.

result = artists.update_many( { :label => 'Hospital' }, { "$inc" => { :plays =>  1 } } )
result.modified_count # Returns the number of documents that were updated.


result = artists.find(:name => 'Aphex Twin').replace_one(:name => 'Richard James')
result.modified_count # Returns 1.

result = artists.replace_one( { :name => 'Aphex Twin' }, { :name => 'Richard James' } )
result.modified_count # Returns 1.

To update documents and return a document via $findAndModify, use one of the three provided helpers: find_one_and_delete, find_one_and_replace, or find_one_and_update. You can opt to return the document before or after the modification occurs.


client = [ '' ], :database => 'music')
artists = client[:artists]

artists.find(:name => 'José James').find_one_and_delete # Returns the document.


doc = artists.find(:name => 'José James').find_one_and_replace(:name => 'José')
doc # Return the document before the update.

doc = artists.find_one_and_replace({ :name => 'José James' }, { :name => 'José' })
doc # Return the document before the update.

doc = artists.find(:name => 'José James').
  find_one_and_replace( { :name => 'José' }, :return_document => :after )
doc # Return the document after the update.


doc = artists.find(:name => 'José James').
  find_one_and_update( '$set' => { :name => 'José' } )
doc # Return the document before the update.

doc = artists.find_one_and_update( { :name => 'José James' }, { '$set' => { :name => 'José' } } )
doc # Return the document before the update.


doc = artists.find(:name => 'José James').
  find_one_and_replace( { '$set' => { :name => 'José' } }, :return_document => :after )
doc # Return the document after the update.

doc = artists.find_one_and_replace(
  { :name => 'José James' },
  { '$set' => { :name => 'José' } },
  :return_document => :after
doc # Return the document after the update.



client =[ '' ], :database => 'music')
artists = client[:artists]

result = artists.find(:name => 'Björk').delete_one
result.deleted_count # Returns 1.

result = artists.delete_one(:name => 'Björk')
result.deleted_count # Returns 1.


result = artists.find(:label => 'Mute').delete_many
result.deleted_count # Returns the number deleted.

result = artists.delete_many(:label => 'Mute')
result.deleted_count # Returns the number deleted.

Write Concern

All write operations in MongoDB are executed with a write concern which is the level of acknowledgment requested from MongoDB for the particular write. More information about write concerns in general is available in the MongoDB manual.

The Ruby driver supports specifying write concern on client, collection, session (for transactions on that session), transaction, GridFS bucket and write stream levels, as well as when manually issuing commands via Database#command.

As of driver version 2.10, all driver objects accepting write concerns do so through the :write_concern option, which should be given a hash with the write concern options. Usage of the :write option is deprecated. In driver versions 2.9 and below, client, collection and GridFS objects took write concern options in the :write option with session and transaction objects employing the :write_concern option.

Below are some examples of passing write concerns to client and colection objects. The :write_concern option can be provided when constructing new client and collection objects, or to the #with methods.

GridFS examples are provided on the GridFS page.

client =[ '' ], database: 'music',
  write_concern: {w: 2})
alt_client = client.with(write_concern: {w: :majority})

collection = client[:artists, write_concern: {w: 3}]
alt_collection = collection.with(write_concern: {w: :majority})

# Uses w: 3
collection.insert_one({name: 'SUN Project'})
# Uses w: :majority
alt_collection.insert_one({name: 'SUN Project'})

Driver versions 2.9 and earlier accepted write concerns on client and collection level via the :write option. This usage continues to be supported for backwards compatibility, but is deprecated:

client =[ '' ], database: 'music',
  write: {w: 2})
alt_client = client.with(write: {w: :majority})

collection = client[:artists, write: {w: 3}]
alt_collection = collection.with(write: {w: :majority})

If both :write and :write_concern options are provided, their values must be identical or an exception will be raised:

# OK
client =[ '' ], database: 'music',
  write_concern: {w: 3}, write: {w: 3})

# Error
client =[ '' ], database: 'music',
  write_concern: {w: 3}, write: {w: :majority})

When #with methods are used to alter the options on a client or collection, the last provided option wins in case of naming differences:

client =[ '' ], database: 'music',
  write_concern: {w: 2})
alt_client = client.with(write: {w: 3})

# => {"w"=>3}

# => nil

When using transactions, write concern is only sent to the server in commit_transaction and abort_transaction operations per the transactions specification. Write concern may be set via the :write_concern option in a with_transaction or start_transaction call, or via default_transaction_options option on a session object. If neither of these is set, write concern of the client is used; note that transactions ignore write concerns of collections that are involved in their operations. Note that when setting the write concern as a transaction option, the :write option is not recognized by any driver version.

client =[ '' ], database: 'music',
  write_concern: {w: 2})
collection = client[:artists, write_concern: {w: :majority}]

session = client.start_session
session.with_transaction do
  collection.insert_one({test: 1}, session: session)

  # Uses w: 2 when committing

session = client.start_session(default_transaction_options:
  {write_concern: {w: 3})
session.with_transaction do
  collection.insert_one({test: 1}, session: session)

  # Uses w: 3 when committing

session = client.start_session
session.with_transaction(write_concern: {w: 3}) do
  collection.insert_one({test: 1}, session: session)

  # Uses w: 3 when committing

When write concerns are inherited, inheritance applies to the entire write concern hash rather than individual elements. For example, j: true is not inherited in the following case:

client =[ '' ], database: 'music',
  write_concern: {w: 1, j: true})
collection = client[:artists, write_concern: {w: 2}]

# => #<Mongo::WriteConcern::Acknowledged:0x47289650367880 options={:w=>2}>

Although CRUD operations accept an options hash, they currently do not recognize the :write_concern option:

client =[ '' ], database: 'music',
  write_concern: {w: 2})
collection = client[:artists, write_concern: {w: :majority}]

# Still uses w: :majority
collection.insert_one({name: 'SUN Project'}, write_concern: {w: 1})

The easiest workaround for this is to use #with to obtain a new collection instance with the desired write concern:

# Uses w: 1
collection.with(write_concern: {w: 1}).insert_one(name: 'SUN Project')

Write concern can also be manually specified in Database#command:

client.database.command(create: 'foo-collection', writeConcern: {w: :majority})

Note that writeConcern here is part of the operation rather than options, and the syntax is the camel case one that MongoDB server recognizes, not the underscore one that Ruby driver uses.

A Note about the BSON Symbol type

Because the BSON specification deprecated the BSON symbol type, the bson gem will serialize Ruby symbols into BSON strings when used on its own. However, in order to maintain backwards compatibility with older datasets, the Ruby driver overrides this behavior to serialize Ruby symbols as BSON symbols. This is necessary to be able to specify queries for documents which contain BSON symbols as fields. Despite this, new documents with symbol type fields should not be stored in the database; instead, use string fields.