split
Definition
split
Changed in version 4.2.
Splits a chunk in a sharded cluster into two chunks. Starting in MongoDB 4.2, shards manage and split chunks based on the chunk size statistics they maintain.
For exceptional circumstances the
split
command does allow administrators to manually create splits. See Split Chunks in a Sharded Cluster for information on these circumstances, and on the MongoDB shell commands that wrapsplit
.Tip
In the
mongo
Shell, this command can also be run through thesh.splitAt()
andsh.splitFind()
helper methods.Helper methods are convenient for
mongo
users, but they may not return the same level of information as database commands. In cases where the convenience is not needed or the additional return fields are required, use the database command.The
split
command must be run in theadmin
database and uses the following form:db.adminCommand( { split: <database>.<collection>, <find|middle|bounds> } ) The
split
command takes a document with the following fields:FieldTypeDescriptionsplit
stringThe name of the collection where the chunk exists. Specify the collection's full namespace, including the database name.find
documentAn query statement that specifies an equality match on the shard key. The match selects the chunk that contains the specified document. You must specify only one of the following:
find
,bounds
, ormiddle
.You cannot use the
find
option on an empty collection.bounds
arrayNew in version 2.4: The bounds of a chunk to split.
bounds
applies to chunks in collections partitioned using a hashed shard key. The parameter's array must consist of two documents specifying the lower and upper shard-key values of the chunk. The values must match the minimum and maximum values of an existing chunk. Specify only one of the following:find
,bounds
, ormiddle
.You cannot use the
bounds
option on an empty collection.middle
documentThe document to use as the split point to create two chunks.split
requires one of the following options:find
,bounds
, ormiddle
.
Considerations
When used with either the find
or the bounds
option, the
split
command splits the chunk along the median. As such,
the command cannot use the find
or the bounds
option to split
an empty chunk since an empty chunk has no median.
To create splits in empty chunks, use either the middle
option with
the split
command or use the sh.splitAt()
command.
Command Formats
To create a chunk split, connect to a mongos
instance, and
issue the following command to the admin
database:
db.adminCommand( { split: <database>.<collection>, find: <document> } )
Or:
db.adminCommand( { split: <database>.<collection>, middle: <document> } )
Or:
db.adminCommand( { split: <database>.<collection>, bounds: [ <lower>, <upper> ] } )
To create a split for a collection that uses a hashed shard key,
use the bounds
parameter. Do not use the middle
parameter for
this purpose.
Warning
Be careful when splitting data in a sharded collection to create new chunks. When you shard a collection that has existing data, MongoDB automatically creates chunks to evenly distribute the collection. To split data effectively in a sharded cluster you must consider the number of documents in a chunk and the average document size to create a uniform chunk size. When chunks have irregular sizes, shards may have an equal number of chunks but have very different data sizes. Avoid creating splits that lead to a collection with differently sized chunks.
Examples
The following sections provide examples of the split
command.
Split a Chunk in Half
db.adminCommand( { split : "test.people", find : { _id : 99 } } )
The split
command identifies the chunk in the people
collection of the test
database, that holds documents that match {
_id : 99 }
. split
does not require that a match exist, in order
to identify the appropriate chunk. Then the command splits it into two
chunks of equal size.
Note
Define an Arbitrary Split Point
To define an arbitrary split point, use the following form:
db.adminCommand( { split : "test.people", middle : { _id : 99 } } )
The split
command identifies the chunk in the people
collection of the test
database, that would hold documents
matching the query { _id : 99 }
. split
does not
require that a match exist, in order to identify the appropriate
chunk. Then the command splits it into two chunks, with the matching
document as the lower bound of one of the split chunks.
This form is typically used when pre-splitting data in a collection.
Split a Chunk Using Values of a Hashed Shard Key
This example uses the hashed shard key userid
in a
people
collection of a test
database. The following command
uses an array holding two single-field documents to represent the
minimum and maximum values of the hashed shard key to split the chunk:
db.adminCommand( { split: "test.people", bounds : [ { userid: NumberLong("-5838464104018346494") }, { userid: NumberLong("-5557153028469814163") } ] } )
Note
MongoDB uses the 64-bit NumberLong type to represent the hashed value.
Use sh.status()
to see the existing bounds of the shard keys.
Metadata Lock Error
If another process, such as a balancer
process, changes metadata while split
is running, you may
see a metadata lock error
.
errmsg: "The collection's metadata lock is already taken."
This message indicates that the split has failed with no side
effects. Retry the split
command.