Docs Menu

Docs HomeMongoDB Manual

Convert a Replica Set to a Sharded Cluster

On this page

  • About This Task
  • Server Architecture
  • Before You Begin
  • Steps
  • 1. Deploy the Config Server Replica Set
  • 2. Deploy the mongos
  • 3. Create an Administrative User for the Sharded Cluster
  • 4. Migrate the admin Database from the Initial Replica Set
  • 5. Restart the Initial Replica Set as a Shard
  • 6. Add the Initial Replica Set as a Shard
  • 7. Update Your Application Connection String
  • 8. Deploy a Second Replica Set
  • 9. Add the Second Replica Set to the Cluster as a Shard
  • 10. Shard a Collection
  • Learn More

Sharded clusters partition data across multiple servers based on a shard key. A sharded cluster scales better than a replica set for deployments with large data sets and high throughput operations.

This tutorial converts a single three-member replica set to a sharded cluster with two shards. Each shard in the new cluster is an independent three-member replica set.

This tutorial uses the following servers:

Hostname
Port
Description
mongodb0.example.net
27017
Member of the initial data-bearing shard, rs0.
mongodb1.example.net
27017
Member of the initial data-bearing shard, rs0.
mongodb2.example.net
27017
Member of the initial data-bearing shard, rs0.
mongodb3.example.net
27018
Member of the second data-bearing shard, rs1.
mongodb4.example.net
27018
Member of the second data-bearing shard, rs1.
mongodb5.example.net
27018
Member of the second data-bearing shard, rs1.
mongodb6.example.net
27017
The mongos, used to connect to the sharded cluster.
mongodb7.example.net
27019
Member of the config server replica set.
mongodb8.example.net
27019
Member of the config server replica set.
mongodb9.example.net
27019
Member of the config server replica set.

The hostnames used in this tutorial are examples. Replace the hostnames used in the example commands with the hostnames used in your deployments.

Important

To avoid configuration updates due to IP address changes, use DNS hostnames instead of IP addresses. It is particularly important to use a DNS hostname instead of an IP address when configuring replica set members or sharded cluster members.

Use hostnames instead of IP addresses to configure clusters across a split network horizon. Starting in MongoDB 5.0, nodes that are only configured with an IP address will fail startup validation and will not start.

To convert a replica set to a sharded cluster:

  1. Deploy the config server replica set.

  2. Deploy a mongos.

  3. Create an administrative user for the sharded cluster.

  4. Migrate the admin database from the initial replica set to the sharded cluster.

  5. Restart the members of the initial replica set as shard servers.

  6. Add the initial replica set as a shard.

  7. Update your application connection string.

  8. Create a second shard and add it to the cluster.

  9. Shard a collection.

Deploy a three-member replica set for the config servers. In this example, the config servers use the following hosts:

  • mongodb7.example.net

  • mongodb8.example.net

  • mongodb9.example.net

1

Select the tab for your authentication mechanism:

Deploy the mongod with your specified configuration:

mongod --config <PATH_TO_CONFIG_FILE>

The config servers use the default data directory /data/configdb and the default port 27019.

2
3

To initiate the replica set, run rs.initiate():

rs.initiate( {
_id: "configReplSet",
configsvr: true,
members: [
{ _id: 0, host: "mongodb7.example.net:27019" },
{ _id: 1, host: "mongodb8.example.net:27019" },
{ _id: 2, host: "mongodb9.example.net:27019" }
]
} )

The preceding command uses the localhost exception to perform administrative actions without authentication.

Important

Run rs.initiate() on just one and only one mongod instance for the replica set.

The mongos provides the interface between the client applications and the sharded cluster.

1
2

Deploy the mongos with your specified configuration:

mongos --config <PATH_TO_CONFIG_FILE>

After you deploy the mongos, use the localhost exception to create the first user on the cluster.

1

Use mongosh to connect to the mongos.

2

Tip

Specify a different username for your sharded cluster administrative user than the usernames in your initial replica set.

Using different names avoids conflicts when restoring the users from the initial replica set to the sharded cluster.

The following db.createUser() method creates a user named admin01 with the clusterManager and restore roles:

use admin
db.createUser(
{
user: "admin01",
pwd: passwordPrompt(),
roles: [
{ role: "clusterManager", db: "admin" },
{ role: "restore", db: "admin" }
]
}
)

After you run the command, the database prompts you to enter a password for the admin01 user.

The admin database contains user and system information. To migrate the admin database from your initial replica set to the sharded cluster, perform the following steps.

In this example, your initial replica set is a three-member replica set. This step updates the initial replica set so that it can be added as a shard to your sharded cluster.

The replica set runs on these hosts:

  • mongodb0.example.net:27017

  • mongodb1.example.net:27017

  • mongodb2.example.net:27017

For sharded clusters, you must set the role for each mongod instance in the shard to shardsvr. To specify the server role, set the sharding.clusterRole setting in the mongod configuration file.

Note

The default port for mongod instances with the shardsvr role is 27018. To use a different port, specify the net.port setting.

1

Use mongosh to connect to one of the members of your initial replica set.

2

Run rs.status() to determine the primary and secondaries:

rs.status()

In the command output, the replSetGetStatus.members[n].stateStr field indicates which member is the primary and which members are secondaries.

3

Warning

This step requires some downtime for applications connected to the replica set secondaries.

After you restart a secondary, any applications that are connected to that secondary return a CannotVerifyAndSignLogicalTime error until you perform the steps in 6. Add the Initial Replica Set as a Shard.

You can also restart your application to stop it from receiving CannotVerifyAndSignLogicalTime errors.

4

Warning

This step requires some downtime for applications connected to the primary of the replica set.

After you restart the primary, any applications that are connected to the primary return a CannotVerifyAndSignLogicalTime error until you perform the steps in 6. Add the Initial Replica Set as a Shard.

You can also restart your application to stop it from receiving CannotVerifyAndSignLogicalTime errors.

After you convert the initial replica set (rs0) to a shard, add it to the sharded cluster.

1

The mongos instance is running on host mongodb6.example.net.

This command authenticates you as the admin01 user you created on the sharded cluster. After you enter the command, enter your user's password.

2

To add a shard to the cluster, run the sh.addShard() method:

sh.addShard( "rs0/mongodb0.example.net:27017,mongodb1.example.net:27017,mongodb2.example.net:27017" )

Warning

Once the new shard is active, mongosh and other clients must always connect to the mongos instance. Do not connect directly to the mongod instances. If your clients connect to shards directly, you may create data or metadata inconsistencies.

After you add the first shard to your cluster, update the connection string used by your applications to the connection string for your sharded cluster. Then, restart your applications.

Deploy a new replica set called rs1. The members of replica set rs1 are on the following hosts:

  • mongodb3.example.net

  • mongodb4.example.net

  • mongodb5.example.net

1

Deploy the mongod with your specified configuration:

mongod --config <PATH_TO_CONFIG_FILE>

Note

When you specify the --shardsvr option for a mongod instance, the instance runs on port 27018 by default.

2
3

Use mongosh to connect to one of the replica set members. For example:

4

In mongosh, run the rs.initiate() method to initiate a replica set that contains the current member:

rs.initiate( {
_id : "rs1",
members: [
{ _id: 0, host: "mongodb3.example.net:27018" },
{ _id: 1, host: "mongodb4.example.net:27018" },
{ _id: 2, host: "mongodb5.example.net:27018" }
]
} )

The preceding command requires the localhost exception to perform administrative actions without authentication.

Important

Run rs.initiate() on just one and only one mongod instance for the replica set.

5

After you deploy the replica set, use the localhost exception to create the replica set's first user.

1

To determine the primary, run rs.status():

rs.status()

In the command output, the replSetGetStatus.members[n].stateStr field indicates which member is the primary.

2

Connect to the replica set primary with mongosh. For example, if the primary is mongodb4.example.net, run this command:

3

Run the following db.createUser() method to create a user named rs1Admin with the userAdmin role:

use admin
db.createUser(
{
user: "rs1Admin",
pwd: passwordPrompt(),
roles: [
{ role: "userAdmin", db: "admin" }
]
}
)

After you run the command, the database prompts you to enter a password for the rs1Admin user.

Add the new replica set, rs1, to the sharded cluster.

1

Run the following command from the command line to connect to the mongos instance running on host mongodb6.example.net:

This command authenticates you as the admin01 user you created on the sharded cluster. After you enter the command, enter your user's password.

2

After you connect to the mongos, add the replica set rs1 as a shard to the cluster with the sh.addShard() method:

sh.addShard( "rs1/mongodb3.example.net:27018,mongodb4.example.net:27018,mongodb5.example.net:27018" )

The final step of the procedure is to shard a collection in the sharded cluster.

1

Determine the shard key for the collection. The shard key indicates how MongoDB distributes the documents between shards. Good shard keys:

  • Have values that are evenly distributed among all documents.

  • Group documents that are often accessed at the same time into contiguous chunks.

  • Allow for effective distribution of activity among shards.

For more information, see Choose a Shard Key.

This procedure uses the number field as the shard key for the test_collection collection.

2

Before you shard a non-empty collection, create an index on the shard key:

use test
db.test_collection.createIndex( { "number" : 1 } )
3

In the test database, shard the test_collection. Specify number as the shard key.

sh.shardCollection( "test.test_collection", { "number" : 1 } )

The next time that the balancer runs, it redistributes chunks of documents between shards. As clients insert additional documents into this collection, the mongos routes the documents to the appropriate shard.

Tip

Schedule the balancing window

When the balancer redistributes chunks, it may negatively impact your application's performance. To minimize performance impact, you can specify when the balancer runs so it does not run during peak hours. To learn more, see Schedule the Balancing Window.

For more sharding tutorials and procedures, see these pages:

←  Convert Sharded Cluster to Replica SetDrop a Hashed Shard Key Index →