Docs Menu
Docs Home
/
MongoDB Cluster-to-Cluster Sync
/ /

Many-to-One Migrations

On this page

  • Syntax
  • Behavior
  • Example
  • Learn More

Important

Cluster-to-Cluster Sync Beta Program

This feature is only available in mongosync beta. To learn more, see Cluster-to-Cluster Sync Beta Program.

Starting in mongosync beta 1.8, you can perform Many-to-One migrations. Many-to-One migrations allow you to sync multiple source clusters simultaneously with a destination cluster. For example, you can consolidate data from many small clusters into a central cluster.

To sync multiple source clusters with one destination cluster, run the following commands when starting mongosync:

./bin/mongosync \
--cluster0 <source-A-connection-string> \
--cluster1 <destination-connection-string> \
--migrationName <string>
./bin/mongosync \
--cluster0 <source-B-connection-string> \
--cluster1 <destination-connection-string> \
--migrationName <string>

To start the sync operation between the source clusters and the destination cluster, see Synchronize Data Between Clusters.

A namespace is a database_name.collection_name combination. You can only sync namespaces that don't conflict.

For example, consider this scenario:

  • Two source clusters S1 and S2.

  • A destination cluster D.

  • Databases named inventory and sales on both S1 and S2.

  • Collections named products, orderLines, orderStatus, and orders on both S1 and S2.

  • You can sync both of these combinations:

    • inventory.products and sales.orderStatus on S1 with D.

    • inventory.orderLines and sales.orders on S2 with D.

  • You cannot sync both of these combinations because they conflict:

    • inventory.products and inventory.orderLines on S1 with D. If D is initially empty, you can sync S1 with D. inventory.products and inventory.orderLines are copied from S1 to D.

    • inventory.products and inventory.orderLines on S2 with D. You cannot sync S2 with D because inventory.products and inventory.orderLines conflict with the namespaces already on D from the scenario in the previous point.

The migrationName string can contain up to 44 alphanumeric and underscore characters. migrationName is appended to the string "mongosync_internal_" to set the migration metadata database name.

For example, if you set migrationName to "cluster_27000_to_cluster_35000_sync", the resulting mongosync metadata database name is "mongosync_internal_cluster_27000_to_cluster_35000_sync".

Many-to-One Migrations is not compatible with the Embedded Verifier. The /start command returns an error if you enable both the verifier and Many-to-One Migrations.

For more information, see Verify Data Transfer.

The following example connects source clusters running on port 27000 and 27001 with a destination cluster running on port 35000. The command also sets the --migrationName option to describe the operations and store migration metadata for each sync.

./bin/mongosync \
--cluster0 "mongodb://localhost:27000" \
--cluster1 "mongodb://localhost:35000" \
--migrationName "cluster_27000_to_cluster_35000_sync"
./bin/mongosync \
--cluster0 "mongodb://localhost:27001" \
--cluster1 "mongodb://localhost:35000" \
--migrationName "cluster_27001_to_cluster_35000_sync"

Back

A->B->C Migrations