Seed with mongorestore

You can use mongodump and mongorestore to seed MongoDB Atlas cluster with data from an existing MongoDB standalone or replica set. For guidance on seeding data from an existing MongoDB sharded cluster, contact Atlas support by clicking Support in the left-hand navigation of the Atlas UI.


Downtime Required

To ensure a completely up-to-date migration, schedule a maintenance window where you can stop all writes to your source cluster. Any write operations issued to the source cluster after the mongodump portion of the procedure completes are not migrated to the target cluster.

You must cut-over your applications to the target Atlas cluster after mongorestore completes the data restoration before resuming write operations. See Connect to a Cluster for complete documentation on connecting to a Atlas cluster.

The total amount of downtime required depends on factors such as the size of data being migrated and the network connection between your source cluster and Atlas. If you have questions or concerns about extended downtime, contact Atlas support by clicking Support from the left-hand navigation of the Atlas UI.

For a guided minimal-downtime migration procedure, see Replica Set Live Migration or Sharded Cluster Live Migration.

Cluster Security

Atlas fully manages database user creation. If the source cluster enforces authentication, use the mongorestore --nsExclude to exclude the admin.system.* namespace. You cannot migrate any existing user or role information to Atlas.


If you use mongorestore with the --oplogReplay option, you will not be able to use the --nsExclude option to exclude the admin.system.* namespace. Instead, go into the dump directory created by mongodump and delete the admin directory before running mongorestore.

For the target Atlas cluster, create the appropriate database users for supporting your application’s usage patterns. Update your applications as part of the cut-over procedure to use the new database users. See Configure Database Users for complete documentation on database user management.


This procedure requires running mongodump and mongorestore on a host in the source cluster. These programs use system resources such as CPU and memory, and may impact the performance of the host.

Consider executing this procedure during non-peak system usage, or during a scheduled maintenance window. If the source is a replica set, you can run this procedure from the host of a secondary member. After stopping writes to the cluster, allow the secondary to catch up to the primary before starting this procedure.

Pipe Behavior

This procedure uses linux pipes to stream the output of mongodump to mongorestore. If the mongorestore process cannot keep up with the mongodump process, you may encounter broken pipe errors.

For guidance on addressing persistent broken pipe errors, contact Atlas support by clicking Support from the left-hand navigation of the Atlas UI.


The following tutorial uses mongodump and mongorestore to upload data from an existing MongoDB cluster to an Atlas cluster:


Optional: Create a database user in the source replica set.


If your source cluster does not enforce authentication, you can skip this step.

If the source deployment enforces authentication, you must provide a database user with privileges to read any database as part of this procedure. See MongoDB Role-Based Access Control for more information on database user privileges.

If no such user exists, create a user in your source MongoDB replica set with the backup role on the admin database.

For example, run the following command in the mongo shell to create the mySourceUser on the admin database and assign it the backup role. For replica sets, you must run this command against the primary.

use admin
     user: "mySourceUser",
     pwd: "mySourcePassword",
     roles: [ "backup" ]

Assemble the mongodump command.

Copy the following template to into your preferred text editor:

mongodump --host '<ReplicaSetName>/<HostList>' \
  --username 'mySourceUser'
  --password 'mySourcePassword'
  --authenticationDatabase 'admin'

Replace <ReplicaSetName> with the name of the source replica set. Replace <HostList> with a comma-separated list of hostnames of the replica set members. You must specify at least one member of the replica set.

For standalone deployments, exclude <ReplicaSetName>/ and specify the hostname of the standalone deployment only. For example, --host ''

Add any additional command line options as required for your cluster. If your source cluster does not enforce authentication, exclude the following options:

Do not run this command yet. Proceed to the next step once you have modified the template.


Set up database user in the target Atlas cluster.

To run mongorestore against an Atlas cluster, you must specify a database user in the Atlas cluster that has the Atlas admin role.

If no such user exists, create the user:

  1. In the Security section of the left navigation, click Database Access. The Database Users tab displays.
  2. Click plus icon Add New Database User.
  3. Add an Atlas admin user.

See Configure Database Users for complete documentation on database user management.


Open the Command Line Tools tab.

To access the Atlas Command Line Tools tab:

  1. From the Atlas UI, use the Context picker in the top-left hand corner to select the Atlas project associated to the cluster you want to connect to.
  2. Click on the button for the cluster. From the picker menu, click Command Line Tools.

You can also click on the cluster in the Atlas UI and select the Command Line Tools tab from the cluster view.


Retrieve and modify the mongorestore connection template.

The Binary Import and Export Tools section of the Command Line Tools tab displays a copyable template with the minimum required options for connecting mongorestore to your Atlas cluster.

The template includes placeholder values for certain options. Copy and paste the template into your preferred text editor and make the following modifications:

  • <PASSWORD> - replace this with the password for the user specified in --username. The template includes a database user for the project as the --username. If you want to authenticate as a different user, replace the value of --username and specify the password for that user in --password.
  • Add --nsExclude and set its value to "admin.system.*".
  • Add --archive.

Your template should resemble the following command:

mongorestore --host myAtlasCluster/,,
  --username myAtlasUser --password 'atlasPassword'
  --authenticationDatabase admin
  --nsExclude 'admin.system.*'

Run mongodump and mongorestore.


Ensure that the host where you are running mongodump and mongorestore is in the project IP whitelist.

To review your project whitelist, click Network Access in the Security section of the left navigation. The IP Whitelist tab displays. For complete documentation on managing your project whitelist, see Project IP Whitelist.

In your preferred text editor, use the pipe | operator to separate the mongodump and mongorestore commands. The final command should resemble the following:

mongodump --host 'sourceRS/source-host1:27017,source-host2:27017,source-host3:27017' \
  --archive --ssl \
  -username 'mySourceUser' -password 'sourcePassword' --authenticationDatabase admin \
| \
mongorestore --host myAtlasRS/atlas-host1:27017,atlas-host2:27017,atlas-host3:27017 \
  --archive --ssl \
  -username myAtlasAdminUser -password 'atlasPassword' --authenticationDatabase admin \
  --nsExclude 'admin.system.*'

Run the completed command from a terminal or shell connected to a host machine on your source cluster.

Upon successful completion of the procedure, connect to your Atlas cluster using the mongo shell and verify the result of the procedure. See Connect via mongo Shell for complete instructions.

You must update your applications to point to the Atlas cluster before resuming write operations. See Connect via Driver for complete instructions on connecting applications to Atlas.

Additional Information

For more information on mongodump and mongorestore, including behavior, options, and examples, see the mongodump reference page and the mongorestore reference page.