Openflow Connector for MongoDB: Maintenance

Reinstall the connector

Use this procedure to move an existing connector to another runtime while preserving its replication state and continuing to write to the same destination tables.

Prerequisites

  • Let any in-progress snapshot loads finish before exporting the connector. The downloaded flow definition does not include queued FlowFiles.
  • Use Openflow Runtime Server version 2026.6.4.18 or later to download the flow definition with component state.
  • Prepare a target runtime that meets the MongoDB connector requirements. Configure source connectivity and destination access as described in Connect to MongoDB.
  • Record the connector’s parameter values and have its credentials available. Sensitive values, such as passwords, aren’t included in the downloaded flow definition.

Warning

The MongoDB oplog must retain the history needed to resume from the saved change-stream position throughout the move. If the resume token expires before the new connector starts, a full data re-sync might be required. Keep the interruption within the source’s oplog retention window.

Move the connector

  1. In the existing connector’s canvas, stop Set Collections for Replication and Read MongoDB Change Stream. Wait for their active threads to finish.
  2. Let the remaining processors finish processing all in-flight FlowFiles. Wait until the connector’s Queued value is zero. If you use a custom Merge Task Schedule CRON, temporarily set it to * * * * * ? so processing doesn’t wait for the next scheduled merge.
  3. Stop all remaining processors, wait for their active threads to finish, and disable the controller services. Keep this connector stopped throughout the move and while the replacement connector runs.
  4. Right-click the connector’s process group and select Download flow definition. Select both options:
    • Export with External Services: includes referenced controller services from parent process groups.
    • Export with Components State: includes the collection replication state and change-stream position.
  5. Import the downloaded flow definition into the target runtime. Don’t start it yet.
  6. Review the imported parameter contexts. Keep the same collection selection, Destination Database, Destination Schema Pattern, and Object Identifier Resolution. Re-enter MongoDB Password and any other sensitive values required by the MongoDB and Snowflake authentication settings. If you changed Merge Task Schedule CRON to drain the old connector, restore the original schedule.
  7. Enable the required controller services and start the imported connector.
  8. Verify that replication resumes and that changes made in MongoDB during the interruption arrive in the destination tables.

Note

Keep the existing destination and journal tables. Don’t remove and re-add collections as part of this procedure; that is a separate operation to restart replication from scratch.