We have identified 4 data integrity issues in the MongoDB Relational Migrator (RM) tool which can cause data loss or corruption without a visible error message.

You may be impacted if:

1. You have migrated with RM to a sharded cluster and your project includes embedded document or embedded array mappings.
2. A table is mapped to more than one collection in the target schema used to migrate.
3. A source table column is used in multiple composite (multiple columns) foreign key relationships.
4. Any fields in the target schema have been renamed to the same name as a different source column.

Relational Migrator version 1.16.0 fixes scenarios 2-4 listed above. Migrations to sharded clusters have been temporarily disabled while we address the first issue.

If you believe any of the situations above impact your RM projects, please:

1. Prioritize checking the affected collections for missing or incorrect data.
2. Plan to re-migrate any affected collections using Relational Migrator v1.16.0.
3. If you’re unsure whether you’re affected or how to check, please open a case with MongoDB Support so we can help you confirm and plan next steps.

For more information regarding each issue, refer to the linked tickets:

• SERVER-126523- Sharded cluster migrations with embedded documents or embedded arrays may cause missing documents. Affects Relational Migrator versions <1.16.0.
• SERVER-126528- Cyclic mappings with an excluded foreign key under an embedded array will be missing. Affects Relational Migrator versions 1.14.1 to 1.15.2 inclusive
• SERVER-126529- Source table column used in multiple composite foreign key relationships can cause mismatched embeddings. Affects Relational Migrator versions <1.16.0.
• SERVER-126530- Conflicting field name choices can cause omitted and mis-named fields. Affects Relational Migrator versions <1.16.0.

If you have any questions or need to confirm whether you have been affected, please open a support case. We apologise for any inconvenience this may have caused.