Hi Anand, Yes, that is correct. The schema versioning is largely additive, and the community aims to preserve backward compatibility whenever possible. In practice, this means we typically add new tables or indexes rather than remove existing ones. Removing structures can introduce breaking changes, so we try to avoid that. However, backward compatibility is not strictly guaranteed. Even seemingly minor changes, such as adding a new column, may require data backfilling or migration steps depending on the deployment. So seeing those tables present in both versions is expected and by design.
Yufei On Tue, Feb 24, 2026 at 1:17 PM Anand Kumar Sankaran via dev < [email protected]> wrote: > Hi all > > My understanding of the schema versioning is that the schema version is > additive. > > i.e, the tables in version 4 build on top of the tables in version 3. > > I see the following tables are repeated in both version 3 and version 4. > > Is that by design? > > | version | Stores the JDBC schema version | > | entities | All entities with their properties and metadata | > | grant_records | Grant records for entities (authorization) | > | principal_authentication_data | Authentication data for clients | > | policy_mapping_record | Policy mappings for targets | > | events | Event log table | > > - > Anand >
