Our source Heroke Postgres v15 source database is experiencing XMIN wraparound. This prevents the Fivetran connector from using XMIN values to identify changed records during incremental replication.

Many of the tables in our source database have change-tracking fields (i.e. last_updated_timestamp) to track when our applications update records.

When we are selecting tables for replication in our Connection Schema, we would like to be able to identify a column for the Fivetran connector to use as an identifier to filter by for each successive incremental replication/sync.

You could optionally suggest that an index be created on this field in the source for more expedient identification of relevant records for replication.

Additionally, even with XMIN wraparound, when writing records to the destination the Fivetran connector should still honor the Primary Key and/or Unique contraints defined in the source database. Our destination does not enforce these constraints but our source does, and as a Fivetran user I would expect the Fivetran ETL application to handle this for me.