dlt handles schema changes silently by default — new columns are added and type conflicts become variant columns without errors. This post shows how to instrument schema monitoring using dlt's `schema_update` metadata, and how `_dlt_loads` and `_dlt_version` tables serve as an audit trail. A step-by-step walkthrough covers baseline loading, detecting new columns, handling type mismatches via variant columns, and tracing schema history through load lineage joins.

3m read timeFrom dlthub.com
Post cover image
Table of contents
The setup: a transactions pipeline with schema monitoring Link iconStep 1: Run the pipeline and check what schema dlt infers Link iconStep 2: Add new fields and check what 'schema_update' reports Link iconStep 3: Load a type mismatch and see it land in a variant column Link iconStep 4: Inspect load lineage with '_dlt_loads' Link iconStep 5: Track schema history with '_dlt_version' Link iconThe takeaway Link icon

Sort: