dbt

Schema Drift Detection via Column Not Found

critical
reliabilityUpdated Feb 19, 2026

Identifies upstream schema changes that break dbt models when columns are renamed, dropped, or have mismatched types, causing runtime Database Errors.

How to detect:

Monitor for Database Errors with messages like 'column does not exist' or 'type mismatch' during model execution. Track error rates correlated with upstream data source refresh times.

Recommended action:

Query information_schema or equivalent to identify actual column names and types. Update model SQL to reference correct column names, add explicit type casts where needed, or use dbt_utils.star() with except clauses to handle schema evolution gracefully.