What causes STAR_WITHOUT_COLUMNS?

SELECT * from a CTE or subquery that itself selects from a table that was dropped or renamed. Schema of the referenced table or view is empty due to a failed schema evolution step. View definition references a deleted column or table and was not updated. Delta table whose schema was cleared by an erroneous REPLACE TABLE operation. Using SELECT * in a streaming query where the schema has not yet been inferred

How do I fix STAR_WITHOUT_COLUMNS?

Run `DESCRIBE TABLE ` to confirm the table schema is populated. If the table is empty-schema, re-run the schema inference or restore from Delta time travel. Replace SELECT * with explicit column names to surface the exact missing column. Recreate or refresh the view if it references dropped columns. For streaming queries, provide an explicit schema rather than relying on inference

Medium severityquery

Power BI Refresh Error:
STAR_WITHOUT_COLUMNS

What does this error mean?

A SELECT * was used but the table or subquery referenced has no resolvable columns. This can occur with empty schema definitions, dropped tables, or malformed CTEs.

Common causes

1SELECT * from a CTE or subquery that itself selects from a table that was dropped or renamed
2Schema of the referenced table or view is empty due to a failed schema evolution step
3View definition references a deleted column or table and was not updated
4Delta table whose schema was cleared by an erroneous REPLACE TABLE operation
5Using SELECT * in a streaming query where the schema has not yet been inferred

How to fix it

1Run `DESCRIBE TABLE <table>` to confirm the table schema is populated
2If the table is empty-schema, re-run the schema inference or restore from Delta time travel
3Replace SELECT * with explicit column names to surface the exact missing column
4Recreate or refresh the view if it references dropped columns
5For streaming queries, provide an explicit schema rather than relying on inference

Frequently asked questions

Can I use SELECT * safely in production Databricks SQL?

It works but is fragile — any schema change in the source will silently change downstream column lists or produce errors. Explicit column lists are safer for production pipelines.