Skip to main content
Version: Next

DATABRICKS

This article focuses on Databricks-specific considerations for users of the datavault4dbt package.

MULTI-ACTIVE HASHDIFF​

From v2.0.0, the multi-active hashdiff on Databricks uses the native LISTAGG function with an explicit WITHIN GROUP (ORDER BY ...) clause. This produces a deterministic, correctly-ordered aggregation without requiring a derived ROW_NUMBER workaround.

LISTAGG(column_name, ',') WITHIN GROUP (ORDER BY multi_active_key)

MINIMUM RUNTIME REQUIREMENT​

LISTAGG ... WITHIN GROUP (ORDER BY ...) requires Databricks Runtime 16.4 LTS or later (or an equivalent Databricks SQL warehouse version). If you are on an older runtime, upgrade before using v2.0.0 multi-active satellites, or the model will fail to compile with a function-not-found error.

ADAPTER SPECIFIC VARIABLE​

The variable datavault4dbt.set_casing can be used to force all column names to be uppercased or lowercased. Allowed values for the variable are upper, uppercase, lower, lowercase.