Hi ATN,

Good question. Based on what I know, dataset_catalog and dataset_schema in the dashboard DAB resource currently apply to datasets that use SQL queries (i.e., the queryLines property), but they do not resolve for datasets that reference metric views via the asset_name property. That lines up with the behavior you are seeing — it works with queryLines but fails with asset_name and an unqualified name.

Metric views in general require fully qualified three-part names (catalog.schema.metric_view_name). The dataset_catalog/dataset_schema substitution does not currently get applied when the dashboard resolves an asset_name reference, which is why you get the TABLE_OR_VIEW_NOT_FOUND error with the empty backticks (``.``.``) — it is not injecting the catalog/schema values at all.

For now, the workaround is to use the fully qualified name in the asset_name property (e.g., catalog.schema.table_name). If you need environment-specific routing, you could use bundle variable substitution to parameterize the catalog/schema portions of the fully qualified name. I do not have a confirmed timeline for when dataset_catalog/dataset_schema will be extended to cover asset_name references, but it is a known gap.

Sources:

• Databricks Asset Bundles resources documentation
• CLI PR #4130 - Add support for configurable catalog/schema for dashboards