Forum Discussion
Semantic Model unable to load from Mirrored Database
Option 1: Create Semantic Model from the Lakehouse (not SQL endpoint)
Instead of using the SQL endpoint, try this:Go to your Lakehouse in Fabric.
Use “New Semantic Model” directly from the Lakehouse table view (not the SQL endpoint).
Select the tables you want and click “Create”.
This method uses Fabric-native metadata and avoids schema errors from the mirrored SQL endpoint.
Option 2: Use Dataflow Gen2 to Load into a Warehouse or Clean View
Use a Dataflow Gen2 to read from the Lakehouse.Store the clean data into a Fabric Warehouse.
Build your semantic model from the Warehouse, which has full SQL support and reliable metadata handling.
Option 1: Create Semantic Model from the Lakehouse (not SQL endpoint)
Instead of using the SQL endpoint, try this:
Go to your Lakehouse in Fabric.
Use “New Semantic Model” directly from the Lakehouse table view (not the SQL endpoint).
Select the tables you want and click “Create”.
This method uses Fabric-native metadata and avoids schema errors from the mirrored SQL endpoint.
Option 2: Use Dataflow Gen2 to Load into a Warehouse or Clean View
Use a Dataflow Gen2 to read from the Lakehouse.
Store the clean data into a Fabric Warehouse.
Build your semantic model from the Warehouse, which has full SQL support and reliable metadata handling.