Forum Discussion
dataflow gen1
Hi remmypocco
Thank you for reaching out to the Microsoft Fabric community forum.
In Dataflow Gen1, this behavior is expected due to how Power Query Online handles data-source privacy. Even if all tables are from the same SharePoint site with identical URLs, credentials, and an Organizational privacy level, the online engine treats each entity in the dataflow as a separate data-source instance.
When you edit the dataflow and perform merge or append operations, Power Query Online re-applies privacy rules at edit time and interprets the operation as combining multiple sources. This privacy check does not happen during scheduled refresh, which is why refresh works with the validated mashup while editing can fail with a privacy error.
Unlike Power BI Desktop, Dataflow Gen1 does not inherit the “Ignore Privacy Levels” option and does not always provide privacy settings in the Service. Because of this, Gen1 cannot reliably support merges across multiple entities, even when they use the same SharePoint site and matching credentials and privacy levels.
With a standard star schema using separate fact and dimension tables, this is a known architectural limitation of Dataflow Gen1 rather than a configuration problem. The main options are to keep all merge logic in a single query chain with one SharePoint connection (which may not be practical for dimensional models), move merge and ID-generation to Power BI Desktop and use Gen1 dataflows only for ingestion, or migrate to Dataflows Gen2, which has a modern engine and consistent privacy handling.
I hope this helps. If you have more questions, feel free to ask and we’ll be glad to assist further.
Best Regards,
Microsoft Fabric Community Support Team.
Thank you for the reply,
For option 1 i cant seem to find anything in power bi service about dataflow privacy service, the settings on desktop ignores privacy.
for option 2 i wont be able to create a single entity that contains all transformation because for my report i have 4 dimension tables and one fact table. however merging is needed to create some IDs in the fact table to connect with the dimension tables. some dimension table are data i get from sharepoint dropped by client as its not in house data. so everything can be the same source however its the same sharepoint sites.
thank you
Hi remmypocco
Thank you for reaching out to the Microsoft Fabric community forum.
In Dataflow Gen1, this behavior is expected due to how Power Query Online handles data-source privacy. Even if all tables are from the same SharePoint site with identical URLs, credentials, and an Organizational privacy level, the online engine treats each entity in the dataflow as a separate data-source instance.
When you edit the dataflow and perform merge or append operations, Power Query Online re-applies privacy rules at edit time and interprets the operation as combining multiple sources. This privacy check does not happen during scheduled refresh, which is why refresh works with the validated mashup while editing can fail with a privacy error.
Unlike Power BI Desktop, Dataflow Gen1 does not inherit the “Ignore Privacy Levels” option and does not always provide privacy settings in the Service. Because of this, Gen1 cannot reliably support merges across multiple entities, even when they use the same SharePoint site and matching credentials and privacy levels.
With a standard star schema using separate fact and dimension tables, this is a known architectural limitation of Dataflow Gen1 rather than a configuration problem. The main options are to keep all merge logic in a single query chain with one SharePoint connection (which may not be practical for dimensional models), move merge and ID-generation to Power BI Desktop and use Gen1 dataflows only for ingestion, or migrate to Dataflows Gen2, which has a modern engine and consistent privacy handling.
I hope this helps. If you have more questions, feel free to ask and we’ll be glad to assist further.
Best Regards,
Microsoft Fabric Community Support Team.