Power Platform Community Forum Thread Details

I have a Power Apps application that uses Azure Databricks with several tables and views defined as datasources.

Our catalogs are environment specific and the connector doesn't support the ability to provide a catalog name as an environment variable. (See my prior question: https://community.powerplatform.com/forums/thread/details/?threadid=e4d91f93-d92e-f111-88b4-000d3a54cb8a) Microsoft Power Platform support has confirmed that there is no support for this today and none is currently planned.

My workaround is to:

export the solution as an unmanaged solution from my developer environment and import it into my target environment as an unmanaged solution
Remove all of the datasources with the lower environment dataset name
Re-add the datasources using the correct dataset name (catalog)

The problem I am encountering is when I attempt to re-add the datasources, not all of the tables or views are showing up in the pick list. And, when I attempt to specify the schema and table name manually, the Azure Databricks connector interface supports this, I receive the following error:

Invalid response from service; the table metadata name prod.speedtomarket.creditcardproduct does not seem to match the name used in the request of <schema-name>.<table-name>. Please contact API owner.

We are using a service principal for the underlying connection. Further, I have verified with the Databricks admin that access to all tables has been granted to the security group to which the service principal has been assigned.

This seems like an issue with the connector or the platform itself to me. Any assistance that the community can provide would be greatly appreciated!

Categories:

Building Power Apps

Connector development

Power Apps Pro Dev & ISV

Apparently, Power Apps interface has a hard display limit (roughly 1,600 objects). When a catalog exceeds this, the UI fails to "discover" the metadata automatically, forcing you to manually enter the full datasource name, including the catalog name.

This approach further uses a simplified naming engine, rather than the legacy naming engine. When the developer is able to select the table from the pick-list, the datasource will be named `<schema-name>.<table-name>`. When manually specifying the table name as `<catalog>.<schema-name>.<table-name>`, the datasource name will be simplified as only `<table-name>`.

You cannot rename these datasources, to align with the function references that existed prior either.