Forum Discussion
Cannot change data source credentials to anything different than current credentials because of SSO
I had written messages 16, 18, 19, in this thread in November 2020, and am finding the same issue as ever.
Again, I had gone through the somewhat painful process of opening a ticket with Microsoft support and only being given the workaround of having IT give my personal account access to the data source in order to get it to refresh.
I would love to hear anyone else's experience of also working with Microsoft support on issue resolution or acknowledging this bug.
Hi! I struggled with this issue for various weeks (the only solution we had found in my company was to give the generic / machine user access to Power BI which I didn't like at all).
But after trying out some things I managed to do it while working from home and connected with VPN. I think it very much depends on your company's implementation of SSO in Azure AD. Please try it out if you have a similar set up and let us know.
-If you ARE connected through VPN, disconnect it.
-In the Power BI workspace go to your Dataflow / Dataset > Edit Credentials.
-The "Pick an account" screen should open in a pop up window. You should see your regular company email address account at the top "Connected to Windows", BUT by now after trying to log in with the generic user in different places you would have it already available to select it below. If not you can also try with "Use another account" but not sure if it will work the same way.
-You then get to a Login prompt with a Password box (in my case this was automatically bypassed when connected through the VPN).
-After putting the generic user's password you would get an error message saying you should activate VPN.
-Activate the VPN
-Refresh the pop up window with the error message
-And BOOM: the actual sign in with the generic user account would happen at this point.
How to check: I guess depends on the source. In my case the source is databricks and the only way I know of telling is to trigger a refresh of the dataset in Power BI, then access to databricks with the generic user and go to the query history to see if any queries related to the dataset were just triggered.
Hope this solutions works for others and not only in my case, but as said it might depends on particularities of each case.