Forum Discussion
Why do I have to restate the Data source credentials after publishing to service?
Hello,
I recently published my dashboard into service and get the error below:
It is easily fixed by just updating it here... but why does it ask me to do this again when I have already updated it in the data source settings for each one in PBI Desktop?
When you publish a .pbix from Desktop, the credentials you saved against your data sources stay on your local machine in the Windows Credential Manager. They are not embedded in the .pbix file and they do not travel to the Service.
The Service has its own credential store, so it asks you to enter them once per data source under dataset settings. After you do, it encrypts them in that store and reuses them for every scheduled refresh. If the source is on-premises, the credentials live on the on-premises data gateway instead.
This is by design, both for security and because cloud refresh runs under a different identity than your Desktop session. Once entered in the Service they should stick, you only need to re-enter them if you change the source, the auth type, or rebuild the dataset.
If that was helpful, please give a thumbs up and mark it as the accepted solution.
Thanks,
Shai Karmani
1 Reply
- Shai_Karmani
Super User
When you publish a .pbix from Desktop, the credentials you saved against your data sources stay on your local machine in the Windows Credential Manager. They are not embedded in the .pbix file and they do not travel to the Service.
The Service has its own credential store, so it asks you to enter them once per data source under dataset settings. After you do, it encrypts them in that store and reuses them for every scheduled refresh. If the source is on-premises, the credentials live on the on-premises data gateway instead.
This is by design, both for security and because cloud refresh runs under a different identity than your Desktop session. Once entered in the Service they should stick, you only need to re-enter them if you change the source, the auth type, or rebuild the dataset.
If that was helpful, please give a thumbs up and mark it as the accepted solution.
Thanks,
Shai Karmani