Forum Discussion
Urgent: Power BI Fabric Dataflow Refresh Failures - MashupException.Error with a 403
:
Your message shows a MashupException with “Microsoft SQL: Internal system error (0xa) (MWC service error: Server responded with error: 403)”. In Fabric Dataflows Gen2, refreshes write through a Warehouse/Lakehouse compute layer (the internal MWC path) and also rely on the DataflowsStagingLakehouse behind the scenes. A 403 almost always indicates an identity/permission problem on either the connection (source or destination) or the staging/output Lakehouse/Warehouse—which aligns with your ownership change last month. In short: something is now refreshing with credentials that no longer have access. See: Manage connections, Workspace identity authentication, and background on staging: DataflowsStagingLakehouse and Dataflow Gen2 staging.
Things to try:
- Rule out a platform incident. Check the Fabric Service Status / Known Issues pages before spending hours: Service Status. If there’s an active incident, note it in your ticket.
- Re-bind every connection with a valid, non-leaver identity.
Go to Settings > Manage connections and gateways. Edit each SharePoint/Web source connection and each Lakehouse/Warehouse destination connection.
Prefer Workspace identity (where available) or a dedicated Entra ID service principal over a human account. Re-authenticate and save. Docs: Manage connections, workspace identity, and Microsoft’s announcement: support for workspace identity in Dataflows Gen2.
- Verify permissions on staging + outputs.
Confirm the DataflowsStagingLakehouse still exists in the workspace and that current admins can open it.
Ensure the identity used by the destination connection has Member/Contributor access to the target Lakehouse/Warehouse (and to any cross-workspace targets). Background on staging behavior: staging lakehouse and staging internals.
- Refresh one small entity and read the refresh history.
Use the Refresh history & Monitoring hub to capture the Request/Activity IDs for your ticket and to confirm whether the 403 is thrown on read (SharePoint/Web) or write (Lakehouse/Warehouse). Monitoring hub. - If your sources are SharePoint/OneDrive:
Re-enter OAuth credentials with an account that actually has site access, and confirm the site isn’t blocked/provisioning (403). See: SharePoint 403 guidance and general OAuth refresh behavior: refresh troubleshooting.
- If you use an on-prem or VNet gateway anywhere in the chain:
Make sure it’s current, and restart the service. Follow Fabric’s gateway guidance: access on-premises data and the install/requirements page: install gateway. Note: features like Fast copy require 3000.214.2+.
If you found this helpful, consider giving some Kudos. If I answered your question or solved your problem, mark this post as the solution.
Hi tayloramy ,
Thanks for your commenting with a listed of recommend ways to fix this issues! But at this time, when all dataflows started to fail in shedule refresh, I and HR Team searched about this issuses and also tried to find all DataflowsStagingLakehouse and transfer the owner to the current Admin of the Workspace which is containing dataflows and data warehouse used for HR Reporting System.
In the middle of 2024, we encountered a similar issue where the lakehouse was owned by someone who had already left the company. Unfortunately, the "Take Over" option was not available at that time, which made the situation more difficult to resolve.
We’re now concerned that a similar ownership issue could arise again, potentially disrupting business operations—especially those tied to HR data and reporting. Although the dataflows continued to function for three days after the HR data specialist (who owned both the dataflows and data warehouse) left the company, the refresh failures and reporting delays still occurred..
So please let me work with HR Department again to config that and let you know if I we have any update of fixed issues smoothly.
Thank alot for your guidance!Hi tayloramy ,
I have checked and tried all things you listed above. Currently, the owner of dataflows/data warehouses/lakehouses is also Admin of the workspace with full access rights.
About the DataflowStagingLakehouse has been taken over by the Admins of the workspace after facing with the issues and after that we have raised ticket to Microsoft Support team.- We have already tried to process somethings:
Access Manage connections and Gateways >> Check status to ensure all connections from Web/SharePoint are Online and active with Signed In of Admin Account with Authentication method: OAuth 2.0 / Privacy level: Organizational. - Inside the dataflows, we have unlinked all connection and re-link again with the Authentication: OAuth 2.0 , some dataflows can refresh successfully some tables inside the dataflows but with the last table which is ready to load to the data warehouse failed again with the error message: MashupException with “Microsoft SQL: Internal system error (0xa) (MWC service error: Server responded with error: 403)
- Re-entered all AOuth account, refresh successfull inside the dataflows, preview data table without errors... Published the dataflow again and still failed in refresh after that with the same error
Thanks again for your resoponse and guidances, but could you please recommend any other solutions which are able use PowerShells to investigate or detect all thing behind the scense and transfer the identity users / owner to the right one to solve the current problems?
Hope to hear from you soon!
Many thanks!Hi BanNguyen9708,
At this point I think more telemetry from your environment is needed to troubleshoot what is happening.
Hopefully the support engineers who pick up your ticket are able to shed some light on it.
Once they are able to help you solve it, please post the solution here so future forum members willl be able to find it.
Sure! I'll but this is really a critical issue because a lot departments in my company rely on the HR data to build the closing month report. But at this time we must do everything manually and waiting for final solutions! If we fix it completely, I'll share in here to other memebers!
Thanks
- We have already tried to process somethings:
