Don't miss your chance to take the Fabric Data Engineer (DP-700) exam on us!
Learn moreWe've captured the moments from FabCon & SQLCon that everyone is talking about, and we are bringing them to the community, live and on-demand. Starts on April 14th. Register now
Hello everyone,
our arrangement is as follows:
- Cross-Tenant shortcuts are in a Lakehouse
- A report connects to these shortcuts via Direct Query
Now, when the report is freshly connected, everything is fine, but after a while it gets sort of stale, and when you attempt to cross filter, an error occurs that seems to simply be a connection fail:
If the report is published to the service, everything is fine. This only occurs from Power BI Desktop (which is still a problem, because we want to edit it there).
Does anyone have an idea?
Solved! Go to Solution.
This is a known pain point with cross-tenant shortcuts and DirectQuery from Desktop. What's happening is that Desktop authenticates to the remote tenant using an OAuth token, and that token expires (typically after ~1 hour). In the Service, token refresh is handled silently in the background — but Desktop doesn't do the same for cross-tenant scenarios, so the connection goes stale and the next query fails.
A few things to try:
Short-term workaround: When it goes stale, go to File → Options → Data source settings, find the Fabric/OneLake connection, and clear the credentials. Then reconnect — this forces a fresh token. It's annoying but confirms the root cause.
More sustainable approaches:
First, make sure you're on the latest Power BI Desktop — Microsoft has been improving cross-tenant token handling and there may be a fix already in a recent release.
Second, check if you can switch the Desktop connection to use the Power BI Service live connection (connect to the published semantic model rather than directly to the Lakehouse shortcut). This routes authentication through the Service layer where token refresh works properly, and you can still edit your report logic.
Third, if you truly need Desktop → Lakehouse DirectQuery, consider whether the cross-tenant shortcut can be replaced with a mirrored copy or a Fabric pipeline that lands data into your own tenant's Lakehouse. That removes the cross-tenant auth entirely from the Desktop path.
If none of those fit, it's worth filing a support ticket — this is a known gap and Microsoft tracks these cases to prioritize fixes
This is a known pain point with cross-tenant shortcuts and DirectQuery from Desktop. What's happening is that Desktop authenticates to the remote tenant using an OAuth token, and that token expires (typically after ~1 hour). In the Service, token refresh is handled silently in the background — but Desktop doesn't do the same for cross-tenant scenarios, so the connection goes stale and the next query fails.
A few things to try:
Short-term workaround: When it goes stale, go to File → Options → Data source settings, find the Fabric/OneLake connection, and clear the credentials. Then reconnect — this forces a fresh token. It's annoying but confirms the root cause.
More sustainable approaches:
First, make sure you're on the latest Power BI Desktop — Microsoft has been improving cross-tenant token handling and there may be a fix already in a recent release.
Second, check if you can switch the Desktop connection to use the Power BI Service live connection (connect to the published semantic model rather than directly to the Lakehouse shortcut). This routes authentication through the Service layer where token refresh works properly, and you can still edit your report logic.
Third, if you truly need Desktop → Lakehouse DirectQuery, consider whether the cross-tenant shortcut can be replaced with a mirrored copy or a Fabric pipeline that lands data into your own tenant's Lakehouse. That removes the cross-tenant auth entirely from the Desktop path.
If none of those fit, it's worth filing a support ticket — this is a known gap and Microsoft tracks these cases to prioritize fixes
If you have recently started exploring Fabric, we'd love to hear how it's going. Your feedback can help with product improvements.
A new Power BI DataViz World Championship is coming this June! Don't miss out on submitting your entry.
Share feedback directly with Fabric product managers, participate in targeted research studies and influence the Fabric roadmap.
| User | Count |
|---|---|
| 51 | |
| 37 | |
| 35 | |
| 19 | |
| 17 |
| User | Count |
|---|---|
| 72 | |
| 70 | |
| 39 | |
| 34 | |
| 23 |