Forum Discussion

Sunflower7500's avatar
Sunflower7500
Helper I
1 year ago
Solved

Error with automatic refreshes for Databricks source: Authentication/authorization error occurred

I have a Power BI semantic model that pulls from Databricks. The automatic refreshes were running perfectly up until yesterday when I started to get this error message. I am able to refresh this repo...
Need Help
  • v-venuppu's avatar
    v-venuppu
    1 year ago

    Hi Sunflower7500,

    Thank you for reaching out to Microsoft Fabric Community.

    Thank you Deku  for the prompt response.

    It’s likely related to how Fabric Capacity handles authentication differently than Pro. PATs sometimes don’t play well in Fabric refreshes due to security model differences. Try switching the Databricks connection to use OAuth2 in Power BI Service instead of a PAT. Also check your dataset’s credentials and gateway settings in the Fabric workspace.

    If those don’t help, then go for opening a support ticket as suggested by Deku.

    I recommend submitting your detailed ticket below:

    How to create a Fabric and Power BI Support ticket - Power BI | Microsoft Learn

     

    If this post helps, then please consider Accepting as solution to help the other members find it more quickly, don't forget to give a "Kudos" – I’d truly appreciate it! 

    Thank you.

     

     

rohit1991's avatar
rohit1991
Super User
1 year ago

Hi Sunflower7500 ,

 

Refreshes from Databricks working fine in Power BI Desktop, but failing with an authentication/authorization error in the Power BI Service. This is becoming more common lately, especially with the newer Fabric capacities and stricter security around tokens. Here’s what usually fixes it:

  1. Switch from PAT to OAuth2 Authentication: Power BI Service (especially with Fabric) no longer plays well with Personal Access Tokens (PAT) for scheduled refreshes. Go into your dataset settings in the Power BI Service, and re-authenticate your Databricks connection using OAuth2 instead of PAT. This single step resolves the vast majority of these errors now.

  2. Re-enter Credentials & Check Gateway: After switching to OAuth2, re-enter your credentials and double-check your gateway settings if you’re using one. Sometimes credentials stored in the Service get stale or don’t update until you re-authenticate.

  3. Double-Check Databricks Workspace Permissions: Make sure your user/service principal still has access to the Databricks workspace and cluster. Sometimes permissions change on the Databricks side and silently break refreshes.

  4. Firewall/VNet Issues: If your Databricks workspace is behind a VNet or firewall, confirm that the necessary Microsoft Power BI Service IP ranges are allowed. Otherwise, the refresh can fail even if everything else is correct.

  5. Token Expiry/Conditional Access: If your organization enforces strict Azure AD token lifetimes or conditional access, the refresh may work intermittently or break when a token expires. Your IT or Azure admin can help check for this.