Forum Discussion
Power BI Refresh Fails with “Unable to Continue Fetch After Reconnect” from Azure Databricks.
- 1 month ago
Hello,
Based on the error and the one-hour refresh time, the most likely culprit is an authentication token expiration. If you're using a Service Principal (SPN) or OAuth for authentication, the token often has a 1-hour lifetime, which would cause the fetch to fail right around the time your refresh is failing. The fix for this is to switch your connection to use a Personal Access Token (PAT) instead. Many users have reported that this resolves the issue because the PAT doesn't have the same hard 1-hour limit.
Best regards,
Daniele
Hello,
Based on the error and the one-hour refresh time, the most likely culprit is an authentication token expiration. If you're using a Service Principal (SPN) or OAuth for authentication, the token often has a 1-hour lifetime, which would cause the fetch to fail right around the time your refresh is failing. The fix for this is to switch your connection to use a Personal Access Token (PAT) instead. Many users have reported that this resolves the issue because the PAT doesn't have the same hard 1-hour limit.
Best regards,
Daniele
- Rabin_011 month agoNew Member
Hi DanieleUgoCopp , Thank you for your suggestion. I have a question regarding the PAT configuration.
In Power BI, when I connect to Azure Databricks, I provide the Server Hostname and HTTP Path. The available authentication options are Azure Active Directory (default - which we prefer currently), Personal Access Token, and Databricks Client credentials. We also use an On-premises Data Gateway to refresh the dataset in the Power BI Service.
I'm not sure where exactly the PAT should be configured. Should I change the authentication method to Access Token in Power BI Desktop when connecting to the Databricks SQL Warehouse, or should it be configured in the Gateway data source in the Power BI Service?
I tested using a PAT in Power BI Desktop by selecting Personal Access Token during authentication, but I'm still getting the same error after publishing and refreshing the dataset.
Could you clarify the correct place to configure the PAT, or whether there are any additional gateway settings that also need to be updated?