Mirrored Azure Databricks catalog not available

[SørenBrandt]

Hi all,

Previously I've had a mirrored Azure Databricks catalog setup and working in my Fabric workspace. But, right now I am struggling to make it work, and am wondering if there's a known issue with this Fabric to Databricks integration, or whether I've gone mad...

My question: How can I troubleshoot Unity Catalog connectivity issues?

I feel quite convinced that I have configured everything as required, but when I attempt to create a Mirrored Azure Databricks Catalog, I get the error message "Catalogs aren't available", with the following in red: "Unable to process response from Databricks. The API returned data in an unexpected format". Is there a way to get more detailed error information that may help me troubleshoot?

BR,

Søren

[v-sdhruv]

Hi @SørenBrandt ,

400 error typically indicates that the API call from Fabric to Databricks failed due to:

Incorrect or missing authentication headers /misconfigured cluster or workspace settings/unsupported request payloads.

You can perform these checks-

1.Check User Permissions-Ensure your account has access to Unity Catalog and validate that the account is active.

2.Review Logs- Look for specific error codes or patterns in Fabric diagnostic logs.

3.Confirm that no firewall or security group is blocking access between Fabric and Databricks.

4. Double-check the Unity Catalog setup and any recent changes.

Refer- Microsoft Fabric Mirrored Databases From Azure Databricks Security - Microsoft Fabric | Microsoft Le...
Also, check-out the limitations here
Limitations in Microsoft Fabric Mirrored Databases From Azure Databricks - Microsoft Fabric | Micros...

Even if get the same error consider raising a Support ticket with detailed logs and session IDs.
Microsoft Fabric service status

Hope this helps!

[SørenBrandt]

Hi v-sdhruv,

Thank you for your suggestions!

Unfortunately, I have been unable to identify any issues in my configuration so far.

However, it turns out that my issue connecting from Fabric to Databricks applies to more than just mirrored catalogs: As a short-term workaround to my problem with setting up a mirrored catalog, I tried to copy data from Databricks to Fabric using a pipeline. This also does not work, and the information available to me is somewhat puzzling because it suggest that the request from Fabric is received and processed by Databricks, but the response is not received or understood by Fabric:

1. On the Databricks side, the log suggests that the Warehouse received the catalog query from Fabric, and returned a set of results.

2. On the Fabric side, the error message indicates that the SQL Warehouse in Databricks does not exist. However, the SQL Warehouse that Fabric claims to not exist is the very one that receives and processes the request from Fabric.

Here's a couple of screenshots to document my observations.

Fabric says:

Databricks says:

Any help would be greatly appreciated!

BR,

Søren

[SørenBrandt]

Following up on my initial question: When I try to create a connection to the Azure Databricks workspace using "Manage Connections and Gateways" in my Fabric workspace, I get the error information below, in case that provides any clues:

Unable to create connection for the following reason: Unable to connect to the data source. Either the data source is inaccessible, a connection timeout occurred, or the data source credentials are invalid. Please verify the data source configuration and contact a data source administrator to troubleshoot this issue.

Details: Web.Contents failed to get contents from 'https://adb-XXXXXXXXX.azuredatabricks.net/api/2.0/clusters/list' (400): Bad Request Please have this information handy if you choose to create a support ticket. Session Id: xxx-yyy-zzz RequestId: xxx-yyy-zzz Cluster URI: https://api.powerbi.com Status code: 400 Time: Thu Aug 14 2025 15:21:55 GMT+0200 (Central European Summer Time)

[SørenBrandt · Accepted solution]

Hi All,

Thankyou for your suggestions!

With help from Microsoft support, I have now identified the root cause.

The issue is a limitation in using guest accounts with Mirrored Azure Databricks Catalogs.

To summarize, I have two Entra Id tenants A and B in my setup, applied as follows:

- Entra Id tenant A hosts my Fabric and Databricks workspaces.

- Entra Id tenant B contains the organisational account I was using to create the Mirrored Azure Databricks Catalog. That is, the account is a guest in Entra Id tenant A.

It turns out that this configuration is not supported by the current version of Mirrored Azure Databricks Catalog.

The fix is to use either a service principal or an organisational member account from tenant A to setup the connection.

Personally, I would appreciate a more general support for multi-tenant setups, as multi-tenancy is a fact of life in my organisation. Therefore, I have often run into limitations of this kind. One day I might even learn to check for this as a matter of course :). 

Best Regards,

Søren

[v-sdhruv]

Hi @SørenBrandt ,

Good to know that you have found the root cause and were able to reach a conclusion that this configuration is not supported by the current version of Mirrored Azure Databricks Catalog.
Thanks for sharing this information with us. This will definitely help other users struggling with the same, figure out the issue.
Can you please mark your post as "Accepted solution" so that other users can also benefit from it?

Thank You.