Mirrored Azure Databricks Catalog - Cannot fetch data after adding tables

[SørenBrandt]

Hi All,

I'm struggling with the following:

1. I have a working Mirrored Azure Databricks Catalog.

2. I add tables to the databricks schema.

3. I select "Manage Catalog" and add the new tables.

4. The Mirrored Azure Databricks Catalog is no longer able to fetch any data, including tables that worked until step 3. After a while, I get the error message "Couldn't access data. Check your permissions for this catalog or try again later."

5. After a few hours, things suddenly start working again. Or, that's at least what happened yesterday. Today, I am again in the situation where the mirror is no longer working and waiting for magic to kick in.

If I create a new mirror item in the same workspace and using the same connection details, it works. Also during the period where the original mirror is not.

Any ideas on what may be going on, and if there's anything I can do myself? It's not practical to create a new mirrored catalog each time tables are added.

BR,
Søren

[v-agajavelly]

Hi @SørenBrandt,

Just checking in did any of those steps help resolve the issue with your mirrored catalog, or are you still running into the same behavior?

Regards,
Akhil.

[v-agajavelly]

Hi @SørenBrandt ,

Thanks @anilgavhane r for the detailed explanation.

@SørenBrandt  did you get a chance to try the suggested steps like re-authenticating the connection, verifying Unity Catalog permissions, or adding tables individually instead of re-selecting the whole schema? Curious to know if that helped stabilize your mirrored catalog.

Regards,
Akhil.

[anilgavhane · Accepted solution]

This behavior typically stems from a temporary desynchronization between Microsoft Fabric and the Unity Catalog in Azure Databricks. When new tables are added and selected via "Manage Catalog," the mirrored item may enter a state where metadata or permissions are not fully refreshed, resulting in the error:

"Couldn't access data. Check your permissions for this catalog or try again later."

Root Causes

• Delayed metadata propagation from Unity Catalog to Fabric
• OAuth token refresh failure or expired credentials
• Schema-level permission inconsistencies for newly added tables
• Fabric caching or sync lag after modifying catalog selections

Recommended Actions

1. Re-authenticate the Databricks connection
   Go to Manage Connections and Gateways in Fabric and re-enter credentials. This often resolves access issues immediately.
2. Verify Unity Catalog permissions
   Ensure the user or service principal has:
• USE CATALOG
• USE SCHEMA
• SELECT on all relevant tables
• EXTERNAL USE SCHEMA (required for mirroring)
3. Avoid toggling entire schemas
   When adding new tables, select them individually rather than re-selecting the whole schema. This minimizes disruption to existing metadata sync.
4. Disable auto-sync if unstable
   If automatic syncing of future catalog changes causes issues, consider disabling it and manually managing table selections.
5. Create a new mirror only as a last resort
   While creating a new mirrored item may temporarily resolve the issue, it’s not sustainable. Focus on stabilizing the existing mirror through connection and permission checks.