The Databricks Genie MCP connection has stopped working by itself

Sourav Gouri 0 Reputation points
2026-05-26T05:04:40.1466667+00:00

Hi Team,

We've been using the Databricks Genie MCP for the Genie Tool in our Copilot AI Agent and it has been working fine but it stopped suddenly since last night IST
We haven't changed any authentication or any related configuration at all.
Now it is showing error in the Tools section as visible in the screenshot below:
User's image

Is there any update going on?

Microsoft Copilot | Microsoft 365 Copilot | Development
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest

  1. AI answer

    2026-05-26T05:04:57.9733333+00:00

    There is no specific outage or product update described in the context for Databricks Genie MCP or its connector to Microsoft Copilot.

    From the available information, only general MCP and connection troubleshooting guidance can be applied:

    1. Validate authentication
      • If the Genie MCP connection uses OAuth (user-to-machine or service principal), re‑validate credentials:
        • For service principals, run databricks auth describe with the client ID/secret to confirm the identity is still valid and has required permissions.
        • If OAuth tokens were revoked or expired, re‑authenticate and update the connection in AI Gateway / MCP configuration.
    2. Verify network configuration
      • Confirm there were no recent changes to IP access lists, firewalls, or private networking that could block the Copilot agent from reaching the Databricks workspace or MCP endpoint.
    3. Check MCP server availability and tools
      • In the Databricks workspace, go to AI Gateway → MCPs and verify the Genie-related MCP server is listed and healthy.
      • Ensure the workspace region still supports Model Serving, because external MCP connections (including Genie and other tools) are only available where Model Serving is supported.
    4. Test with another MCP client
      • Use an MCP test client such as MCP Inspector to connect to the same Databricks MCP server. If it works there but fails in Copilot, the issue is likely on the Copilot side; otherwise, it is likely on the Databricks side.
    5. Collect diagnostics and open a support ticket
      • Gather:
        • Authentication method (OAuth vs PAT / service principal)
        • MCP server URL
        • Exact error message shown in Copilot (including any correlation IDs)
        • Network configuration details (IP restrictions, firewall rules)
      • Provide these details to Databricks or Microsoft support so they can check for any backend regression or regional incident.

    Because the context does not mention a known service incident matching this symptom, the next step is to follow the above troubleshooting steps and then contact support with the collected diagnostics.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.