I am trying to add searchservice connection in my foundry project but getting below error Error creating connectionService temporarily unavailable. Please try again later.

Kaveri R Goudgaon 0 Reputation points
2026-09-06T06:57:00.8+00:00

Foundry's identity added managed identity. For keyvault added keyvault secrets officer and key vault reader to foundry's managed identity. I was able to add keyvault connection successfully but when i tried adding searchservcie connection getting : Error creating connection Service temporarily unavailable. Please try again later.

Foundry Tools
Foundry Tools

Formerly known as Azure AI Services or Azure Cognitive Services is a unified collection of prebuilt AI capabilities within the Microsoft Foundry platform


1 answer

Sort by: Most helpful
  1. AI answer

    2026-09-06T06:57:21.7933333+00:00

    For an Azure AI Search connection in Foundry, the most relevant checks are the connection method, authentication type, and Search-side permissions.

    1. Recreate the connection from the Foundry portal
      • In Microsoft Foundry, make sure New Foundry is on.
      • Go to ManageProject detailsConnected resourcesAdd connection.
      • Select Azure AI Search.
      • Select the correct Search service and choose the authentication method.
      • Confirm the connection appears in Connected resources.
    2. Prefer Microsoft Entra ID authentication
      • Foundry supports different authentication methods depending on connection type.
      • Using Microsoft Entra ID can require specific Azure role-based access permissions.
      • For keyless authentication, the Foundry account’s system-assigned managed identity must have the required roles on the Azure AI Search service.
    3. Verify the Azure AI Search endpoint and auth details
      • In the Azure portal, open the Azure AI Search service.
      • From Overview, confirm the endpoint format is https://<searchServiceName>.search.windows.net.
      • If using key-based authentication, go to SettingsKeys, enable Both if needed, and use an admin key.
      • Do not store real keys in source control. Use a secure store such as Azure Key Vault.
    4. If creating the connection by CLI, use the expected payload For Microsoft Entra ID:
         {
           "properties": {
             "category": "CognitiveSearch",
             "target": "https://{searchServiceName}.search.windows.net",
             "authType": "AAD"
           }
         }
      
      For API key:
         {
           "properties": {
             "category": "CognitiveSearch",
             "target": "https://{searchServiceName}.search.windows.net",
             "authType": "ApiKey",
             "credentials": {
               "key": "{searchAdminKey}"
             }
           }
         }
      
      Then run:
         az cognitiveservices account project connection create \
           --resource-group <resource-group> \
           --name <foundry-resource-name> \
           --project-name <project-name> \
           --connection-name <connection-name> \
           --file connection.json
      
    5. Important distinction from the Key Vault success
      • A successful Key Vault connection does not confirm Azure AI Search permissions are correct.
      • Azure AI Search requires its own endpoint and either:
        • key-based authentication, or
        • keyless authentication with Microsoft Entra ID and RBAC assigned to the project managed identity.
    6. Most likely next action in this scenario
      • Check that the Foundry project managed identity has the required Azure AI Search access for keyless authentication.
      • If the portal still returns Service temporarily unavailable, try creating the same connection with Azure CLI using the JSON payload above. That helps confirm whether the issue is portal-side or connection configuration.

    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.