715-123420: An error occurred. Please reach out to support for additional assistance.

Error code 715-123420 is a generic deployment failure message in Foundry that does not map to a specific documented root cause in the provided guidance. The next step is to use the standard deployment troubleshooting flow to uncover the underlying error.

Use these steps to narrow down and resolve the issue:

1. Check deployment logs in Foundry
   1. Open the project in Foundry and go to Models + endpoints.
   2. Select the failing deployment.
   3. Review the deployment logs and status messages for a more specific error (for example, quota, image build, or network-related errors).
2. Check for quota or capacity issues
   1. In the Azure portal, open the subscription used by the Foundry project.
   2. Check quota for the VM SKU or model being used.
   3. If usage is at or near the limit, request a quota increase or choose a different VM SKU with available quota.
   4. For managed compute deployments, follow the “Deployment fails with quota exceeded error” steps: verify quota, request an increase, or switch SKUs.
3. Verify model availability and region
   1. Confirm that the model ID used in the deployment is correct and still available in the catalog.
   2. If the deployment has been stuck or failing for a long time with obscure errors, try deploying the same model in a different region, as region-specific capacity or availability issues can cause failures without clear messaging.
4. Check environment image build (if the error mentions image or environment) If the detailed error mentions Environment Image, ImageBuildFailure, or similar:
   1. Use the guidance under Error: Deployment failure to locate the environment image build logs.
   2. Option 1: From the Foundry project, navigate to the connected storage account and open Data Storage → Containers, then drill into the container and folders indicated in the error message to find image_build_aggregate_log.txt.
   3. Option 2: In Azure Machine Learning studio, go to Endpoints, select the endpoint, then open the linked Environment and review its build logs.
   4. Use the log details to fix missing packages, invalid Docker configuration, or other environment issues, then redeploy.
5. Validate networking and permissions
   1. For managed compute deployments, verify that the hub’s managed network settings allow access to required resources (storage, registries, etc.).
   2. If tools or prompt flows are involved and you see AuthorizationFailed or ToolLoadError in the detailed logs, assign the Azure ML Data Scientist role to the endpoint’s managed identity as described in the deployment troubleshooting steps, then wait a few minutes and retry.
6. If the deployment remains in provisioning or fails repeatedly
   1. Confirm the deployment is not stuck in a long-running provisioning state; if it is, cancel/delete the deployment and recreate it.
   2. Optionally reduce instance count or change VM SKU and redeploy.

If, after checking the detailed deployment logs, quota, region, environment build, and networking/permissions, the only information available is still the generic 715-123420 message, the issue likely requires direct platform investigation. In that case, capture:

• The deployment name and ID
• Subscription ID and region
• Timestamp and correlation IDs from the deployment logs

and provide them to Microsoft support as the error message suggests.

References:

• How to troubleshoot your deployments and monitors in Microsoft Foundry portal (classic)
• How to deploy and infer with a managed compute deployment (classic) (ai-foundry-portal)
• How to deploy and infer with a managed compute deployment (classic) (python-sdk)