removed silent failure for sdk

Author: Mohamed Zeidan

src/sagemaker/hyperpod/cli/cluster_stack_utils.py

@@ -347,9 +347,10 @@ def _handle_stack_deletion_error(error: Exception, stack_name: str, region: str,
         return False  # Should re-raise
 
     # Handle CloudFormation retain-resources limitation
+    # Always re-raise for SDK usage to ensure clear exceptions
     if retain_resources and "specify which resources to retain only when the stack is in the DELETE_FAILED state" in error_str:
         _handle_retention_limitation_error(stack_name, retain_resources, region, message_callback)
-        return True  # Handle gracefully, don't re-raise
+        return False  # ensure SDK gets the exception
 
     # Handle other deletion errors
     _handle_generic_deletion_error(error_str, message_callback)

src/sagemaker/hyperpod/cluster_management/hp_cluster_stack.py

@@ -624,11 +624,12 @@ def sdk_confirm_callback(message: str) -> bool:
         except Exception as e:
             error_str = str(e)
 
-            # Handle CloudFormation retain-resources limitation gracefully for SDK
+            # Handle CloudFormation retain-resources limitation with clear exception for SDK
             if retain_resources and "specify which resources to retain only when the stack is in the DELETE_FAILED state" in error_str:
                 error_msg = (
                     f"CloudFormation limitation: retain_resources can only be used on stacks in DELETE_FAILED state. "
-                    f"Try deleting without retain_resources first, then retry with retain_resources if deletion fails."
+                    f"Current stack state allows normal deletion. Try deleting without retain_resources first, "
+                    f"then retry with retain_resources if deletion fails."
                 )
                 logger.error(error_msg)
                 raise ValueError(error_msg)