Propogate the system error to RH (#2296)

* fail the run for error propogation

* adding missing pkgs

assets/training/model_management/src/azureml/model/mgmt/utils/exceptions.py

@@ -10,6 +10,9 @@
 from azureml._common._error_definition.azureml_error import AzureMLError  # type: ignore
 from azureml._common._error_definition.system_error import ClientError  # type: ignore
 
+from azureml.core.run import Run  # type: ignore
+from azureml.automl.core._run import run_lifecycle_utilities
+
 
 class ModelImportErrorStrings:
     """Error strings."""
@@ -207,6 +210,11 @@ def wrapper(*args, **kwargs):
                 logger.error("Exception {} when calling {}".format(azureml_exception, func.__name__))
                 for handler in logger.handlers:
                     handler.flush()
+
+                # fail the run
+                run = Run.get_context()
+                run_lifecycle_utilities.fail_run(run, azureml_exception, is_aml_compute=True)
+
                 raise azureml_exception
             finally:
                 time.sleep(60)  # Let telemetry logger flush its logs before terminating.

assets/training/model_management/tests/dev_conda_env.yaml

@@ -11,6 +11,7 @@ dependencies:
   - azureml-mlflow~=1.51.0
   - azure-ai-ml~=1.8.0
   - azureml-core~=1.52.0
+  - azureml-automl-core~=1.52.0
   - azureml-telemetry~=1.52.0
   - azureml-evaluate-mlflow~=0.0.19
   - langcodes