[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[GitHub] Fokko commented on a change in pull request #3658: [AIRFLOW-2524] Add Amazon SageMaker Training

Fokko commented on a change in pull request #3658: [AIRFLOW-2524] Add Amazon SageMaker Training
URL: https://github.com/apache/incubator-airflow/pull/3658#discussion_r207548867

 File path: airflow/contrib/operators/sagemaker_create_training_job_operator.py
 @@ -0,0 +1,98 @@
+# -*- coding: utf-8 -*-
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements.  See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership.  The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License.  You may obtain a copy of the License at
+#   http://www.apache.org/licenses/LICENSE-2.0
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# KIND, either express or implied.  See the License for the
+# specific language governing permissions and limitations
+# under the License.
+from airflow.contrib.hooks.sagemaker_hook import SageMakerHook
+from airflow.models import BaseOperator
+from airflow.utils import apply_defaults
+from airflow.exceptions import AirflowException
+class SageMakerCreateTrainingJobOperator(BaseOperator):
+    """
+       Initiate a SageMaker training
+       This operator returns The ARN of the model created in Amazon SageMaker
+       :param training_job_config:
+       The configuration necessary to start a training job (templated)
+       :type training_job_config: dict
+       :param region_name: The AWS region_name
+       :type region_name: string
+       :param sagemaker_conn_id: The SageMaker connection ID to use.
+       :type aws_conn_id: string
 Review comment:
   @srrajeev-aws In this case you would just kick off multiple operators in parallel. This is inherent of the concept of a DAG, if the training jobs don't have any dependencies on each other, they will just run in parallel. The only flexibility that the decoupling of the kicking of the job, and monitoring the job is in the case when you don't care about the outcome of the job. This is also analoge to Druid, an indexing job can take up to a couple of hours.
   Having a separate operator and sensor would make the DAGs unnecessarily complicated, since in practice you will always use them as a pair.

This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:

With regards,
Apache Git Services