Sagemaker Pipeline strange warning message

Question

Using Sagemaker's Python SDK 2.11 when I run my pipeline, I see this strange warning message: 
```
/personal_dir/lib/python3.8/site-packages/sagemaker/workflow/pipeline_context.py:233: UserWarning: Running within a PipelineSession, there will be No Wait, No Logs, and No Job being started.
```
 Before, I ran the exact same pipeline script with LocalPipelineSession without any problems and without any kind of weird warning messages.

This is how I am creating the PipelineSession object:
```
def get_session(region, default_bucket):
    boto_session = boto3.Session(region_name=region)
    sagemaker_client = boto_session.client("sagemaker")

return PipelineSession(
        boto_session=boto_session,
        sagemaker_client=sagemaker_client,
        default_bucket=default_bucket
    )
```
Im getting the region in the following way:
```
import boto3

region = boto3.Session().region_name
```
I have tried to search the web for the meaning of that warning message, but could not find anything. What does that warning message means?? Am I doing something wrong and what can I do to make that warning disapear

Accepted Answer

Hi, this warning is simply to clarify that running in a pipeline session will defer execution of jobs - generating pipeline step definitions instead of kicking off the jobs straight away.

For example calls such as `Estimator.fit()` or `Processor.run()` when using a pipeline session won't **start a job** (or wait for it to complete, or stream logs from CloudWatch), just prepare a definition to build up a pipeline that can be started later.

If you're already familiar with how PipelineSession works, I would say you can ignore it :-)  If not, can refer to the [SDK docs here for more details](https://sagemaker.readthedocs.io/en/stable/amazon_sagemaker_model_building_pipeline.html#pipeline-session).

Could be that there's an inconsistency between `LocalPipelineSession` versus `PipelineSession` in showing the message? Or that you disagree this message should be at warning level... Either way I'd suggest raising an issue on the [SageMaker Python SDK GitHub](https://github.com/aws/sagemaker-python-sdk) might be a good way to log that feedback with the team!

Answer

https://repost.aws/questions/QUm9Ml6PX2QdOA5VIaDMblQg/sagemaker-pipeline-strange-warning-message#ANRkaRozOcQZaJO_YiZJBiew

Sagemaker Pipeline strange warning message

相關內容