Empty output job folder

Question

Hello!

I have setup local JupyterLab instance with SageMaker extension in order to run jobs in the AWS and now I see that jobs are started with proper .ipynb file in the input S3 folder but do not see any result because no files with "output" prefix exist and nothing to [download](/media/postImages/original/IMegGzLkp1QWqX-z5EQ-GdKQ).

- Everything is configured following https://aws.amazon.com/blogs/machine-learning/schedule-your-notebooks-from-any-jupyterlab-environment-using-the-amazon-sagemaker-jupyterlab-extension/
- Roles, permission are double checked

Could you give me a direction? because I have no idea what I should check more or where I can find relevant logs to have a glance under the hood.

Best,
Dani

Answer

Same question on the StackOverflow:
https://stackoverflow.com/questions/78333128/empty-output-job-folder

I provided an answer there.

Answer

I just followed the blog post and I was able to run a notebook job successfully.

If you successfully run the job, you will see following files under the S3 bucket you specified.

```
$ aws s3 ls s3://sagemaker-automated-execution-123456789012-us-east-2 --recursive
2024-04-17 17:57:49       9480 notebookjobtestip-notebookjobtest-bf50211d-2024-04-18-00-57-47/input/notebook-job-test.ipynb
2024-04-17 18:00:14      13948 notebookjobtestip-notebookjobtest-bf50211d-2024-04-18-00-57-47/output/output.tar.gz
```

I am sharing my input configurations:

- Compute type : ml.m5.large
- Image : arn:aws:sagemaker:us-east-2:429704687514:image/sagemaker-base-python-38
- Kernel : python3
- Role ARN : arn:aws:iam::123456789012:role/SagemakerJupyterSchedulerRole
- Input Folder : s3://sagemaker-automated-execution-123456789012-us-east-2/
- Output Folder : s3://sagemaker-automated-execution-123456789012-us-east-2/

Please try similar configurations and check if it works.

Empty output job folder

Contenuto pertinente