search

aws / amazon-sagemaker-feedback

Amazon SageMaker Public Feedback Dashboard
Creative Commons Attribution Share Alike 4.0 International
4 stars 1 forks source link

Custom UIs served by Python processes in JupyterLab or VSCode inaccessible using Sagemaker Studio #35

Open airlangga-gunawan-faculty opened 6 months ago

airlangga-gunawan-faculty commented 6 months ago

Product Version

Issue Description

The data scientists in my company have a need to create rough UIs, dashboards and data visualisations in Sagemaker. This was possible in Studio Classic, but now seems broken on Studio?

Expected Behavior

Using the JupyterLab or Code Editor space, execute a Python script that creates a webserver that serves a UI on localhost:. A good minimal example is the Dash Plotly minimal Dash app: https://dash.plotly.com/minimal-app I would expect to be able to access the UI that the app renders on a URL that uses https://<base path>/proxy/<some port>

Observed Behavior

https://<base path>/proxy/<some port> does not work. More to the point, I now have no idea how to access any Flask/FastAPI app that is ran within Sagemaker.

Product Category

JupyterLab, Code-Editor

Feedback Category

Error Handling and Debugging

Other Details

No response

rekha-seshadrinathan commented 5 months ago

Thank you for reporting this.