Dash on AWS Sagemaker


Dash works fine on my local Jupyter Notebook/Lab, but now I need to make it work on a remote environment (AWS Sagemaker), and things are a bit more complex.

I have Jupyter Studio with Dash 2.17.0 installed. The computing instance that hosts Jupyter seems to reside outside the subnet I have created for it, so it seems challenging to open the extra port for Dash.

Has anyone managed to get this setup to work or could point me to some helpful information? Many thanks.

Hello @edsgarmoore,

Welcome to the community!

I dont have any experience hosting Jupyter, why do you want to run this in Sagemaker instead of a dedicated app and server?

@tphil10 might be able to help?

This is a common problem because sagemaker is a managed service so you don’t have access to the underlying instances OS or file system, only the sagemaker api. I think the ultimate solution is going to be some kind of proxy/port combination to make it work.

Did you try anything from the open ticket with plotly?

I’m not a network expert but what happens if you try and use the jupyter notebooks default port?

I unfortunately don’t have any examples of this because i don’t use sagemaker (yet). I know people run a variety of frameworks in there though like tensorflow or aim ui which do require different ports so there has to be a way to do this.