building a custom jupyter notebook container that supports python 3.10

[anannra]

the off the shelf notebooks only support python 3.7. This doesn't work for us as many of the libraries we use require a minimum of python 3.8 or higher.

Is there a way to create a customer notebook where we can configure the correct environment for our needs?

[kvandres]

Good day @anannra,

Welcome to Google Cloud Community!

You can try using a custom image when creating the notebook, I think this would be the ideal solution in your situation. By doing this, you can install any necessary dependencies and the version of Python that you need. You can use these guides: https://cloud.google.com/vertex-ai/docs/workbench/managed/custom-container
https://cloud.google.com/vertex-ai/docs/workbench/user-managed/custom-container

You can also check these guides as they might be helpful to your case but please note that these are third party links and Google doesn't support them.
https://stackoverflow.com/questions/57322838/requirements-for-launching-google-cloud-ai-platform-not...
https://stackoverflow.com/questions/58697538/how-do-you-set-up-a-python-3-8-kernel-on-gcps-ai-platfo...

In case, you haven't built your container, you can use this link as a guide: https://cloud.google.com/build/docs/building/build-containers

Additionaly, a feature request was already submitted which is to upgrade the python version to 3.8 and above, you can track the request using this link: https://issuetracker.google.com/issues/235826369

Hope this helps!

[anannra]

Thanks. I tried a bunch of these with no luck.

I eventually found a solution here : https://medium.com/google-cloud/upgrade-google-vertex-ai-workbench-notebook-python-version-64ee1d8b4...