HTTP 405 when accessing Contents API from DataSpell

JupyterHub Zero to JupyterHub on Kubernetes
1

We are testing connecting to Jupyterhub on Kubernetes using DataSpell, but it looks like the API URLs aren’t quite as expected.

For example, the “configured server” in DataSpell is:
https://jupyterhub.myhost.com/?token=mysecretusertoken

Then when i create a file in DataSpell a 405 appears in the DataSpell logs:

2022-01-31 15:18:27,978 [1949333] ERROR - intellij.openapi.progress.Task - Error with connection to Jupyter: server responded with 405 on Request(method=PUT, uri=https://jupyterhub.myhost.com/api/contents/test.py-upload-24SzJvvFWJ50kWwMu5b0XZPDUyH, headers=[(Content-Type, application/json; charset=UTF-8)], body=java.io.ByteArrayInputStream@53c27d49, context=Context(mayInteractUi=true)). [200, 201] was expected.

Is it possible to connect to a Jupyterhub pod using an external IDE or am I barking up the wrong tree?

2

I’m not familiar with DataSpell, but it’s possible to access the jupyter-server API on JupyterHub. For example:

$ curl localhost:8000/user/demo/api/contents -H 'Authorization: token <token>'

{"name": "", "path": "", "last_modified": "2022-02-06T17:30:11.565506Z", "created": "2022-02-06T17:30:11.565506Z", "content": [], "format": "json", "mimetype": null, "size": null, "writable": true, "type": "directory"}

This assumes your jupyter-server is already running.

3

Thanks manics. I was hoping the API was just automatically compatible with whatever Dataspell expects in order to be able to share the notebook with the IDE. haha.

Do you have experience / recommendations for using any other external tools to sync with Jupyterhub on Kubernetes? Some of our users prefer using an external IDE instead of notebooks-in-a-browser.

4

I don’t, I usually use JupyterLab for editing notebooks :slight_smile: