How to stop VS Code interactive python (Jupyter) extension from starting the ipykernel on random ports in remote docker?

Question

I am developing remotely through the Remote-SSH + Remote-Containers plugins in VS Code, in other words I am ssh-ing into a remote server that runs a container with my configuration. Sometimes I need to do some quick visualisation or testing, so I like to use jupyter notebooks.

When I created the container I exposed only the 8888 port. If I start a notebook from command line, everything works fine, the jupyter server gets automatically launched at my container's 127.0.0.1:8888 and I can open it in the browser on my local machine. However, when I want to open the Jupyter interactive session using the Jupyter VS Code extension I can't do it, the interactive window opens and hangs saying it's

Waiting for the Jupyter session to be idle

I can see however, that whenever I try to open the interactive Jupyter window, it doesn't automatically start at the 8888 port, but on random different ports. I assumed that is the problem. Is my diagnosis correct? The question then would be: how to force the Jupyter VS code extension to use 8888 port for the interactive window?

Print screen of the problem