New Virtual environments not listed in Kernels until i reload VS Code

[DonJayamanne]

• Run notebook, & view kernels
• Create virtual environment
• Close notebook, open a notebook, the new venv is not dipslayed in the kernel picker
• Need to reload

[rchiodo]

I think we need a refresh button like python has:

Us:

Python:

[kylepierce]

This bug has made vscode notebooks unusable for me. I cannot select the appropriate virtualenv.

[DonJayamanne]

Here's the feature request for Python:

• Expose API for when the list of interpreters have changed vscode-python#16676
• Expose API for when the list of interpreters have changed vscode-python#16683

[DonJayamanne]

Depends on #7583
Suggestion

• Refactor notebook controllers to use the new Python extension api (loading interpreters as and when they are discovered)
• Big change, hence pointless trying to do that in this debt week

[rchiodo]

This should be fixed.

[cpoptic]

As of June 2022, this issue is not fixed. Please re-open this issue and keep it open until a dependable, reliable fix has been implemented.

To users encountering this issue, one high-probability method I've found solves this persistent issue is to:

• uninstall Jupyter Extension
• uninstall Python Extension
• Reload VS Code
• Re-install Jupyter Extension
• Re-install Python Extension
• Open a Jupyter notebook
• Select the desired kernel from the upper right corner of your notebook.

Uninstalling extension, reloading VS Code, and re-installing those extensions is currently the only dependable way of solving the mystery of missing kernels/environments in VS Code Jupyter notebooks, which is a less than ideal solution.

[rchiodo]

@cpoptic can you open a new issue? Creating a venv and refreshing the list of interpreters in the python extension works for me.

Perhaps we need a 'force refresh' button for kernels too. But kernels update when we detect from the python extension that a new environment was created.

Once I clicked refresh on the 'Select Interpreter' dialog, my kernel showed up:

[guidorice]

@rchiodo the problem seems to be with the "Select kernel" dialog which pops up when you run a notebook in VSCode.

• The list of venvs is incomplete, or it apparently doesn't see the venv which the current Python interpreter is using in VSCode.
• The list of venvs cannot be refreshed.
• It doesn't have an affordance to enter a path.

edit:

There seems to be some condition which triggers the current venv to appear in the picklist for 'select kernel'. Cannot quite figure out what that is. Have tried installing ipykernel, have tried running jupyter notebook , etc.

[rchiodo]

@guidorice I added a new issue for you (referencing your comment).

[Mascobot]

Same issue here as @guidorice. The environments list seems to be incomplete (both from conda and virtualenv), and something seems to trigger some old envs that suddenly appeared, but the list is still incomplete. The normal interpreter works fine, but the issue only happens when you open Jupyter Notebook and try to "Select Kernel". No refreshing is available. So far I tried re-installing ipykernel, re-installing the JN extension, and saving the notebook inside the virtualenv environment folder. No success.

[haimengzhao]

Same issue here as @guidorice and @Mascobot. I'm currently using this work around:

1. Manually execute jupyter notebook in a terminal with correct python environment
2. Open the Command Palette and select Jupyter: Specify Jupyter Server for Connections
3. Copy the URL http://localhost:8888/?token=blahblah from the terminal to it
4. Now it will appear in the Kernel list as a (Remote) Jupyter Kernel.

[Mascobot]

Sweet. Nice work around. Just did it and it works.

[fabioxxt]

who still has this problem, I found that closing the ipynb file tab and opening it again solves it, if not, close the tab again and with the second mouse button, click on the ipynb file and open it with the terminal

[pablomassirer]

The only two solutions for me that seems to work until now is closing and reopening the file as mentioned above, or specifying the interpreter path in the "Python: Select Interpreter" from the Command Palette.

[DonJayamanne]

Sorry everyone about this issue.
However we've fixed a big gap in the environment discovery by using the same API that Python extension uses to discovery Python environments.
You can try that out in the latest Jupyter relase (some time tomorrow or this week)
Optionally you can try that now with the latest vscode insiders and pre-release version of the Jupyter extension.

If this still does not work, please do ping here and we'll happily re-open this issue and look into it, or please feel free to create a new issue,.

[dmnburrows]

Hi, I'm still getting this issue - I have upgraded jupyter, and when I SSH into a remote server and launch a notebook I am unable to select a kernel. When I load the notebook it never detects a kernel, and when I press on the kernel button on the top right nothing happens. If i search for my kernels in the search bar at the top nothing appears. I don't have this issue when I am loading notebooks on my local machine. Any help would be much appreciated!

[DonJayamanne]

@dmnburrows Please could you file a new issue for this, thanks
When creating the new issue please provide the logs below
Please could you enable logging as follows:

• Open settings & go into Jupyter->Logging
• Change the value to verbose
• Reload VS Code,
• Attempt to repro the issue & then copy the output from the Jupyter output panel.