Can't change remote jupyter notebook kernel

[mrober01]

Hi,
-pointed here from stackoverflow-
With the new version i am able to connect to a remote jupyter session and that works great, but there does not appear to be a way to change the remote kernel. This feature would be very helpful.
Thanks,
Mike

[u8sand]

This is problematic--vscode is defaulting to a kernel that isn't even python for me.

Edit: My current workaround is starting the notebook myself with

jupyter notebook --KernelSpecManager.whitelist="['python3']"

causing python3 to be the only option.

With specification of the kernel itself, this might enable interactive support to different languages as well given that there is an available kernel.

[jayBana]

Is this something similar to what the Hydrogen package provides in Atom?
e.g. after configuring the remote Jupyter Server it gives the following options when selecting the "Connecting to Remote Kernel":

It would be great to have the ability spacify the kernel e.g. in conda environments.

[williamjqk]

If you support this feature, I would give up Atom and use vscode!

[vnijs]

Hydrogen also has an option to "remember" multiple jupyter servers the user might want to connect to. The first choice is then "which remote" and then what kernel. Very handy

[CermakM]

+1 For this feature

[nlspoon]

Bumping for this feature. Having the ability to reconnect to and/or select which remote jupyter server and which kernal to use would be very useful to me and my team.

[elyase]

This would be great to have. BTW is this possible locally?

[sfhansen]

Another +1 for this feature!

[kcyu2014]

+1 for this feature

[langley]

+1 for this feature Running / connecting to kernels in other languages would be VERY useful!

[erinxocon]

+1 for this feature! I am loading a custom kernel installed by django-extensions to have the ORM initialized inside of the notebook. This would be grand!

[QuinRiva]

+1 for this feature. I run multiple kernels with different packages installed.

[DonJayamanne]

Thanks everyone for adding your interest. However please could you vote on the main comment (👍 ). This way its easier for us to check the reactions (interest) on this issue, else we'd have to scroll down to look for +1 or the like.

[rchiodo]

Remote doesn't seem to show up in our new (about to release) UI. Investigating

[DonJayamanne]

Validated

[mparq]

Hi @DonJayamanne / @rchiodo, I see this issue is closed. We're trying to use a local docker container running jupyter in vs code and have the same issue regarding wanting to select different kernels on the remote server. Should this be possible now or is it possible in the new UI?

[rchiodo]

It should be possible. Remote kernel selection should work. For VS code stable anyway. VS code insiders is using something that doesn't have remote support just yet.

Please open a new issue if you're using VS code stable and you can't pick a remote kernel.

This is what it shows for me:

Those first 3 kernels being already running kernels.

[mparq]

Ahh, you're right! I see what the issue was. I'm on VS code stable. I was using the functionality for running a cell directly from a .py file which was able to use the Jupyter Server I had configured from the jupyter extension.

From the .py file and the Interactive window, I can't see Jupyter: Select a Kernel in the command pallette, but if I export the .py file as an .ipynb, I do get that command and I can select remote kernels properly.

But this made me check back and it looks like I still have the kernel selection available through the UI in the python interactive window and I can select a kernel properly from there.

Is it a bug that Jupyter: Select a Kernel isn't available in the interactive window? I can create a new issue if so.

[rchiodo]

Is it a bug that Jupyter: Select a Kernel isn't available in the interactive window? I can create a new issue if so.

Yes seems like an over sight. Please enter an issue.