VSCode notebooks connected to ARE sessions- interrupting a cell shuts down the kernel

Hi everyone!

I mainly use VSCode to run Jupyter notebooks connected via SSH to an ARE session. One difference I’ve noticed compared to using ARE in the browser is that when I interrupt a cell execution in VSCode, the entire kernel shuts down, rather than just stopping the cell. Whereas, the browser version of ARE will just stop that cell - which is far more desirable.

I was wondering whether anyone else has noticed this behaviour, or if there is a setting or workaround in VSCode to prevent the kernel from shutting down on interrupt? I’m working with very large datasets and experimenting with efficiency (e.g. using Dask and preprocessing), so being able to interrupt a long-running cell without restarting the entire notebook would be really helpful.

Thanks,
Alanah

Unfortunately I don’t have a good answer for you, but a tangential piece of information you may want to be aware of: sometimes dask takes less time to do the exact same thing a second time, so while optimising you may want to restart the kernel anyway.

As another workaround, it might also help to play around with optimising a small subset of your data so even if it’s really inefficient it doesn’t take too much of your time?

Hope someone else here can give a more complete suggestion/solution
@paocorrales @sam.green ?