Cell hangs on execution until interrupt sent, then it executes

[TomNicholas]

Description

Cell execution hangs until an interrupt command is sent, at which point the cell actually begins execution.

Reproduce

I don't know how to reproduce this unfortunately, it seems to happen randomly on the HPC-hosted JupyterHub I am using. It does seem to happen more often after non-trivial amounts of data have been loaded into memory in that session...

Expected behavior

Obviously it should execute the command when I execute the cell, not when I interrupt it. I'm pretty sure that nothing is happening until I send the interrupt command, because I tried this with a cell that contained a command to save a series of files, and none of the files started appearing until the interrupt command was sent.

Context

• Operating System and version: the client (my laptop) is running MacOS. Jupyterlab is run on the National Center for Atmospheric Research's HPC system Casper (on a login node)
• Browser and version: Chrome Version 120.0.6099.199 (Official Build) (arm64)
• JupyterLab version: 4.0.8

Troubleshoot Output

Paste the output from running `jupyter troubleshoot` from the command line here.
You may want to sanitize the paths in the output.

Command Line Output

Paste the output from your command line running `jupyter lab` here, use `--debug` if possible.

Browser Output

Paste the output from your browser Javascript console here, if applicable.

[davidbrochart]

I'm wondering if it is a display message issue. Could you try and see if print(ds_to_reproject) has the same issue (that would make it a stream message)?

[TomNicholas]

print(ds_to_reproject) has the same issue, but so does literally just entering 1 in the cell.

Also see this, which I presume means this is not just about what is sent back to display in the notebook, but what is actually executed on the backend?

I tried this with a cell that contained a command to save a series of files, and none of the files started appearing until the interrupt command was sent.

[TomNicholas]

Also note that sometimes I have to press interrupt multiple times to get it to execute, but sometimes the second time I press interrupt it does actually interrupts the now-running cell, stopping the computation I am trying to start. The In[*] state gives me no way to distinguish these two cases until after the execution has been cancelled. Obviously this is incredibly frustrating for trying to actually get any work done 😩

[krassowski]

@TomNicholas can you install jupyterlab-kernel-spy and share a screencast (or series of screenshots) demonstrating the details of kernel messages before and after running, and after interrupting?

Possibly related to #14831