Identify notebook being run by kernel

JupyterHub Zero to JupyterHub on Kubernetes
1

If I run a notebook on jupyterlab I can switch to the terminal and run ps to see the kernel running. I’ll get something like:

    196 ?        Ssl    0:00 /srv/paws/bin/python3 -m ipykernel_launcher -f /home/paws/.local/share/jupyter/runtime/kernel-b5238fdc-3b0b-4622-968d-daa97d7f15b4.json

Though what file is actually being run? In this case it is a file called python-loop.ipynb, though I don’t see that when I run ps nor do I see reference to it in /home/paws/.local/share/jupyter/runtime/kernel-b5238fdc-3b0b-4622-968d-daa97d7f15b4.json

Is there some way to see what files were used with running processes?

Thank you!

2

You can find out what file is opened in the .json file :slight_smile:

3

Hi @spookster thanks for the reply! I think I would need some help decoding the json file as I don’t see the expected “python-loop.ipynb” file mentioned in it.

@PAWS:~$ ps ax
    PID TTY      STAT   TIME COMMAND
      1 ?        Ssl    0:06 /srv/paws/bin/python3 /srv/paws/bin/jupyterhub-singleuser --LabApp.collaborative=true
     74 ?        Ssl    0:00 /srv/paws/bin/python3 -m ipykernel_launcher -f /home/paws/.local/share/jupyter/runtime/kernel-69906426-ccd4-47bd-9888-e20efa40758a.json
     91 pts/0    Ss     0:00 /bin/bash -l
     96 pts/0    R+     0:00 ps ax
@PAWS:~$ cat /home/paws/.local/share/jupyter/runtime/kernel-69906426-ccd4-47bd-9888-e20efa40758a.json
{
  "shell_port": 44609,
  "iopub_port": 47119,
  "stdin_port": 49983,
  "control_port": 42175,
  "hb_port": 43483,
  "ip": "127.0.0.1",
  "key": "427a81ca-d9d079e7dea72752aeed7939",
  "transport": "tcp",
  "signature_scheme": "hmac-sha256",
  "kernel_name": "python3",
  "jupyter_session": "beaf2b2a-ce4c-45cf-baf2-e2b13f03514f"
}@PAWS:~$

Thank you!

4

huh?

that’s weird, my jupyter_session entry is the path of the open file…

what version of jupyterlab are you running?

5

We’re on jupyterlab==3.4.8 (been having some slowness issues in going to 4. Though could try if you think the listing would be there on something newer)

6

What happens when you run this in the working notebook?

import os

print(os.getenv('JPY_SESSION_NAME'))
7

It prints dfe3cf8d-02e8-44e0-a8b7-39379f905a9e
Which matches in the file from ps:

}@PAWS:~$ cat /home/paws/.local/share/jupyter/runtime/kernel-d5aff487-36c7-4631-b776-79e3f0692f54.json | grep dfe3cf8d-02e8-44e0-a8b7-39379f905a9e
  "jupyter_session": "dfe3cf8d-02e8-44e0-a8b7-39379f905a9e"
8

I cannot try this on my side but here is what I’d do:
Run this on a terminal in home directory (update the id if necessary :slight_smile: ):

grep -rnwl . -e "dfe3cf8d-02e8-44e0-a8b7-39379f905a9e"

Maybe you will find another file which then also contains the path to the open notebook?

You can also find the paths to all open notebooks in your workspace file @ ~/.jupyter/lab/workspaces/
But in my case there is are no session strings which connect the PIDs with the paths of open notebooks.

9

Sure enough there it is in the file.

 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2cff9077-f7e7-40f7-afee-515694375ce7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dfe3cf8d-02e8-44e0-a8b7-39379f905a9e\n"
     ]
    }
   ],

I think that should work for me. Thank you for your help!

1 Like