Embedded server in Jupyter notebook with port forwarding

leggitta · December 24, 2024, 3:10pm

I’m trying to embed my bokeh app in a Jupyter notebook. The following example works fine when Jupyter is running on my local machine (Mac laptop) …

github.com

bokeh/bokeh/blob/3.6.2/examples/server/api/notebook_embed.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Embedding a Bokeh server in a Notebook\n",
    "\n",
    "This notebook shows how a Bokeh server application can be embedded inside a Jupyter notebook. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import yaml\n",
    "\n",
    "from bokeh.layouts import column\n",

This file has been truncated. show original

But I typically run Jupyter on another machine (linux desktop) and connect via port forwarding …
ssh -NL 8889:localhost:8889 [email protected]

When I try to run the notebook_embed.ipynb example on my desktop via my laptop, the output doesn’t render. I tried changing the notebook_url argument to the local address and get the following

Bryan · December 24, 2024, 3:54pm

It’s not clear to me that it will be possible to open a websocket through an SSH tunnel, in general. But I don’t use SSH tunnels regularly, so perhaps they are more capable than I am aware of. Nevertheless, a first action to try is the one given in the error message: setting the environment variable as described, on whatever machine is actually running the notebook kernel.