Python Notebook node nonfunctional

[bguseman]

Using a workflow with a 'Read DataFrame' node (which runs successfully) connected directly to a 'Python Notebook' node, the Python Notebook is completely nonfunctional under the Pyspark kernel. The notebook is accessible, but the Pyspark kernel looks like it is constantly busy, and attempting to run a code cell does nothing. Restarting the kernel also fails.

Please see the included log snippets. The jupyter log came from inside the running notebooks container. The console log and the headless jupyter snippet were written to the startup console.

I am running on a Mac

console.log
jupyter_headless.log
jupyter.log

[metasim]

@jaroslaw-osmanski Any direction on how we might further debug/address this is really appreciated. This is a major blocker for us right now, and we're out of ideas on how to isolate the problem.

[jaroslaw-osmanski]

#51 should solve it.
Quick fix: downgrade tornado to 4.1 on sessionmanager container.

[jaroslaw-osmanski]

@bguseman @metasim Can you confirm that it is solved in master?

[metasim]

@jaroslaw-osmanski We'll test it today or Monday. Been at a conference.

[metasim]

@jaroslaw-osmanski I can confirm that this is now working after a fresh build. Thanks so much for addressing this!