Cannot reach jupyter notebooks on the deep learning VM

[Pradhyo]

I first created a VM on 4/1/2018 and then again now but still not able to connect. More details here -
https://stackoverflow.com/questions/49603966/cant-reach-jupyter-notebooks-on-azure-deep-learning-virtual-machine

When I tried to run jupyter notebook -

[C 04:00:12.963 NotebookApp] Bad config encountered during initialization:
[C 04:00:12.963 NotebookApp] No such notebook dir: ''/dsvm/Notebooks''

Today I installed an X2Go client and found the jupyter notebook icon on the desktop but starting that threw this error -
Failed to run Jupyter.desktop
Failed to parse the desktop file: Key file contains line '/anaconda/pkgs/notebook-5.4.0-py27_0/info/icon.png' which is not a key-value pair, group, or comment

Looks like that file calls jupyter notebook which throws the same error as before.

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: 27c1a5f0-b0df-88f4-feaf-90bc17eeb69e
• Version Independent ID: 3d5b36f7-e892-8d1d-fa14-a2e5d407a25c
• Content: Provision a Linux (Ubuntu) Data Science Virtual Machine on Azure
• Content Source: articles/machine-learning/data-science-virtual-machine/dsvm-ubuntu-intro.md
• Service: machine-learning
• GitHub Login: @bradsev
• Microsoft Alias: bradsev

[udooz]

@Pradhyo, it is a configuration. Instead run Jupyter Lab. The problem what I am facing is user credential is invalid.

[YutongTie-MSFT]

@Pradhyo Thank you for your feedback, we have assigned the issue to the author of the document. He will update as necessary.

@udooz Thank you for your response and feedback. Can you please open a separate doc comment for your question if necessary?

@bradsev Jupyter notebook is mentioned in the post. Can you look into this feedback to determine if further guidance should be applied to the document? Thanks.

[cjgronlund]

#in-progress

[cjgronlund]

@gopitk Please review this issue and submit a fix as appropriate. @ me on your response, pls.

[gopitk]

Sorry. This was a bug that was intreoduced in the most recent version. You can overcome this till our next release in the following ways:

1. Access Jupyter server from your client machine directly using JupyterHub. The URL would be https://:8000/ . If you have the default self signed cert instead of a real certificate you may get a warning. If you get past the warning you will be asked to login. Here you can use the local DSVM account credentials.

2. If you want to run "jupyter notebook" from within a X2Go session, you will have to regenerate the Jupyter config file as the one currently is incorrect (and what we will fix in our next release).

sudo -s
source /anaconda/bin/activate py35
jupyter notebook --generate-config config=/usr/local/etc/jupyter/jupyter_notebook_config.py

Sorry about the issue. Hope one of these workarounds works for you.

[cjgronlund]

@gopitk Thank you for this answer. Could you submit a change to the docs to cover this issue, unless the new release is imminent? Thank you!

[cjgronlund]

@MicrosoftDocs/azure-cxp-triage Please assign this to @gopitk
Bradsev is no longer with the company.

[Pradhyo]

@gopitk Thanks for getting back.
I tried both suggestions before posting here but Chrome on my client machine couldn't reach the VM.

I played around with different values in the jupyter config file but I didn't try the entire config file you mention. I will update here when I try this.

I remember setting allow_origin = "*" but it was still unreachable.

[gopitk]

For port 8000 (suggestion#1) when you try to access JupyterHub remotely from your client machine browser you must use HTTPS. You have to get past cert warning if you are using default self signed certs that came with the DSVM instance. Can you confirm that? Another reason - It is possible your Azure subscription admin may have set some client access policies that may be preventing access or your network may have rules around port 8000.

In Option (2) you must login to the DSVM using the X2Go graphical desktopm regenerate the jupyter config and start "jupyter notebook" locally on the DSVM X2Go session. After a few seconds it shoudl open a browser to show you jupyter.

[j-martens]

This appears to have been solved. #please-close for now.

[LukeDevs]

This is still an active issue.

Following the steps @gopitk outlined above still does not resolve the user authentication error. I am able to see the Jupyter Notebook externally over the internet, but the DSVM user password is not accepted at the password prompt.

It is unclear what password it is requiring.

Remote Desktop to the DSVM via X20Go also prompts me with the password challenge that I cannot pass.

[dennisyar]

@LukeDevs The DSVM credentials worked for me (local connection to anaconda through port 8000)

[DaveVoyles]

I just noticed this issue as well. I know it worked fine for me before, too.

[gopitk]

@DaveVoyles - Can you please describe your usage scenario? i.e Are you using Jupyterhub and accessing noteboosk externally or are you trying running "jupyter notebook" from a bash locally on a X2Go session? What exactly is the error or behavior? Also when did you create the instance of the VM.

BTW - We just published a new VM image fixing the the error above "NotebookApp] No such notebook dir: ''/dsvm/Notebooks''" when you try to run "jupyter notebook" on shell.

New instances of Ubuntu Data Science VM or Deep Learning VM will have these fixes.

[DaveVoyles]

Sure thing.

The other day, I was able to run my Jupyter Notebooks, which I had saved in a folder on my desktop. When I logged in today I clicked the Jupyter button to launch the app, and it threw the error above. That was on an Ubuntu DL VM.

I deleted that machine and just created an Ubunutu Data Science VM, and it works fine.

[gopitk]

Thanks @DaveVoyles for confirming our fix is working!

#please-close