L4T-ML:r35.1.0-py3 Jupyter Notebook password issue

[michael.janus]

cannot login with a password to the jupyter notebook that is automatically started in the docker image. The ‘nvidia’ password is not accepted and you can only log-in with the token. When I try to set a new password with the token (via the jupyter web page), still password is not accepted.

[AastaLLL]

Hi

Based on the document, it should be nvidia.

https://catalog.ngc.nvidia.com/orgs/nvidia/containers/l4t-ml

Connecting to JupyterLab Server

…
Note that the default password used to login to JupyterLab is nvidia.

We are checking this issue internally.
Will share more information with you later.

Thanks.

[AastaLLL]

Hi,

Confirmed that we can reproduce this issue internally.

We are checking this with our internal team.
Will share more information with you once we got feedback.

Thanks.

[dusty_nv]

Hi @michael.janus, since the authentication behavior of Jupyter has changed in the latest versions, please try running jupyter server list in your docker terminal, and it will give you a token / URL for your JupyterLab instance. If you navigate your browser to that URL you should be able to login.

[michael.janus]

yes, that is how I indeed accessed the notebook in the end. However, resetting the password using the token does not work (the jupyter page accepts changing it, but it’s not persisted; maybe a file permission issue?). Furthermore, accessing the jupyter notebook from another machine via the 20+ character token is hardly an option.

[dusty_nv]

@michael.janus

However, resetting the password using the token does not work (the jupyter page accepts changing it, but it’s not persisted; maybe a file permission issue?).

Do you mean after you shut the container down, it doesn’t remember the password after you start it again? In that case, you can use the docker commit command to save the state of the container.

Alternatively, if you know how to disable the tokens, you could change the l4t-ml dockerfile and rebuild it, or build your own container based on it. Sorry that I’ve not been able to figure out how to disable it.

@michael.janus

Furthermore, accessing the jupyter notebook from another machine via the 20+ character token is hardly an option.

I would recommend starting an SSH session to your Jetson from your other machine and using that SSH terminal to either start the container or attach to the container, then copy+paste the token from there.