How to open Clara Train - Jupyter notebook browser login token/password?

Hello Everyone,

I am able to see the Jupyter Lab page but when I key in the login token from the cmd line output message,it throws an error message stating that “Invalid Credentials”. Can you help me fix this?

Even when the token is provided in the URL, it still prompts for password/login token?

This is the output message that I got after executing the docker command

" To access the notebook, open this file in a browser:
file:///root/.local/share/jupyter/runtime/nbserver-1-open.html
Or copy and paste one of these URLs:
http://(c07978cb23c0 or 127.0.0.1):8888/?token=1234567891ee345a505ddc83e78fbb3584550a535"

Please note that I am trying to open this browser locally in desktop. So I have protforwarded from 8888 to 7878.

As I am able to see the Jupyter Lab screen and it prompts for password, I guess the issue is with the login token/password. I am not able to use this login token either for login or for setting up password. Can someone help me?

Hi ephssmk,

you mention that you tried to access the notebook locally in your desktop. Do you also run the docker locally? Or do you want use a remote docker container?
In the later case, do you ssh in the remote machine with the option “–L”? There is a nice (general) “how-to” here: https://github.com/schoenemeyer/docker-nvidia-basics

Hi,

Thanks for the response. I am able to access the Jupiter notebook locally. Yes, our docker is running in remote server but after port forwarding am able to access the jupyter notebook locally. I mean I am able to see the main page of jupyter lab where it prompts for password / login token

When I key in the token from the URL displayed in the command line window, I get an error message as invalid credentials. Do you think the issue is still with port forwarding it the login token?

Thanks
Selva

I connect from my host(desktop) to remotge server (GPU) where the docker runs. So, I have used Putty SSH client to do port-forwarding. I open the browser in localhost:7878 (I forwarded the 8888 (displayed from command output shown above ) port to 7878 (localhost) using Putty). However, I am getting password /login token error. Is it again a port forwarding issue? I am attaching few screenshots to help you understand the issue better

I am attaching the images

  1. Command_output image - This shows the result of docker command

  2. Jupyter notebook mainpage - This shows the jupyter notebook which is opened locally in my desktop

  3. port forwarding putty

Can images be only attached and not pasted in comment window?

Command_output..JPG


Port_forwarding_putty.JPG

Command_output..JPG

Port_forwarding_putty.JPG

Can anyone help me with this? Is the password issue ‘Invalid Credentials’ due to the port forwarding issue? If that’s the case, I shouldn’t be even seeing the Jupyter Lab in my local desktop browser. Am I right?

Hi,

Does your launch command look something like this?

docker run --runtime=nvidia --rm -p 8787:8888 nvcr.io/nvtltea/med/tlt-annotate:v0.1-py3 jupyter notebook --allow-root --ip 0.0.0.0 --port 8888

This seems to work for me when I open the browser to the remote machine where this was launched, to port 8787.

Hello Alvin,

We still face the same issue. I use the below docker command from the clara train TLT documentation to run the Jupyter notebook examples

docker run --runtime=nvidia -it --rm -v /home/selva/brdemo:/mnt/brdemo -p 8888:8888 -w /opt/nvidia/medical/segmentation/examples/brats nvcr.io/nvtltea/med/tlt-annotate:v0.1-py3 jupyter notebook /opt/nvidia/medical/segmentation/examples/brats --ip 0.0.0.0 --allow root --no-browser

The above command produces the expected output and let’s me know the URL and port number where the Jupyter notebook is running. As we don’t have UI to view Jupyter notebook in remote, I port forwarded it to local using Putty
When I type in “localhost:7878” in URL, I am taken to the Jupyter Lab main page where it prompts me for the password. When I key in the login token from command output, it throws an error message as “Invalid credentials”

Am I missing anything basic here?

Why are you port forwarding? Does typing the address of the remote host:port not work from your local browser? (E.g. http://:/)

No, it doesn’t work.

Can you also help me with the below two posts?

https://devtalk.nvidia.com/default/topic/1050853/clara-train-ai-assisted-annotation-sdk-new-/error-when-executing-engine_pool-json-docker-command-annotation-server/

https://devtalk.nvidia.com/default/topic/1051306/clara-train-ai-assisted-annotation-sdk-new-/end-to-end-flow-for-clara-train-aiaa-and-tlt/

Thanks
Selva