Connecting to Sherpa dashboard

[goravkaul]

I ran sherpa code in my jupyter notebook on laptop(Windows) which connects to a server(Ubuntu) and Sherpa study gave following link for dashboard: http://172.17.0.2:8880.

On server, there is docker running which was started with below command:

nvidia-docker run -it --rm -v /home/user/workdir/:/root/mywork -p 8888:8888 -p 6006:6006 -p 8880:8880 tensorflow/tensorflow:latest-gpu-py3 "/bin/bash"

-p 8888:8888: used for Tensorflow
-p 6006:6006 : used for Tensorboard
(I tried both with and without -p 8880:8880 option)

For tunneling I ran below command:
plink -ssh -T -L 8880:server:8880 user@server

But nothing gets displayed when I try to open the dashboard link, it shows connection timed out.

[goravkaul]

For now, I found a workaround.
I modified function _run_web_server(self,port) in file core.py by passing hard coded values to port and host variable in call to multiprocessing.Process api.

proc = multiprocessing.Process(target=app.run,
                                       kwargs={'port': 5000,
                                               'debug': True,
                                               'use_reloader': False,
                                               'host': '0.0.0.0',
                                               'threaded': True})

And created below plink on my laptop:
plink -ssh -T -L 5000:remote_server:5000 sagar@remote_server