Brief info, I already running until lesson 3 with my local PC Ubuntu 16.04, GTX 750 Ti inside. Everything runs well, and every time I need to access to the jupyter notebook, I just type jupyter notebook on the command line, and it opens new tab in Firefox.
Now I just get access to my school cluster GPU, Ubuntu 16.04 with GTX 1080 Ti inside (yay, no more memory error limitation). Anaconda, Keras, Theano & Jupyter notebook I could install properly and everything run well. But now I can’t just open my Firefox on local PC and access localhost, right? because I try it and it didn’t work, haha.
I look on some of tutorial and question on the forum, but either it’s on a home server, or cloud server (AWS, Google, Azure), while none questioning about a cluster. So could anyone point out what is the right step by step that I need to do in order to access my jupyter notebook that running on cluster?
Try this instead and see if it works. You need to portforward the port 8888 for the jupyter notebook to make it work. Then when you create a tunnel with the portforward the localhost:8888 will work on your pc
EDIT: not sure what the port -p 8403 is need for so i just kept it. Feel free to remove it if not needed. You can use any other port for jupyter as long as your specify the port option but otheriwse the default is 8888
Thanks, Nell!
It’s worked, need to change the port to 8889 though since Jupyter Notebook said they can’t use 8888, no idea why. And port -p 8403 I didn’t have any clue too what this, just keep it anyway.
In case any other student looking forward to this problem, what I did
After that run
jupyter notebook --no-browser, check which port it’s using, 8889 in my case
Open another terminal, input Nell suggestion, and remember to change the port number if necessary.
ssh -p 8403 -L localhost:8889:localhost:8889 myusername@subdomain.myschool.edu
then type in password
After that open Firefox in your local machine and access localhost:8889, it should be connected now.
you may have a zombie process still holding onto port 8888 … run the command
netstat -pano | grep 8888
and see which process is holding on to port 8888. If you’re the only one using the server then you should be able to kill -9 … if it’s a shared server then maybe someone else is already using the port so be careful
