My lab has
(public address: 133.3.133.133, host name: my.lab.host.name)
lab-gateway (private address: 192.168.0.1)
|__labgpu01 (private address: 192.168.0.3)
|__labgpu02 (private address: 192.168.0.4)
|
...
and
- I am a user without sudo
- Each user has their home, and are the same for every servers in the network.
How can I setup jupyterhub so I can use those gpus at my home ?
- which spawner should I use ?
- tutorial suitable for this situation ?
Extra info
- I always do this to get into gpuserver
localuser@local ~ $ ssh lab
remoteuser@lab-gateway ~ $ ssh labgpu01
remoteuser@labgpu01 ~ $
with .ssh/config
Host *
AddKeysToAgent yes
Host lab
HostName my.lab.host.name
User remoteuser
- To access the gpuserver from our home, other suggest me use ssh tunnel
remoteuser@lab-gateway ~ $ ssh -L 8888:localhost:9000 remoteuser@my.lab.host.name -t ssh -L 9000:localhost:9000 labgpu01
remoteuser@labgpu01 ~ $ jupyter notebook
with
c = get_config()
c.NotebookApp.ip='*'
c.NotebookApp.password = 'sha1:...'
c.NotebookApp.open_browser=False
c.NotebookApp.port=9000
and then in my browser localhost:8888
but I just succeed once and then I don’t what I did, this way didn’t work anymore…