0
Context
I'm trying to improve my working workflow.
Currently I have a desktop (that, basically, has a GPU) and a notebook. I want to exploit the desktop even when I'm working with the notebook remotely (they both run Linux)
In my university only the port 80 is open to outside connections.
Current solution
My desktop has an SSH server and I'm doing some port-forwarding in my NAT wanaddr:80 -> localaddr:22
to avoid the restriction in the University network.
Moreover, I'm deploying the code and running with a remote interpreter using IntelliJ IDEs, it works like a charm.
Problem
I have the need to use more services remotely, e.g. tensorboard
, MLFlow
, jupyter
.
The difficulties arise since I must use the port 80 on the WAN side, so I can't just port-forward other ports.
Attempted solution
Assuming that the domain name of my desktop is desktop.dev
.
I'm trying to redirect specific sub-addresses to the correct services, e.g.
ssh.desktop.dev:80 -> <?> -> localhost:22
tensorboard.desktop.dev:80 -> <?> -> localhost:8888
jupyter.desktop.dev:80 -> <?> -> localhost:8000
I tried to use Caddy but later on I understood it supports only HTTP requests, I'm looking into Traefik but I'm not getting how to do it easily.
Can you point me in the right direction?
Thank you, it's working! – Luca – 2019-06-28T07:45:26.083