Port mapping between local and remote targets

Tested on a Linux 5.4.0-94-generic on 18/01/2022

When we are developing software using several machines we want to access resources across several machines. (e.g. access a Jupyter notebook server running on a docker container on a remote host)

I will use the example use case above for the walk-through shown below.

This line below is run from my local PC

ssh -N -L localport:ip_of_the_resource_seen_by_the_remote_host:port username@ip_address
ssh -N -L 8887:localhost:8888 ganindu@192.168.0.33

IPV6

ssh -N -L localport:ip_of_the_resource_seen_by_the_remote_host:port username@ip_address%network_interface
ssh -N -L 8887:localhost:8888 g@fe80::4bdc:e743:27d1:bf03%enp0s41f7

8887 : Port in my local PC that I want to map to
localhost:8888: Source IP:PORT as as seen by my host computer
-N: Prevent executing other shell commands (helps to keep the shell standalone)
-L: Local mapping setup (check manual entry for ssh)

Note: you can also map multiple ports in a single line:

In the the example we are mapping localhost (127.0.0.1) ports of the remote host to the localhost ports of the connecting PC. They can be any host (IP or text URL)

ssh -N -L 12345:localhost:12345 -L 54321:localhost:54321 pi@proxy55.rt3.io -p 32897

Use-case example with a slightly different syntax:

Imagine this deepnotes website is hosted on a remote PC’s localhost (e.g.127.0.0.1:4001/deepnotes/) and this remote pc is accesible via ssh jon@654.32.1.87
Type ssh -L 127.0.0.1:4000:127.0.0.1:4001 jon@654.321.1.87 in a new terminal to create the tunnel (keep terminal active to keep the tunnel alive).
then access the remote webpage by inserting 127.0.0.1:4000/deepnotes/ in the adress bar from the local pc.

Note: This is same as ssh -N -L 4000:localhost:4001 jon@654.321.1.87

Note: Sometimes port names are fiddly because they might expects incoming data from a certain port so if stuff goes bad with arbitrary mapped port ids map to the same id and see if it works. (ssh -N -L 4001:localhost:4001 jon@654.321.1.87)

You can extend this to other use cases as Jupyter notebook servers or ipython kernels.

P.S.

Alternatively you can do something like ssh -N -L localhost:8887:localhost:8888 username@computer1 or ssh -N -L 8887:localhost:8888 username@computerx too.

source 1: Connecting to a Jupyter Notebook on a remote Linux machine with an SSH tunnel
source 2: Accessing a Jupyter notebook server through reverse port forwarding