Please suggest a workflow for working on a remote machine.
by SegFault1 from LinuxQuestions.org on (#56297)
My organization has a very powerful machine that I use to run my ML models. It runs on some version of Ubuntu. Due to this ongoing pandemic, I have to work from home. This machine is not public facing and therefore, I VPN into my organization network using OpenFortiVPN. And then I SSH into that machine.
The problem is that my internet is not very good and/because it a mobile network that I tether on my machine (and that's why I get a dynamic IP evertime I am connected to the network). However, that machine at my organization is connected to very good network. The datasets that I work with are hundreds of GBs. I cannot download the dataset and work on my machine. The dataset and computing power reside on the remote machine, not on my machine.
Suggest a workflow for this setup. The problems I face are,
Admittedly, I'm new to SSH, and VPNs and networking so, if there is something obvious that I'm missing in my setup, please tell me.


The problem is that my internet is not very good and/because it a mobile network that I tether on my machine (and that's why I get a dynamic IP evertime I am connected to the network). However, that machine at my organization is connected to very good network. The datasets that I work with are hundreds of GBs. I cannot download the dataset and work on my machine. The dataset and computing power reside on the remote machine, not on my machine.
Suggest a workflow for this setup. The problems I face are,
- When I SSH into the machine, I only get one terminal. I need many terminal---to edit my code for the model, to run my model, to examine datasets, etc. I don't want to exit my editor every time I want to AWK into my dataset or, run my code. The models take a lot of time to train and run, I don't want to sit and look at my cursor, I'd rather look at my code. I can finally turn to TMUX or Screen, but I want to keep them as a last resort.
- My dataset also contains images, and while examining dataset, I sometimes need GNUPLOT. SSHing into the machine doesn't give me a X.
- I tried mounting the remote machine using SSHFS to my local machine. That solves some problem, but now my local machine becomes slow due to a slow connection to the remote machine. And viewing images on that mountpoint downloads the entire directory which my network cannot support.
- Due to my choppy internet, that key press has a noticeable lag in the SSH session. As someone who doesn't even use a DE (only tiling-WMs or sometimes even without X) to work fast, this lag gets on my nerve.
- I don't know why, but if I leave my SSH terminal for too long without activity, the terminal hangs. Maybe due to bad connection, but not sure.
Admittedly, I'm new to SSH, and VPNs and networking so, if there is something obvious that I'm missing in my setup, please tell me.