Local SSH vs OOD

SSH vs OOD

	SSH	OOD
Network	UVA Network/VPN required	No network requirement
Execution	On front-end, no running code	On production nodes, can run code
Resources	ijob resources specific to terminal	No need for ijobs, have resources already
Notes	Must take extra care to use properly	Some extensions may be unavailable

Extension Availability
- But if one isn’t available just reach out to us!
Local preferences
- But you can export the settings.json file to HPC
No queue wait time
- But if you are requesting appropriate, interactive resources the wait time will be minutes at our busiest

We strongly urge our users to use OOD.

Those resources are available and best for our users.

All functionality is the same and you will always have access to compute resources.

You will not need to worry about being overwhelmed with processes or fighting with ijob terminals.

Search for “Remote – SSH” in the Marketplace.

It is not necessary but it is a tool to make it much easier.

Connect to Host > Add New SSH Host…

ssh -YA mst3k@login.hpc.virginia.edu

Modify your config file

You can later open and edit the config file as you need

Now that you are connected you can open up your workspaces you’ve created, open your folders, etc.

You can see your connection in the bottom right

You still stay connected so long as you stay on the network

Open a new terminal

ijob -c 6 -A hpc_training -p interactive --time=00:30:00

Starts an ijob on the interactive partition with 6 cores for 30 minutes

Will need to wait for resources to be available

An ijob is specific to the terminal you start it in, you only have access to those resources in that terminal

The job ends if you close the terminal

This is for running code via terminal commands

If you want to run Python code you need to “Run Python File” first, then type in ijob command in that new terminal

Different method for JupyterLab

DO NOT RUN JOBS FROM IJOBS

Start an ijob, then load JupyterLab and start a notebook server:

module load miniforge jupyterlab
jupyter-notebook --no-browser --ip=0.0.0.0

Copy the URL that starts with http://.

In VS Code, select Existing Jupyter Server, paste the URL, and press Enter.

Select your kernel from the kernels we offer or create your own kernel from a conda environment.

VSCode can create many processes on the front-ends.

If you are going to use SSH you must use it properly.

Disconnect your session every time you are finished.

Do not leave it running in the background.

In most cases, the OOD interactive app is superior and preferred to SSH.

Last updated on Apr 14, 2026