Logging into GMU (ORC) Computing

Overview

Teaching: 15 min
Exercises: 5 min

Questions

• Why do I need to login to another computer?

• How do I login to GMU computers?

Objectives

• Explain why it is necessary to login to another computer

• Explain what computer systems exist at GMU to log into

• Explain how to login to the GMU computers from a student’s computer

Why do I need to login to another computer instead of using my own computer?

In climate science most of our data analysis work is primarily done on either: 1. Workstations (individual computers with multiple CPUs, large internal memory and access to large data storage) 2. Clusters (groups of dozens to ~100 connected workstations, often called “nodes”), or 3. Supercomupters (very large clusters with hundreds or thousands of cores)

Such high-powered computers are almost always running a version of the Unix or Linux operating systems. These types of computers have more computational capability, more memory, and much more disk (or hard drive) space than a personal computer. We need this to handle the very large datasets used in climate data analysis.

Our research computers at GMU already contain many of the datasets we will work with. These datasets are too large for your home computer and would take weeks to months to download over the Internet. The Center for Ocean-Land-Atmosphere Studies (COLA) in the Department of Atmospheric, Oceanic, and Earth Sciences (AOES) has petabytes (i.e., thousands of terabytes, millions of gigabytes) of climate data from models and observations.

Linux/Unix computers use a command line interface that allow us to automate our data processing tasks much faster than a graphical user interface (GUI) that you may be familiar with in Windows or Mac OS.

GMU Computing - ORC and the clusters

The GMU Office of Research Computing maintains two separate clusters:

1. The HOPPER Cluster is ORC’s primary compute cluster, which provides support for both batch jobs via Slurm and interactive computing. The cluster includes: 74 compute nodes each with 192 GB of memory, and two GPU nodes each with 1 TB of memory.
2. The ARGO Cluster is a ~2000 core Linux cluster also providing a scheduled job environment using Slurm. There are a number of different node configurations on ARGO ranging from 16 cores to 32 cores per node. Each node provides a minimum of 4 GB/core, however it has some nodes with 512 GB of RAM and one node with 1.5 TB of RAM. Additionally there are also several kinds of GPU nodes.

We will use HOPPER in this class.

Setting up Software for your computer

There are several ways to connect to HOPPER. By far the easiest is via the Dashboard in your browser. This requires no special software installation or setup on your laptop. There are a number of commonly-requested apps provided from the dashboard, and a link to a more complete list of apps. In this course, we will be using two of the options listed:

1. The HOPPER Desktop - a Linux GUI OS interface to HOPPER, and
2. Jupyter Lab - an interface to create and run Python notebooks on HOPPER

As with all things GMU, you will have to log in with your Mason ID and password. If you have your browser setup properly, you will only have to do this every 14 days (all secure GMU pages including the ORC Dashboard should recognize the same authentication).

Connecting to HOPPER via the Dashboard

If you do not already have an ORC account, please see the new user information. The pre-course survey should have informed us whether you already have an account.

1. Click on the “HOPPER Desktop option”.
2. You will be moved to a page called “Interactive Sessions” and given options to launch an interactive desktop session.
   • Number of hours: Your session will expire automatically after the selected number of hours. Since this class lasts 1:15, select 2 hours (you may want to set a longer time when you are working outside of class).
   • Number of cores: Select 1. We will not do any massively parallel computing that requires more than one core. We will run some Python packages that take advantage of multiple CPUs simultaneously within the same core.
   • There is an option to be notified by email when your session becomes available. For such a small request, your session is typically ready after only a few seconds - it is not necessary to request email notification unless you are making a very large request (many nodes) for which resources may not be immediately available.
3. A new browser tab will open and present you with a virtual Linux desktop. Click on the “terminal emulator” icon at the bottom to launch a session in a terminal window. Your session will start in your home directory on HOPPER.

Connecting to HOPPER via ssh

As an alternative to the HOPPER desktop, you may access HOPPER via a terminal window on your laptop running secure shell (ssh) software. The software differs based on what type of computer you have.

Mac OS

For a Mac computer, use software called Xquartz, which takes advantage of the fact that Mac OS is built on top of Unix.

Windows

For a Windows computer, use software called MobusXterm.

Linux

The default Shell for Linux operating systems is usually bash. On most versions of Linux, it is accessible by running the (Gnome) Terminal or (KDE) Konsole or xterm, which can be found via the applications menu or the search bar. If your machine is set up to use something other than bash, you can run it by opening a terminal session and typing bash.

Download and install the correct ssh software for your computer

If you have not already done so, it would be a good idea to download the correct ssh software for your computer

Follow the instructions on your computer to install the software

On macOS

Launch the XQuartz software you downloaded and select Shell-> New Window from the menu in the upper left. A window will appear with a command line prompt, waiting for you to type.

To connect to HOPPER, type the following and replace username with your Mason ID:

$ ssh -Y username@hopper.orc.gmu.edu

Enter your password when prompted.

On Windows

1. Launch the MobusXterm software you downloaded.
2. Click Session->SSH
3. In the Remote host box, enter hopper.orc.gmu.edu
4. Check the Specify username box and enter your username
5. Click OK
6. Enter your password when prompted
7. Select No when asked to save your password.

If this is the first time you have logged in, you will be required to change your password.

Password Policies

• Keep your password secure and in a safe place - remember, this is your password for everything at GMU, not just this computer account!
• Do not share your password.
• No one should ever ask you for your password. If you ever get an email or other electronic message asking for your password DO NOT respond to it, it is a phishing attempt.
• Do not store your password on your laptop in plain text.
• Passwords must be changed every 6-months. You will get an alert from the university when the time is approaching.
• Follow the Mason Responsible Use of Technology Policies

Key Points

• You need to use GMU computers for this class

• The GMU computer that you will log into is a cluster called HOPPER

• Secure (ssh) software is used to login to GMU computers