This workshop assumes some prior familiarity with Hyak Klone. However, if you are completely new to Hyak or Open OnDemand (OOD), there are two things you must do first:
- Log in to Klone at least once so your home directory is created
- Become aware of Hyak storage locations and quotas so you do not accidentally fill your home directory
Before starting this tutorial, you should have the following in place:
- A UW NetID
- An active Hyak account for Klone
- Two-factor authentication (2FA) enabled
- Access to an SSH client on your local computer
If you do not yet have a Hyak account, you can request access for training purposes:
Klone demonstration account request form
Students should apply for a Student Technology Fee (stf) account through:
UW policy requires two-factor authentication (2FA) for all UW-IT services, including Hyak. Confirm that 2FA is enabled before attempting to log in:
UW Two-Factor Authentication Info
If you are participating from off campus, you must connect to the UW VPN before accessing Open OnDemand follow these steps:
Establishing your connection to the University of Washington VPN
Helpful resources:
To initialize your account and create your home directory, log in to Klone once using SSH:
ssh UWNetID@klone.hyak.uw.eduA successful login will display a welcome banner. Once you see this, your home directory has been created.
📝 NOTE: Too many failed login attempts may result in a temporary IP ban (one hour).
After logging in, you will land in your home directory:
[UWNetID@klone-login01 ~]$Your home directory:
- Is intended for configuration files and small scripts
- Has a strict 10 GB storage quota
- Should not be used for large datasets or software installations
Resource: Hyak Home Directories Video
For this workshop, you will work in:
/gscratch/scrubbedThis location is designed for active computation and provides substantially more space.
📝 NOTE: Scrubbed storage is temporary scratch space for active computation. Files are not backed up and are automatically deleted after 21 days of inactivity.
Create your personal folder there and navigate to it:
mkdir /gscratch/scrubbed/$USER
cd /gscratch/scrubbed/$USER📝 NOTE: The
$USERvariable automatically expands to your username.
Next, clone the git repository for this tutorial into your personal folder:
git clone https://github.com/UWrc/jupyter-tutorial.gitJupyter launched through Open OnDemand starts in your home directory by default.
To make it easy to access high-capacity storage from within Jupter, we create a symbolic link in your home directory that points to your scratch workspace.
ln -s /gscratch/scrubbed/$USER ~/scratchThis creates a directory called scratch in your home directory that points to your workspace for this tutorial. View that it was set up properly by listing your home directory with the long format option, which will show the symbolic link and its true path.
ls -l ~Now, when you launch Jupyter through Open OnDemand, you can easily navigate to ~/scratch and work directly in your scratch storage.
📝 Important notes about symbolic links:
- A symbolic link is not a copy of the directory — it is a pointer
- Deleting the link does not delete the target directory or its contents
- You can safely remove the link later with:
rm ~/scratchYour data in
/gscratch/scrubbed/$USERwill remain intact.
These setup steps ensure that interactive applications launched through Open OnDemand can access appropriate storage locations without risking home directory quota issues.