If you find yourself needing more to do more than what's possible within OnDemand–such as installing new software, doing parallel programming, etc.–we recommend working with the clusters through shell access and linux commands, as explained in our Guide to the Princeton Research Computing Clusters.
Before working on the cluster, is it important to understand where different types of files should be stored in the folder hierarchy. Therefore, before proceeding, we highly recommend reviewing our Data Storage page for this information.
Select
“Home directory”
Or
"/scratch/network/<user>" (for Adroit) or "/scratch/gpfs/<user>" (for other clusters)
Click “Upload”
Select “Browse files” to upload a file
Navigate to the file you’d like to upload. Note you can select several
Click “Upload file”
Your file has now been uploaded to the directory you specified on Adroit.
You can view the file by clicking on its filename in the Files interface.
NOTE: When uploading files through the OnDemand portal, each upload is limited to a size of 2 GB. If you need to upload larger files, we recommend using Linux commands, as described on our Transfer Files page. You can enter these Linux commands inside the Terminal, which you can open as described in our Running a Terminal from OnDemand section further down this page. Reach out to [email protected] if you need help with this.
Select
“Home directory”
Or
"/scratch/network/<user>" (for Adroit) or "/scratch/gpfs/<user>" (for other clusters)
Navigate to your file
Select the menu option with 3 vertical dots and a downward facing arrow next to your file:
Select “Edit” from the dropdown menu that appears.
You are now in the editor interface for your file. Populate it with any contents you need. Once you’re done, hit “Save” in the upper lefthand corner of the Editor interface.
To run jobs using the OnDemand interface, you need to create a slurm script for your code. To learn more about how to submit jobs to the clusters, we recommend reviewing our Guide to Princeton's Research Computing clusters.
To run commands on the Adroit head node, for example, browse to MyAdroit and choose "Cluster" then "Adroit Cluster Shell Access". The same applies for the other clusters with OnDemand access.
Users must follow the 10-10 rule on the head node of any cluster. The 10-10 rule says that you can use up to 10% of the resources of the machine for up to 10 minutes. The head node of each cluster is shared by all users so this rule prevents someone from monopolizing the machine. The head nodes should only be used for light work such as installing software, transferring files and doing short test runs. You may be contacted by a system administrator if you fail to observe the 10-10 rule.
To begin an interactive session with Jupyter, RStudio, MATLAB, or Stata on our clusters, click on "Interactive Apps" in the top menu bar of your web browser. Specific intsructions for each of these programs can be found in the sections below.
When launching Jupyter, for example, you will need to click on "Interactive Apps" and then "Jupyter". You will need to choose the "Number of hours", "Number of cores" and "Memory allocated". Set "Number of cores" to 1 unless you are sure that your script has been explicitly parallelized. Click "Launch" and then when your session is ready click "Connect to Jupyter". Note that the more resources you request, the more you will have to wait for your session to become available. When your session starts, click on "New" in the upper right and choose "Python 3.7 [anaconda3/2019.10]" from the drop-down menu.
Note that Mathematica can also be used via OnDemand. If you only need a single CPU-core then consider using the Princeton Virtual Desktop which is maintained by central OIT. For more, see our Mathematica page.
You can open interactive sessions on the visualization node of each cluster. To do so, click on "Interactive Apps" menu, and then select an option with "Desktop (Visualization Node)" or a similar name.
Navigate to
/home/<your-netid>/ondemand/data/sys/dashboard/batch_connect/sys/<interactive-app-name>/output
Be sure to insert the relevant information for the content in <>'s within the path name.
For example, if your netid is janedoe, and you're looking for logs from your jupyter session, you would navigate to
/home/janedoe/ondemand/data/sys/dashboard/batch_connect/sys/jupyter/output
Navigate to the folder that has the time and date that matches the session of interest. The folder name will look something like
e1692794-4951-4cb5-910a-2e416b7461e8
In other words, a mess of numbers and letters.