Skip to content

SSH access to Purdue AF

Although the web-based JupyterLab interface is the recommended way to access Purdue AF, it is also possible to connect to your session from a terminal via SSH.

Note

SSH connects you to your running AF session, not to a separate login node — you must start the session in a web browser first. If you would rather work in a full local IDE, see Access via VSCode-based IDEs.

Instructions

  1. Start an AF session in the normal way

    In a web browser, log in to Purdue AF and start a session.

  2. Obtain an authentication token

    1. Go to File → Hub Control Panel.
    2. Click the Token tab in the top left of the page.
    3. Click Request new API token to obtain the token string — you will need it in the next step.

  3. Log in to your session from an external terminal

    ssh <username>@jupyterhub-ssh.cms.geddes.rcac.purdue.edu
    • If you are using a CERN or FNAL account, remember that your username must include the -cern or -fnal suffix, respectively — see Login methods and usernames.
    • Instead of a password, paste the token obtained in the previous step.

  4. (Optional) Set up an alias in .bashrc

    By adding the following line to ~/.bashrc or ~/.bash_profile on your local machine, you can avoid copy-pasting the token every time. Note: you will need the sshpass utility installed on the local machine.

    alias purdue-af='sshpass -p <token> ssh <username>@jupyterhub-ssh.cms.geddes.rcac.purdue.edu'

    Now simply running the purdue-af command will immediately connect you to Purdue AF, provided that you have already started a session.

Caveats

Tip

If you encounter any formatting issues while working at Purdue AF via an SSH connection (for example, broken lines in the Vim text editor), try running the resize command.

No SFTP / scp support

At the moment, the SSH server at Purdue AF does not support the SFTP protocol, which means that you cannot use scp commands to download and upload files from your local machine.

Use the web interface to transfer files instead — see Uploading and downloading files:

  • upload files by drag-and-dropping them into the Jupyter file browser, or via the "Upload" button at the top of the file browser window;
  • download single files via the right-click menu;
  • download a directory by right-clicking on it and selecting "Download as an Archive".