Skip to content

Code of Conduct

Day to day guidance

PARCC’s flagship cluster, Betty, supports hundreds of users daily. Access to the system is provided in the form of multiple gateways: 3x login nodes for day to day, 2x Globus nodes for data transfers, and more. These gateway nodes serve as the interface between users and cluster resources. To maintain a stable and collaborative computing environment, all users are strongly encouraged to follow best practices and ensure their activities do not negatively impact system performance or other users.

Do Not Run Jobs on the Login Nodes

The login nodes are shared by hundreds of users and are intended for lightweight tasks such as job monitoring, job submission, file editing, and occasional code compilation. Running compute-intensive jobs directly on the login nodes is strictly discouraged, as it can degrade performance for all users.

If you need to perform a quick test to verify input files or check that an application runs properly, ensure the task completes within a couple of minutes. For more thorough testing, request an interactive session using the interact command. Additionally, avoid activities that place heavy load on the file systems, such as using rsync or copying large volumes of data between file systems. These actions can significantly impact system performance and other users’ workflows. Please use Globus (FAQs) to transfer large amounts of data. In short, do not run jobs or perform intensive computing on login nodes. Your account may be suspended and you will lose access to the queues if your activities are impacting other users.

✅ Permitted Activities on Login Nodes

The login nodes are intended for lightweight, short-duration tasks. The following activities are allowed:

  1. Request an interactive session
    Use the “interact” command (see FAQs or User Guide) to launch a session for testing or running applications.
  2. Compile code
    Compiling software (e.g., using make) is permitted.
    • Use caution with parallel builds:
      • make -j4 is generally acceptable.
      • Avoid using large values (e.g., make -j20), which can impact other users.
  3. Check your jobs
    • Example: “sqme” to view your current jobs.
  4. Edit and manipulate files
    • Use editors like vim, nano, or simple shell commands to manage files and scripts.
  5. Submit jobs to the scheduler
    • Example: sbatch script-name
  6. View output files
    • Example: more filename, less filename, or tail filename

❌ Prohibited Activities on Login Nodes

To ensure system stability and fairness, the following activities are not allowed on the login nodes:

  1. Running computational jobs directly
    • Do not execute binaries like ./a.out or any other application from the login node.
  2. Heavy file operations
    • Avoid using rsync or copying large numbers of files between file systems.
  3. Running interactive applications like MATLAB
    • Launch such applications through an interactive session instead. or using OpenOnDemand.

PARCC Terms of Use

1. Definitions

User

An individual affiliated with the University of Pennsylvania (holding an active PennKey) who has been granted an account to use the Betty’s resources.

Account

A user-specific login (userid) that provides access to Betty’s computing resources. Coldfront Portal Tutorial

Allocation

A specified number of service units (SUs) assigned to a particular project. A single account (userid) may have access to one or more allocations.

Principal Investigator (PI)

A user, typically a faculty member, who manages one or more project allocations. PIs have the authority to grant users access to these allocations. They may also delegate responsibilities by assigning proxy roles to trusted collaborators.

2. Accounts and Allocations

  • PIs can create and manage projects through the Coldfront Portal Tutorial
  • PIs can assign proxy status to other users for assistance in managing projects, allocations, and users.
  • All users must must be added via the Coldfront Portal by their PI or a proxy.
  • PIs may sponsor external users by requesting guest PennKeys, when appropriate.

3. Acceptance of Terms

By using any PARCC resources, including those on the Betty cluster, all users (including external users) agree to abide by these Terms of Use.

  • These Terms may be updated at any time. Continued use of PARCC resources requires acceptance of the most current version.
  • Violations of these Terms may result in suspension or termination of the user’s account.

4. Security

  • Account sharing is strictly prohibited. Each account is for individual use only and must not be shared with others, including members of the same research group.
  • Password security is the responsibility of the user. Betty uses a multi factor user authentication based on two-of-three: Kerberos, Duo or SSH key.
  • Users must report any discovered security flaws to the PARCC administration immediately and must not exploit such vulnerabilities.

5. Usage Guidelines

  • Resources must be used solely for the research and computational tasks outlined in the associated allocation(s).
  • All data and software used on the system must be legally obtained and comply with applicable licenses and terms of use.
  • All compute-intensive tasks must be submitted through the resource manager (Slurm)  to compute nodes.
  • Users must not run high-resource applications or long-running processes on the login nodes.
  • Bypassing the job scheduler to access compute nodes directly is prohibited. Any such activity must be reported and not exploited.
  • Users are responsible for ensuring their use of resources complies with U.S. export control regulations, including securing necessary licenses.

6. Publications and Acknowledgment

  • All publications that rely on data generated or software developed using PARCC  resources must acknowledge the allocation under which the resources were used.
  • PIs must follow applicable proposal and reporting guidelines.
  • Publications should be uploaded to Betty’s coldfront Portal as soon as they become available.