HPC-one web portal

Single point access to HPC resources with a web browser

 

 

✓ Command-line free access to HPC resources

✓ File Management and Data Transfer

✓ Job script template, Job submission and Job Usage Monitoring

✓ Access to Interactive Desktop and Graphical Application without doing any software configuration (e.g. VNC, X11-Forwarding, Port Forwarding etc)

 


Access Requirements

    • Access to HKU network (prior connection to HKUVPN2FA needed for off-campus access)
    • Modern Web browser (Google Chrome version 90+; Firefox version 78+; Microsoft Edge (Desktop devices are recommended as the web layout is not intended for mobile devices)
    • Active HPC2021 User Account

 

Web access

 

  • Use a web browser to open https://hpc2021-io2.hku.hk
  • Log in with an active HPC2021 username and password
    Home page with

    1. Disk Quota Usage
    2. Resource Usage Efficiency of recent job
    3. Quick Launch button for featured applications
  • Log out properly by closing all browsers tab as the “Logout” button at top right corner is not working

File Manager

Top Panel: Files > Home Directory

 

  • Browse files/folders where user has access using the “Files” app
  • Allow file operations (Copy, Paste, Rename, Delete) and transfers (Upload, Download with drag and drop of files/folders < 10GB)
  • Direct editing of plain text files with syntax highlighting


Jobs: Active Jobs

Top Panel: Jobs > Active Jobs

  • List details of queuing/ running jobs, user may view SLURM job logs and cancel a job

 

Job Composer

Top Panel: Jobs > Job Composer

SLURM job script templates

    • Create new jobs from desired templates and modify them for specific needs
    • Or create your own templates

Job History

Top Panel: Jobs > Job History

  • Historical  job resource usage report allows users to review the resource usage of finished jobs such that better estimate of resource requirement can be obtained for tuning of resource request in subsequent jobs.

Job Resource Usage

Top Panel: Jobs > Job Resource Usage

  • Real time job resource usage report gives user up-to-date information about a job’s resource usage during the execution.

Web SSH Terminal

Top Panel: Clusters > HPC2021-io2 Shell Access
  • For those tasks that command line access may be more convenient, a web-based terminal would do the job
  • N.B. Closing a browser tab of web terminal would close the session
  • Session timeout would occur in idle terminal

Interactive Jobs

Request session

Top Panel: Desktop > Remote Desktop

 

  • Interactive jobs are the most unique feature of HPC-one
  • Users request a SLURM job to run an interactive application ( Remote Desktop in this example) on a compute node
  • Users will receive an email notification (if email address is provided) when the job is running

 

Launch Session

  • Once an interactive job is ready, user may click the “Launch …” button to open the application window in a new browser tab

 

  • The interactive application is running in a compute node in the cluster and displayed through a web browser
  • Resizing the web browser would automatically trigger resizing of the application window
  • For copy/paste between your device and application, “Clipboard” would do the job

Delete Session

  • Closing a browser window of an interactive job or any network/power hiccups in your computer would not terminate the job
  • When you no longer need the session, click the corresponding “Delete” button of an interactive job in the “My Interactive Session” to terminate it

Interactive Applications Available

 


Resource Usage

Interactive apps are primarily for visualization code development and hence up to 24 hours of wall time is allowed. For production run of developed codes, it is recommended to submit them as batch jobs to SLURM scheduler. Below is an example on how to convert a jupyter notebook into native python codes for batch executions.

# Convert a jupyter notebook into python code (.ipynb becomes .py)
$ jupyter nbconvert --to python cell2location_cpu.ipynb


#Put the following in a SLURM script and submit to SLURM to run as batch job

ml miniconda/py39/4.10.3
conda activate /share1/cell2location/0.0.6
python cell2location_cpu.py