Skip to main content
Print

Intro to ARCC

Workshop video: Intro to ARC resources: HPC computing/analysis tools for research and education

Information on the access and usage of the Advanced Research Computing Cluster (ARCC) High Performance Computing (HPC) Service at the University of Cincinnati.

Off Campus Cluster Access

The ARC cluster is only available from on campus, or off campus using the UC VPN. If you are off campus, you will first need to connect to the UC VPN, and then connect to the cluster as normal.
Read the VPN Instructions.
Please contact us at ARC_INFO@.uc.edu for any further questions regarding the usage of the cluster.

Login and File Transfer

Once your ARC account is created, you will be notified that your access has been provisioned.
Request an account here:

Windows Login

To login on Windows machines, you will need to use an ssh client. UC recommends you use PuTTY which is a free and open-source ssh application. You can download Putty.

Once installed, connect to: arcc.uc.edu

If prompted, verify the SSH fingerprint is valid: 


ARCC.UC.EDU valid SSH key fingerprints

ECDSA key
MD5: 9c:6c:c8:76:a1:f2:ce:89:ca:99:52:1c:dd:9b:08:07
SHA256: F0Ol1ut8DowsTxNDaIDtwtVzqA/TCyTvBWxbNwspROw

ED25519 key
MD5: 51:58:2e:e9:44:08:9a:5e:83:39:85:be:15:68:6d:96
SHA256: jdn5SYwZTCAihsK8V9PrCJvlNiqFl5sx2Uu4e5bo7gY

RSA key
MD5: e8:66:5e:ea:ef:cc:18:a6:34:91:d8:92:bf:70:c4:d2
SHA256: MxRI8uO5oJ0EEJAjk+7pTm7AP/m5NGQZ4QHw+IVuEfg

File Transfer

To transfer files to and from ARCC on a Windows machine, you will need to use an FTP client. UC recommends you use the WinSCP which is a free and open-source FTP client. You can download WinSCP.

location of ARC file

Linux Login

On Linux, you can login using terminal. To do so, enter the following string in your terminal: ssh username@arcc.uc.edu

Linux File Transfer

To transfer files to and from ARCC on a Linux machine, you can use the built in File manager. On an ubuntu-18.04, the ‘Other Locations’ option in the file manager can be used to ‘Connect to Server’ which will mount the user home directory. A similar option should be available for all the different flavors of Linux in the file manager.

Mac Login

On Mac, you can login using terminal. To do so, enter the following string in your terminal: ssh username@arcc.uc.edu

Mac File Transfer

To transfer files to and from ARCC on a Mac, you will need to use an FTP client. UC recommends you use FileZilla which is a free FTP client for Mac.

You can download FileZilla here: https://filezilla-project.org/

Command Line

You can also transfer files to and from your machine using command line. To transfer from your machine to arcc, scp -r /path/to/files username@arcc.uc.edu:/destination/path/in/arcc To transfer from arcc to your machine,

scp -r username@arcc.uc.edu:/path/to/files/in/arcc /destination/path/in/your/machine Note that the ‘-r’ flag stands for recursive and is for transferring directories.

Open Ondemand

After you have logged into the terminal for the first time, you can then use the OnDemand web portal. You can access OnDemand dashboard here: https://ondemand.arcc.uc.edu From here, you can edit, upload/download files, create/view jobs and cluster status. We are currently working on adding interactive apps, Jupyter Notebooks is currently available. 

NOTE: The OnDemand portal is in a early release status, please report any issues to ARC_INFO@uc.edu 

Loading Software into the Environment

To find out the software installed on the cluster, use the following command:

“Module avail”

L- Indicates the module currently loaded

D- Indicates the default

D indicators

To load software into the cluster, use the following command:

“module load software/version”

To unload software, use the following command:

“module unload software/version” SLURM Resource Manager

The cluster uses the SLURM resource manager to manage how jobs are submitted to the compute nodes. Please refer to the following web page to get a basic understanding of the slurm commands.

http://arcc.uc.edu:81

Running jobs using SLURM

Jobs are best run in batch mode. Please use a script to submit your jobs. A typical slurm script is given in the web page and looks something as follows.

#!/bin/bash
# set the partition where the job will run
#SBATCH –partition=low
# set the number of nodes
#SBATCH –nodes=1
# set the number of CPUs to use per node
#SBATCH –ntasks-per-node=1
# set max wallclock time
#SBATCH –time=100:00:00
# set name of job
#SBATCH –job-name=test123
# mail alert at start, end and abortion of execution
#SBATCH –mail-type=ALL
# send mail to this address
#SBATCH –mail-user=user@mail.com

# run the application “appCPU”
appCPU

Please choose your resources (number of nodes, number of tasks, wall time) carefully, as they will impact the job priority, which in turn will effect when the job starts during heavy traffic.

Some users have priority on the “krg” partition of the cluster and jobs running on that partition will get preempted when those priority users run their jobs on those nodes. The jobs that were already running on those nodes will get requeued. In light of this, please checkpoint your jobs frequently, and run your applications in such a way so as to resume from the latest checkpoint, in case of a requeue.

Partition low is the default when a partition is not specified by the user. It is recommended that users not specify the nodes to run their jobs, unless absolutely necessary, and let slurm decide the best location to run the job by specifying just the partition.

Table of Contents