FAQ

1. I am new to Cheaha, what is the best way to get started?
2. How do I access my home directory on Cheaha from the desktop?
3. How do I request Cheaha technical support?
4. What is our standard operating procedure (SOP) for storage on Cheaha?
5. What classes of storage are defined on Cheaha?
6. What is the Service Level Agreement for servers hosted with UAB IT?
7. How to check disk quota on Cheaha?

Here are a few suggestions:

• Request a new cluster account by following these instructions.
• Familiarize yourself with the Quick Start documentation.
  • Especially regarding SLURM, which is how you submit jobs to compute nodes.
  • And Environment Modules, which is how you access installed software (including software packages comprising many common scientific and bioinformatics pipelines).
• Sign up for the hpc-announce mailing list.
  • PAIRS partners have access to the #hpc channel in our BioITX Slack team workspace, where traffic on this mailing list is posted automatically for your convenience and to facilitate discussion or answering questions.
• Attend training sessions offered monthly by Research Computing.
  • Selected topics from 2018 included:
    • May 2018
      • Logging on to the new system, migrating data, and the SLURM queuing system.
    • Jun 2018
      • Git basics and parallel programming scripts
      • Open session for the HPC users to ask questions,
    • Jul 2018
      • Git basics and advanced parallel programming scripts.
      • Open session for the HPC users to ask questions,
    • Aug 2018
      • Parallelism using array jobs and bash scripting as well as ways to run your Matlab and R code in parallel.
    • Sep 2018
      • Logging on to the new system, migrating data to/from Cheaha, and the SLURM queuing system.
      • Q&A session, where users can ask about the difficulties they are facing on Cheaha.
    • Oct 2018
      • Virtual environments for Python and Anaconda, Using MATLAB with GPU.

Home directories on Cheaha are available via SMB share, which can be mounted to the file system.

For example, on a Windows computer, you can Map Network Drive to \\files.uabgrid.uab.edu\<blazerid>.

I suggest an e-mail to SUPPORT@LISTSERV.UAB.EDU, which is the support request mailing list for UAB HPC, monitored by Research Computing.

This has an advantage over an e-mail to askit@uab.edu. While both will create a ticket in ServiceNow, if you e-mail SUPPORT@LISTSERV.UAB.EDU, the ticket will be routed directly to Research Computing, which should eliminate the risk of delays due to mis-routed ServiceNow tickets.

For your labs, I suggest requesting "project storage". Here are conventions we are following:

• Directory name of $SHARE_PROJECT/<foo>lab, where foo is investigator name or some other moniker.
  • Examples
    • $SHARE_PROJECT/biokdd-lab
    • $SHARE_PROJECT/chonglab
    • $SHARE_PROJECT/daylab
• Unix group of same name.
  • Lab members are added to this group.
• Unix permissions on directory of designated investigator as owner and lab Unix group as group.
  • Examples
    • drwxrws--- 3 jakechen biokdd 512 Apr 24 2017 biokdd-lab
    • drwxrws---+ 5 zchong chonglab 512 Nov 29 16:06 chonglab
• The default quota is 50 TB.

We have also established project storage at $SHARE_PROJECT/bioitx.

• We host the public data described at Hosted Data here.
• Access restricted to bioitx Unix group members.
• Partners are added to this Unix group as part of welcome package when joining PAIRS.

See https://docs.uabgrid.uab.edu/wiki/Cheaha_GettingStarted#Storage for official documentation.

On the UAB IT Server Management page in ServiceNow, under the Included tab, there is a link to a Service Level Agreement PDF document.

The mmlsquota command (and related mmfs tools) is no longer in your PATH by default so you will either need to add /usr/lpp/mmfs/bin to your PATH variable or use the full path.

Here are a couple examples:

• /usr/lpp/mmfs/bin/mmlsquota --block-size=auto -v -j bioitx data
  • Returns quota for the "bioitx" Fileset.
• /usr/lpp/mmfs/bin/mmlsquota --block-size=auto -u $USER data:user
  • Returns user quota.

A big TY to Mike Hanby in Research Computing for his help on this FAQ item.