Information on using the LSUHSC-NO clusters
Getting Started
LSUHSC-NO operates two High-Performance Computing clusters — Tigerfish and TigerShark. Both are Linux-based clusters accessed via Secure Shell (SSH) and are behind our institutional firewall. Both are available to all LSUHSC-NO researchers and research partners free of charge.
To request an account on either cluster please email dmadir@lsuhsc.edu with your name, PI name, department, and a brief description of your research needs.
If you are on campus, you must be connected to the LSU-Secure wireless network or an on-campus Ethernet port. If you are off campus you first connect to the Ivanti Secure Access Client VPN. Once connected open your Terminal (Mac or Linux) or MobaXterm (Windows) and enter:
ssh username@tigerfish.lsuhsc.edu - for Tigerfish
ssh username@tigershark.lsuhsc.edu - for TigerShark
You will be prompted to enter your password. The first time you log in use the password provided when your account was created. After your first login please change your password immediately using the command passwd. Use the command exit to disconnect.
Running Jobs
Your initial login places you on the head node. This node is responsible for allocating cluster resources. You should not run any compute intensive jobs on the head node as this will overburden it and degrade performance for everyone.
Both clusters use the SLURM scheduler to manage computing resources. All jobs must be submitted through the srun command for interactive jobs or sbatch command for batch jobs. These commands place your job in a queue until resources are available.
When submitting a job you specify a partition based on your resource needs. On Tigerfish the default partition is node, with himem for large memory jobs and gpu for GPU jobs. On TigerShark the default partition is compute, with lowmem for standard jobs, himem for large memory jobs up to 1.1 TB, and gpu for GPU and deep learning jobs.
Resources for learning to use the SLURM scheduler are available on the Resources page.
Appropriate Cluster Behavior
Usage: Both clusters are shared institutional resources and should only be used for work-related activities. Protected health information is strictly prohibited from being stored on either cluster. You should only use de-identified research data and never place any PHI on the clusters.
Data Storage: These clusters are computational resources, not data storage servers. Please delete large data sets after computing activities have been completed. Users consuming excessive storage will be warned and may be subject to having data deleted or disk quotas enforced.
Data Persistence and Backup: Data stored on these clusters is not backed up and could be lost at any time due to hardware failures. You are responsible for your own backups. It is strongly recommended that you transfer your results back to your own lab storage once analyses are completed.
Be a Team Player: We rely on our user community to be good citizens. Please choose sufficiently complex passwords and do not become a resource hog. Account usage will be monitored and misbehaving users will be warned prior to having restrictions placed on their account or credentials revoked.