Processor availability is on a first-come, first-serve basis.
To enable access to the cluster to all users, the following guidelines will be implemented:
On Nagini, users may employ up to 64 cores simultaneously.
On Faradis, users will have access to 32 processors at a time.
Users wishing to use more than the above allotted processors must first notify the systems administrator by email.
Arrangements for special jobs with additional requirements should be discussed with the systems administrator.
Please use qlogin to reach a free compute node for interactive jobs.
Users should run jobs on compute nodes and not on the head nodes.
Processes using excessive resources on the head node may be terminated without prior notice.
Special storage needs exceeding 100GB should be discussed with the systems administrator in advance.
We would like to emphasize that we will continually review usage and accept suggestions to revise policies as necessary.
We look forward to working with you and getting your feedback on the value the facility provides. We will also work with you in preparation of funding requests.
Practical aspects of the implementation of the usage policies:
Do not run any CPU-intensive processes on the head nodes especially Faradis, as we all have to share that one.
Use qsub to submit jobs to queues via scripts
Use qstat -f to check the status of your jobs and monitor their progress.
If you see a high load average (over 5) or low load (below 0.2) on a node, please contact the system administrator.
If you have a jobs that requires special resources (e.g., more than 1 month to complete, large storage space, many parallel processors, etc.) please contact the systems administrator.
If you have any questions please contact: kevin.kelliher@mssm.edu.