Hecate Usage Guidelines

• All jobs will be submitted using the PBS batch queuing system.
• The system uses the wallclock time to calculate the scheduling. The default time is 1 minute.
• The maximum wall clock time that can be specified is 15 days (360 hours). Jobs requesting more than this will be rejected.
• Any jobs requiring more than 4 GBytes of memory per CPU must allocate extra CPUs to cover this memory requirement as there is only 4 GB/CPU. For example, if there were a single CPU job requiring 6GB, then allocation should request ncpus=2.

Queuing System

Hecate uses the OpenPBS/Torque queuing system with the Maui scheduler.

The current setup allows up to 64 CPUs to be allocated to a job. Actual job allocations are prioritized on two factors: 1) the amount of the total machine a group owns, and, 2) the length of time requested for runs, with shorter jobs given higher priority.

For the fairshare ownership priority, effective usage is calculated over the past 8 days and job priority is adjusted to move waiting jobs up or lower in priority based on this level of usage. This does NOT mean that a group's inactivity will allow jobs to run immediately, just that when considering which job to run next that those jobs are adjusted accordingly.

Home Directories and Scratch Space

There is a quota on the home directory. View this with the command "quota -v" to determine how many KB are in use and where the limits are currently set.

Scratch space is available in /scr0 and /scr1. This space is a shared space. Files are NOT backed up so move any important files to long term storage (your home directory, another machine). Also note that, someday, these scratch directories will be cleaned nightly to purge files older than 15 days.

Please remember that these are shared resources for all users.

Last change Mar-09-2007 wgw