Jobs will be controlled through the batch system using Slurm.
- Node sharing. No node sharing.
- Allocations. No allocation controls.
- Best effort to allocate nodes of same CPU speed.
- Max time limit for jobs will be as outlined in the QOS definitions below.
- Scheduling is set based on a current highest priority set for every job, plus backfill.
- Fairshare boost in priority at user level. Minimal boost to help users who haven't been running recently. Our Fairshare window is two weeks.
- Small relative to time small short jobs are given a boost for length of time in the queue as it relates to the wall time they have requested.
- Reward for parallelism. Set at the global level
- Special Reservations - upon request.
Partition Name Access Accounts Node/core count Memory Features Node Specification apexarch restricted all accounts in protected environment 10/80, 6/72 16384, 24576, 49152, 49451, 98902 c8, c12 aa[001-016] Total 16/152
- Job priorities. Majority of a job's priority will be set based on a quality of service definition
or QOS. The following initial QOSs to be defined:
QOS Priority Preempts Preempt Mode Flags Maxwall apexarch 1000 apex-freecycle cancel 3-00:00:00 apex-freecycle 1 cancel NoReserve 3-00:00:00