On 1/24/22 11:17 am, Tom Harvill wrote:
We use a 'fair share' feature of our scheduler (SLURM) and have our decay half-life (the time needed for priority penalty to halve) set to 30 days. Our maximum job runtime is 7 days. I'm wondering what others use, please let me know if you can spare a minute. Thank you!
We use Slurm but we don't use fairshare, instead we have a priority threshold which jobs have to age to before they can get a forward reservation on nodes (they can of course backfill before then). We configure things so that jobs age at ~1 priority point per minute and then set our QOS's so that the start time is a certain amount of time away from that threshold.
We also set things up so that only 2 jobs per user+account+qos association can age, and once one starts running the next in line will begin ageing.
All the best, Chris -- Chris Samuel : http://www.csamuel.org/ : Berkeley, CA, USA _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit https://beowulf.org/cgi-bin/mailman/listinfo/beowulf
