Tom,
While an HPC -related topic (and some that aren't HPC-related) are fair
game in this forum, have you considered posting this question in the
slurm-users mailing list?
https://slurm.schedmd.com/mail.html
While I'm sure there's significant overlap between the two mailng lists,
might get additional feedback there.
The max time limit on my cluster is 48 hours, and the half-life is 15 days.
I personally thing your halflife should be 2-4x as long as your max.
time limit to give the scheduler a fair chance of equalling out cluster
access. If you have a SLA that dictates when fairshare will be evaluated
(in a monthly report to your beancounters, for example), I think your
half-life should be the same size or smaller. I have no rigorous
mathematical basis for that, or any though experiments to justify it. If
someone else has done that, or know of papers (including vendor
whitepapers) about it, please share!
Prentice
On 1/24/22 2:17 PM, Tom Harvill wrote:
Hello,
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!
Tom
Tom Harvill
Holland Computing Center
University of Nebraska
_______________________________________________
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
_______________________________________________
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
