Actions

Fair-share scheduling policy

From ALICE Documentation

Revision as of 13:14, 20 April 2020 by Dijkbvan (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Fair-share scheduling policy

Each user submitting jobs is allocated a share of computing resources. If a given user's jobs have exceeded their share the last 2 weeks, their default job priority will be decreased relative to a user who hasn't consumed their full share for that period. Heavy usage further in the past is also considered but has less impact on job priority (older jobs and usage levels have exponentially less impact than recent ones).

Additionally, as GPUs are a scarce resource, each user is limited to 8 GPU jobs either running or queued. Further GPU jobs are blocked until running GPU jobs complete, allowing other users to queue jobs ahead of the blocked jobs.

Partitions/queues
Partition MaxNodesPerUser MaxJobsinQueue (running and waiting) MaxTime
cpu-short no-limit 100 2 hours
cpu-medium 18 30 1 day
cpu-long 15 30 7 days
gpu-short no-limit 100 2 hours
gpu-medium 8 16 1 day
gpu-long 7 14 7 days
mem 15 30 7 days
test no-limit 4 1 hour
playground-cpu no-limit 1 infinite
playground-gpu no-limit 1 infinite
notebook-cpu no-limit 1 infinite
notebook-gpu no-limit 1 infinite