Skip to content

Job Priorities

In our computing cluster, 3 levels of job priority have been established to help organize the workload: idle, low, and high. By selecting a priority level, you are essentially queuing your task in a way that reflects its urgency or resource demands. Here's a breakdown:

  1. idle : you can submit up to 100 idle jobs per user; these will bear the lowest priority score, hence are ideal for non-urgent or resource-light tasks, as they'll be processed whenever the system resources are less occupied.

  2. low : you can have up to 8 low jobs per user in the queue ; these have an intermediate priority score ; they're a balanced choice for tasks that are somewhat urgent but not entirely critical.

  3. high : this is the top-tier level where you can submit only 1 job per user, but with the highest priority score ; it's suited for urgent or resource-intensive tasks that need to be attended to promptly.

In essence, you get to choose between these three priority levels based on the urgency and the resource demands of your tasks, helping to ensure that critical jobs get the necessary resources and are completed in a timely manner.

How to specify a priority level

When submitting job, you can add the --qos flag to specify what priority level you want to attribute to that job:

#!/bin/bash
#SBATCH --qos=low
#SBATCH --ntasks=1
#SBATCH --gpus-per-task=1
#SBATCH --cpus-per-task=4
#SBATCH --mem=5G
#SBATCH --time=4:00:00
#SBATCH --container-mounts=/data/bodyct:/data/bodyct \
#SBATCH --container-image="doduo1.umcn.nl#uokbaseimage/diag:tf2.10-pt1.12-v1"

See job priorities when running squeue

  1. append the following line to your ~/.bashrc file : export SQUEUE_FORMAT="%.18i %.9P %.8j %.8u %.2t %.10M %.10q %.6D %.20R"
  2. reload your bash settings via source ~/.bashrc

From then on, you should see job priorities when runnning squeue.