Submitting array jobs

When you want to submit a large number of jobs (batch or parallel jobs) to the system at once, please use the array job feature.

To execute as an array job, run sbatch with the -a option like this: -a 1-6:2. The -t 1-6:2 specifies a minimum index number of 1, a maximum of 6, and :2 to apply every other index number, resulting in 1,3,5. These index numbers can be seen at the end of each line in the squeue output as task IDs.

xxxxx-pg@at022vm02:~$ sbatch -a 1-6:2 ./arraytest.sh 
Submitted batch job 744
xxxxx-pg@at022vm02:~$ squeue
             JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
             744_1 parabrick arraytes xxxxx-pg  R       0:03      1 igt010
             744_3 parabrick arraytes xxxxx-pg  R       0:03      1 igt010
             744_5 parabrick arraytes xxxxx-pg  R       0:03      1 igt010

The task ID can be referenced within the job script through the SLURM_ARRAY_TASK_ID environment variable. This allows each task within the array job to perform different operations based on the value of SLURM_ARRAY_TASK_ID.

Tasks start executing as soon as a computer becomes available (they may not necessarily run in parallel). If computers are available, tasks can run in parallel, but you can limit the maximum concurrency with a percentage notation.

Other examples:

• sbatch -a 1-10%5 job_script.sh

  • Submits a total of 10 tasks from $SGE_TASK_ID 1 to 10. If computers are available, it sequentially executes up to 5 in parallel.

• sbatch -a 10 job_script.sh

  • Submits a single task with $SGE_TASK_ID of 10.

• sbatch -a 2-10:2 job_script.sh

  • Specifies skipping every other index number starting from 2 to 10 (task-ids 2,4,6,8,10).

When submitting an array job, the following environment variables are available for use.

Slurm Environment Variables for Array Jobs

The environment variables accessible from within an array job include: