HPC Grid Knowledge: Why Walltimes are Important

A maximum walltime for a job is required under all conditions. Jobs submitted without providing a time limit will be rejected by the scheduler. Walltime is required in order to keep the job wait times reasonable.  Without a walltime, resources would be tied up indefinitely by some users leaving other users unable to submit any work. The primary QoS has a maximum walltime of 30 days. The secondary QoS has a maximum walltime of 90 days. The job duration limits are tuned in an effort to get the best value from the hardware we have. This prevents jobs from going defunct and forgotten about by the job owner indefinitely consuming resources that could be used by others. 

A significant portion of the available resources is purchased by grants, departments, individual researchers, etc. Walltime is required to ensure the respective owners have timely access to their hardware.  Users who have been granted permission by the purchaser of specific equipment have higher priority, longer walltimes, and/or other perks for accessing those resources.

This can be specified with the following directive:

-t, --time=<time>

In a job script:

#SBATCH -t, --time=<time>

Acceptable time formats include "minutes", "minutes:seconds", "hours:minutes:seconds", "days-hours", "days-hours:minutes" and "days-hours:minutes:seconds".