In our algorithm, the jobs' effective priority changes with time. So we should periodically run the scheduler even without changes in the database, since some jobs might have become urgent.
I think this is relatively low-prio task since this would matter if we leave the system running for like 6-8 hours without any new jobs coming, and IIRC Florian mentioned the average runtime of tasks is like 6 hours so our algorithm will be re-triggered again.
In our algorithm, the jobs' effective priority changes with time. So we should periodically run the scheduler even without changes in the database, since some jobs might have become urgent.
I think this is relatively low-prio task since this would matter if we leave the system running for like 6-8 hours without any new jobs coming, and IIRC Florian mentioned the average runtime of tasks is like 6 hours so our algorithm will be re-triggered again.