this is something I’ve been battling with. Although the scheduler has 10 threads, only 1 thread works on delayed_jobs, and some of those delayed_jobs take 10 seconds to process. So 1700 delayed jobs with no extras being added might take half an hour to clear.
I’ve been documenting the things i’ve tried here
[SOLVED] Zammad is hammering elasticsearch to reindex the same 10 users (it’s not really solved, I should change the subject line)