I started this process with the simple question: how many jobs are in the queue.
I then realized I wanted to know more... how many have run, how many completed with throwing an exception (or vv), how long did each job take (minimums, maximums, mode and average period).
How long does each job wait before it starts...
Much of this data would be easy to get from the delayed_jobs table; it's clear that the data gets cleaned up.
It seems that if I could hook into that process, I could move the data to another table, and then run some statistics on it before deleting it.
Has anyone else looked doing this kind of thing?