This seems like a specifically designed system without hard assurances that the 'mail will get through' if I have queries scheduled for e.g. longer than ~a few days - if a client restarts on a regular basis there's no continuity handled by launchd, it's just magic 8 ball 'try asking again', delayed to the point that it makes my server look broken …and I will fully admit I'm holding it rong, but I had majority 7-day query intervals and was getting one out of 4 or half of the results I would have expected, and c++ is not quite easy to follow, control-flow wise, to really understand why it was behaving this way