Server resources maxed, accounts flagged several time 😬🤦, server admins mildly disgruntled as I whimper up and ask... please can I be unsuspended again?
Outcome is, I've refactored a bunch of processes to allow for more granular control. I can now choose how often and how many items to process from the inbox & outbox queue's, choose when to run maintenance and data integrity scripts (which are now separated from ongoing cron job needs), plus view a very loggy log of the processed queue ( UI tweaks coming around that for sure. )
I also implemented a light migration, so that if I make some sort of data changes ( which is very often as I uncover more and more with ActivityPub ), it will only run once, and only once, during a deployment.
Deploy often; break things as I go. That's what personal experiments are all about.