Started working on something only to find a performance issue in a @pixelfed background job that warms redis sets with follower ids, those with many followers cause the job to timeout and stuff like this is hard to catch until you reach a certain scale.

Love solving these and I can’t stress enough to other fedi devs, there is no such thing as premature optimization when you build something that will integrate with an ecosystem with millions of users. Plan your database schema with this in mind!