it was on some temporary AWS service like lambda or something? (We had eight ECS tasks on AWS, all running five instances of our backend), but, regardless logs should be somewhere persistent.

If they weren't, that should be the first thing you fix.