No idea how this akkoma instance is still standing.

How do people handle rebooting these large services to minimize downtime? Just crank the reboot quickly?