This is a bug that is deeply routed in starmades netcode, schema has confirmed it for us.
The issue is that the networking thread locks up when a single client fails to respond (such as on logins, logouts, when people experience minor internet dropouts). The server will then sit and wait until they respond, everything running like physics and factories, except the bit that talks to clients. when it affects 1 client, it affects all of them. This has been around for along time, but only since a recent patch has it become a problem (I assume schema made a change to the netcode somewhere). Understand that we play on the server too, we are just as pissed as anyone else, but we have taken every step we can up to having a script check the server is still responding, and if not, zip the logs and a thread dump and send them to schema.
Every server see's it, some more than others, depends on a number of things. there is nothing we can do about it. And I'm sure gravypod and the rest of us would like to see this thread/video removed.