Having the same issue here. Never experienced NIC hang ups “mid-flight”, but whenever rig is getting restarted (via console, web page, app or by hashrate watchdog) there’s a chance that instead of it rebooting I’ll get a network killer. Not sure if it happens on shutting down process, or on early stages of booting, but at some moment before “tso off gso off” command kick in. Encountered that problem with mobos mentioned in topic and also with GA-Z170X-Gaming 7’s. Recent Mikrotik controlled switch and old 24-port dumb DLinks are somewhat resistant to the problem, so I don’t get entire network incapacitated, but small part that consists of cheap 8-port switches goes down, with LEDs on those switches blinking rapidly (instead of normal semi-random blinking). I wasn’t able to catch anything with a sniffer on affected segment, so I have no idea what affected NIC dumps into network. Guess gonna have to resort to using USB ethernet adapter.
UPD: So I caught the moment of NIC hanging up, it happens on shutdown part of rebooting process, this is the last thing that appears on screen before network symptoms appear. OS hangs up as well. Any ideas?
UPD2: Turns out this TIME_WAIT thing happens unrelated to using e1000e NIC, just experienced it while using USB ethernet adapter.
So when it happens. it just triggers network problem on e1000e controllers. Still not sure whan might cause it. I run TRM with --watchdog_disabled, but sometimes use HiveOS’ hashrate watchdog. Port 65078 seems to be port that hiveos uses to get stats from TRM via API, and it feels it have something to do with stuck TRM processes being unable to close used sockets. Might try using other miner.