2

2013 Mac Pro. Using both built-in ethernet ports for 2 separate networks, one a Storage LAN and one an Internet connection. On heavy disk I/O (Disk First Aid, array to array copies, etc) the Storage LAN interface just stops working after a while. No framing errors are logged, the routing table is unaffected (until the arp cache times out), the interface counters increment when I ping, and there are no logged events, but all ARP entries disappear and packets never leave the interface. The Internet interface (which was lightly loaded) is fine. Weirdly, even unplugging and replugging the cable does not reset the interface, and neither does changing IP, resetting the switch port on the other end, or changing flow control. I used to have to reboot to reset the interface, but now I found that switching from 1Gb to 10G and back clears the problem.

I'm dying here, however, because it is not reproducible reliably, and ChronoSync hangs so bad when the shares disappear that I have to force reboot with the power button (force quit or a kill -9 of the parent ChronoSync process doesn't work), so I imagine some underlying mount process hanging is the culprit here.

High Sierra 10.13.3 (17D47)

  • Please add your system version in the question itself and tag your question respectively... – klanomath Feb 3 '18 at 11:34
  • Next time this incident happen, make a netstat -I interface a netstat -A, then try ifconfig interface down ; sleep 30 ; ifconfig interface up to check if this isn’t safer and faster than the off;on button which is a FS heavy damage. – dan Feb 3 '18 at 12:07
  • I've seen something similar, but on Windows servers - one with the Ethernet on the main board and another that had a quad Ethernet adapter where one port failed. In both scenarios, replacing the faulty hardware solved the issue. In your case, you would have to replace the Port Board Unfortunately, I can't test as I don't have a Mac Pro, but this would be my guess as to what the issue is. – Allan Feb 3 '18 at 13:52
  • @Allan, I do not suspect it is a bad port, as I have found several postings as far back as 2010 referencing the exact same issue. I would not rule out, however, that the port is affected by heat, but there is really no way to tell that with the facilities I have. – Shane Pseudoelvis Guthrie Feb 6 '18 at 12:24

You must log in to answer this question.

Browse other questions tagged .