PFSense LAN Link Down - Every Once a while

[starchx]

Hi Guys,

Just set up a new PFSense box for our data center production servers.

Not sure why, it was good and running smooth for first 2 weeks, but then the LAN link start to fail once every 2-3 days randomly.

2.0.1-RELEASE (i386)
built on Mon Dec 12 17:53:52 EST 2011
FreeBSD 8.1-RELEASE-p6
IP: 192.168.3.1

When it failed,
1, all internal servers are not accessibly from Public. (of course, their connection to PFSense is down).
2, I tried to ping 192.168.3.1 from one internal server and Times out. That means no connection to PFSense from Internal LAN.
3, I can't access the PFSense Web UI either from internal server

To quick fix this, I simply unplug the LAN network cable from PFSense LAN Network Card and Plug back in. PFSense will print these message below, and then everything back to normal.

The worst thing is I don't see any messages logged in System Log and I can't debug this issue.
Anyone can please help?

Thanks,
tim

Apr 1 14:48:26 check_reload_status: Linkup starting em1
Apr 1 14:48:26 kernel: em1: link state changed to DOWN
Apr 1 14:48:27 kernel: em1: Watchdog timeout – resetting
Apr 1 14:48:27 kernel: em1: Queue(0) tdh = 936, hw tdt = 905
Apr 1 14:48:27 kernel: em1: TX(0) desc avail = 31,Next TX to Clean = 936
Apr 1 14:48:30 check_reload_status: Linkup starting em1
Apr 1 14:48:30 kernel: em1: link state changed to UP
Apr 1 14:48:31 php: : Hotplug event detected for lan but ignoring since interface is configured with static IP (192.168.3.1)
Apr 1 14:48:35 php: : Hotplug event detected for lan but ignoring since interface is configured with static IP (192.168.3.1)
Apr 1 14:48:35 check_reload_status: rc.newwanip starting em1
Apr 1 14:48:35 kernel: arp: 192.168.3.26 moved from 56:0f:55:d7:18:3f to 0e:f7:4a:5e:4d:80 on em1
Apr 1 14:48:39 kernel: arp: 192.168.3.27 moved from 7a:18:15:02:56:4a to 5e:eb:4e:72:8f:7b on em1
Apr 1 14:48:40 php: : rc.newwanip: Informational is starting em1.
Apr 1 14:48:40 php: : rc.newwanip: on (IP address: 192.168.3.1) (interface: lan) (real interface: em1).
Apr 1 14:48:40 apinger: Exiting on signal 15.
Apr 1 14:48:41 check_reload_status: Reloading filter
Apr 1 14:48:41 apinger: Starting Alarm Pinger, apinger(42458)

[cmb]

Is your state count anywhere near the maximum when it happens?

Are you getting ARP replies still, and is the MAC in hosts' ARP cache correct?