I'm scratching my head. I recently configured a server with two identical Intel NICs inside a DMZ. This morning at around 8:05, I successfully SSH'd in through the internal interface (via my LAN-based PC) and had access to the outside world as well. When I got into work at 9:30 or so, while I still had external access (eth1), I wasn't able to SSH in, nor could I ping out to the LAN (eth2). Restarting the interfaces or rebooting had no impact.
Beginning at 9:22, every fifteen seconds or so, syslog spits this out ("Virtualization" is the server's hostname):
Jul 4 11:50:46 Virtualization kernel: [ 6239.229261] e1000e: eth2 NIC Link is Up 100 Mbps Full Duplex, Flow Control: Rx/Tx
Jul 4 11:50:46 Virtualization kernel: [ 6239.229380] e1000e 0000:07:00.0 eth2: Link Speed was downgraded by SmartSpeed
Jul 4 11:50:46 Virtualization kernel: [ 6239.229383] e1000e 0000:07:00.0 eth2: 10/100 speed: disabling TSO
Jul 4 11:50:47 Virtualization kernel: [ 6240.024231] e1000e: eth2 NIC Link is Down
lspci -vv shows me that the two Intel NICs are seen by the system as identical in every way.
The cables and connections are all well tested and weren't touched between 8:05 and the time I came in.
Any ideas?
Thanks,
Best Answer
Look for problems with the connecting switch or similar equipment. In my case the connecting port on my Zyxel GS1100-24 switch had suddenly died (for good). I moved the cable over to an unused port in the switch and that made it work.
Repeted message in my kern.log while connected to the faulty switch port (enoch beeing my hostname):