e1000e Detected Hardware Unit Hang / Reset adapter unexpectedly

[nitaish]

Hello Team,

I have a Proxmox server 3.4-3 where I am facing a network driver issue. On server eth0 showing error "Detected Hardware Unit Hang" and stop giving network response. To resolve the same I need to reboot Main node.

I start facing this issue today only and it has occurred four time since morning.

I found on other forums such as mentioned below where they mentioned that it could be Network Driver issue :-

http://serverfault.com/questions/19...rking-driver-problems-galore-where-do-i-start

Plz give some suggestion regarding the same.

Many Thanks in advance.

 

[RobFantini]

were there any recent updates done right before the problem started?

If not it could be a hardware issue.

 

[nitaish]

There was no update done. We tried to change the eth port and it worked properly for more than 12 hours. The problem arose again. The network connectivity of the node goes off and after we restart the networking service, the network of the node reconnects, but the VMs cannot connect to the network. Only after rebooting the node, the network connection of the VMs return. The problem is appearing every few hours.

I found the following from the dmesg log

e1000e: eth1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx
vmbr0: port 1(eth1) entering forwarding state
e1000e 0000:06:00.0: eth1: Detected Hardware Unit Hang:
e1000e 0000:06:00.0: eth1: Reset adapter unexpectedly
vmbr0: port 1(eth1) entering disabled state
e1000e: eth1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx
vmbr0: port 1(eth1) entering forwarding state
e1000e 0000:06:00.0: eth1: Detected Hardware Unit Hang:
e1000e 0000:06:00.0: eth1: Detected Hardware Unit Hang:
e1000e 0000:06:00.0: eth1: Reset adapter unexpectedly
vmbr0: port 1(eth1) entering disabled state
e1000e: eth1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: Rx/Tx

 

[RobFantini]

Is the nic built in or a card?

try adding or replacing the nic is what I'd do.

It seems like an hardware issue.

PS:
In the old days we'd use updated e1000 drivers from intel on a regular basis. When kernel was updated I'd install a newer e1000 module.
However since the hardware was working fine with your current software I do not think an updated e1000 kernel module is worth the effort.