Ubuntu: Multiple NICs, one used only for Wake-On-LAN
Posted
by
jcwx86
on Server Fault
See other posts from Server Fault
or by jcwx86
Published on 2012-03-20T09:58:06Z
Indexed on
2012/03/20
11:31 UTC
Read the original article
Hit count: 340
This is similar to some other questions, but I have a specific need which is not covered in the other questions.
I have an Ubuntu server (11.10) with two NICs. One is built into the motherboard and the other is a PCI express card. I want to have my server connected to the internet via my NAT router and also have it able to wake from suspend using a Magic Packet (henceforth referred to as Wake-On-LAN, WOL). I can't do this with just one of the NICs because each has an issue - the built-in NIC will crash the system if it is placed under heavy load (typically downloading data), whilst the PCI express NIC will crash the system if it is used for WOL. I have spent some time investigating these individual problems, to no avail.
My plan is thus: use the built-in NIC solely for WOL, and use the PCI express card for all other network communication except WOL. Since I send the WOL Magic Packet to a specific MAC address, there is no danger of hitting the wrong NIC, but there is a danger of using the built-in NIC for general network access, overloading it and crashing the system.
Both NICs are wired to the same LAN with address space 192.168.0.0/24. The built-in ethernet card is set to have interface name eth1 and the PCI express card is eth0 in Ubuntu's udev persistent rules (so they stay the same upon reboot).
I have been trying to set this up with the /etc/network/interfaces file. Here is where I am currently:
auto lo
iface lo inet loopback
auto eth0
iface eth0 inet static
address 192.168.0.3
netmask 255.255.255.0
network 192.168.0.0
broadcast 192.168.0.255
gateway 192.168.0.1
auto eth1
iface eth1 inet static
address 192.168.0.254
netmask 255.255.255.0
I think by not specifying a gateway for eth1, I prevent it being used for outgoing requests. I don't mind if it can be reached on 192.168.0.254 on the LAN, i.e. via SSH -- it's IP is irrelevant to WOL, which is based on MAC addresses -- I just don't want it to be used to access internet resources.
My kernel routing table (from route -n) is
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
0.0.0.0 192.168.0.1 0.0.0.0 UG 100 0 0 eth0
169.254.0.0 0.0.0.0 255.255.0.0 U 1000 0 0 eth0
192.168.0.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
192.168.0.0 0.0.0.0 255.255.255.0 U 0 0 0 eth1
My question is this: Is this sufficient for what I want to achieve? My research has thrown up the idea of using static routing to specify that eth1 should only be used for WOL on the local network, but I'm not sure this is necessary.
I have been monitoring the activity of the interfaces using iptraf and it seems like eth0 takes the vast majority of the packets, though I am not sure that this will be consistent based on my configuration. Given that if I mess up the configuration, my system will likely crash, it is important to me to have this set up correctly!
© Server Fault or respective owner