2 interfaces, same subnet, conflict.

t3kn0lu5t · 02-27-2006, 02:38 AM

I've discovered a problem with my Debian setup. I have a server with two interfaces connected to my router, both interfaces have ips on the same subnet, and when they are both up I can't get pings or ssh back from either IP.. to illustrate.

y.z.149.121 - Router, routing y.z.149.120/nm255.255.255.248
|
|--y.z.149.123 sparkplug - eth0
|
|--y.z.149.122 ts1 - eth1
|--y.z.149.124 ts1 - eth0

After setting all this up and getting back home from the datacenter I noticed I couldn't ssh into either IP. I couldn't get a ping back either.. I was able to ssh into sparkplug, and from there I could ssh into either ts1 IP, and could ping as well. I also noticed if I shut down a single interface on ts1, then everything started working normally for the other interface.

So since I don't know a whole lot about routing I figured that's where the problem probably is. Here is some background info.

/etc/network/interfaces:

Code:

auto lo
iface lo inet loopback

# The primary network interface
auto eth1
iface eth1 inet static
        address         y.z.149.122
        netmask         255.255.255.248
        gateway         y.z.149.121
        network         y.z.149.120
        broadcast       y.z.149.127

auto eth0
iface eth0 inet static
        address         y.z.149.124
        netmask         255.255.255.248
        gateway         y.z.149.121
        network         y.z.149.120
        broadcast       y.z.149.127

and ifconfig:

Code:

eth0      Link encap:Ethernet  HWaddr 00:D0:B7:48:D9:20
          inet addr:y.z.149.124  Bcast:y.z.149.127  Mask:255.255.255.248
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:18 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:1080 (1.0 KiB)  TX bytes:0 (0.0 b)
          Interrupt:9 Base address:0xecc0 Memory:fdfff000-fdfff038

eth1      Link encap:Ethernet  HWaddr 00:B0:D0:46:40:30
          inet addr:y.z.149.122  Bcast:y.z.149.127  Mask:255.255.255.248
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:3493 errors:0 dropped:0 overruns:1 frame:0
          TX packets:689 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:238519 (232.9 KiB)  TX bytes:93726 (91.5 KiB)
          Interrupt:5 Base address:0xec00

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:8 errors:0 dropped:0 overruns:0 frame:0
          TX packets:8 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:560 (560.0 b)  TX bytes:560 (560.0 b)

and the output of route:

Code:

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
y.z.149.120 *               255.255.255.248 U     0      0        0 eth1
y.z.149.120 *               255.255.255.248 U     0      0        0 eth0
default         y.z.149.121 0.0.0.0         UG    0      0        0 eth0
default         y.z.149.121 0.0.0.0         UG    0      0        0 eth1

I don't understand, I'm sure there's just something I'm missing. Any help is very appreciated.

Thanks.

t3kn0lu5t · 02-27-2006, 04:09 PM

Still no takers? Bump

drakebasher · 02-27-2006, 04:54 PM

I don't hink you can have 2 NICs in one box with both on the same sub-net. I haven't seen anything definite to say that it is impossible, but whenever I have seen someone ask about that problem (as I did once), the solution is always the same: put them on different subnets. Or maybe this page has the answer.

t3kn0lu5t · 02-28-2006, 04:09 AM

Well, I ended up solving the problem by aliasing, eth0 for one IP and eth0:0 for the other.

A coworker thought that maybe since I had two default routes that the system wouldn't know how to route outgoing traffic.

I would still like to know why this doesn't work.. just for future reference.

baldy3105 · 02-28-2006, 09:01 AM

Your server is by default operating as a router. A router is an IP network boundary, therefore by definition putting two interfaces into the same subnet is a no-no.

And your co-worker is right. You have two routes for the connected network x.y.149.120 and two default routes. The server will simply use the first route-table match it finds in each case which will lead to unpredictable behaviour as you have found.

If you did this on a Cisco you would get Load Balancing, but on a linux box I'm not sure what happens but I'm not surprised by the behaviour your getting.

The only ways to use two nics to the same network is either bond both nics into a trunk or use one active and one standby to provide redundancy.

t3kn0lu5t · 02-28-2006, 09:39 AM

Thanks, that's what I wanted to know!