Setting 2 IPs on Ubuntu 12.04 with two network adaptor card

Question

There is a VPS runing ubuntu 12.04.3 server on vmware esxi 5.
It has 2 valid static IPs and two network cards eth0 and eth1. but only one of the assigned ips is pingable from outside. (both ips ang gateway is pigable from inside).
it seems that I receive packets on dead IP, but don't reply them. ifconfig will returns:

eth0      Link encap:Ethernet  HWaddr 00:50:56:03:43:2b  
          inet addr:91.121.247.155  Bcast:91.121.247.155  Mask:255.255.255.255
          inet6 addr: fe80::250:56ff:fe03:432b/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:179 errors:0 dropped:0 overruns:0 frame:0
          TX packets:58 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:20976 (20.9 KB)  TX bytes:7246 (7.2 KB)

eth1      Link encap:Ethernet  HWaddr 00:50:56:05:11:46  
          inet addr:91.121.247.148  Bcast:91.121.247.148  Mask:255.255.255.255
          inet6 addr: fe80::250:56ff:fe05:1146/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:174 errors:0 dropped:0 overruns:0 frame:0
          TX packets:66 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:20181 (20.1 KB)  TX bytes:8317 (8.3 KB)

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:65536  Metric:1
          RX packets:68 errors:0 dropped:0 overruns:0 frame:0
          TX packets:68 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:5884 (5.8 KB)  TX bytes:5884 (5.8 KB)

and ip route:

default via 188.165.247.254 dev eth1 
188.165.247.254 dev eth1  scope link 
188.165.247.254 dev eth0  scope link 

also ip nei :

fe80::12bd:18ff:fee4:5040 dev eth0 lladdr 10:bd:18:e4:50:40 router REACHABLE
fe80::ee30:91ff:fee0:dfc0 dev eth0 lladdr ec:30:91:e0:df:c0 router REACHABLE
fe80::12bd:18ff:fee4:5040 dev eth1 lladdr 10:bd:18:e4:50:40 router REACHABLE
fe80::ee30:91ff:fee0:dfc0 dev eth1 lladdr ec:30:91:e0:df:c0 router REACHABLE

but arp -a returns NO OUTPUT at all.

this is my /etc/network/interfaces

# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).

# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
    address 91.121.247.155
    netmask 255.255.255.255
    broadcast 91.121.247.155
    post-up route add 188.165.247.254 dev eth0
    post-up route add default gw 188.165.247.254
    post-down route del 188.165.247.254 dev eth0
    post-down route del default gw 188.165.247.254



auto eth1
iface eth1 inet static
    address 91.121.247.148
    netmask 255.255.255.255
    broadcast 91.121.247.148
    post-up route add 188.165.247.254 dev eth1
    post-up route add default gw 188.165.247.254
    post-down route del 188.165.247.254 dev eth1
    post-down route del default gw 188.165.247.254
    dns-nameservers 213.186.33.99

the problem is 91.121.247.155 is not pingable from outside. and i don't know why.

Solution

The problem is that you can't have two 'default' gateways by default. How would The kernel know which interface to use if it's being told to use two different interfaces.

A quick search online found a recent post showing how to do it by creating a second routing table.

Without knowing what limitations you may be working with, why use two separate interfaces? From the looks of it, these two IPS are actually part of the same block. Why not just use an alias on eth0 for the second IP? If it's for bandwidth reasons you should consider setting up link aggregation(LACP, set up with the bonding kernel module) so both addresses can take advantage of the extra (virtual) nic. Assuming each virtual nic is connected to a separate real NIC (going to the same switch) you would also get redundancy for both IPs should one of the cables get pulled or a card goes bad on the VMware host.

Comments

[Admin]

I'm surprised any of them work with your configuration. Your default gateway isn't in your subnet (with a netmask of 255.255.255.255).

[Mike]

Why not setup an ip alias and just use a single device? Is there a reason for 2?

[r004]

you answer was illuminating; I guess I should go with IP aliasing and create eth0:0.

[r004]

you answer was illuminating; i would go with IP aliasing and I do eth0:0. Now there is a couple of questions. first what should i do with the eth1; should I just leave it be. the second Q? is how should i write the directives for eth0:0. post-up /sbin/ifconfig eth0:0 91.121.247.148 netmask 255.255.255.255 broadcast 91.121.247.148 post-down /sbin/ifconfig eth0:0 down is it right?

[Admin]

You'll want to search elsewhere for that unfortunately. I spend my time in the RHEL/CentOS world. This howto looks like it would work. As for eth1, if it is connected to the same switch (I would imagine it is if it's vmware); I'd just shut it down for now. If you want to use bonding later you can but if these two virtual interfaces are connected to the same network under VMware, there's not much point in using it unless you need more throughput (and the VMware host actually has a faster uplink like 10Gbit).

[r004]

How to IP Alias in with this condition; I did the above and I got nothing. still 148 is not pingable from the outside.