Im configuring centos with load balance with ip route and ip rule

Eth0 192.168.1.5
Eth1 192.168.5.128
[root@host1 testscripts]# ip route
192.168.5.0/24 dev eth1  scope link  src 192.168.5.128
192.168.1.0/24 dev eth0  proto kernel  scope link  src 192.168.1.5
169.254.0.0/16 dev eth1  scope link 
[root@host1 testscripts]# ip route show table wan2
default via 192.168.5.2 dev eth1  src 192.168.5.128
[root@host1 testscripts]# ip route show table wan1
default via 192.168.1.1 dev eth0  src 192.168.1.5
[root@host1 testscripts]# ip rule show
0:      from all lookup 255
32762:  from 192.168.5.128 lookup wan2
32763:  from all to 192.168.1.5 lookup wan1
32765:  from 192.168.1.5 lookup wan1
32766:  from all lookup main
32767:  from all lookup default

ping –I 192.168.1.5 www.google.com

I'm not an expert on this product but I might write a few things that might help.

With this form of load balancing (sometimes called clustering), each real node has a fixed ip address that doesn't ever move. These are the REAL ip addresses. There is a VIP (virtual ip address) that all the clients use that is the address which, in the event of a failure, switches from one real box to the other. The clients aren't aware of the real IP addresses, only the virtual one.

I also notice that Centos users often configure a 'director' service on a third separate machine but I don't believe this is mandatory.

Hope that helps. Perhaps a Centos expert will contribute to this thread.

Anyway, you need to configure a VIP. Search Google if you need help.

Pinging one of the real ip's will only ever get a response from that particular box.

The problem is solved.
resolv.conf was pointing to second interface dns ip. That’s why packets travels to second interface .