Cman binds to wrong ip address when 2 NICs on subnet with Red Hat High Availability Addon

Environment

• Red Hat Enterprise Linux 6 (With High Availability or Resilient Storage Add-ons)
• High Availability Cluster with 2 or more nodes

• 2 or more network interfaces active on cluster nodes

  • At least 2 network interfaces are on the same subnet. For example, in the following example, eth1 and bond0 are both on 10.0.0.0/24:

    $ grep -B2 -e 10.0.0.XX -e 10.0.0.YY sos_commands/networking/ip_address 
    3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
        link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
        inet 10.0.0.YY/24 brd 10.0.0.255 scope global eth1
    --
    6: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP 
        link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
        inet 10.0.0.XX/24 brd 10.0.0.255 scope global bond0
    

  • One of the interfaces on the same subnet is configured to be used for cluster communication:

    # cman_tool status | grep "Node addresses"
    Node addresses: 10.0.0.YY