Iptables – Secure IPTables rules for Corosync

I have two HA load balancers (hollywood and wolfman) running Corosync and Pacemaker. The eth1 interfaces are connected to the WAN, and the eth0 interfaces to the LAN, using a virtual IP as the gateway for the back end servers. The eth1 IP of hollywood is xxx.xxx.195.45, and the eth1 IP of wolfman is xxx.xxx.195.46. The bindnetaddr in Corosync is xxx.xxx.195.32, the same as the WAN's network address, and the Corosync port is the default 5405.

The relevant IP tables rules on both servers are:

*filter

--flush

:INPUT DROP

--append INPUT --protocol udp --destination-port 5404 --jump ACCEPT
--append INPUT --protocol udp --destination-port 5405 --jump ACCEPT

This setup seems to work fine, but initially I added --in-interface eth1 and --source xxx.xxx.195.46 to wolfman, and --source xxx.xxx.195.45 to hollywood. Most of the time this seemed to work, but rebooting the passive balancer sometimes killed communication between the load balancers, writing these errors to syslog:

[TOTEM ] Totem is unable to form a cluster because of an operating
system or network fault. The most common cause of this message is that
the local firewall is configured improperly.

So it seems that either my simplistic belief that all the Corosync traffic is directly between the two load balancers over eth1 is wrong, or that something else is causing a problem.

I'd like to lock port 5404/5405 down in IPTables to just the cluster. What do I need to do to make this happen?

Edit: corosync.conf as requested. This is all default Ubuntu other than the bindnetaddr.

# Please read the openais.conf.5 manual page

totem {
        version: 2

        # How long before declaring a token lost (ms)
        token: 3000

        # How many token retransmits before forming a new configuration
        token_retransmits_before_loss_const: 10

        # How long to wait for join messages in the membership protocol (ms)
        join: 60

        # How long to wait for consensus to be achieved before starting a new round of membership configuration (ms)
        consensus: 3600

        # Turn off the virtual synchrony filter
        vsftype: none

        # Number of messages that may be sent by one processor on receipt of the token
        max_messages: 20

        # Limit generated nodeids to 31-bits (positive signed integers)
        clear_node_high_bit: yes

        # Disable encryption
        secauth: off

        # How many threads to use for encryption/decryption
        threads: 0

        # Optionally assign a fixed node id (integer)
        # nodeid: 1234

        # This specifies the mode of redundant ring, which may be none, active, or passive.
        rrp_mode: none

        interface {
                # The following values need to be set based on your environment
                ringnumber: 0
                bindnetaddr: xxx.xxx.195.32
                mcastaddr: 226.94.1.1
                mcastport: 5405
        }
}

amf {
        mode: disabled
}

service {
        # Load the Pacemaker Cluster Resource Manager
        ver:       0
        name:      pacemaker
}

aisexec {
        user:   root
        group:  root
}

logging {
        fileline: off
        to_stderr: yes
        to_logfile: no
        to_syslog: yes
        syslog_facility: daemon
        debug: off
        timestamp: on
        logger_subsys {
                subsys: AMF
                debug: off
                tags: enter|leave|trace1|trace2|trace3|trace4|trace6
        }
}