heartbeat+ipfail will keep restarting and won't work

I've run into a very strange problem with heartbeat + ipfail.

For demonstration purposes I want to setup a cluster made of two virtual
machine hosted under vmware 5 using heartbeat + drbd + samba. I use
ubuntu 5.04 with kernel 2.6.10-5-386 on the host and guest machines,
heartbeat version is 1.2.3-3ubuntu1.

Everything goes well, with the exception that when heartbeat is started
at boot time it complains that the ping node is dead (which is not: it
is alive and pingable from the cluster nodes) and a heartbeat shutdown
is triggered. Then heartbeat restarts, complains about the ping node
being dead, and cycles like that forever :-(

The strange thing is that a
        killall -9 heartbeat ; /etc/init.d/heartbeat start
completely solves the problem.
On the other side, giving instead a:
        /etc/init.d/heartbeat restart
won't work.

I tried writing a shellscript that does the
        killall -9 heartbeat ; /etc/init.d/heartbeat start
Well, if I launch it manually from the shell, it works. if I
set it up to be launched at boot time, the trick won't fix
heartbeat's behaviour anymore.

I'm attaching the syslog from the two clusternodes as well as
the config files.

Thank you,

Francesco