Cluster shutdown on web farm when using clustered filesystem

Hi,
I have a number of servers running mongrel cluster all of whom feed off
of a shared filesystem (OCFS2). Each cluster has its own
mongrel_cluster_web2.yml file with distinct PID and Log file names. In
theory they shouldn’t be aware of each other’s presence.

What I have found is that when I want to test a new set of code on one
server and I shut one server’s mongrel cluster down, all of its peers
shutdown as well. Any ideas as to how this is possible?

I am running mongrel 1.1.4 on Centos5.1, fully patched.

Thanks,
Frank