This smells Ubuntu specific – https://bugs.launchpad.net/ubuntu/+source/xen-3.1/+bug/1167281
I was able to resolve the issue by replacing this ‘msg n’ with this ‘tty -s && mesg n’ in /root/.profile.
However, its hacky little fixes like this that tend to bite you unless you build them into your process and document them well. That said, we respun our cluster with CentOS 6.6 and all is well out of the box.