Weird TFTP behaviour preventing reboot of previously configured machines

I recently restored Foreman 1.7 from backed up DB copy and most everything
went fine.

The significant difference in the restore was that the foreman server went
from RHEL 6.5
to RHEL 7.0.

Basically here is what happens. If any Foreman managed machine gets
rebooted it freezes or goes into a
reboot loop after the PXE menu timeout. The last few lines printed on the
screen are:

Booting from local disk…
PXE-M0F: Exiting Intel Boot Agent.

Here's why this is weird.

  1. If I delete said host from foreman, disable the tftp server(xinetd) or
    the DHCP server the host will boot just fine.
  2. If I rebuild the host the TFTP portion works just fine and the host will
    once again freeze in the same spot
    on the first reboot after build.
  3. I see no errors in the journalctl.
  4. Removing the host from foreman and adding it back using 'New host'
    doesn't solve the problem.

What could this be? This affects machines that already have a running OS
and the only place I can see where
something might be misconfigured is in the TFTP template that get sent to
the host because when I remove that
step the same hosts boot into the OS normally.