This configuration is not supported (neither by Cern nor by ETM!).
...but it does appear to work :-)
And it is fast too.
Paul