New node hardware can't easily, if at all, be added to existing cluster, which is running an old OS.

There are several possible approaches related to this option, but only this first one seems "best"

1A: Install new OS on Lancaster's cluster. Add Crane's new nodes to that cluster.

This is very similar to the work we are (8/2013) doing with Collum's cluster. Takes a lot of time and coordinating, but is mostly a known quantity.
If choose to do this, how and when to fit this in, since can't do immediately.

Other, "crazier" ideas under this general option:

1B: Buy compatible (old? eBay?) nodes.

May be more attractive if can obtain more nodes, even though they individually would be less powerful and already worn through their warranty and obsolescence cycle.

1C: Invest time and staff resources to continue trying to build a capsule which would work with current, old OS