upgrade options page on graceful failover is dangerously incorrect

The upgrade options page seems to be written at way too low a level. For instance, the "graceful failover" bit says that GSI indexes aren't perturbed, but that assumes that you don't need to upgrade the nodes where they live. Realistically, when a user wants to upgrade a cluster, they want to upgrade the WHOLE cluster. Only if someone clicks through on the link for graceful failover would they see that other services may have negative impact.

Then there's "Option #2". While it's possible to use XDCR to move data to another cluster, running the same application servers on the other cluster simultaneously is nearly impossible. Any connectors to other systems would need to be reconfigured. Then there's the fact that XDCR is relatively resource intensive for a cluster, so you probably can't just add that to an existing cluster.