Interesting... I had auto-deploy running in a lab as a POC with 5.0. No issues to report and we decided to move forward with it.
In the meantime 5.1 was released with cached mode, the prod environment was built with 5.1. The first cluster that I converted had local disk so I turned on cached mode... I have been frustrated since then... My major problem is inconsistancy/ odd behavior and so on...
I had my case escalated with VMware and now have a pretty engineer on the case and we have daily calls.
Reading your thread I wonder if most of the problems are not related to cached mode. and regarding the reboot message. VMware has isolated the settings that are modified that are generating this error message. In our case they are the Net/TCP values recommended by NetApp for NFS.
Another observation is that the settings sticks to the correct value between reboots... So it seems that there is a flag on these settings to say a reboot is required and that is what is causing the error
Amusing but if I recall correctly, applying these settings through the netapp VC plug-in never requires a reboot.
G.