I'm running SCOM 2012 R2 and SCVMM 2012 R2 and I'm using v3.2.7510 of the SCVMM mgmt pack.
I'm getting my self into a loop where I can't seem to monitor all the Hyper-V hosts and my VMM server at the same time. I can only monitor one or the other never both.
I can get to a state where all my hosts are monitored as below

But my VMM server is not monitored.

At this point I attempt to get the VMM server into a monitored state by running the "Flush Health Service State and Cache" task. Within minutes my VMM server is in a monitored state but all my hosts are no longer in a monitored state.

As soon as I flush the cache on any of the hosts above in attempt to get them into a monitored state my VMM server goes back into a not monitored state. And this loop persists ad finitum.
Looking at the OpsMgr log on all the servers involved show no errors or warning whatsoever.
Any ideas.