What you're trying to do is more to do with DR and business continuity planning. Sounds like you need to figure out a DR plan, that maps out and defines the order in which you restart services after an outage, and identifies any weak points. Also look at how to better protect those core services during such outages (ie via UPS's, physical DNS servers that are separate to your virtualization environment are just a couple of obvious suggestions). Also look at what you can do to remove single points of failure to make your infrastructure more resilient.

Certainly not putting things in DNS to try and resolve these dependency concerns is not the right way to solve this.  

HTH