I've tested this with a late 7.0 or maybe an early 7.1 and I think there is no big difference to the current upstream
So if the DC, or its physical system, crash.
-> all server instances configured on the DC are dead
-> each HC will write periodical messages that the DC can not be found
-> you lost the management of the DC
But your application will still processed!
If the DC is up again all HC's will reconect to the domain and everything will work as before.
What I've not checked is the behaviour if you have changes to one/many of the HC's configuration during the DC is down, but I think this will not be very common