In a situation of DR induction, in which you decide to use 2 Virtual Centers (one production and one DR site, just mirroring storage), the main problem is the use of dvswitch. In fact, precisely the architecture of vSphere, when you go to register on the DR site (even for testing purposes for example) before production mirrored the datastore and then turned off the machines that we are on, they show big problems on the virtual network cards with error "invalid Backin" which means that they do not find references to the new vCenter dvportgroup. And God forbid...
Thanks to the functions created by the LUCD about dvswitch this problem can be overcome, although work is required due to the fact that virtual machines with multiple network cards in PROD should have exactly the same networks allocated on DR.
The script can be improved especially in the csv export side to this: I left in my case because they are regularly scheduled in other scripts ... but if you want you can assign to array you're working command directly without exporting... and then re-importing.
Following is the "scriptone"
