Is there a way to reliably keep Windows Server domain member servers talking to domain controllers on Amazon EC2 if the instances are stopped/started?
We have a number of testing servers on EC2 that are joined to Active Directory domains. Right now the way I have these working is that they all belong to the same security group, and then I have the DNS Server on each member server set to the Amazon-DHCP assigned internal IP address of the domain controller.
The issue here is that if you stop the domain controller and start it back up, the system does not necessarily receive it's old DHCP-assigned IP Address lease, and receives a new IP (understandably so) In turn, the member servers have no idea how to find the Active Directory domain as name resolution fails.
Setting up an elastic IP for the domain controller, then setting the DNS server to that value does not work, as my understanding is traffic functionally goes "outside" and then back "inside" as it traverses the public IP, and thus I'd need massive ranges of ports opened up on the security group (bad idea I think ...)
Is there any kind of way that a system can find another EC2 instance (by instance ID), then set your instance's DNS Server to that other instances current internal IP Address, or another method by which my Active Directory communication can persist across instance starts/stops?