First, here is some information that may or may not be relevant:
Host Server
HP Proliant DL120 G7, quad-core, 16GB RAM, 250GB OS drive, 2 1TB data drives in a mirror. (VM guests reside on the mirror)
NIC = HP NC-Series Intel E1q Gigabit Ethernet, driver current.
OS = Server 2008 R2 x64, Enterprise. Roles = Hyper-V, all updates current + SP1 + microsoft update.
SERVER IS HOSTED AT A REMOTE DATACENTER.
Guest Servers
Server 2008 R2 x64, Enterprise. Primary Role = Active Directory/Domain controller, DNS, all updates current + SP1 + microsoft update, CONNECTIVITY PROBLEMS.
Server 2008 R2 x64, Enterprise. Exchange 2010 installed, all updates current + SP1 + microsoft update, joined to domain, CONNECTIVITY DEPENDENT ON DNS.
Server 2008 R2 x64, Enterprise. Exchange 2010 Edge Transport Role installed, all updates current + microsoft update, not joined to domain. External DNS. NO CONNECTIVITY PROBLEMS.
NO GUEST VMs ARE IN PRODUCTION ENVIRONMENT YET. NO NETWORK LOAD IS PLACED ON THEM ASIDE FROM RDP SESSIONS FROM MYSELF.
Issue Breakdown
Upon installing and configuring guests, connectivity is fine. They can access the internet and the internal network. Upon leaving and returning to work, the DC guest has lost internet connectivity, which in turn causes the Exchange guest to lose connectivity,
because it uses the DC for DNS. RESTARTING THE DC GUEST VM DOES NOT RESTORE CONNECTIVITY. Changing connectivity settings for the guest does not restore connectivity. Restarting the host server does not restore connectivity. Reinstalling the
host NIC driver does not restore connectivity. Removing the Hyper-V role and re-installing that role does not restore connectivity. Attempting to install the KB in the host that everyone says to install results in the error message stating "...does not apply
to this server". I don't think that KB applies in general anyway, as the network load on the host and guests is negligible.
I disabled offloading, to no effect. And, having read another article on here about an hour ago, which suggested switching from auto-negotiation to 1gig full duplex, I tried that (unfortunately) and the network connection went down, which I need for RDP,
so now I'm spending money on a support case for our Datacenter techs to log into the server and change it back to auto-negotiation. So, while that's in the works, I can't do anything else on the server until I can establish RDP again, or drive a few hours
and access the server directly.
I've been having this issue for the past 3 days, and have been researching for as long, and every solution I've tried, short of going to the server and installing a new NIC, has failed. Is there something I'm missing? Or do I really have to put a new NIC
in? (I'm also aware that it's better to have dedicated NICs for my VMs, but it's not an option at this point. I'd like to first be able to confirm reliable operation of the system as a whole before I make further purchases.)
Any thoughts?