Hyper-V Host: Dell PowerEdge R620 Server with 8 NIC, 256GB RAM connected to iSCSI SAN running Windows 2008 R2 SP1
Guest Machine(s): Windows 2008R2 or 2012 servers (total of 8 servers)
Symptom: Windows 2008 R2 server drops the network connection. Cannot PING or access the server. Tried rebooting guest server, but still not able to connect. So changed the network connection via Hyper-V Management console to another virtual adapter, and the system comes back online immediately.
Logical thinking: The NIC must be bad. But the problem continued at random times (like once or twice a week - especially in the morning with heavy network traffic load). So, I switched the network back to the original Virtual NIC, and the system runs fine.
So it seems that the Hyper-V randomly "kills" the NIC when it is utilizing heavy I/O load. And only way to fix it at this time is to swap the virtual NIC (5 sec but need to do manually). Tried add another NIC to provide additional bandwidth, but same problem exists.
Is there some sort of patch available to address this? Or maybe registry update to IGNORE the heavy I/O on a Guest OS?
I looked into http://support.microsoft.com/kb/2263829/en-us but this seems to be a different issue as the article states that the machines comes online when rebooted where as that is not the case on my part.