Hi. We have Windows 2008 R2 clusters with two vNICs, one for production (NIC0) and the other for the Windows cluster heartbeat (NIC1). Both NICs are set to DHCP in vCD, but the IPs were configured manually in the OS. (This due to the import process of migrating our VMs into the cloud). All of these VMs operate normally, with the exception of the Windows clusters.
In vCD, we noticed that on NIC0, the IP reported for the Windows clusters is the 169.x.x.x that Windows clustering assigns to the virtual "Microsoft Cluster Failover Virtual Adapter". (vCenter reports all three IPs) The binding order in Windows is set to Production, Heartbeat, Failover adapter.
Under these circumstances, NIC0 appears to "flap" in that vCOPS sees NIC0 reporting the production IP address (and less frequently the heartbeat address!), but then it changes the 169.x.x.x address and Windows clustering subsequently sharts-the-bed due to network connectivity issues.
As a workaround, we power down the VM, set NIC0 in vCD to Static-Manual and input the production IP. This seems to overcome the problem without turning on Guest Customization! We had assumed that the network setting in vCD would be irrelevant if set to DHCP, but something is breaking in either vCD or vCenter and we can't seem to pinpoint it. Our best guess is that the 169.x.x.x is getting associated with the MAC of NIC0, but local ARP doesn't reflect this.
The networks for NIC0 and NIC1 are both direct connection externals in vCD.
Any insight into this problem would be greatly appreciated.
Thanks!