We're running two servers in a datacentre, call them WEB1 and WEB2 in unicast NLB mode.
Each server has a public NIC, the NLB adapter and a private NIC for communication between hosts. We have a VLAN between the public NICs so the cluster IP is routed to both hosts. I can access the Virtual IP just fine, and hosts can talk on the separate private network. The public NICs are connected up to a Cisco switch with a VLAN between two ports, one for WEB1 and one for WEB2 (all done by the datacentre).
Intermittently on WEB2, I lose connection from my local machine and the Event Logs show multiple:
NLB cluster []: NLB detected duplicate cluster subnets. This may be due to network partitioning, which prevents NLB heartbeats of one or more hosts from reaching the other cluster hosts. Although NLB operations have resumed properly, please investigate the cause of the network partitioning.
WEB1 works just fine and doesn't suffer from these issues. Removing WEB2 from the NLB cluster leaves everything working fine (with just the single WEB1 left).
Perhaps unrelated but for WEB1 & WEB2, the interface metric for the private NIC is 10 while the public NIC is 20 (although 'route print' shows much higher values around 270-300). [seems this is correct according to http://blogs.technet.com/b/tristank/archive/2004/06/08/150734.aspx]
Really hoping someone can help as NLB seems a great solution!
Thanks ever so much