We have a mixed cluster of ESXi hosts running 4.1, 5.0 & 5.1 which is in the process of being all upgraded. The issue has been around before the upgrade when the mix was just 4.1 and 5.0s.
A particular server (running Win2K8 R2), losses connectivity to its datamovers IPs. The server still responds to ICMP pings and can successfully ping everything else in and outside it’s subnet. The issue is normally resolved by VMotioning the server again to another host in the cluster (doesn’t seem to matter what version of host) then ICMP connectivity to the DMs is restored which are on a different subnet. We’ve replaced the Vnic with the same E1000, no change and then replaced with a VMXnet3 which worked until a Vmotion then lost connectivity completely to the server. Only a shutdown and replacing the vmxnet3 with the E1000 and then another vmotion resolved the issue.
A workaround is to add a 2nd NIC in the datamover VLAN to the server and then it works fine but wish to find a solution to the problem and run with 1 nic and 1 IP address.
Any ideas?