So the other day I discovered a client were having issues with their Dell blades where connection was being dropped between a vCenter Server (virtualised) and the ESXi host it was residing on….
It looks like traffic gets dropped if the VC and the Mgmt port of the ESXi host uses a N-Par on the same physical LOM.
From the only article I’ve found online, it seems it’s due to how ARP works…. (broadcast to all ports apart from itself)
http://communities.vmware.com/thread/341266?start=0&tstart=0
I wonder if this is because N-Par gets configured in the server BIOS and so gets presented out to the PowerConnects as 8 MACs over 1 physical port?!?
I don’t ever remember encountering this issue with HP Flex10 or FlexFabric…… and I believe this could be because the NIC partitioning happens on the Virtual Connect modules rather than on the actual blades itself……
Anyways, if anyone out there has a solution (me being non-network savvy) feel free to drop a reply!