Hi,
I'd be grateful for some help. I've just built a Server 2012 (not R2) cluster on our EqualLogic SAN. It all went well at the start; on the hosts I had a LAN team (2-port), CSV team (2-port), live migration team (2-port) and 2 x individual iSCSI ports, both on the same VLAN and subnet. This worked fine apart from doing the cluster validation, which complained that both host NICs were on the same subnet. I did some digging around and found that actually, Microsoft do recommend 2 different subnets if possible. The hosts were also running the Dell Host Integration Tools for MPIO.
I reconfigured the hosts, EqualLogic units and switches so I had 2 different VLANs with different IP subnets. Clustering was now happy, but the EqualLogics really weren't- they lost all of their MPIO and were complaining that the SAN group couldn't ping some hosts (which is rubbish, because I tested this from the CLI and the group could ping both subnets).
Anyway... I've just put everything back to how it was (i.e. a single VLAN with a single IP subnet) and the EqualLogics are happy again- each host NIC has one connection to one of the EqualLogic iSCSI ports, making a total of 4 connections for each LUN- but, as expected, the clustering is now complaining that the host NICs are back on the same subnet (this is the only real warning brought up by the Cluster Validation test- the others were all to do with insufficient MPIO paths and similar).
Can anyone shed any light on this? Given that everything seems to work fine with a single iSCSI subnet, can I just ignore that cluster warning? Does the MS best practice only apply if using Microsoft MPIO, and that using a third party tool negates the need for this best practice?
Many thanks!