Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm looking at putting in a small cluster spanning two locations on the same campus.

The vSphere hosts in each location would have a vSwitch connected to the production LAN, and I'd also be using a physical dedicated iSCSI LAN which would have switches in both locations with dedicated 10gbps fibre between both.

If the iSCSI fibre fails both hosts would be up and able to ping the other, but one host would not be able to see the iSCSI shared storage.

I can't find a guide that details how to configure HA in the situation above.

Thanks in advance.

share|improve this question
up vote 2 down vote accepted

There's nothing to configure. If a host loses its storage then it can't maintain its VMs, at which point if it has VMKernel access to one or more hosts it'll announce itself as down at which point those hosts able to restart the down hosts last known VMs will do so according the the HA plan and policies. Once the downed host gets its storage back it'll rejoin the HA cluster at which point its capable of taking load, either manually or via DRS.

Oh and don't forget that the VMKernel NICs have to be on the same VLAN, but I'm sure you've thought of that.

Hope this helps.

share|improve this answer
Thank you for that, so rough scenario: "vSwitch0, pNICs on main LAN plus service console" and "vSwitch1, pNICs on iSCSI LAN, multiple VMKernels". If the iSCSI link between the sites fails the service consoles on vSwitch0 will still be able to ping each other, but the VMKernels won't be able to ping each other ALTHOUGH the server at the primary site will still be able to access the shared storage (as the SAN has quorum to decide where to failover to). – Hutch Oct 17 '10 at 17:23
Yes, basically that's right, you've got it. Although I'd stick to ESXi if I were you, the SC is going away and totally unrequired. – Chopper3 Oct 17 '10 at 18:36
Thanks, poor terminology on the SC on my part, I meant the management IP of the ESXi box. I think when I have the kit and can play with it I'll be much better off, for now it's mostly the theory that's proving difficult to back up with documentation. So are you suggesting have a VMKernel on each host on the production LAN purely for heartbeat? – Hutch Oct 17 '10 at 19:02
No, it won't work like that, if I were you I'd simply have one vSwitch with two teamed physical NICs, then have your vmkernel, iSCSI traffic and VM traffic on that one vswitch. Setup right you should get about 2Gbps of traffic plus be able to handle the failure of a single NIC, cable or switch port. Come back to us if you get more questions ok. – Chopper3 Oct 17 '10 at 20:50

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.