NLB Clustering and the Quorum Disk Under 2003 Enterprise

Daniel333

So I am on lab 4.1 and 4.2 in the Sybex 70-284 book (ISBN0782143385). I am required to create a NLB cluster and install two servers into it. I am using Vmware Workstation 6. All VM Servers are 2003 Server Enterprise.

I've joined them to my domain, created a DNS record and installed second NIC's into their own unrouted network.

When I attempt to add a second node into the cluster I am receiving a message about access to the quorum disk. Honestly, I am not clear on what this quorum disk is supposed to be. I first thought it was some sort of shared drive or something they could use. But after some google time I discovered it's something to do with iSCSI/SAN.

I am not clear on how I am supposed to replicate this in VMware? Shoot, or even in real life. I have a few boxes around I can install 2003 on if I want, but they are all IDE. Let's just say it's not in the budget to buy a SAN for home, as much as I want.

Here is the exact message I get at the end of the Add Nodes Wizard.
MAIL: A Multi-node cluster cannot be created because there is not a quorum-capatible resource to all nodes.
- Error 0x900713de: The Quorum disk could not be located by the cluster service.


Any help would be appriciated!

astorrs

You are saying you want to create an NLB (network load balancing) cluster, but you are attempting to create a cluster using Microsoft Clustering Services. Which is it?

If you do want to create a MSCS cluster, have a look at this article and it will show you what additional steps you need to take to share a local SCSI (virtual SCSI it can be on IDE) disk between the two virtual machines.

Daniel333

Oh yes! Not NLB *shakes fist at lack of coffee*, MSCS all the way.

Reading now, thanks.

thanhinfo

in my opinion, i think disk quorum isn't working. you can access to the quorum disk again.

royal

Can't do shared disks in VMware Workstation. Well you can with some hacking, but if you knew how to do that, you'd know what you were doing for the Quorum. In other words, the Cluster disks have to be shared disks that both disks can access but neither node can have access to the disks at any given time. This means both nodes cannot be started at the same time until the disks are owned by at least one cluster node as it prevents other nodes from accessing it.

HeroPsycho

You don't need to hack necessarily. Just load something like OpenFiler in a VM...

Mmartin_47

Quroum disk must be external, such as a shared SCSI disk.

astorrs

Or you could just follow the link in my original reply and it will walk you through the process of sharing a virtual SCSI disk - you know the reply from 2 months ago when the question was originally asked. :P

undomiel

astorrs wrote:

Or you could just follow the link in my original reply and it will walk you through the process of sharing a virtual SCSI disk - you know the reply from 2 months ago when the question was originally asked. :P

But that would require actually listening to you for once! Baby steps, baby steps.

HeroPsycho

astorrs wrote:

Or you could just follow the link in my original reply and it will walk you through the process of sharing a virtual SCSI disk - you know the reply from 2 months ago when the question was originally asked. :P

Or, we could all get in a big circle, throw up, bark at the moon, and follow the yellow brick road!

(What's funny is that's what my algebra teacher would say in high school when someone screwed up the same way several times in a row. You can't make that up!)