Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I am trying to setup a simple environment (virtual) to prove that when a Node in a cluster goes down, the other picks up.

After much pain I think I may have this however I would love for people here to see if I have setup the system correctly?

I want to make sure that for example, the disks are correctly configured. It's a simple 2 Node cluster on WS2008 R2.

Configuration

DC

1 Private NIC

IP: 192.168.1.2 Sub: 255.255.255.0 DNS Server: 127.0.0.1

Node 1

2 NICs. 1 NIC is a public nic.

Private Nic 1: IP: 192.168.1.3 Sub: 255.255.255.0 Gateway: 192.168.106.1 <-- this is a VMWare thing I believe DNS: 192.168.1.2

Node2

2 NICs. 1 NIC is a public nic.

Private Nic 1: IP: 192.168.1.4 Sub: 255.255.255.0 Gateway: 192.168.106.1 <-- this is a VMWare thing I believe DNS: 192.168.1.2

VSAN

So I am using Starwind to create iSCSI storage/targets on this virtual machine. Oh - I dont have much clue about iSCSI!

VSAN IP: 192.168.1.5 Sub: 255.255.255.0 Gateway: 192.168.106.1 DNS: 192.168.1.2

I have 4 targets here.

Clustering Info

There are 2 nodes in this cluster, as you gathered from above. ClusterIP: 192.168.1.16 ClusterName: w2k8r2-clstr

Now, in regards to the disks, I see them as different volumes.

Cluster Disk 1 is a Witness in Quorum Available Storage: Cluster Disk 2, 3, 4

I see that if I reboot Node 1, the disks appear in my computer on Node 2 If Node 2 is rebooted, I see the disks appear in my computer on Node 1

is this the correct behavior? Have I missed something? wrong configuration?

Thats all for now. I REALLY hope someone can help, I would appreciate it :)

share|improve this question
add comment

1 Answer

up vote 1 down vote accepted

Your networking is quite messed up, and you don't tell anything about your private NIC configuration, but I'll give it a try.

First of all, you can't use a default gateway of 192.168.106.1 on a computer which is connected to the 192.168.1.0/24 network; there's just no way that computer could actually reach the gateway. If VMware has automatically configured your virtual network to use the 192.168.106.0/24 subnet, you need to either stick to it or configure it to use the same subnet you want to use for your VMs (192.168.1.0/24). For the sake of simplicity, let's just assume you don't change anything: then, your main network should use the 192.168.106.0/24 subnet, and your virtual servers should be configured accordingly.

Then there's the iSCSI setup; while it is possible to run iSCSI on the same network used for public cluster access, this is definitely not a proper setup; you should use a dedicated network for iSCSI. You aren't saying where you are actually running your StarWind software, but chances are you are running it in another VM. You should add another NIC to that VM and to the two cluster nodes, and use those NICs for iSCSI (BTW, if it's a Windows Server 2008 R2 VM, you don't need the StarWind software: Microsoft offers a free iSCSI target software for this system; see here). You should configure your iSCSI target software to offer its services on the iSCSI IP address, and then configure the cluster nodes to use that IP address as their iSCSI target.

Finally, you are mixing the terms "private" and "public" here; in cluster terminology, the "public" network is the one through which flows all communications between the nodes and the rest of the world, while the "private" one is only for the cluster nodes to talk between themselves. This means the DC should be on the "public" network, not on the "private" one.

Your setup should include three networks (public, private and iSCSI); as I said above, the main network should use the same subnet where your default gateway resides; you are free to choose any other subnet for the other two networks.

In this example, I'll use the following networks:

Public: 192.168.106.0/24
Private: 192.168.10.0/24
iSCSI: 192.168.20.0/24

Your setup should look like this:

DC:
1 NIC, public network: 192.168.106.2/24, DNS 192.168.106.2, default gateway 192.168.106.1

iSCSI Server:
1 NIC, public network: 192.168.106.5/24, DNS 192.168.106.2, default gateway 192.168.106.1
1 NIC, iSCSI network: 192.168.20.5/24, no DNS, no default gateway

Node 1:
1 NIC, public network: 192.168.106.3/24, DNS 192.168.106.2, default gateway 192.168.106.1
1 NIC, private network: 192.168.10.3/24, no DNS, no default gateway
1 NIC, iSCSI network: 192.168.20.3/24, no DNS, no default gateway

Node 2:
1 NIC, public network: 192.168.106.4/24, DNS 192.168.106.2, default gateway 192.168.106.1
1 NIC, private network: 192.168.10.4/24, no DNS, no default gateway
1 NIC, iSCSI network: 192.168.20.4/24, no DNS, no default gateway

share|improve this answer
    
Thank you. I appreciate your detailed reply. As you can tell, I know almost nothing about these things. I need to reset up the clustered environment as its completely messed up due to configuration/setup using starwind iSCSI. you are correct, I am running this on a seperate VM which is joined to the domain. Can you recommend some iSCSI software that works without much issues? I tried the target software however it did not seem to detect anything. And yes i am running WS2008 R2 –  Ahmed ilyas Apr 28 '12 at 17:13
    
I noticed that for Node1, Node2 and the iSCSI server you have specified no DNS or gateway - why not? Clustering requires this surely? I will forget about using the VMWare network it installs on my host PC as the gateway. –  Ahmed ilyas Apr 28 '12 at 17:24
    
What do you mean by "I tried the target software however it did not seem to detect anything"? It does not need to detect anything at all, you just create LUNs (as VHDs) and expose them through iSCSI. –  Massimo Apr 28 '12 at 21:06
    
When a server has multiple NICs, the DNS and gateway settings only need to be specified on the NIC that will actually be used to access them. –  Massimo Apr 28 '12 at 21:08
    
Thanks. I created the VHD's sure but then through the iSCSI initiator, when I enter the IP of the iSCSI NIC, it didnt find anything to bind to in terms of the VHD's. i recreated the environment/cluster and seems to be ok - for now. no random failures of disks and still using starwind. so confused my head hurts. –  Ahmed ilyas Apr 28 '12 at 22:20
add comment

Your Answer

 
discard

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.