Building a Host Cluster with Hyper-V Beta 1


Well by now you have heard the news that Microsoft shipped Hyper-V Beta 1 early!!

Like many of you I have been testing away and thought I would share the steps for creating a host cluster in Windows 2008 and Hyper-V. For those that have built host clusters in Virtual Server, you will find the process in Windows 2008 and Hyper-V much more streamlined and user friendly. One major improvement is the integration of Hyper-V as a cluster aware application and native support for making a virtual machine highly available.

So what are the 10 steps you have to go through to build a host cluster using iSCSI shared storage?

1) Build nodes with Hyper-V
2) Build Virtual Network Switches in Hyper-V
3) Establish iSCSI Target Quorum and Data drives
4) Use iSCSI initiator on each node to connect to targets
5) Install Failover Clustering on each node
6) Run Validate on the cluster nodes
7) Create Cluster
8) Build a VM to make highly available
9) Make the VM highly available
10) Test it by moving the VM to another node

image

Node1

LAN IP address = 192.168.0.170
Heartbeat IP address = 10.10.10.1

Node2

LAN IP address = 192.168.0.171
Heartbeat IP address = 10.10.10.2

Cluster IP Address = 192.168.0.181

Step 1:

Building the nodes with Windows 2008 Enterprise or Data Center edition is a simple process.  Once you have the OS installed, you need to add the Hyper-V role in Server Manager (and reboot)

Step 2:

Once you reboot and Hyper-V installation completes, you need to launch the Hyper-V management console from the Administration Tools menu on each node.

[Do this on both nodes] On the right hand side, click the Virtual Network Manager, and create a new virtual network switch. For simplicity make it a private network and call it Private. The name must be the same on both nodes.

Step 3:

Go to the iSCSI target (hardware or software based) and build two disks enabled for shared access.  One disk needs to be 500 MB or larger to hold the cluster configuration info, and other disk needs to be big enough to hold the virtual machine, say 10 GB minimum. Be sure to enable the option for shared or cluster access to the disks.

Step 4:

On NODE1 launch the iSCSI initiator software.  Discover the iSCSI target via name or IP address and then establish a connection to the Data and Quorum disks that were established in step3.

image image

Once the disks are connected to Node1, use Disk Management to initialize and format them as NTFS volumes. Assign driver letter Q: to the small disk and a drive letter like S: to the large disk.

Now go to Node2 and use the iSCSI initiator to connect the target disks, you do not need to do anything else.

Step 5:

From Server Manager, install the feature called Failover Clustering on each node. and when done, launch the Failover Clustering Management console .

image

Step 6:

Verify that Node1 has current ownership of the disk resources (although either disk could have ownership) and then launch the Validate process from Node1 specifying Node1 and Node2 are the nodes of the cluster, run all tests.  If issues are identified, the validate report will provide details on what the issue(s) are and how to resolve them.

Click Validate a Configuration

image

Provide the names of the two nodes

image

Execute the validation process

image

Get a report back on the progress, if successful you should see all green checkmarks

image

Step 7:

Once you have completed a successful validate process, you can now build the cluster.  It is a four step process

1) Select Create Cluster from the Failover Cluster Management console.
2) Specify the nodes of the cluster
3) Specify the name and IP address of the cluster
4) Execute the create process

image

image

image

Step 8:

Using the Hyper-V Management console, create a virtual machine called TestVM on Node1 that can be made highly available.  Use Windows 2008 for the operating system.  The integration components are automatically installed for you.

Do not start TestVM, it must be in the powered off state to make it highly available.

Step 9:

Make TestVM highly available. This allows you to have a planned or unplanned migration of a VM between nodes.

Click Configure a Service or Application

image

Select Virtual Machine from the list of available services or applications

image

Select TestVM as the VM to make highly available

image

Execute the process

image

Success!!

Now bring the VM online, right click the VM in the list and select “Bring this service or application online”

image

Step 10:

Once it is online you can test the fail over by moving the VM to Node2.

Right-click the Application and select “Move this service or application to another node”, select Node2

image

Watch the process

1) The VM is placed in saved state
2) The resources are moved to Node2
3) The VM is resumed from saved state

image

image

Now the VM has successfully failed from Node1 to Node2.

You have successfully built and tested a 2-node host cluster using Hyper-V.  Note that you can have up to 16 nodes in a Hyper-V host cluster.

Hope this helps you understand Hyper-V host clustering……:-)


Comments (23)

  1. Anonymous says:

    Great article thanks. When will the guest cluster document be available?

  2. Anonymous says:

    Many applications are now required to run 24X7 with little or no downtime. They are also expected to

  3. Anonymous says:

    Are you into big iron?  Well, if you are the heavy metal sort and get your jollies by building clustered

  4. Anonymous says:

    In response to larwin.  you probably did one of the following:

    A) you only have one drive added instead of the reqired two and have a virtual machine installed on it.

    or

    B) you do have two drives but you installed a virtual machine on the drive that has the cluster folder on it.

    I hit this bump too and couldn’t figure out why since clustering was something i’m a noob in. But i’m still learning. Let’s see if I can explain…

    The reason why Rob said one drive with 500mb and another drive that’s bigger is because the cluster service writes information to that 500mb drives and becomes reserved and inaccessable to everything else.  So your Virtual machine must go to the other drive(s).  You can dink around with permissions all day and it’ll get you no where.  This virtual machine must reside on a different drive than the cluster drive.  So what does this mean?

    For every virtual machine you plan on making high availabile, you need that many more extra shared drives + the cluster drive.  The reason is because these drives will go offline and online as they migrate around the cluster so if host A has a virtual machine on a drive, host B won’t see this drive until host A migrates the virtual machine to host B. Then at the point host A will see this drive as offline and host B sees it as online.  On the same note, try not to use these drives as like means of temporary storage because you’ll probably wonder why it "disappered".

    With iSCSI, making drives is pretty easy, the JBOD I use is called the EMC AX150i SP fully populated with 16x 250gb drives.  You can RAID 5 the whole thing (with hot spare) then make a drive and assign only 1gb to it for the cluster drive and then a boat load of smaller drives for multiple high availability virtual machines.  You can also use Fibre drives too.

  5. Anonymous says:

    My current project involves being the only dedicated technical resource on the Virtualization RDP Team.

  6. Anonymous says:

    Lately it’s been very quiet on my blog. There are a couple of things to that. First and foremost there

  7. Anonymous says:

    Larwin,

    Sorry I have been very busy and not paying enough attention to my blog. You issue is that you must place a VM that you want to make highly available on the shared disk resource of the cluster.

  8. Anonymous says:

    Can you direct me to some overview level information about clustering?  There’s a lot of specific information and some conflicting terminology in this area.  A search for clustering provides information about high performance computing (HPC), web server clustering, terminal server clustering, a/a – a/p failover clustering but little about how each of those compare and where they’re appropriate.  

    Thanks for the specific instructions, I’ll be using them once I’ve firmed up my knowlege a bit.

  9. Anonymous says:

    Ok, so earlier this week I sat the Hyper-V beta exam. Beta exams are a full suite of questions, in this

  10. Anonymous says:

    How to Build a Host Cluster with Hyper-V Beta 1

  11. Anonymous says:

    Hyper-V is a cluster-aware feature within Windows Server 2008, offering native support for VM high availability. And now that beta 1 is available, you can check it out yourself. So what are the 10 steps you have to go through to build a host cluster using

  12. Anonymous says:

    If you haven’t seen this one yet – Robert Larson has done an excellent post on how to configure Hyper-V

  13. Anonymous says:

    Love the hyper-v articles, especially with so little documentation out there.   I’m about to test CCR with Exchange 2007 on hyper-v.

  14. Anonymous says:

    Greetings! Below are the best of the questions and answers that occurred during our TechNet Webcast entitled,

  15. Anonymous says:

    Hi, thank you for this great and helpful introduction. Without this article, i was not able to imagine how-to start a clustering of VMs.

    But i always got an error while tryin´to make a vm HA.

    " The disk path ‘C:programdatamicrosoft….hyper-v’ is not available, not accessible or cannot be managed by the cluster.

    Any ideas from where this message spring from?

  16. Anonymous says:

    A couple of weeks ago I did a Live Meeting broadcast for some folks over at IBM and have been finding

  17. Anonymous says:

    I’ve had a few links I’ve been meaning to post up for some time now, so here’s a small collection of

  18. Anonymous says:

    Läs Robert Larsons blogginlägg om hur man skapar ett kluster med Hyper-V, screenshots ingår. http://blogs.technet.com/roblarson/archive/2007/12/17/building-a-host-cluster-with-hyper-v-beta-1.asp

  19. Anonymous says:

    (updated 8/26/08) My current project involves being the only dedicated technical resource on the Virtualization

  20. Anonymous says:

    Be  sure to also check this other post with a similar solution, using a Windows Server 2008 Core install for the two Hyper-V and including details on the configuration of the iSCSI Target: http://blogs.technet.com/josebda/archive/2008/04/14/snw-demo-windows-server-2008-core-hyper-v-and-failover-clustering-with-screenshots.aspx

  21. Anonymous says:

    If you haven't seen this one yet – Robert Larson has done an excellent post on how to configure Hyper

  22. Anuj says:

    thanks mate i am looking for it

    regards

    Anuj

    http://www.systechblog.com