Failover Clustering and Network Load Balancing Team Blog
So they asked me to write for a blog. Well, I’m new here, and what can a new guy write about? The basics! But the target audience for this blog is not the faint of heart, so, let’s start with a NEW basic thing for everybody: “Creating a Cluster in Windows Server 2008”.
First a small introduction:
Clustering is available in Windows Server® 2008 Enterprise and Windows Server® 2008 Datacenter editions. The improvements to failover clusters (formerly known as server clusters) are aimed at simplifying clusters, making them more secure, and enhancing cluster stability. Cluster setup and management are easier with the new MMC Snap-In management interface, and complexity is reduced by providing the user with a simple interface to create, manage and use their failover cluster. Setup is more straightforward, with fewer steps and less configuration. Also, Cluster setup is fully scriptable so that you can automate your deployment. Security and networking in clusters have been improved, as has the way a failover cluster communicates with storage.
Does your current hardware works? Just run the new validation wizard. It will run tests to determine whether your system, storage, and network configuration is suitable for a cluster. These tests include specific simulations of cluster actions, and fall into the following categories: System Configuration tests, Network tests and Storage tests.
What else is new? Well, GUID partition table (GPT) disks and partitions of up to 2 TB are supported in failover cluster storage. And, if you’re like me, you’re asking what is a GPT? Here is the answer: GPT disks provide increased disk size and robustness. Specifically, GPT disks can have partitions larger than two terabytes (that’s a lot) and have built-in redundancy in the way partition information is stored, unlike master boot record (MBR, bye old friend) disks. With failover clusters, you can use either type of disk.
Validating a Potential Cluster Configuration:
Ok, so let’s start with this cluster business. I’ll create a simple cluster to show you the reduced numbers of steps. Let’s suppose you have two servers, running Win2k8 Enterprise already, with the “Failover Clustering” feature installed. So, hit Start, type failover, and there it is, the management interface, just hit enter. (Off topic: I love this new Start Menu, don’t you?)
And then, the new MMC Snap-In for “Failover Clustering Management” appears:
These is your main “Cluster Control Panel”. From here you can create, manage and even shutdown your cluster, along with a lot of other options. This is also where you can run the Validate a Configuration Wizard, to make sure your servers’ hardware conform to the basic requirements. As a best practice, we’ll do this first to make sure the hardware configuration will work and be supported.
Just click the link Validate a Configuration… and the new Validate a Configuration Wizard appears. This info page remind you of the tests that will be performed and has links on how to prepare your hardware.
Once read, hit Next and select all the tests (additional storage devices and nodes exponentially increase the time to completion, so if, for example, you have tested the storage before you may want to skip it). Again, select Next and you will have a review page, listing your servers and all tests that will be performed on them.
Hit Next and it starts! See it running on my second server:
This can take some time. Cofee anyone?
When it’s done, you see a new page, with a review of all tests, and can hit the View Report button, to get a nicely formated web page, with all tests details, like below:
A copy of this report is stored in your %windir%\cluster\reports directory if you need to access it later.
OK, so as expected your hardware is perfectly clusterable. What? Some issues? Failed? OK, there is a link on the tests review page that explains the validation tests and gives you some hints of what to do. This part is done, hit finish and we are back on our “Cluster Control Panel”.
Creating a Cluster
So, hit the link Create a Cluster…, and the Create Cluster Wizard appears, with some useful information regarding clustering. After reading, hit Next and the fun begins. This is where you will enter the servers that will participate in your “soon-to-be-formed” cluster (remember this in W2k3?). Just put the name and hit add for each one of them. I already did it for my two test machines:
Once you are done listing your servers (you can even put just one, but that is not a cluster, isn’t it?), hit Next, and the new validation screen appears, asking if you want to validate your configuration. Select No, as we already know our servers passes all tests.
So here it is finally, your cluster name. This is how everybody in your company/store/business, will access your cluster (consequently your file server). Give it a comprehensive name (my is SIbagy-FileServ) and hit Next.
See below the name? Wow, that’s great! Everything is setup automatically for us! Didn’t I tell you we have a more straightforward interface, with fewer steps and less configuration?
So we get to a review page, with our cluster name, the nodes that will participate on it, even the IP addresses! Hit Next.
Some time passes by, some messages, some blue bar filling up and…
We created a new cluster! You can even check a detailed report with the View Report button. When you press Finish, you go back to our Cluster Control Panel, with our two-node cluster created and ready to configure a highly-available service or application.
I will be describing how to set up a simple file server in my next post, so stay tuned for more fun next time I blog to you guys.
Clustering is available in Windows Server 2008 Enterprise and Windows Server 2008 Datacenter editions
I need to create the cluster over a WAN - each office will have a server and a NAS.
Can you please provide the article information.
Windows Server Failover Clustering does not support NAS. We support Fibre Channel, SAS, and iSCSI to permit block-level writing and persistent reservations.
Clustering & HA
Trying to set up NLB on two servers with 2008 server standard. They both have two NICs. They both have a Server Farm NIC and a NLB NIC. I can set them both up one at a time and they finish converging, but when I put the second server in and it initiates convergence again, the MX03 server never finishes converging. They are in the same subnet, being bridged together via an L2TP tunnel. The Ports the NLB NICS are using and the NLB NICS themselves are manually set to 100 Full. The Cluster Port resources are consistent. It is always MX03 that does not finish converging. If I stop the other server MX02 in the NLBMGR.exe, then the MX03 server will converge, but if I start up MX02 again, then MX03 will return to the converging state and never finish.
You ask a tough question as it depends on your configuration. Here's some troubleshooting links to get you started:
The Cluster Team
In windows 2008 r2 -- can clustered shared volumes (CSV) be backed up using Tivoli Storage Manager by running an agent on clustered nodes?
For NPS, which requires stateless configurations for IP traffic, you probably want to use Network Load Balancing (NLB) for your high-availability solution.
Here's an introduction to Failover Clustering which should be helpful: msevents.microsoft.com/.../WebCastEventDetails.aspx
If you are asking for design help, you should work with Microsoft Consulting Services.
Clustering | Microsoft
a quick question, a create a two node cluster on Windows Server 2008 R2 with FC shared storage 3 drives including Q for quorum
next we add the other two disk on the cluster but with no other configuration on cluster
durinng test (disconnecting lan and FC cables) for failover only the Q drive was move to the other node, i think this is normal because we did not proceed with other cluster configuration, right?
the other 2 drives will be used for the SQL team to install SQL Cluster
You will get all the answers to your question by posting in the cluster forum: social.technet.microsoft.com/.../threads
I'm not sure what you mean by not adding configuration to the other disks. You need to make sure you add the disks to the cluster and run Validation. If you don't see any errors, then it is a supported configuration. If you lose SAN connectivity from one node (which owns a disk) the disk will move to another cluster node.
I really got a lot out of this post and the TechNet webcast. I didn't quite understand how the wizard would handle static IP's but I am sure once I stand up my first cluster, it will be clearer.
Any thoughts clustering a virtual node with a physical node? We thought this could reduce hardware while utilizing our SAN.
If your cluster detects DHCP during the configuration wizard, it won't ask for a static IP Address, although these can be configured at anytime.
You can mix physical and virtual machines, so long as the solution passes the 'Validate' tests. If one node is virtualized, your storage will be restricted to using iSCSI, as this is required for Hyper-V. Also make sure that you don't put your physical machine and VM on the same node, because if your physical machine crashes, you would also lose your VM (and hence the entire cluster).
Good day, and would like to know if I am able to create clustering in Windows server 2008 standard version? Whereby I am having this error message prompted "the computer does not have failover clustering feature installed", where I have already installed the failover clustering feature in all nodes in the environment.
Kindly advice on the issue above.
The failover cluster feature is available in Windows Server 2008 Enterprise and Windows Server 2008 Datacenter. The feature is not available in Windows Web Server 2008 or Windows Server 2008 Standard.
Hello, MSFT guru!
Could you please advice me? I have 2 data centers (DC) - Primary and Backup, linked by FC. Also I have a SAN (EMC) in each DC. I plan to migrate our old fileserver (without clastering) to clustered solution. I have 3 servers for this project. Plan to set first and second servers at Primary DC and third server to Backup DC.
I've read a doc (download.microsoft.com/.../WS2008%20Multi%20Site%20Clustering.doc) about multi-site clustering. As I understand the best way for me is to create Node Majority cluster and set node weight for server at Backup DC to 0. In case of one server in Primary DC fail we'll have second server active. In case of Primary DC fail I should activate third server manualy (net start clussvc /fq). Is it a good solution or I misunderstand something important?
Many thanks for your time.
The Quorum configuration of Node Majority Cluster is good enough for you.
You don't have to change the Node Vote Weight of the 3rd server at your Backup Site down to 0.
The goal of setting the Node Vote Weight to 0 is to ensure that when Backup Site goes down then it should not take the Primary Site along with it.
However, in your case with 2+1 Server configuration if your Backup Site goes down you still have enough majority Quorum votes to have the Primary Site up and running.
If you plan to add a 4th Server in your Backup Site, then you would want to set the 4th Server (in your Backup Site)'s Node Vote Weight down to 0. This way you get the same experience as with the 2+1 configuration.