Strategies for Monitoring Failover Clusters (Part 1)

by [Published on 1 Sept. 2011 / Last Updated on 1 Sept. 2011]

This article explains how to use System Center Operations Manager to monitor your failover clusters.

If you would like to read the other parts in this article series please go to:

Introduction

Failover clusters are essential to providing fault tolerance within a Windows datacenter. Once a failover cluster has been created however, you have to have a method of monitoring the cluster’s health. Otherwise, you may never even know when a failover has occurred. In this article series, I will show you how you can keep tabs on your failover clusters by using System Center Operations Manager 2007 R2.

Why is Cluster Monitoring a Big Deal?

In my introductory paragraph, I said that unless you have a cluster monitoring solution in place you might never even know when a failover has occurred. To be perfectly clear about what I mean, I want to point out that Windows Server 2008 does show you the cluster’s health.

To show you what I mean, take a look at Figure A. This screen capture was taken from my production Exchange Server 2007 clustered mailbox server. As you can see in the figure, the Failover Cluster Management Console clearly shows that the cluster named E2K7CMS is online. It also shows that the various Exchange Server related resources are also online.


Figure A: The Failover Cluster Management Console shows that my cluster is online.

If you look at Figure B, you can see that I have selected the Failover Cluster Management Console’s Nodes container. Upon doing so, the console displays the health of every node in the cluster. In this case, both of my cluster nodes are online.


Figure B: The Failover Cluster Management Console displays the health of each node in the cluster.

Since the Failover Cluster Management Console displays the cluster’s health, you might be wondering why any additional cluster monitoring techniques are even necessary. The reason why additional monitoring capabilities are needed is because the Failover Cluster Management console will work in a pinch, but it is completely inadequate for ongoing cluster health monitoring.

Imagine for a moment that the Failover Cluster Management console was your only tool for keeping tabs on the cluster’s health. Now let’s pretend that a failover occurs. If you happen to glance at the monitor while you are walking through the datacenter then you would probably notice the failover. Otherwise, the problem might go undetected.

Just for the sake of argument, let’s pretend that you have the Failover Cluster Management Console displayed in a place where it is highly visible. That being the case, let’s also assume that you look at the console often enough that any failure would be immediately obvious to you.

Even under such an ideal situation there is still one major problem with using the Failover Cluster Management console as your only monitoring tool. If you look back at the two previous screen captures you will notice that in both cases I expanded the cluster node (which is named ProdCluster.production.com in this case) and then clicked on individual sub containers to reveal the cluster’s health. All of these sub containers are cluster specific. In other words, if I were to create multiple clusters, which is a very common thing to do, then the console would not be able to simultaneously display each cluster’s health. You would be forced to manually select the cluster that you want to view.

Using System Center Operations Manager

Microsoft’s solution to the failover cluster monitoring dilemma is to use System Center Operations Manager to monitor the health of your failover clusters. The biggest advantage to using this approach to cluster monitoring is that if a failure does occur, then System Center Operations Manager can notify you of the failure rather than leaving you to eventually find the failure on your own.

System Center Operations Manager is not able to provide cluster level monitoring capabilities on its own. For that it requires the Windows Server Cluster Management Pack. This management pack can be used to monitor the health of Windows Server 2003, 2008, and 2008 R2 clusters. You can download the management pack for free here. Keep in mind that although the management pack itself is free, you must still license System Center Operations Manager to manage the servers that make up your failover clusters.

Installing the Management Pack

The management pack consists of an MSI file that is less than 1 MB in size. When you double click on this file, Windows launches a standard installation wizard that asks you to accept a license agreement and then verify the installation path. Be sure to make note of the installation path because you will need to know where the management pack files are located so that you can import them.

Once you have worked through the installation wizard, you will need to import the management packs. To do so, open the System Center Operations Manager 2007 R2 console and then select the Administration option. After doing so, choose the Management Pack option from the console tree and then click on the Import Management Packs link, found in the Actions pane.

When the Import Management Packs dialog box opens, click the Add button and then choose the Add From Disk option. When you do, you will see a message warning you that some management packs may have dependencies that cannot be located locally. The message goes on to ask you if you want to search the online catalog for the dependencies. Go ahead and click Yes. If you click No, it will cause problems later on.

Windows should now display a dialog box that allows you to specify the management packs that you want to import. Navigate through the file system until you locate the folder where the management packs reside, as shown in Figure C. There should be six different management packs in the folder.


Figure C: You must specify the location where the management pack files reside.

Select all of the management pack files and click Open. When you do, you will see a warning message for each of the management packs, as shown in Figure D. This warning message indicates that some dependencies are missing from the server, but can be imported from the online catalog.


Figure D: You will see a warning message for each management pack.

To fix the problem, click the Resolve link next to a management pack. Doing so will cause Windows to display a dialog box telling you which dependencies are required. Click the Resolve button shown in Figure E and the problem will be fixed. You must repeat this process for any additional warning messages that are displayed.


Figure E: Click the Resolve button to resolve the issue.

Once all of the issues have been resolved, click the Install button and the management packs will be imported into System Center Operations Manager. When the process completes, click Close. You can verify that the necessary management packs have been installed by going to the Monitoring tab and looking for the Microsoft Windows Cluster container, which is shown in Figure F.


Figure F: If the Microsoft Windows Cluster container exists then the management packs have been imported successfully.

Conclusion

Now that the management packs have been imported, it is time to begin monitoring our server. I will show you how to deploy the necessary agents and perform cluster monitoring in Part 2.

If you would like to read the other parts in this article series please go to:

Advertisement

Featured Links