Windows

Perform daily cluster management tasks with Cluster Administrator

Deploying new technology can be fun, but the challenge is often working with it on a daily basis. Deploying Windows 2000 clusters is no exception. Read Jim Boyce's take on how to perform daily cluster management tasks.

Once you’ve set up your Windows 2000 cluster, you may think the hard part is over. Then, you find the real challenge is managing daily tasks, like monitoring and configuring network interfaces. You can perform these tasks—and all cluster management tasks—through the Cluster Administrator. Here, I’ve created a short guide on executing some of the more common tasks including bringing nodes online, taking nodes offline, and managing individual resources and group resources.

Configuring network properties
The Cluster Configuration/Networks branch in the Cluster Administrator gives you access to the network interfaces for monitoring and configuration. Click on the Networks branch to view in the right pane the status for all interfaces. Cluster Administrator shows the interface name, status, subnet mask, role, and description. Click an interface under the Networks branch to view additional information, such as location, adapter type, and address. To change properties for an interface, right-click the interface under the Networks branch in the left pane and choose Properties. The resulting properties page allows you to change the interface name, enable or disable it for the cluster, and specify the role of the interface (public, private, or mixed).

Managing nodes
The nodes in a cluster appear as separate branches in the Cluster Administrator, listed by their individual server names. You can perform several actions on nodes, such as pausing and resuming the node, stopping or starting the Cluster service on the node, or evicting the node from the cluster.

When you pause a node, existing resources and groups stay online, but the node can’t bring additional resources or groups online. For example, you might need to pause a node to perform maintenance on it. Resuming the node allows it to resume full responsibility in the cluster, including taking on resources from a failed node.

You can take a node offline and either keep it in the cluster or evict it from the cluster. You would use the former when you need to perform routine maintenance and the latter if you need to take the node out of the cluster permanently or need to perform a major overhaul, such as changing hardware. To take a node offline without evicting it from the cluster, right-click the node in the Cluster Administrator and choose Pause Node. Then, open the Active Groups container, right-click a group, and choose Move Group to move it to the secondary node. Repeat the process for all other groups. After the groups are successfully moved, perform the maintenance on the server, and then in the Cluster Administrator, right-click the node and choose Resume Node to bring it back online.

If you need to evict a node from the cluster, open the Cluster Administrator, right-click the node, and choose Stop Cluster Service. After the service stops, right-click the node and choose Evict Node. Take the server offline and maintain/rebuild as needed. Then, bring the server back online, reinstall the Cluster service, and rejoin the cluster.

Managing resources
There are several tasks you might need to perform on a regular basis for resources. For example, perhaps you must perform an upgrade on a node and need it to fail over to the secondary so you can take it offline. Or maybe you want to move resources from one group to another, bring a resource online, or take one offline.

When you take a resource offline, it becomes unavailable to clients. For example, you might take your file share virtual server offline to add or remove applications. To take a resource offline, open the Cluster Administrator and expand the group in which the resource is located. If you want to take a single resource offline, right-click the resource in the right pane and choose Take Offline. Right-click it again and choose Bring Online to restore it. You can also take an entire group offline or bring it online by right-clicking the group in the left pane and choosing either Bring Online or Take Offline, as needed.

In some cases, you might want to move a resource from one group to another, say for administrative reasons, to simplify your cluster structure or to change the way resources fail over. You might also decide to add a dependency to a resource and need to move the resource into the group where the dependent resource is located. Whatever the case, you can move a resource without taking it offline. Locate and right-click the resource and choose Change Group; then, select the group to which you want it moved.

When it comes time to upgrade a system, you can fail the groups on a node over to the secondary to allow you to add new software or apply updates. You might also want to fail groups to test their behavior and dependencies. To fail a group, locate and right-click the group in the Cluster Administrator and choose Initiate Failure. Repeat the process for all of the other resources hosted by the node, and then pause the node and perform the upgrade. Resume the node when you’re finished with the upgrade, and MSCS will take care of failing back the resources. Of course, this assumes that you’ve configured them to allow failback.

Conclusion
Administering a new Windows 2000 cluster can give a network administrator a headache. And learning to perform certain tasks can be time-consuming, especially when you’re in unfamiliar territory. By familiarizing yourself with the everyday tasks I’ve discussed, though, you should be able to get through the day without much hassle.
0 comments

Editor's Picks