Removing a Node
Applies To: Windows Compute Cluster Server 2003
Removing a compute node from a compute cluster can only be done if the node is not running any jobs. If a node is paused, jobs run to completion, but no new jobs are scheduled on that node.
Deleting a node removes the node from the Job Scheduler and the cluster management infrastructure. Deleting a node also removes information about the cluster head node from the registry in the following key:
"LogfileDir"="C:\\Program Files\\Microsoft Compute Cluster Pack\\LogFiles\\""ClusterName"="<ClusterHeadNode>"
In this example,
> is the actual name of the head node.
Deleting a node from a cluster does not remove the Compute Cluster Pack software from the node local drive nor does it delete the computer account of that node from Active Directory. These actions must be performed by the administrator.
To remove a compute node
Run the Compute Cluster Administrator.
From To Do List, navigate to the Node Management page.
Select the node or nodes you want to remove from the cluster.
Right-click the node or nodes selected and click Remove, or click Remove in the Action pane.
The View Summary page displays a summary of the node or nodes that you designated for removal. If you are satisfied with the results, click Finish.
The Progress page is displayed briefly, then the Result pane displays whether the remove node operation was successful
If you are going to completely remove a compute node from the network infrastructure, do not forget to remove the computer account of the node from Active Directory.
The head node cannot be removed from a compute cluster. Compute Cluster Pack can be uninstalled on a head node, effectively removing that role from the server.