Microsoft KB Archive/247327

From BetaArchive Wiki
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

Article ID: 247327

Article Last Modified on 11/1/2006



APPLIES TO

  • Microsoft Windows NT Server 4.0 Enterprise Edition
  • Microsoft Windows NT 4.0 Service Pack 4
  • Microsoft Windows NT 4.0 Service Pack 5
  • Microsoft Windows NT 4.0 Service Pack 6
  • Microsoft Windows NT 4.0 Service Pack 6a



This article was previously published under Q247327

SYMPTOMS

After you add resources to a cluster server with many resources (close to 1,600 or more), the Cluster service may start and appear to function but you may be unable to start Cluster Administrator. You may still be able to ping cluster IP addresses, connect to some virtual servers, and so on. Virtual servers other than the server with the cluster name may be available to connect to by using UNC paths, but not the cluster name itself. Attempting to connect to Cluster Administrator by using the "." option may or may not start Cluster administrator. If Cluster Administrator does start this way, it will likely stop responding (hang).

CAUSE

There is a "functional" limitation to the number of resources Cluster Server can support. Cluster Server can only support 1,600 resources. However, it does not prevent you from adding more resources when you reach the resource limit. By adding more resources, you run the risk of unpredictable behavior, including but not limited to: resources stopping and coming back online for no reason, virtual servers suddenly not starting, virtual servers stopping for no reason, and the inability to administer the cluster by using Cluster Administrator.

RESOLUTION

There are two methods to work around this behavior.

Method 1

Shut down one node and stop the Cluster service on the other (running) node. Then, recover the Clusdb file from a tape backup that was created when you were able to use Cluster Administrator. Note that all changes to the cluster that have been made since that point are lost. You should restart the Cluster service with the /fixquorumswitch or rename the current Chkpoint files from the MSCS folder on the quorum drive.

Method 2

Manually remove resources until you can once again administer the cluster by using Cluster Administrator. This may require a few deletions or several hundred. To manually remove resources from the command line, you need to first stop the resource and then delete it. There is no wildcard for deleting resources; you must delete them one at a time. The commands to stop and delete resources are:

cluster [cluster_name] res [resource_name] /off

cluster [cluster_name] res [resource_name] /del


STATUS

This behavior is a side-effect of putting too many resources on a cluster server. There is no resolution other than to stay below the 1,600-resource limit.

MORE INFORMATION

This behavior is a direct result of having a functional limit to the maximum possible resources instead of a physical limit. The most common occurrence of this behavior occurs on cluster servers that have many file shares (possibly for user home folders and roaming profiles). Instead of having hundreds or thousands of individual resources for all the file shares, use the advanced sharing option that was introduced in Service Pack 4. For additional information, click the article number below to view the article in the Microsoft Knowledge Base:

194831 SP4 Cluster Shares Must Be Reset to Recognize Added Subdirectories



Additional query words: mscs server cluster

Keywords: kbprb KB247327