Displaying cluster administrative domains

You can manage and monitor the status of cluster administrative domains in your high-availability environment.

This graphical interface provides the ability to list the MREs along with the global status for each resource. Detailed information can be displayed by selecting an MRE. This information includes the global value for each attribute that is associated with the MRE, along with an indication whether the attribute is consistent or inconsistent with the domain. If the global status of a monitored resource is inconsistent, the administrator should take the necessary steps to determine why the resource is inconsistent, correct the problem, and resynchronize the resource.

If the resource is inconsistent because an update failed on one or more nodes, information is kept for the MRE that can help you determine the cause of the failure. On the node where the failure occurred, a message is logged with the MRE as to the cause of the failed update. On the other nodes, there is an informational message logged internally, which tells you there was a failure, along with the list of nodes where the update failed. These messages are available through the PowerHA graphical interface or by calling the Retrieve Monitored Resource Information (QfpadRtvMonitoredResourceInfo) API. Failure messages are also logged in the job log of the administrative domain job.

After the cause of the inconsistency is determined, the resource can be resynchronized, either as a result of an update operation on the node where the failure occurred, or by ending and restarting the administrative domain. For example, an MRE for a user profile could be inconsistent because you changed the UID for the user profile on one node in the administrative domain, but the UID you specified was already in use by another user profile on one of the nodes. If you change the value of the UID again to something that is not used by another user profile within the administrative domain, the change will be made by the cluster administrative domain on all nodes and the global status for the user profile MRE is set to consistent. You do not need to take any further action to resynchronize the user profile MRE.

In some cases, you need to end and restart the cluster administrative domain CRG in order for the inconsistent resources to be resynchronized. For example, if you change the UID for a user profile that has an MRE associated with it, but the user profile is active in a job on one of the other cluster nodes in the administrative domain, the global value for the MRE associated with the user profile will be set to inconsistent because the change operation failed on the node where the user profile was active in a job. In order to correct this situation, you need to wait until the job has ended and then end the cluster administrative domain. When the administrative domain is started again, the global value for each attribute that is inconsistent will be used to change the resource to a consistent state.

The global status for a monitored resource is always set to failed if the resource is deleted, renamed, or moved on any node in the domain. If this is the case, the MRE should be removed because the resource is no longer synchronized by the cluster administrative domain.

When you restore a monitored resource on any system that is part of a cluster administrative domain, the resource is resynchronized to the global value currently known in the cluster administrative domain when the cluster administrative domain is active.

The following restore commands result in a resynchronization of system objects: RSTLIB, RSTOBJ, RSTUSRPRF, and RSTCFG. In addition, RSTSYSINF and UPDSYSINF result in a resynchronization of system values and network attributes. To resynchronize system environment variables after running the RSTSYSINF or UPDSYSINF commands, the peer CRG that represents the cluster administrative domain must be ended and started again.

If you want to restore your monitored resources to a previous state, remove the MRE that represents the resource that you want to restore. Then, after restoring the resource, add an MRE for the resource from the system where the restore operation was done. The cluster administrative domain will synchronize the monitored resource across the domain by using the values from the restored resource.

To display a cluster administrative domain using the PowerHA graphical interface, follow these steps:

  1. In a Web browser, enter http://mysystem:2001, where mysystem is the host name of the system.
  2. Log on to the system with your user profile and password.
  3. Click PowerHA from the IBM® Navigator for i window.
  4. On the PowerHA page, click Cluster Administrative Domain.