Synchronization of monitored resource

Synchronization of monitored resources occurs when monitored resources are changed on nodes which have been defined in the cluster administrative domain.

During this synchronization process, the cluster administrative domain attempts to change each resource with attributes whose values do not match its global values, unless there is a pending change for that resource. Any pending change is distributed to all active nodes in the domain and applied to each affected resource on each node. When the pending changes are distributed, the global value is changed and the global status of each affected resource is changed to consistent or inconsistent, depending on the outcome of the change operation for the resource on each node. If the affected resource is changed successfully on every active node in the domain, the global status for that resource is consistent. If the change operation failed on any node, the global status is set to inconsistent.

If changes are made to the same resource from multiple nodes while the cluster administrative domain is inactive, all of the changes are propagated to all of the active nodes as part of the synchronization process when the domain is started. Although all pending changes are processed during the activation of the cluster administrative domain, there is no guaranteed order in which the changes are processed. If you make changes to a single resource from multiple cluster nodes while the cluster administrative domains inactive, there is no guaranteed order to the processing of the changes during activation.

If a node joins an inactive cluster administrative domain (that is, the node is started while the cluster administrative domain is ended), the monitored resources are not resynchronized until the cluster administrative domain is started.
Note: The cluster administrative domain and its associated exit program are IBM-supplied objects. They should not be changed with the QcstChangeClusterResourceGroup API or the Change Cluster Resource Group (CHGCRG) command, or unpredictable results will occur.

Start of changeAfter a cluster node that is part of a cluster administrative domain is ended, monitored resources can still be changed on the inactive node. When the node is started again, the changes will be resynchronized with the rest of the cluster administrative domain. During the resynchronization process, the cluster administrative domain applies any changes from the node that was inactive to the rest of the active nodes in the domain, unless changes had also been made in the active domain while the node was inactive. If changes were made to a monitored resource both in the active domain and on the inactive node, the changes made in the active domain are applied to the joining node. In other words, no changes made to any monitored resource are lost, regardless of the status of the node. You can specify the synchronization option to control synchronization behavior.End of change

Start of changeIf you want to end a cluster node that is part of a cluster administrative domain, and not allow changes made on the inactive node to be propagated back to the active domain when the node is started (for example, when ending the cluster node to do testing on it), you must remove the node from the administrative domain peer CRG before you end the cluster node. End of change