Hitachi

uCosminexus Application Server Operation, Monitoring, and Linkage Guide


9.2 Overview of Management Event Notification and Management Action

If failure, resource depletion, and other phenomena occur on the J2EE server or batch server in the management domain, Management Server is notified of them via events called Management events. Issuance of Management events is triggered by messages that are output during operation of the J2EE server or batch server. By defining the operations, to be performed when a Management event is notified, in the Management Server, the action can be executed automatically when the management event occurs. This action is called a Management action.

Important note

The following messages that are output by the J2EE server or batch server do not trigger issuance of Management events:

  • KDJE90001-E

  • KDJE90002-E

  • KDJE90003-E

  • KDJE90005-W

  • KDJE90006-W

  • KDJE90009-W

For details on the messages, see the manual uCosminexus Application Server Messages.

The following figure shows the flow from the time a Management event is issued until a management action is executed in the Management Server:

Figure 9‒1: Management event and action

[Figure]

In the example, a management event is issued when the message is output to indicate that the threshold value specified in the resource depletion monitoring functionality has been exceeded. The flow of processes in the figure are explained below:

  1. A management event is issued when the threshold value specified in the resource to be monitored is exceeded.

  2. The management event is notified to the Management Server.

  3. The process is executed automatically according to the management action definitions defined in the Management Server.

    In the definition of the management action, define the action corresponding to the Management event sent from the J2EE or batch server. The management action must be defined in advance.

If the resource depletion monitoring functionality is used to perform memory monitoring, symptoms that can trigger Full GC can be detected. If a symptom that can trigger Full GC occurs during memory monitoring, the symptom is reported to Management Server as a Management event. At this time, if an action that shuts down and restarts the services on the J2EE server or batch server is defined as a reaction to reception of that Management event, the action is automatically performed for the server. This helps to prevent the termination of the request processing.