3.3.4 Triggers for sending alarms

(a) Using "State changes for the alarm" for the notification setting

If you select the Notify when the state changed and State changes for the alarm radio buttons when creating or editing an alarm, an alarm event is sent when the status of the alarm changes. The following example shows the appropriate settings.

Item	Detailed item	Setting
Basic information	Notify when the state changed	Selected
	State changes for the alarm	Selected
	Evaluate all data	Cleared
	State changes for each record instance	Cleared
	Always notify	Cleared
Alarm Conditions	Record	Logical Disk Overview (PI_LOGD)
	Abnormal condition	ID="*" AND % Free Space < "10"
	Warning condition	ID="*" AND % Free Space < "30"

The alarm status of a record is determined based on the results of evaluating the instances in that record against the criteria specified as alarm conditions in the alarm definition. The highest severity level among the evaluation results is set as the alarm status of the record.

Figure 3-24 Monitoring the alarm status (example of monitoring disks) illustrates an example of sending alarms when an agent is monitoring the operating status of disks. In this example, the record that collects the performance data regarding the disks has two instances (for drive C and drive D).

If the agent detects an abnormality in drive C, the alarm status is set to Abnormal and the agent sends an alarm to the manager. If the agent detects an abnormality in drive D later, the agent does not send an alarm because the alarm status used for monitoring the disk is already Abnormal. If the status of drive D returns to Normal, the alarm status remains Abnormal and the agent does not send an alarm. When the statuses of both drives C and D return to Normal, the alarm status of the record changes to Normal and the agent sends a Normal alarm to the manager.

Figure 3‒24: Monitoring the alarm status (example of monitoring disks)

(b) Using "State changes for each record instance" for the notification setting

If you select the Notify when the state changed and State changes for each record instance radio buttons when creating or editing an alarm, an alarm event is sent when the status of a record instance changes. The following example shows the appropriate settings.

Item	Detailed item	Settings
Basic information	Notify when the state changed	Selected
	State changes for the alarm	Cleared
	Evaluate all data	Cleared
	State changes for each record instance	Selected
	Always notify	Cleared
Alarm Conditions	Record	Logical Disk Overview (PI_LOGD)
	Abnormal condition	ID="*" AND % Free Space < "10"
	Warning condition	ID="*" AND % Free Space < "30"

Suppose an agent is monitoring the operating status of disks. In such a case, the record that collects the performance data regarding the disks has two instances (for drive C and drive D). If the agent detects an abnormality in drive C while monitoring the operating status of instances, the agent sends an alarm status indicating Abnormal. If the agent detects an abnormality in drive D later, the agent sends another Abnormal alarm even though the alarm status has already been set to Abnormal because the status of the instance has changed. The agent also sends a Normal alarm when the status of the instance returns to Normal.

Figure 3‒25: Monitoring the operating status for each instance (example of monitoring disks)

The following versions of products are required to enable this function:

• PFM - Manager version 10-10 or later

• PFM - Base version 10-10 or later

• PFM - Web Console version 10-10 or later

• PFM - Agent version 10-00 or later and PFM - RM version 10-00 or later^#

#

Required to bind an alarm table for sending alarms in units of instances.

Note

Select State changes for each record instance if you want to use a single alarm to monitor multiple items, or instances. For more detailed examples see, 3.3.5 Considerations for monitoring the alarm status in units of instances.

To create multiple alarms with alarm conditions specified for each instance, select State changes for the alarm rather than State changes for each record instance. If State changes for each record instance is selected, the statuses of all instances of each alarm are checked, which takes more time as the number of alarms increases.