Job Management Partner 1/Performance Management User's Guide

6.9.3 Notes on evaluating alarms

Limitation on the number of instances evaluated in an alarm
If multi-instance records are collected in PFM - Agent or PFM - RM, the maximum number of instances handled by a collection is 32,767. When alarms are bound to PFM - Agent or PFM - RM, up to 32,767 instances are evaluated. More than 32,767 instances cannot be evaluated.

Interval for evaluating the alarm
An alarm is evaluated at fixed intervals. The interval is the record collection interval for each agent. For details on the collection interval of each record, see the appropriate chapters (the Collection Interval value of each record) that describe records of each PFM - Agent or PFM - RM manual.
To modify the record collection interval, perform the following procedure:

From the browser of the monitoring console, log on to PFM - Web Console.

In the tab frame of the main window, select the Services tab.

Select the monitoring agent that the alarm is bound to.

In the method frame, select Properties.

Expand the Detail Records or Interval Records folder.

Change the value of the Collection Interval property.

The record (performance data) collection interval changes to the value you set.

Alarm condition-dependent alarm evaluation
Details of an alarm evaluation depend on the alarm conditions and the type of record to be evaluated. The following table describes the differences among alarm evaluations for various combinations of alarm conditions.

Table 6-4 Differences among alarm evaluations for various alarm conditions

Conditional expression Record type Always All Alarm evaluation (notification)

When Monitor whether the value exists is not selected Single-row record^#1 N N

If the Abnormal condition is met, and the previously reported alarm was other than Abnormal (red), the Abnormal (red) alarm is reported.

If the Abnormal condition is not met but the Warning condition is met, and the previously reported alarm was other than Warning (yellow), the Warning (yellow) alarm is reported.

If neither of the above cases is met, and the previously reported alarm was Abnormal (red) or Warning (yellow), the Normal (green) alarm is reported.

N Y

Y N If either the Abnormal or Warning condition is met, the Abnormal or Warning alarm is reported, regardless of any previously reported alarm.

Y Y

Multi-row record^#2 N N

If one instance is found that meets the Abnormal condition, and the previously reported alarm was other than Abnormal (red), the Abnormal (red) alarm is reported for that instance.

If no instance is found that meets the Abnormal condition, one instance is found that meets the Warning condition, and the previously reported alarm was other than Warning (yellow), the Warning (yellow) alarm is reported for that instance.

If none of the collected instances match the cases above, and the previously reported alarm was Abnormal (red) or Warning (yellow), the Normal (green) alarm is reported.

Note: As soon as any instance is found that meets the condition, the alarm evaluation is ended. Therefore, all of the collected instances are not always evaluated.

N Y

If, after all of the collected instances are evaluated, one or more instances are found that meet the Abnormal condition, and the previously reported alarm was other than Abnormal (red), the Abnormal (red) alarm is reported for each of those instances.

If, after all of the collected instances are evaluated, no instance is found that meets the Abnormal condition, one or more instances are found that meet the Warning condition, and the previously reported alarm was other than Warning (yellow), the Warning (yellow) alarm is reported for each of those instances.

If none of the collected instances match the cases above, and the previously reported alarm was Abnormal (red) or Warning (yellow), the Normal (green) alarm is reported.

Note: Because all of the instances are evaluated, more than one alarm can be reported in one interval.

Y N

As soon as one instance is found that meets the Abnormal condition, the Abnormal alarm is reported based on that instance, regardless of any previously reported alarm.

As soon as one instance is found that meets the Warning condition but no instance has been found that meets the Abnormal condition yet, the Warning alarm is reported based on that instance, regardless of any previously reported alarm.

Note: As soon as any instance is found that meets the condition, the alarm evaluation is ended. Therefore, all of the collected instances are not always evaluated.

Y Y For each instance that meets either the Abnormal or Warning condition, the Abnormal or Warning alarm is reported.
Note: Because all of the instances are evaluated, more than one alarm might be reported in one interval.

When Monitor whether the value exists is selected Multi-row record^#2 N N All of the collected instances are checked for the value specified in Alarm Wizard - The value whose existence is to be monitored, and if no such value is found (the condition is not met), the Abnormal (red) alarm is reported.
Note: The alarm notifying no operation is reported only once.
If no instance is collected, the alarm is not evaluated.

N Y

Y N All of the collected instances are checked for the value specified in Alarm Wizard - The value whose existence is to be monitored, and if no such value is found (the condition is not met), the Abnormal (red) alarm is reported.
Note: The alarm is reported every time.
If no instance is collected, the alarm is not evaluated.

Y Y

Legend:

Always: Whether Always notify is selected

All: Whether Evaluate all data is selected

Y: Used (selected)

N: Not used (not selected)

#1

Single-row record refers to a single-instance record.

#2

Multi-row record refers to a multi-instance record.

The alarm evaluation method is explained for various alarm conditions below.

When you set a value whose existence is to be monitored:

When you set a value whose existence is to be monitored, all fields in the specified record of PD and PI record types are evaluated to check for the specified value. If the value is not found, the alarm is reported once per interval.

When you set alarm conditions:

When you set alarm conditions, multiple records are collected in one interval for the record of PD and PI record types to be evaluated in this alarm. By default, as soon as the first instance is found that meets the conditional expression, the alarm is reported and the evaluation is ended. Therefore, all the performance data is not always evaluated. To evaluate the performance data of the PD record type in an alarm, select Evaluate all data in the Advanced settings tab of Alarm Wizard - Main Information.

How alarms are evaluated differently when Damping is enabled
In addition to the differences among alarm evaluations for various alarm conditions, if you set Damping, other differences are added to the alarm evaluation. The following table describes the differences among alarm evaluations for various alarm conditions with damping.

Table 6-5 Differences among alarm evaluations with damping

Damping Always All Alarm evaluation (notification)

Y N N

The alarm is reported only when the status of the alarm changes from the previously reported status.

Based on the instance that indicates the highest severity at the time of reporting, the status of the alarm is reported.

Note: Because the status of the alarm is determined by evaluating the damping condition, the status of the alarm might differ from the threshold of the reported instance.

Y N Y

The alarm is reported only when the status of the alarm changes from the previously reported status.

If the status is Warning or Abnormal, the status of the alarm is reported based on all of the instances that meet the status condition at the time of reporting the alarm.

Note: Because the status of the alarm is determined by evaluating the damping condition, the status of the alarm might differ from the threshold of the reported instance.

Y Y N The instance that indicates the highest severity at the time of reporting the alarm is reported.

Y Y Y

From the records that were collected at the same time, the instance that indicates the highest severity at the time of reporting the alarm for each record is reported.

Legend:

Always: Whether Always notify is selected

All: Whether Evaluate all data is selected

Y: Used (selected)

N: Not used (not selected)

The following table describes the differences when alarms are reported.

Damping When the alarm is reported

n/m The alarm is reported when the threshold is exceeded n times during m evaluations of the alarm. Subsequently, the alarm is reported every time the threshold is exceeded n times during m evaluations of the alarm.

n/n^# If the threshold continues to be exceeded, the alarm is reported every n evaluations.
This is useful when you do not want alarms that have exceeded a threshold to be continuously reported.

#: If Always notify is selected and the threshold is exceeded during the first collection after collection is started, the alarm is issued regardless of the interval.

Conditional expression	Record type	Always	All	Alarm evaluation (notification)
When Monitor whether the value exists is not selected	Single-row record^#1	N	N	If the Abnormal condition is met, and the previously reported alarm was other than Abnormal (red), the Abnormal (red) alarm is reported. If the Abnormal condition is not met but the Warning condition is met, and the previously reported alarm was other than Warning (yellow), the Warning (yellow) alarm is reported. If neither of the above cases is met, and the previously reported alarm was Abnormal (red) or Warning (yellow), the Normal (green) alarm is reported.
N	Y
Y	N	If either the Abnormal or Warning condition is met, the Abnormal or Warning alarm is reported, regardless of any previously reported alarm.
Y	Y
Multi-row record^#2	N	N	If one instance is found that meets the Abnormal condition, and the previously reported alarm was other than Abnormal (red), the Abnormal (red) alarm is reported for that instance. If no instance is found that meets the Abnormal condition, one instance is found that meets the Warning condition, and the previously reported alarm was other than Warning (yellow), the Warning (yellow) alarm is reported for that instance. If none of the collected instances match the cases above, and the previously reported alarm was Abnormal (red) or Warning (yellow), the Normal (green) alarm is reported. Note: As soon as any instance is found that meets the condition, the alarm evaluation is ended. Therefore, all of the collected instances are not always evaluated.
N	Y	If, after all of the collected instances are evaluated, one or more instances are found that meet the Abnormal condition, and the previously reported alarm was other than Abnormal (red), the Abnormal (red) alarm is reported for each of those instances. If, after all of the collected instances are evaluated, no instance is found that meets the Abnormal condition, one or more instances are found that meet the Warning condition, and the previously reported alarm was other than Warning (yellow), the Warning (yellow) alarm is reported for each of those instances. If none of the collected instances match the cases above, and the previously reported alarm was Abnormal (red) or Warning (yellow), the Normal (green) alarm is reported. Note: Because all of the instances are evaluated, more than one alarm can be reported in one interval.
Y	N	As soon as one instance is found that meets the Abnormal condition, the Abnormal alarm is reported based on that instance, regardless of any previously reported alarm. As soon as one instance is found that meets the Warning condition but no instance has been found that meets the Abnormal condition yet, the Warning alarm is reported based on that instance, regardless of any previously reported alarm. Note: As soon as any instance is found that meets the condition, the alarm evaluation is ended. Therefore, all of the collected instances are not always evaluated.
Y	Y	For each instance that meets either the Abnormal or Warning condition, the Abnormal or Warning alarm is reported. Note: Because all of the instances are evaluated, more than one alarm might be reported in one interval.
When Monitor whether the value exists is selected	Multi-row record^#2	N	N	All of the collected instances are checked for the value specified in Alarm Wizard - The value whose existence is to be monitored, and if no such value is found (the condition is not met), the Abnormal (red) alarm is reported. Note: The alarm notifying no operation is reported only once. If no instance is collected, the alarm is not evaluated.
N	Y
Y	N	All of the collected instances are checked for the value specified in Alarm Wizard - The value whose existence is to be monitored, and if no such value is found (the condition is not met), the Abnormal (red) alarm is reported. Note: The alarm is reported every time. If no instance is collected, the alarm is not evaluated.
Y	Y

Damping	Always	All	Alarm evaluation (notification)
Y	N	N	The alarm is reported only when the status of the alarm changes from the previously reported status. Based on the instance that indicates the highest severity at the time of reporting, the status of the alarm is reported. Note: Because the status of the alarm is determined by evaluating the damping condition, the status of the alarm might differ from the threshold of the reported instance.
Y	N	Y	The alarm is reported only when the status of the alarm changes from the previously reported status. If the status is Warning or Abnormal, the status of the alarm is reported based on all of the instances that meet the status condition at the time of reporting the alarm. Note: Because the status of the alarm is determined by evaluating the damping condition, the status of the alarm might differ from the threshold of the reported instance.
Y	Y	N	The instance that indicates the highest severity at the time of reporting the alarm is reported.
Y	Y	Y	From the records that were collected at the same time, the instance that indicates the highest severity at the time of reporting the alarm for each record is reported.

Damping	When the alarm is reported
n/m	The alarm is reported when the threshold is exceeded n times during m evaluations of the alarm. Subsequently, the alarm is reported every time the threshold is exceeded n times during m evaluations of the alarm.
n/n^#	If the threshold continues to be exceeded, the alarm is reported every n evaluations. This is useful when you do not want alarms that have exceeded a threshold to be continuously reported.

[Contents][Back][Next]

[Trademarks]