Tutorial: Defining a threshold

Thresholds are the alerting mechanism for potential and actual problems with your managed resources. Use the tutorial to learn the basic steps for defining a threshold to raise an alarm when the condition occurs.

About this task

This tutorial uses the Linux OS agent for showing you how to define a threshold in the Threshold Manager and view the raised alarm in the Application Performance Dashboard. Your user ID must have View permission for the Threshold Manager to complete these steps.

Procedure

  1. From the navigation bar, click icon System Configuration > Threshold Manager.
  2. Click the Data Source Type list box and select the Linux OS data type.
    The thresholds that were defined for the Linux OS agent are displayed in the table.
  3. Click icon New to open the Threshold Editor for defining the threshold.
  4. Define a threshold to raise an alarm of icon Unknown severity when the average CPU is under 75%:
    1. In the Name field, enter CPU_average_below_75_percent.
    2. In the Description field, enter Threshold tutorial
    3. Leave the Severity, Interval, and Required consecutive samples fields at their default values.
    4. In the Data set field, select KLZ CPU Averages.
    5. In the Conditions field, click iconNew and add the comparison in the dialog box that pops up:
      1. In the Attribute field, select CPU_Usage_Current_Average
      2. In the Operator field, select Less than
      3. In the Value field, enter 75
      After you click OK, the attribute and comparison are displayed in the Conditions field.
    6. In the Group assignment field, select the Linux OS system group.
    7. Click Save to complete the definition and return to the Threshold Manager page.
    CPU_average_below_75_percent is displayed in the list of thresholds that are defined for the Linux OS data source.

Results

You defined a threshold that raises an alarm when the average CPU usage on any of your Linux OS managed systems is under 75%.

What to do next

  • View the event:
    1. From the navigation bar, click icon Performance > Application Performance Dashboard.
    2. In the My Components summary box, click the Events link.
      The summary box for My Components with the mouse pointer on the Events link
    3. In the Events tab that opens, look for the CPU_average_below_75_percent threshold in the list. It can take 1 or 2 minutes for the alarm to be raised. If the CPU average is over 75 percent, however, no alarm is raised.
  • Edit the threshold:
    1. From the navigation bar, click icon System Configuration > Threshold Manager.
    2. Click the Data Source Type list box and select the Linux OS data type.
    3. Select the CPU_average_below_75_percent threshold from the list and click iconEdit.
    4. Change any of the fields for a condition that you want to be alerted of. For example, you might change Name to CPU_high_warning, change Severity to Warning, edit the condition to be Greater than 75, and add a condition: Less than 95. To edit the comparison, click iconEdit, change Operator to Greater than and click OK. To add the new condition, leave the Logical operator field at the default And (&), click iconNew for a new condition, select Operator Less than, and enter Value 95.
      Edit Condition and Add Condition dialog boxes
      Name, CPU_high_warning; Description, CPU average between 75% and 95%; Severity, Warning; Condition, CPU_Usage_Current_Average between 75 and 95.
  • Review the predefined thresholds for your agents and adjust any comparison values as needed for your environment.
  • Create new thresholds to raise alarms on other conditions that you want to be alerted of.