Scenario: Job monitor for CPU utilization

 

Situation

You are currently running a new application on your system, and you are concerned that some of the new interactive jobs are consuming an unacceptable amount of resources. You would like the owners of the offending jobs to be notified if their jobs ever consume too much of the CPU capacity.

You can set up a job monitor to watch for the jobs from the new application and send a message if a job consumes more than 30% of the CPU capacity.

Configuration example

To set up a job monitor, you need to define which jobs to watch for, what job attributes to watch for, and what the monitor should do when the specified job attributes are detected. To set up a job monitor that accomplishes this goal...

1. In iSeries™ Navigator, expand Management Central > Monitors, right-click Job monitor, and select New Monitor...

2. On the General page, enter the following values:

   1. Specify a name and description for this monitor.

   2. On the Jobs to monitor tab, enter the following values:

      1. For the Job name, specify the name of the job you want to watch for (for example, MKWIDGET).

      2. Click Add.

3. Click the Metrics tab, and enter the following information:

   1. In the Available metrics list, expand Summary Numeric Values, select CPU Percent Utilization, and click Add.

   2. On the Threshold 1 tab for the metrics settings, enter the following values:

      1. Select Enable trigger.

      2. For the threshold trigger value, specify >= 30 (greater than or equal to 30 percent busy).

      3. For Duration, specify 1 interval.

      4. For the i5/OS^® trigger command, specify the following:

         SNDMSG MSG('Your job is exceeding 30% CPU capacity')
         TOUSR(&OWNER)

      5. Click Enable reset.

      6. For the threshold reset value, specify < 20 (less than 20 percent busy).

4. Click the Collection Interval tab, and select 15 seconds. This will override the Collection Services setting.

5. Click the Actions tab, and select Log event in both the Trigger and Reset columns.

6. Click the Servers and groups tab, and select the servers and groups you want to monitor for this job.

7. Click OK to save the new monitor.

8. From the list of job monitors, right-click the new monitor and select Start.

Results

The new monitor checks the QINTER subsystem every 15 seconds, and if the job MKWIDGET is consuming more than 30 percent of the CPU, the monitor sends a message to the job's owner. The monitor resets when the job uses less than 20% CPU capacity.

 

Parent topic:

Scenarios: Performance