Hands-on scenario: Solving a z/OS performance problem

You can use the System Health dashboard to quickly identify problems in your mainframe environment, navigating from a filtered list of performance or automation events to detailed performance data or automation status to help you to isolate issues faster.

About this task

In this best practice, you will identify, isolate and resolve real problems in a z/OS® environment using Service Management Unite. Firstly, you detect z/OS performance problems (such as high CPU and an excessive wait for an enqueue) and analyze the situation. Next, you learn how to cancel address spaces to restore service again.

Hands-on video:

You can watch this video and follow the guide to perform actions in the video at the same time. Click here (https://ibm.biz/BdYAUV) to watch the video in full screen.

Procedure

  1. On the Welcome page, select Monitor System Health dashboard.
    The Monitor System Health page is displayed. Note that component z/OS shows critical events on the left hand side.
  2. In the Health Status widget, select the z/OS component to only display the events related to z/OS in the Events widget.
  3. Right click one of the error events. In the drop-down menu, select View LPAR Details.
    The LPAR Details page opens showing key performance and status data for this z/OS LPAR.
  4. The Enqueue and Reserve Summary widget displays information about all global enqueue conflicts and reserves for the system. Right click on the displayed entry and select Isolate Problem.

    The problem isolation page opens. It shows detailed information about the selected enqueue conflict. In the Enqueue and Reserve Details widget, you can see the address space that owns the resource, and the address space that is waiting for it. The Wait Time shows how long the task of the waiting address space has been waiting for the resource.

  5. Maximize the Suggested Actions widget on the page using the widget menu.

    The Suggested Actions widget lists typical problems and suggested solutions. One recommended action is to cancel the job that holds the ENQ.

  6. Switch back to the LPAR Details page.
    1. On the LPAR Details page, right click again on the ENQ job listed in the Top 5 CPU Uitlization widget. The drop-down menu provides access to common commands.
    2. Select Cancel ENQ Address Space. The Issue Command page opens with the Cancel command pre-filled.

      MVS CANCEL ENQ,A=00BF

    3. Click Go on the Issue Command page.
  7. Switch back to LPAR Details page. Refresh Top 5 CPU Uitlization , Address Space Bottleneck and Enqueue widgets.

    The job ENQ no longer exists in the table.

  8. Repeat step 7 to 8 to cancel ENQ2 as well, which takes over as CPU hog.

Results

The address spaces with excessive CPU usage is canceled, and the service is restored.