IBM Support

Instructions for collecting OMEGAMON Inspect data for tasks using high cpu

Troubleshooting


Problem

How do I analyze tasks that are using high cpu using OMEGAMON Classic?

Resolving The Problem

Logon to an OMEGAMON Classic (Realtime Collector) session and follow the steps described below.
Refer to the powerpoint slides attached and referenced in these steps.

Inspect slides.pptInspect slides.ppt
  1. Make sure your userid has authority to issue Level3 commands (PEEK).
    If using internal security, issue /PWD on the top line and enter the Level=3 password when prompted.
  2. (Slide 1) On the ZMENU screen, on the top line, type /LOGON and hit enter to start session logging.
    You will see VTM LOG OM/DEX on the top line if successful.
  3. On line 2 column 1 type two periods followed by a space and hit enter to clear the screen.
  4. (Slide 2) Type the following commands on the screen starting on line 2 column 1 and hit enter
    JOBN jobname
    ELAP
    TCP2
    TCP2.R
    -PEEK jobname
    XTCBS
    This will provide baseline information about the amount of cpu in use by the job and individual TCBs.
  5. Wait 10-15 seconds and hit enter again.
  6. (Slide 3) Overtype the XTCBS command with DDNS and hit enter.
  7. Save the names of the DSNs displayed under the STEPLIB DD
    Cut-and-paste to notepad or write them down for use in step 11 below.
  8. (Slide 4) Clear the screen and type the following commands and hit enter.
    INSP JOB(jobname) SAMPLES(0) INTERVAL(5) START
    ICPU
  9. (Slide 5) This will start the data collection in the background. You want to let this run until the samples used count is at least 400. The higher the number of used samples the better the diagnostic data. For tasks that do not use a lot of cpu this could take up to an hour. Hit enter every so often to check the count. It is also useful to try to let the sampling run at least 15-30 minutes so that special processing that occurs at those intervals (historical collection or RMF interval processing) is included in the data.
  10. (Slide 6) On the line with the ICPU command, type HOT after TCB(*). That will sort the data by most percent and remove 0 percent entries.
  11. (Slides 7-8) While you are waiting for the sample count to increase, this is a good time to issue some other commands that are needed for processing the inspect data. You will need to use the dataset names that you previously save in step 7 above. At the first blank line on the screen, after the ICPU display, enter the following command and hit enter. You will need to issue that command for all the names you saved. You can enter all the commands at one time, or individually.
    MLIB ADD(dataset name1)
    MLIB ADD(dataset name2)
    If you do not do this step, then when you drill down to the CSECT level, you will see $PRIVATE as the csect name and support will not be able to determine where the cpu is occurring. After they have all been accepted you can remove those commands from the screen.
  12. (Slides 9-18) Once you have a sufficient number of samples, it's time to start drilling down for details.
    In general, you will want to drill down on the 3 highest items in each category.
  13. Place the cursor on any part of the address of the first TCB in the list and hit enter
  14. Place the cursor on any part of the address of the first LMOD in the list and hit enter
  15. Place the cursor on any part of the address of the first CSECT in the list and hit enter
  16. Overtype the GR(nn) value on the ICPU command line with GR(2) and hit enter.
    For large CSECTs you may get an error due to not enough rows. Increase the GR value until the error is no longer issued.
  17. Place the cursor on the < sign under the ICPU command to go back one level.
  18. Repeat steps 15-17 for the top three CSECTs, then back up to the LMOD level.
  19. Repeat steps 14-18 for the top three LMODs, then back up to the TCP level.
  20. Repeat steps 13-19 for the top three TCBs.
  21. (Slide 19) Type STOP after the INSP command and hit enter.
  22. 22. (Slide 20) Perform steps 3-5 again to get the current cpu information.
  23. Type /LOGOFF on the top line and hit enter.
  24. Type /LOGOUT on the top line and hit enter.
  25. (Slide 21) The log will be found as a SYSnnnn file under the OMEGAMON realtime Collector task.

Send in the inspect log along with the complete output of the job that was inspected.

[{"Type":"MASTER","Line of Business":{"code":"LOB35","label":"Mainframe SW"},"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Product":{"code":"SSGHN2","label":"IBM OMEGAMON for z\/OS"},"ARM Category":[{"code":"a8m0z00000008mYAAQ","label":"OMEGAMON for zOS-\u003ETroubleshooting"}],"ARM Case Number":"","Platform":[{"code":"PF035","label":"z\/OS"}],"Version":"All Versions"}]

Document Information

Modified date:
21 October 2022

UID

swg21964210

Page Feedback