Detecting Real CPU Looping Address Spaces with OMEGAMON XE for zOS
JoeWinterton 110000BXHR Comments (3) Visits (2928)
With OMEGAMON XE for zOS the KM5_CPU_Loop_Warn Situation allows alerting of possible cpu looping address spaces on an LPAR. This unique function allows you to find these jobs before they use lots of your valuable CPU processor time. I have seen from our default settings of this situation to many false positive results while monitoring our LPARs. What I have done to reduce greatly these false positives is to change the formula by editing the situation. I have changed the warning limit from 95% to 98%. This means the address space must be using cpu or waiting for cpu 98% of the time vs 95%. Now I also changed the PERSIST in the ADVANCED tab from 2 to 3 cycles. What this does is to wait for 15 minutes (Each cycle is 5 minutes) from the current 10 minutes to raise the alert. One thing we know about cpu looping address spaces is they will continue to loop till cancelled, so waiting 5 more minutes is not a big deal. You can change the PERSIST to a bigger number also if that helps you. So in the attached you can see the changes I made with the situation editor to the KM5_CPU_Loop_Warn situation. Then remember to distribute it out to your active LPARs. Now just let it do its job and catch those naughty looping address spaces in action so you can cancel them.