Vernooij, CP (ITOPT1) - KLM wrote: >I had 2 incidents recently with Resource Groups, for which I cannot find a >solution.
>We have a Resource Group in our Sysplex, which covers several LPARs. The group >is defined as "Capacity is in Service Units with Sysplex Scope.". Several >jobclasses in several LPARs are controlled in this RG and this has been >running well for years. Care to share the details about that specific RG? Perhaps there is a setting or two? >About 2 weeks ago, I noticed a complete stop of the jobs in the RG on one >LPAR, while the total SU consumption on the Sysplex was well below the maximum >and jobs on the other LPARs were running normally. It seemed WLM had lost >track of the utilization and/or the jobs on that LPAR and kept op capping them >100%. A reactivate of the policy made WLM reinitialize or recalculate the >situation and all jobs started running again and the systems ran again as >expected. Did something changed? PTF, changes in Parmlib, WLM, etc? Just don't get mad at me please, but in such cases I would ask around just to be sure. [1] >Today we had the same problem and it was again solved by a reactivate of the >current WLM policy. I find it weird. Is that only on that one and only LPAR, but never on the others? What about the other LPARs, are they running at full speed without delays, etc? If you means 'reactivate', do you mean - re-activate across SysPlex or just on the problem LPAR? Are the other LPARs using the same RG setup? >This is z/OS 1.13. >Anyone an idea where to look further? I wish I have something, but perhaps you could see if there are any delays at all. I will ask my favourite IBMer about this. We got nearly every month problems about 100% CPU, GRS lockout, delays, etc, but usually we can resolve it, or it 'goes away'. Groete / Greetings Elardus Engelbrecht [1] - Dec 2014 - Our CPU went 100% and everything grinded at a halt - up to the whole CEC - After tracing and detective work - even after the programmers swear high and low they did *not* *changed* *anything* - we found the culprit - Some second-hand *$$hole wrote a 'Christmas animation' (15 screen updates) for every user who logs on. Now many users got p*ssed off about that animation and log off and logon back - guess - that just made it worse... ;-D ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
