z/OS MVS Planning: Operations
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Exploiting dynamic CPU addition

z/OS MVS Planning: Operations
SA23-1390-00

This section describes the software-side actions after you perform the dynamic CPU addition on the hardware side. For more information about how to update the LPAR definition, see the "Dynamic Capacity Upgrade on Demand" topic of System z10™ Enterprise Class Processor Resource/Systems Manager Planning Guide.

After you add CPUs to an LPAR, message ISN011I will be issued for each CPU that is dynamically added.

HiperDispatch=NO environments support only the first 64 CPUs being brought online. When HiperDispatch=NO, you can dynamically add a CPU after the 64th CPU has been added, but z/OS® does not allow the CPU to be brought online while the system is running HiperDispatch=NO. If a CPU is dynamically added after the 64th CPU is brought online when HiperDispatch=NO, the system issues ISN012E to indicate HiperDispatch=YES is required to bring the CPU online. For information about HiperDispatch, see "HiperDispatch Mode" in z/OS MVS Planning: Workload Management and "IEAOPTxx (OPT) Parameters" in z/OS MVS Initialization and Tuning Reference.

Then you need to check if the newly-added processors are offline or not available. Issue the DISPLAY M=CPU command to view the status of the newly-added processors, and the status shown will be either OFFLINE or NOT AVAILABLE. The actions to take are as follows:

  1. Change the NOT AVAILABLE status to OFFLINE: either deconfigure one or more CPUs offline from one or more other logical partitions as needed, or perform a physical machine upgrade to add more physical processors to the machine configuration.
  2. Bring offline processors online: use the CONFIG CPU command.

For more information about the CONFIG command, see z/OS MVS System Commands.

With the CPU addition, you might want to adjust the trace buffer size of your system. The way that you defined the buffer size for the previously existing processors will impact the buffer size set aside for the newly-added processors.
  • If you used the TRACE ST,nn command to define the size of the per-processor trace buffer, the newly-added processors will get the same size you assigned to the previously existing processors. If the total size of all the trace buffers exceeds a system-defined maximum, the size per processor will be reduced.
  • If you used the TRACE ST,BUFSIZ=nn command to define the overall size of the trace buffer, the total BUFSIZ value will still be honored and the per-processor buffer size will be calculated based on the new number of processors, thus reducing the per-processor size by a value related to the number of newly-added processors compared to the number of previously existing processors.
To resize the total trace buffer size, use the TRACE ST,BUFSIZ= command. For more information about the TRACE ST command, see z/OS MVS System Commands.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014