On Thu, 5 May 2016 14:22:23 -0500, John McKown wrote: >On Thu, May 5, 2016 at 2:07 PM, phil yogendran wrote: > >> We recently performed a DR exercise and on one of the LPAR's, the CSA was >> smaller by about 800k. Parmlib specification in DR and production were >> identical. The LPAR affected is our designated 'network' box so there are >> no apps or subsystems like CICS or IMS running on it. >> >> Any thoughts on what could make CSA shrink at IPL, particularly when >> parmlib was not changed? The IODF in DR would have been different but I >> can't see that affecting CSA. >> > >CSA rounds to 1M boundries. A bigger IODF could be due to more devices, >which mean more UCBs, and so less CSA, assuming that the 1M boundry didn't >get bumped up to the next higher boundry. Just a possibility.
John has identified the reason for the difference. CSA and ECSA are the last of the common areas to be allocated at IPL time. The size of CSA is adjusted upward so that it will start on a 1 MB boundary, and the size of ECSA is adjusted upward so that it will end on a 1 MB boundary. Both your CSA and ECSAS are larger than what you have specified in IEASYSxx This is described in detail in the Init and Tuning Reference. -- Tom Marchant ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
