Everyone: The issue involving an apparent duplicate CPU serial number which I posted about last week has been resolved. Jim Mulder of IBM z/OS System Test provided valuable information and clarification on this issue, as follows:
------------------------quote "It looks like the answer is that when running under LPAR (which is always the case for any z/Architecture machine except the old z800 or z900), the STIDP data is not guaranteed to be world-wide unique, because LPAR replaces the plant of manufacture nybble with LPAR zone identification information. If you want an identifier which is world-wide unique, you need to look at the SYSIB 1.1.1 section of the data returned by the STSI instruction. (under z/OS, you can use the CSRSI callable service to obtain this instead of issuing the STSI instruction yourself). The Sequence Code from SYSIB 1.1.1, together with the Machine Type from STIDP, should provide world-wide uniqueness. I expect that a future edition of Principles of Operation will be updated to clarify this state of affairs." ------------------end quote So, for ISVs that maintain data bases of licensed customer CPU serial numbers, it's possible for two different customers with the same IBM processor model type to have identical CPU serial numbers, IF you only consider the rightmost four characters of the CPU serial number when creating that customer's data base entry. If you consider the rightmost FIVE characters of CPU serial number (which you CANNOT do except when the box is executing in non-LPAR (i.e. BASIC) mode), the CPU serial numbers ARE unique within IBM CPU model type. Jim points out above that all current IBM boxes execute z/OS in LPAR mode. I am NOT considering CPU serial numbers in use when z/OS is executing as a z/VM guest. You z/VM guys will have to sort that out. -- Mike Shaw MVS/QuickRef Support Group Chicago-Soft, Ltd.
