This is *not* a counter-example to Mark's point that the 'building LPAR' should make no difference but an observation that in 'bronzeplex', some things work differently from a gold?/platinum? plex. Our bronzeplex is truly a parallel sysplex that meets all IBM requirements for you-know-what. However, almost nothing is shared except for the few XCF volumes and a handful of others. In particular, each subplex has a separate RACF data base. One subplex has two fully sharing members, the other subplex has only one member. All three members belong to the parallel sysplex.
For administrative reasons, we use RACF sharing in the two-member subplex. Since the other subplex has a different RACF data base, it cannot share. CFRM policy defines IRRXCF00_P001 and IRRXCF00_B001 for RACF sharing--with DISPOSITION DELETE--but only the two-member RACF data base has the SHARE bit set. If the non-sharing system is IPLed first, these structures will not be used and may not even be allocated. But when the first sharing system is IPLed, the structure is allocated and used. Whichever sharing member is IPLed first, the structure will look the same because they use a single set of attributes in the CFRM policy. In our DR setup scenario, the non-sharing member always comes up first. The sharing-member(s) have never had a problem coming up afterwards. There are many cases where a defined structure is not actually used by all members--in some cases by any member--but the order of system IPL does not matter. The only possible case I can imagine is if a CF LPAR runs out of storage and fails to allocate one or more structures beyond a certain point. That's an abnormal circumstance that should be avoided in the first place. Likewise, in a POR sequence, CF LPARs should be initialized before z/OS LPARs to ensure that CF structures will be available for exploiters. . . . J.O.Skip Robinson Southern California Edison Company Electric Dragon Team Paddler SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile jo.skip.robin...@sce.com -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Mark Brooks Sent: Friday, December 05, 2014 11:27 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: LPAR IPL order within a Sysplex Hi, Other than the issue of wanting the last system that goes down to be the first system IPLed into the sysplex (so as to avoid the operator I,J,R prompt), I'm not aware of any technical reason that would require a particular IPL order. In particular, I don't understand why there would be any motivation to "ensure structures built in the coupling facility will always be built by a particular LPAR". To me, the very notion is anathema. We could never deliver on the promise of high availability if there was a dependency that one particular system had to be the one to create the CF structures. It also suggests that one would need to wait for all applications that create CF structures to do their allocations before any other system could be IPLed into the sysplex. Does someone have an example of a CF Structure that would demand the sort of enforcement that is being suggested? Mark A. Brooks z/OS Sysplex design and development 845-435-5149 T/L 8-295-5149 Poughkeepsie, NY mabr...@us.ibm.com ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
