You are articutlaing the complications well. As you mentioned: We looked at LPARNAME, CEC and serial number but those are also ephemeral over time but could be used to trace provenance.
Looking into the XCF/SYSPLEX System ID as that seems to have the stability I’m looking for. It sounds like an odd request but when your looking to analyze a lot of data over a period of time its really a necessity. Thanks for your thoughts. Matt Hogstrom > On Jun 5, 2024, at 12:27, Al Sherkow Digest <a...@sherkow.com> wrote: > > My SCRT auditing product has to keep track of z/OS Images and back in 2003 I > used the 4-Char SMFID, the machine type, and the machine serial number. I > included the machine type because in the 1980s sites could upgrade hardware > while maintaining the serial number to avoid some licensing issues. That > worked fine, until I encountered a site with two SMFID of SYSA. One was also > set to 1999 (during 2005) for regression testing that had been developed for > Y2K testing and they were still using it. Honestly that blew my mind. It was > in the same sysplex and on the same machine with different LPARNAMEss. > > As Mark Brooks wrote earlier adding the 8-char XCF/SYSPLEX SystemID helped. > That was how this site had 2 SYSAs, I also added the LPARNAME. This handled > everything I encountered. > > The remaining problem was linking SMF records together. The "older" records, > such as 0 (IPL), 7 (data loss), 90 (system status) and others only have the > 4-char SMFID. For example it is non-trivial to match which SYSA IPL was > related to which z/OS image with SMFID of SYSA. > > Al Sherkow > > ---------------------------------------------------------------------- > 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