<snip> Since a similar topic comes up now and then, let me make yet another pitch for a 'hybrid' process. We follow the REPLACEMENT path except that we never actually use the supplied versions of crucial installation-centric components. That is, at run time, we use our own master catalog, our own JES spool, our own RACF database, our own PARMLIB/PROCLIB, and so on. The ones that get created during ServerPac install are discarded. For example, on one system running under R13, the master catalog has a creation date of 1983.193. That pretty much dates to the introduction of ICF catalogs. Trying to tailor such ServerPac components for each new release leads to much churning and hand wringing. As well as questions to IBM Main. ;-) </snip>
Exactly the process I use. That is the 24 hour part of the 60-80 hour migration specified in my previous post. In fact, in my case, the SERVER PACK/CBPDO system is IPL'ed exactly once, to prove it was installed properly. All further testing is done after the SERVER PACK/CBPDO system has been "migrated" into a "near clone" of our existing systems. This "near clone" is then migrated through devtest/qa/production stages for further bullet proofing. HTH, ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
