On Fri, 17 Jul 2009 16:06:29 -0700, Guy Gardoit <[email protected]> wrote:
<snip> > >However, even after such diligence, if there is a production problem after >implementing, why would I want to try to RESTORE one or more troublesome >PTFs?? At worst, just IPL the production system(s) from the previous resvol >(set) until the problem can be determined and fixed. Surely, another resvol >(set) (including OMVS etc) without the mantenance has been preserved?! I agree that IPLing from an old sysres set is an option if needed but several points: 1) If we aren't having the problem described by the PE, I can just RESTORE it and the next clone / rolling set of IPLs will not have the PE. We can do IPLs once or twice a month (not every LPAR... rolling IPLs). 2) Quite often there is a fix included that may be critical that you don't want to back off along with the bad one by IPLing from an old sysres set. That maintenance could even a required level of support for hardware that can't be backed off. For example, you applied z10 required maintenance along with the other RSU or HIPER maintenance, upgraded your processors, now one of those other PTFs went PE. You can't just IPL from an old sysres set. > If >worse comes to worse, (this has never been necessary but I always prepare >for it), restore your staging (or whatever) SMP/E environment until a fix is >found then re-do all the maintenance. This is pretty drastic but CYA is >pretty important too. You're right, it's pretty drastic and not necessary if you follow what most people consider a best practice - that is, not to perform ACCEPT until the next apply cycle or at least until the maintenance has been running in production for a reasonable amount of time (and no, I don't consider a week or 2 a reasonable amount of time). > >Usually, a fix will turn up either by my action or someone else's after >which it can then be put through the maintenance 'cycle'. > >Everyone may have differing opinions, but they are just opinions. I see no >inherent 'danger' in ACCEPTing major maintenance as long as you have a back >out plan of some sort. I'd prefer not to rely on the RESTORE function from >prior ugly experiences. > You don't have to rely on RESTORE even if you delay accept. Taking the opposite point of view, there is no harm at all by delaying ACCEPT. It gives you more options and an easier backout of a bad PTF. There are opinions and more than one way to do things. But there are also generally accepted (no pun intended) best practices on how to maintain software with SMP/E. Mark -- Mark Zelden Sr. Software and Systems Architect - z/OS Team Lead Zurich North America / Farmers Insurance Group - ZFUS G-ITO mailto:[email protected] z/OS Systems Programming expert at http://expertanswercenter.techtarget.com/ Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
