Yes, an IPL is incredibly disruptive. There are times, however, where it is needed to prevent even greater disruption from catastrophic system failures. It seems to me that reallocating a dataset on the linklist is one such time.
On Sun, Apr 9, 2023 at 7:39 AM Peter Relson <rel...@us.ibm.com> wrote: > Bill G wrote > <snip> > Would the dataset be de-allocated at IPL so I might redefine my dataset? > Or would there be allocation across the sysplex for other Lpars? > </snip> > All the allocations are local to the system. The only safe time to do what > you want is while any system(s) using that data set are not IPL'd. It is > > Doug F wrote > <snip> > Take LLA down. > Take the dataset out of the active LNKLIST > Reallocate it. > Put it back. > Start LLA > </snip> > > That is incomplete and therefore cannot be considered correct. The missing > step is unpredictably dangerous and you take it wholly at your system's > risk. Only you can decide if that risk is worth taking. > > The 2nd step above would really be stated as making sure that nothing is > running with a LNKLST set (there could be multiple) that has the current > data set in it. In most cases you would do that by defining a single new > LNKLST set that does not have the data set. If, within this window, work > needs that data set you'll need to have a copy of the data set within the > new LNKLST set. Once defined, then you would activate this new LNKLST set. > > Then you get to making sure that no address space is using a LNKLST set > that has the data set in it. > That is the unpredictably dangerous step. SETPROG LNKLST UPDATE will > happily do what you tell it to do. > Don't rely on your system to stay up after that happens. Will it? Very > likely. Will it stay up without encountering problems? Very likely. Can you > tell if something bad is going to happen in advance? No. > > Peter Relson > z/OS Core Technology Design > > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > -- Jay Maynard ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
