It's part of that generic MVS principle from at least the 1970's, if not earlier, that things that can be shut down cleanly before an IPL should be, to minimize partially completed transactions, partial updates to a collection of related data sets, database updates that haven't yet reached a commit point, records written to output data sets but not yet flushed from buffers to DASD, subsystems left in a state which makes restart more difficult, etc, etc.
When you have no idea what some system address space is doing under the covers, you have no way to know whether it has potential to be a problem or not, unless lack of exposure is explicitly stated somewhere, or it is just impossible to terminate. And obviously you don't want system address spaces that you can't easily terminate if they could in turn prevent the normal shutdown of JES or some other system address space that benefits from a controlled shutdown. Yes, a forced IPL or "pulling the plug" with things still running certainly resets all running address spaces, but the potential for increased DB2 recovery time and potential need to recover damaged or inconsistent tables and data sets, and the possibility of JES job output on DASD not even accurately reflecting what batch job step was running at the point of interruption when restarting rudely interrupted jobs makes that a less-than-optimum approach that can potentially delay by many minutes the availability of a system for production usage after an IPL. At a minimum you want to be able to quiesce any address spaces that are processing end-user transactions or jobs, or processing service requests in support of that workload (and the number of z/OS address spaces in that latter category seems to be continually growing). Every SysProg I've known always believed you minimized your pain by always doing a controlled shutdown of as much as possible before "pulling the plug", but I suspect most have also encountered occasional cases where some software, hardware, or environmental issue didn't offer much of a choice. Joel C Ewing On 8/24/20 11:05 AM, Steve Smith wrote: > I'm not clear on why you all would care about shutting them down for an > IPL. IPL is 100% effective in removing unwanted processes. > > sas > -- Joel C. Ewing ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
