>I agree... but in my experience we have as many or more customers who >balk at "put the following command in xxx". Before we updated our doc >some years ago, we got a number of complaints that we as a vendor >obviously didn't understand how the Real World works, we must be >developers working in a lab environment who have never run in a >production shop, etc.
Exactly. I am one of those customers, especially if the context says that the address space is supposed/required to stay up all the time. Of course no address space *ever* takes a terminating abend. Then how do you (re)start? And how do you *timely* restart? Document the startup dependencies and the startup commands (and the message id of your asid that says that the asid is now fully functional) and if you want to be real good, also indicate where to put stuff in the WLM hierarchy. The latter is something that IBM, too, always neglects to mention. >It gets worse if you want to have them run a UNIX program. If you tell >them to run a started task with PGM=BPXBATCH, someone asks "can't I >just run it from a shell prompt or using cron the way we run 'normal' >stuff?". If you say something about a shell prompt, the complaint is >that this is a production shop, we don't use shell prompts, etc. And what you apparently all forget: Document the shutdown procedure!!!! Again including dependencies. It does no good if an asid needs TCPIP but since that wasn't a startup requirement, TCPIP might be gone when this asid attempts to terminate, and it can't because the services it depends upon are already gone. So the said just hangs because it was too stupid in the first place to test for TCPIP availability. This is *especially* important in a UNIX environment, as there are installations (ours is one of them) that do a f bpxoinit,shutdown=forks as part of normal shutdown for IPL. That kicks the feet of anything unix out from under them pretty fast, but also (usually) makes it hard for them to terminate in the first place. Not to mention that *some* of them (like WAS) have a distinct need for shutdown in certain sequence, and if that isn't done, the filesystems can get corrupted at *IPL*. I am harping on this, as I get to investigate the 'system xyz did not shut down' all the time, and it is usually because *someone* forgot to mention *some* requirement. Bets regards, Barbara ---------------------------------------------------------------------- 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
