It is indeed not required to XAUTOLOG a recovery machine if a "single user GCS", it won't be used, it has no role to play.
Maybe REXX code behind the GROUP panels aren't clever enough. In fact, what you enter there is saved into a "fn GROUP" file, but will also be copied in "fn ASSEMBLE" and that is assembled, yileding "fn TEXT". Maybe the Assembler macros are more clever and don't require a recovery userid for "single user GCS" systems. I never used the GROUP panels anymore since VM/SP R6, I used the ASSEMBLE file, what allowed me at that time to make overrides to make GCS variants depending on the target system. Hence the panels are not fresh in my mind. 2009/4/21 Schuh, Richard <[email protected]> > Are you saying that in a single user environment, it is unnecessary to > even log the specified recovery machine on? If not, is it even necessary to > specify a real id? > It appears that what is being implied, but not stated in this thread, is: > > 1. For a single user system, you must specify an arbitrary recovery machine > userid simply for purposes of having it included in the NSS. (Why can it not > be blanks in this instance?) > 2. That userid does not have to exist because it will never be logged on. > (Does anything bad happen if it is defined and someone logs it on and IPLs > CMS? Probably not, but it most likely would be a bad idea for it to be > allowed to IPL GCS.) > > I have, over the years, frequently heard, "Do not infer anything, > especially when dealing with IBM systems. If it is not explicitly stated in > the documentation, you cannot assume that it is so." Here, it appears that > we must trust that our inferences match the intended implications. > > > Regards, > Richard Schuh > > > > > ------------------------------ > *From:* The IBM z/VM Operating System [mailto:[email protected]] *On > Behalf Of *Kris Buelens > *Sent:* Tuesday, April 21, 2009 1:17 PM > > *To:* [email protected] > *Subject:* Re: Question on SERVICE EXEC and PUT2PROD > > I'd say the recovery userid you specify will be ignored. > > In fact, in such a "single user GCS" userid you could run RSCS, VTAM, VSCS, > and AVS together, they are all multitasking. An abend in one would cause > the death of them all. That's -I guess- why the help mentions "test > environment". > The GCS recovery machine is mainly meant to be able to cleanup the common > storage gotten by a member when that dies. E.g. when VTAM starts, it tells > the recovery machine to run program ITSxxxx when it would die, it would > FREEMAIN the storage gotten by VTAM. In a "single user GCS" system there is > no such problem of common storage that must be freed in case of abend. > > 2009/4/21 Schuh, Richard <[email protected]> > >> Rather than GROUP=NO, there is an entry for "Single User Environment >> (YES or NO)". Above that line, there is one that asks for "RECOVERY MACHINE >> USERID (required)" >> >> The note dealing with the single user entry reads: >> >> >> 5. Will the GCS system run in a single user environment? >> >> Enter YES to create a single user environment. Specify NO, to create a >> >> group environment with common shared storage. >> >> Applications that do not require group communications and that run >> >> without the use of shared common storage can run in a single system >> >> environment. These applications will run without the overhead of group >> >> initialization and multiple virtual machines. For example, a test >> >> environment requiring a single user and not group communications will >> >> be able to run without the overhead of group initialization and the >> >> multiple virtual machines. >> >> >> >> I take that, "the multiple virtual machines," in the last two statements >> to mean other machines like VTAM, ACS and buddies, but does not include the >> recovery machine (which is, per the display, required). The note related to >> the recovery machine question states that the GROUP command does check the >> recovery machine userid for validity, "but does not prevent you from saving >> an incorrect entry." What does that mean? You can put in something that >> would be acceptable as a userid, even though there is no such id in the >> directory, or it must be an existing id, even including one that is NOLOG? >> >> All of that begs the question, how do you accomplish running without the >> recovery machine? Do you simply enter an id and never log it on? I know that >> in multiple user mode, GCS refuses to come up in RSCS or any other machine >> in the group if the recovery machine is not already on. The documentation is >> perfectly clear (to the one who wrote it). >> >> >> >> Regards, >> Richard Schuh >> >> >> >> >> ------------------------------ >> *From:* The IBM z/VM Operating System [mailto:[email protected]] *On >> Behalf Of *Kris Buelens >> *Sent:* Tuesday, April 21, 2009 12:03 PM >> >> *To:* [email protected] >> *Subject:* Re: Question on SERVICE EXEC and PUT2PROD >> >> You need to build a special GCS for it. Try this in MAINT >> VMFSETUP ZVM GCS >> GROUP >> or maybe GROUP GCSRSCS >> Group presents panels to define GCS >> >> 2009/4/21 Schuh, Richard <[email protected]> >> >>> Where is this discussed? When I search the books on the RSCS shelf for >>> GROUP=NO, all I get is a blank page. Neither do I get any hits when >>> searching the "z/VM Group Control System" manual. >>> >>> Regards, >>> Richard Schuh >>> >>> >>> >>> > -----Original Message----- >>> > From: The IBM z/VM Operating System >>> > [mailto:[email protected]] On Behalf Of Alan Altmark >>> > Sent: Tuesday, April 21, 2009 10:30 AM >>> > To: [email protected] >>> > Subject: Re: Question on SERVICE EXEC and PUT2PROD >>> > >>> > On Tuesday, 04/21/2009 at 01:21 EDT, Kris Buelens >>> > <[email protected]> >>> > wrote: >>> > > For GCS, even SERVICE is a bit disruptive: the GCS saves system is >>> > replaced, >>> > > what means an IPL GCS will fail untill the GCS recovery >>> > machine (hence >>> > all >>> > > other GCS users) are restarted. Q NSS USERS GCS will >>> > reveal which of >>> > your >>> > > users might be affected. Without SNA, RSCS is probably the >>> > only user. >>> > >>> > For non-SNA networks, I prefer to run RSCS in single-user >>> > (GROUP=NO) mode so that there is no recovery machine and it >>> > can be treated like a stand-alone CMS user. >>> > >>> > Alan Altmark >>> > z/VM Development >>> > IBM Endicott >>> > >> >> >> >> >> -- >> Kris Buelens, >> IBM Belgium, VM customer support >> >> > > > -- > Kris Buelens, > IBM Belgium, VM customer support > > -- Kris Buelens, IBM Belgium, VM customer support
