No system is perfect, and I have always thought that this area we are discussing is one of the more annoying weaknesses in CMS. >From the earliest days, PROFILE EXEC was a user's gun, so to speak.
In shops where I have worked thus far, we have tried to reign-in service machine maint employing a variety of tricks, including such slick VM tricks as a shared 191. At one time, a shared 191 was not a problem for VM TCP/IP service v-machines. Not so now. [grrr!!!] Response to Jim's point follows. On Mon, 5 Dec 2005, Jim Bohnsack wrote: > What SYSPROF did you refer to in the "sniplet" below? Is that SYSPROF EXEC > from the S-disk? I know you were talking about TCPIP but I didn't find a > SYSPROF on any of the 4TCPIP40 disks. SYSPROF EXEC on the S-disk gets > serviced and is a pain in the lower back. I think that just about everyone > has a SYSPROF2 or something like that. The last RSU I put on z/VM 4.4 hit > SYSPROF. The official IBM SYSPROF EXEC would benefit from a call to an > exec such as TCPRUNXT or SYSPRRUN and a cast in concrete guarantee that > SYSPRRUN or SYSPROF2 would never be "maintained". SYSPROF EXEC is a great improvement from the old days. The only way I have found to "hook it" is to make a mod (which is more painful here than for the TCP/IP PROFILEs because SYSPROF EXEC is on the 190 with that delicate DCSS relationship). Having thus hooked SYSPROF, we call our own LCLPROF. I very much agree with Alan's point that writing run-time mods to the A disk is the right thing. Trouble from my perspective is taht the A disk might well be temp, and the "PROFILE EXEC is servicable turf" rule makes that more ... complicated. If a service v-machine needs to write no permanent files, but does need a R/W disk, then temp makes a huge lot of sense. If you apply the same local quirk to ALL service v-machines (and could them in the TCP/IP suite!) then a shared 191 with a shared PROFILE makes a huge lot of sense. This is a ten+ year old argument and I have no expectation of changing IBM's mind. So take this as an FYI rather than a complaint ... maybe. -- R;
