Indeed, DIRMAP has a table of known sizes. Ages ago, I used to zap some
entry in this table when a customer had larger disks than their DIRMAP
module knew. I just finished an HTML document for my customer to describe
how we ("we" = myself & my customer's sysprogs) manage the CP directory. I
can send it to anyone asking it. I explain for example why I disabled
DIRMAINT's .xxxx. minidisks, and how I did that.
Kris,
IBM Belgium, VM customer support
VM/ESA and z/VM Discussions <[email protected]> wrote on
2005-11-18 19:09:41:
> I had always done this as well, but I always used the last real
> cylinder, rather than the first one beyond the actual size of the
> volume. The problem, of course, is that you may actually be using
> that cylinder, so you have to remove entries in the $EOV$ userid if
> that occurs. And, if you're not using that cylinders, the unused
> space at the end of the device is shown as 1 cylinder too low. These
> aren't enormous problems, but Rich's method seemed superior, because
> theoretically it should work regardless of whether the last cylinder
> is in use, so I tried it. It works fine in DISKMAP, but in DIRMAP it
> causes a problem. DIRMAP produces the following:
> 3336 3338 003 Gap
> $EOV$ 0A00 R 3339 3339 001
> 3340 4364 1025 Gap
> Kris earlier mentioned that Dirmap "somehow" ignores full-pack
> minidisks when creating overlap messages. I suspect this logic is at
> work here. I gave it a mdisk that's beyond the bounds of a real 3390
> (which only goes to cylinder 3338), so Dirmap apparently determined
> that I have a model of 3390 that has 4365 cylinders. Dirmap must
> have a table of actual devices, and that table must show that there
> is a model of 3390 that has 4365 cylinders. (I know there's a very
> large 3390, but somehow I thought it was bigger than 4365 cylinders.)
> So if you use Dirmap, you probably have to have your $EOV$ user stay
> within the confines of your real device.
> - Tom.
> ---------------------------------------------------------
> At 04:51 AM 11/18/2005, you wrote:
> >I solve this problem by coding a dummy directory entry $EOV$ that
> >owns one cylinder just beyond the last cylinder of the volume for
> >each volume on the system. This works just like $ALLOC$ and give me
> >free space after the last allocated minidisk on the volume.
> >
> Tom Cluster
> County of Sonoma
> Santa Rosa, CA
> (707) 565-3384 (Tuesdays and Wednesdays only)
