One side remark (David surely knows, but his text below may not be explicit enough) when you'd need a real IPL with disks starting at cylinder 1, just copying cylinder 1 to 0 will not be sufficient. One needs to "move" the whole disk up one cylinder. So DDR: COPY 1 END REORDER 0 And this for all packs that the system to IPL will use. Failure to do so may even cause data loss.
2010/10/1 David Boyes <dbo...@sinenomine.net> > > ty for this information, but I do not follow how the last cylinder of any > pack on the IIS being unused allows you to name the real packs anything you > want while still retaining the default names for the 2nd Level system. > Could you explain in a little more detail? > > Think of it this way: by making the IIS systems deliberately short 1 cyl > for each pack, you can do the following: > > On level 1, each disk has a unique label in real cyl 0, always, no excuses. > Defining a minidisk from 1 to END on that pack gives you a virtual cyl 0 for > your guest that can be the default IBM label set (eg 540RES or the like) > without ever interfering with the real system or risking the possibility > that the 1st level system will accidentally pick up a duplicate volser and > use the wrong one. You can just restore the IIS to the minidisk, and it’ll > Just Work. > > You can have as many second level systems as you want in this > configuration, and they can ALL have the default labels. You just can’t > directly IPL from them because the boot loader is in cyl 1 instead of 0. I > solve that problem by having a process for DDR (or flashcopy) of cyl 1 to > real cyl 0 from test to my “boot set” of volumes (I have one set for > current, one set for pending IPL that I alternate between, so backout is > just IPL from the other set if something goes totally casters up). It’s less > confusing for IBM if the standard label set is there, and less confusing if > you have to maintain a lot of similar systems. > > 6.2 is probably going to change a lot of that methodology – lots of new > things coming, and we’ll have to see what still works and doesn’t work at > that point. > > > > -- db > -- Kris Buelens, IBM Belgium, VM customer support
