I really prefer unique volsers. One IPL with an errorneous volume coming online in a given VM system may destroy important minidisks, and yield to a corrupted SFS or DB2 for example. And, if the volsers are unique, one can -with precautions- access the minidisks of one VM using another VM system. For example to repair a broken SYSTEM CONFIG. All you need to know is the place of the source directory of the target system and use DEFINE MDISK. For my former customer we used this a lot, not to repair broken things, but to copy DB2 databases from one VM system to another, in these cases we used DIRENT to read the MDISK definition from the object directory, so we were absolutely sure to have the latest directory information. For occasional use, we had two execs, one of which requires you have access to the source directory. With things in place, R/O access to a minidisk of another VM system was as easy as a LINK & ACCESS command.
As response to Thomas' question: yes, DIRECTXA verifies the disk label before attempting to write the object directory. 2009/1/30 Quay, Jonathan (IHG) <jonathan.q...@ihg.com>: > You don't *have* to do this. I have five VM images and just use system > config to define what images see what device addresses and thus segregate > which duplicate volsers go with which system images. > > ________________________________ > > From: The IBM z/VM Operating System on behalf of Mike Rydberg > Sent: Wed 1/28/2009 11:01 AM > To: IBMVM@LISTSERV.UARK.EDU > Subject: Re-labeling CP-Owned volumes on a live system > > > > I am assisting with a migration from a z9 to a new z10, which runs several VM > LPARs. One problem is that a couple of the VM systems use the same DASD > volsers on different device addresses. > > These volsers had been isolated by creative IOCP configuration on the z9, > segregating the duplicate volumes/devices from each of the VM LPARs, but the > problem is such that it needs to be resolved now, so that each VM system has > unique CP volsers. > > > > The VM system that we chose to re-label does not use dirmaint, so I'll be > hand editing of the USER DIRECT. > > > > It's been a long time since I have done this, so I would like to know if I > have forgotten any of the key steps to re-label all the CP Owned volumes from > a live system. > > Here's the steps that I think will accomplish this. > > > > While the system is up and running: > > > > - update the CP Owned list in the MAINT CF1/2 SYSTEM CONFIG > > > > - xedit a copy of the USER DIRECT and change the CP volsers on all the MDISK > statement to the new volsers. > > (ie: change /OLDRES/NEWRES/ *, c /OLDSPL/NEWSPL/ *, c /OLDPAG/NEWPAG/ *, c > /OLDW01/NEWW01/ *, etc) > > > > - use CPFMTXA to change the labels on all the old volsers to the new volsers. > This can be done after the system is shutdown, if that matters. > > > > - place the updated USER DIRECT online (DIRECTXA USER DIRECT), note that I'm > not sure if it matters whether this is done before or after the volume > re-label step. > > > > - Shutdown and IPL VM with the device address for NEWRES. > > > > I have backups of all the volumes already. We have other VM systems with > access to these DASD volumes for recovery if things don't go well or that can > assist in this task. > > > > Is there anything obvious with that I may have missed or a better order for > these steps? > > > > Thanks > > > > Mike > > > > > > > -- Kris Buelens, IBM Belgium, VM customer support
