On Fri, 22 Nov 2013 12:13:34 -0600, Donald Likens <[email protected]> wrote:
>We have 5 isolated systems running clones of our z/OS operating system. Our >current >zFS root file system has not been controlled and now we are now using a >Serverpac to >upgrade. I am planning to implement the maintenance procedure at the bottom >of this message but first here are my questions: > >1) How do you control your root file system so that it is not updated? Read only. Even if you use a shared file system configuration. >2) How do you clone your root file system? Either DFDSS or FDRCOPY from the version that is mounted at a service mount point where maintenance is applied. (no different than SYS1.LINKLIB or any other sysres DSN). If you use FDR, be aware that you should add a step in your copy process to quiesce the zFS files prior to copy. >3) Is it your understanding that /u and /usr/local are made available by IBM >for our use? Yes, but since /usr/local is in the read only root (as distributed by IBM), I have a symlink to /etc/local. Glad IBM fixed that issue for CRON (and MAIL+UUCP) in z/OS 1.13 by having the distributed z/OS root use symlinks to /var. One less step to do for each OS upgrade. (but I still have to do it for /usr/local) >4) Does anyone see improvements to my planned procedure that follows? I'll let others comment. There is more than one way to do this. I've used an "A/B" on smaller systems years ago when DASD was really hard to come by. In recent years, I never IPL from the SMP/E set of data sets. I clone, then IPL for testing. You can see some sample cloning jobs on my web site. > >With z/OS 1.13 we will name all USS file systems (including the root zFS ie. >SYS1.OMVS.ROOT) that > migrate with system >maintenance to have a HLQ of SYS1 and that nothing other > than what is shipped > by IBM is located in the root zFS. Exception! To make this less impacting, we > will add mount points already > created to the root system each upgrade but future mount points will be > >added to the /u or /usr/local > directories. We will have a two volume SYSRES (one containing PDSs the other > containing USS files). >The implementation process will be to IPL from SMPR13 and SMPF13, Make sure >all is working >well, Copy SMPR13 and SMPF13 over the production system res volumes, >re-catalog the > SYS1 zFS datasets, and IPL. To support not having anything in the root, all > future non-IBM mount > points will be in the /u or /usr/local directories will be pointing to the > SYS.U.zFS and SYS.local.zFS > respectively. We will also explore making the root system read only on all > systems other > than the Tech LPAR (for system maintenance). > The z/OS Planning for Installation manual has information about data set placement and Appendix D specifically talks about making a copy of your system (cloning), so you may want to review that. BTW, I also have some info on this on my web site for creating / sharing a read only root. It is not the same steps for a sysplex shared file system config (where the root should also be read only) but is still basically valid for systems that don't use sysplex sharing of z/OS unix file systems. Now that zFS can be indirectly cataloged, I guess it is more relevant again. When I wrote that doc there was no such thing as zFS. Regards, Mark -- Mark Zelden - Zelden Consulting Services - z/OS, OS/390 and MVS mailto:[email protected] ITIL v3 Foundation Certified Mark's MVS Utilities: http://www.mzelden.com/mvsutil.html Systems Programming expert at http://search390.techtarget.com/ateExperts/ ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
