On Tue, 27 Apr 2010 12:15:13 -0500, Bruce Burgdoff wrote: >Our upcoming migration from HFS to zFS for the root file systems is causing a >new problem for us.
Up through z/OS 1.10, I kept our root ZFS data sets on a dedicated volume cloned in parallel with the main sysres. (While I think of this as a "clone" process, in fact all files are processed using DFDSS logical processing at the data set level with ALLDATA(*) ALLEXCP to maintain identical data set allocation characteristics.) While the root ZFS was always cataloged, the data set name had the serial number of the matching sysres volser in the ZFS data set name, as others have described. In z/OS 1.10, I had three ZFS data sets comprising my composite root file system - z/OS itself, another for XML, and a third for ZOSMF. At z/OS 1.11, I've decided to make a change. I prefer to have all my hfs elements in one ZFS container for my root file system. At z/OS 1.11, this ZFS container became 10,400 cylinders. FWIW - this includes large products such as XML, Java and ZOSMF which I prefer to manage along with z/OS. Anyway, due to the > 4GB size of the resulting ZFS, as of z/OS 1.11 I have converted to an SMS managed ZFS data set for the root file system. The data set exists in our production SMS pool, with management class instructions that say "leave me alone" (no backup, no migrate). By moving to the SMS pool, the ZFS can be Extended Format which relieves the 4GB size limit. And, I'm no longer constrained to a single volume - if a future release of z/OS grows further, the new ZFS structure will grow along with it with no issues. FWIW Brian ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
