<----- snip -----> We were early users of XRC circa 1998. In our first few tests, we had lingering problems because, with mirrored CFRM data set, the newly IPLed systems could never get over the loss of the two CFs left back in the smoking hole. They were convinced that some magic would bring the lost CFs back to life. I put the problem to Curt Jews at SHARE--in SCIDS of course--and he recommended *not* to mirror the CFRM data set. Voila.
The crux is to IPL the first system with an empty, freshly formatted CFRM data set. There is now no trace of the lost CFs ever having been in use. <----- snip -----> I remember a lot of problems in case of a failing CF or migration to new CF(s) in the 90's. But, the functionality is much better now. Recently we moved a sysplex from 2084 to 2094 (four systems and two ICFs on the same CEC). We did not use a freshly formatted CFRM couple dataset. Instead, before the upgrade we activated a new CFRM policy, the old policy containing CFs CFOLD1 and CFOLD2 and the new policy containing CFs CFOLD1 and CFNEW2. After the successfull upgrade we activated the actual CFRM policy, containing CFs CFNEW1 and CFNEW2. That worked without problems and we had a failback possibility all the time. And as you wrote, there is now no trace of the lost CFs ever having been in use. Neither we used freshy formatted (new) XCF couple datasets nor freshly formatted (new) CFRM couple datasets. Zaromil ---------------------------------------------------------------------- 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