IBM Mainframe Discussion List <IBM-MAIN@BAMA.UA.EDU> wrote on 01/31/2007 10:50:23 AM:
> The non-SMS volumes for the esoteric in question > are all mounted STORAGE (all 35 of them). > > We do not have any program products installed that > are intercepting (x37). > > The abend I am most familiar with did not involve > pre-allocation via an IEFBR14. It was defined as > below (dsn has been changed) in a production job. > > //AHY1OUT DD DSN=HLQ.SLQ.TLQ(+1), > // DISP=(NEW,CATLG,DELETE), > // UNIT=(ACHDA,6), > // SPACE=(CYL,(1000,500),RLSE), > // DCB=(HBOPM.PF,RECFM=VB,LRECL=3924,BLKSIZE=0,BUFNO=14) > > The file used the primary of 1000, then grabbed > two secondaries of 500 cylinders on two other > volumes, then ran out of space as the remaining > 3 volumes that were allocated all had less than > 339 cylinders on them. There were other volumes > available with 500 contiguous cylinders but they > were not assigned to the file. Why were these > candidate volumes allocated if they had less than > 500 cylinders while others with ample space were not ? > Brent, I am seeing a little clearer. I thought you did get an allocation of 339 cylinders on one or more of the volumes. When you allocate a non-SMS managed dataset using unit count greater than one, you will get all devices allocated at step initation. If the dataset is SMS managed, only one device will be allocated and the remaining will be converted to an asterisk. When you need to extend to a new volume, SMS will then pick a volume that will accomodate your allocation request. For non-SMS datasets, your devices are set without regard to space availability. What can you do to prevent/reduce this problem? 1. Increase the unit count to the number of volumes in the pool up to a maximum of 59 and hope for the best. 2. Convert this pool (or dataset) to SMS which will work. 3. Get a third party product that will dynamically add a volume when you need it. Regards, John ---------------------------------------------------------------------- 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
