On Tue, 6 May 2008 08:25:42 -0400, John Kington wrote: >Natasa, > > >> Hello, >> in order to prevent releasing of unused space for some datasets, I >> allocated it >> yesterday with MC that has Partial release parameter set to 'No'. >However, >> today when I look at some of those datasets, they look like the space was > >> first released (primary extent is 1 cyl, allocated was 200), and then as >the >> data was loaded it took 11 secondary extents. This is exactly what I >wanted >> to prevent. Can someone explain to me how this happened? >> >> General Data Current Allocation >> Management class . . : MCNORLSE Allocated cylinders : 221 >> Storage class . . . : MEDIUM Allocated extents . : 12 >> Volume serial . . . : SSP122 >> Device type . . . . : 3390 >> Data class . . . . . : DCDYNV Current Utilization >> Organization . . . : PS Used cylinders . . : 204 >> Record format . . . : FB Used extents . . . : 12 >> Record length . . . : 3000 >> Block size . . . . : 27000 >> 1st extent cylinders: 1 >> Secondary cylinders : 20 >> Data set name type : SMS Compressible : NO > >The fact that the allocated cylinders is greater than the used cylinders >(221 > 204) leads me to believe that space was not released from the >dataset (allocated cylinders would equal used cylinders). The 1st extent is >just the size of the first extent and not the primary space allocation >amount. Each allocation can be up to five extents, even more if you are >using the storage space constraint relief features of SMS. I recommend you >look at the fragmentation index on your volume(s) and run some defrags. > It seems implausible that if the volume is so fragmented that only one cylinder was available for the first extent, 20 cylinders were available for each of 11 following extents. Other possibilities:
o The data set was migrated and recalled between allocation and use. o You didn't say what program populated the data set. I have an open PMR on very similar misbehavior by /bin/cp which has resulted in New Function APAR PK64372. (Development saw no way to repair the problem without changing a default behavior.) -- gil ---------------------------------------------------------------------- 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
