I understand now. Tape sortwks do work. My company used to have several jobs using them.
On Thu, Jun 7, 2018 at 1:44 PM David Betten <[email protected]> wrote: > Just a few thoughts on ways you might be able to reduce the sort work > requirement. > > Since the input is on tape, do we know if DFSORT is getting accurate file > size information? Do you see DFSORT messages in the output like UNKNOWN > FILE SIZE or INTERMEDIATE MERGE...? > The reason I ask is that if DFSORT doesn't know the file size or > incorrectly estimates, it can under allocate virtual storage which makes > the sort very inefficient and causes an increase in sort work space > requirements. > > > Another option to consider would be splitting the input file, sorting each > segment separately, then merging the outputs. Not ideal, but still might > run faster than a single large sort using tape sort work. > > If you could post the DFSORT messages from the job, we might be able to > offer some better suggestions. > > > Have a nice day, > Dave Betten > z/OS Performance Specialist > Cloud and Systems Performance > IBM Corporation > email: [email protected] > > > IBM Mainframe Discussion List <[email protected]> wrote on > 06/07/2018 02:14:09 PM: > > > From: Jesse 1 Robinson <[email protected]> > > To: [email protected] > > Date: 06/07/2018 02:15 PM > > Subject: Re: SORTLIB DD > > Sent by: IBM Mainframe Discussion List <[email protected]> > > > > Let me reiterate. The problem job tries to allocate more DASD work > > space than *exists* on the system. SORTIN is on tape--multiple > > files. We have the capability of putting more volumes online > > temporarily, but this is a major PITA and requires intervention from > > the Storage boys. I'm hoping that tape SORTWK will get the user over > > the occasional hump for this ad hoc non-production job. It does not > > have to perform well. It just has to work. > > > > . > > . > > J.O.Skip Robinson > > Southern California Edison Company > > Electric Dragon Team Paddler > > SHARE MVS Program Co-Manager > > 323-715-0595 Mobile > > 626-543-6132 Office ⇐=== NEW > > [email protected] > > > > > > -----Original Message----- > > From: IBM Mainframe Discussion List [mailto:[email protected] > > ] On Behalf Of R.S. > > Sent: Thursday, June 07, 2018 2:49 AM > > To: [email protected] > > Subject: (External):Re: SORTLIB DD > > > > Get rid of SORTLIB DD > > Get rid of SORTWKnn DD > > Use dynamic sortwork datasets, optionally set the number of datasets > > via OPTION DYNALLOC Don't use tapes for sortwork > > > > BTW: > > What is a size of input data? > > How much space do you have for temp datasets? > > How much memory can the job use? > > > > > > My €0.02 > > > > -- > > Radoslaw Skorupka > > Lodz, Poland > > > > > > > > > > > > > > W dniu 2018-06-05 o 18:31, Jesse 1 Robinson pisze: > > > We have a DFSORT job that wolfs down enormous amounts of SORTWK > > space. It has been exceeding the DASD capacity on the system where > > it runs, so we advised the user to point SORTWK to tape instead of > > DASD. Now it fails with > > > > > > IEC130I SORTLIB DD STATEMENT MISSING > > > IEF472I CIHM373 STEP010 CIHM373 - COMPLETION CODE - SYSTEM=000 > USER=0063 > > > > > > IBM doc indicates the need for SORTLIB with a 'tape sort'. We have > > no working example to share with the user. My question: what should > > DD SORTLIB point to? SMPE puts load modules into > > > > > > SYS1.SORTLIB > > > SYS1.SICELINK > > > > > > Should the user specify only the first one or both? I hate to drag > > them into a sysprog guessing game. > > > > > > . > > > . > > > J.O.Skip Robinson > > > Southern California Edison Company > > > Electric Dragon Team Paddler > > > SHARE MVS Program Co-Manager > > > 323-715-0595 Mobile > > > 626-543-6132 Office <===== NEW > > > [email protected]<mailto:[email protected]> > > > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
