MVSFORUMS.com

lenovo · Posted: Wed Jan 18, 2012 10:14 am Post subject: Strange SORT problem

Hi,
I have a strange SORT problem with the production jcl. A single dataset is being used by 3 different sort steps.
The sort criteria is similar in all the 3 steps. From the last the 3 or 4 days, the third step is failing with the
reason sort capacity exceeded and first 2 steps are fine. See below for the example

superk · Advanced Joined: 19 Dec 2002 Posts: 684 Topics: 5

Why on earth is that being not done as a single job step?

lenovo · Posted: Wed Jan 18, 2012 11:26 am Post subject:

I completely agree with you but thats an existing design. So i can't comment on that.

seagull · Beginner Joined: 13 Sep 2006 Posts: 5 Topics: 0

Where and how are output-1, output-2 and output-3 defined? Are they all defined with the same amount of space? Are they all being defined on the same disk pack, and then the first two take up all the available space before the third file is created?
_________________
"Programming today is a race between software engineers striving to build bigger and better idiot- proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning."

kolusu · Posted: Wed Jan 18, 2012 12:29 pm Post subject:

lenovo,

I would suggest to get rid of the 3 steps and code this single step

Frank Yaeger · Posted: Wed Jan 18, 2012 12:39 pm Post subject:

lenovo,

SORT CAPACITY EXCEEDED means you don't have enough work data set space. I'm guessing the reason for that is you allocated work data sets (//SORTWKdd DDs) in each step and that space is not being released between steps (regardless of the RLSE parameter). I suspect if you just removed the //SORTWKdd DD statements, DFSORT would work fine using dynamically allocated work data sets which it releases between steps.

We always recommend the use of dynamically allocated work data sets rather than JCL SORTWKdd DD statements. For more information, see:

http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/ICE1CA60/A.0?DT=20110608113434

And, of course, this could easily be done in one DFSORT step with OUTFIL statements - that would certainly require less work space. Are you not allowed to suggest a better way to do something? If you can make a change to the SPACE parameters and/or remove the //SORTWKdd DD statements, why can't you just change the job to a single step?
_________________
Frank Yaeger - DFSORT Development Team (IBM)
Specialties: JOINKEYS, FINDREP, WHEN=GROUP, ICETOOL, Symbols, Migration
DFSORT is on the Web at:
www.ibm.com/storage/dfsort

Anuj Dhawan · Posted: Thu Jan 19, 2012 4:29 am Post subject:

Some how I believe that what you say here on this board, lenovo, about this probelm is your percption about the problem and not the actual thing becasue I've seen SORT CAPACITY EXCEEDED at times and that has a solution. Specially when you use DfSort and Kolusu and Frank are on this thread; said that - suggest you share the SYSOUT of failed job with us.
_________________
Regards,
Anuj

lenovo · Posted: Mon Jan 23, 2012 10:40 am Post subject:

Thanks all for the inputs. I have changed the job in one step.