Joined: 15 Feb 2005 Posts: 7129 Location: San Jose, CA
Quote:
Since it is a ICETOOL tool, I didnt provide the SORT work space.
Unless your site has changed the default, ICETOOL will use dynamically allocated work data sets in this situation. I'm not sure exactly what causes the message you received, but if it's just a matter of not providing enough work space with your site's default number of work data sets, you can increase the number of work data sets used like this:
Code:
//DFSPARM DD *
OPTION DYNALLOC=(,n)
/*
You can set n to 8 or 16 or whatever you need (up to 255). Of course, if you really don't have the space available that you need, that won't help.
As Frank has indicated if you do not have enough work space available it does not matter how you code your JCL. You may need to consult with your systems programmer responsible for DFSORT or SMS to tell you how much work space is available.
I would stick with dynamic allocation and let DFSORT manage sort work space allocation.
Are your input records on disk or tape? If they are on tape or a non-SMS disk unit, add use the OPTION control statement to specify estimated number of records - FILSZ=Ennnn and the average record length - AVGRLEN=n. Please refer to the DFSORT Application Programming Guide for specific syntax.
If there is not enough work space available to perform a disk sort, you will have to use tape sort. For that please refer to the DFSORT Application Programming Guide for specific syntax.
You can provide JCL SORTWKs, but the first four characters of the ddname has to correspond to the four characters in the USING(cccc) operand. For example, if you have USING(ABCD), then you have to use ABCDWKdd DD statements.
Joined: 15 Feb 2005 Posts: 7129 Location: San Jose, CA
Quote:
Are your input records on disk or tape? If they are on tape or a non-SMS disk unit, add use the OPTION control statement to specify estimated number of records - FILSZ=Ennnn and the average record length - AVGRLEN=n.
You do NOT have to specify FILSZ unless you have unmanaged tape data sets. You do not have to specify AVGRLEN unless the actual average record length is significantly different from LRECL/2.
Quote:
If there is not enough work space available to perform a disk sort, you will have to use tape sort.
We recommend NEVER using a tape work data set sort. It's a very old, very inefficient technique that doesn't support most of the DFSORT functions. Instead, when necessary, you can split up the input file with COPY and SPLIT, sort the smaller files and then merge them.