Page 1 of 1

Identifying "Missing" Records

PostPosted: Mon Mar 19, 2012 10:21 pm
by Thomas Ramseier
Hi All
I'm faced to a problem with SMF-records. I need to find out which records are in FILE1 but nor in FILE2 and vice versa. Optionally, I'd like to keep the records that are identical in both files in a third file. I *thought* this or a very similar problem was discussed in a former "smart DFSORT-Tricks" doc but couldn't find it anymore.
Both files contain SMF30-records, subtype 2 or 3, Recfm=VB, no spanned records, but they can go up to the full length of 32K.

Thanks a ton for any advice and have a great week

Thomas

Re: Identifying "Missing" Records

PostPosted: Mon Mar 19, 2012 10:41 pm
by Frank Yaeger
DFSORT is not a comparison utility per se. It can do some comparison operations but the maximum number of bytes for the comparison is about 4K. It cannot compare 32K records. As the author of the Smart DFSORT Tricks paper, I can assure you that no such trick ever appeared in it.

Re: Identifying "Missing" Records

PostPosted: Tue Mar 20, 2012 12:13 am
by Thomas Ramseier
Hi Frank
Thank you for your response and for the hint about the 4k. Could the comparison be done for less than the 1st 4k per record?
Background: We are currently transforming our SMF-process from SYS1.MANx-datasets to log streams. Since we are in transition, both collecting processes run in parallel and we recognize slight differences in the post processing, especially in the accounting totals. There's a plethora of records written every day and we try to find out which records are "missing".
Thanks again

Re: Identifying "Missing" Records

PostPosted: Tue Mar 20, 2012 1:00 am
by Frank Yaeger
Could the comparison be done for less than the 1st 4k per record?


DFSORT's JOINKEYS function can be used for that kind of thing. For examples of using JOINKEYS for various types of comparisons, see the "Create files with matching and non-matching records​" Smart DFSORT Trick at:

http://www.ibm.com/support/docview.wss? ... g3T7000094

Re: Identifying "Missing" Records

PostPosted: Tue Mar 20, 2012 1:42 am
by Thomas Ramseier
Frank Yaeger wrote:
Could the comparison be done for less than the 1st 4k per record?


DFSORT's JOINKEYS function can be used for that kind of thing. For examples of using JOINKEYS for various types of comparisons, see the "Create files with matching and non-matching records​" Smart DFSORT Trick at:

http://www.ibm.com/support/docview.wss? ... g3T7000094


Thank you, I'll try JOINKEYS.

Cheers

Thomas