PO Generator Performance

[ssolberg]

Noticed a SAR 5610326 that is similar to what I am experiencing. We have always searched on both the Vendor # and the first Cat Code (PRP1). We recently added quite a few more part masters to our system but they (technically) should not have impacted this inquiry but obviously they did. We still have the same number of parts that fall into the Vendor # and Cat Code selection as before we loaded new part masters. But... if I make the same selections now, the inquiry takes "forever" (1+ hours) just to bring up a few dozen parts. If I only select on the Vendor #, parts pop right up but many more than I need to look at. As soon as I include PRP1 to the selection (which now should just be a "subset" of the parts that took merely a minute to display) it takes over an hour to display. It's killing us and wondering if anyone else has resolved this.

I looked at the code for P43011 and it clearly has to do with the logicals that are used for the various flavors of selections. Any of the other selections appears to go over a variety of logicals that are only over the F4102 file. If you choose to include a search over the category codes (PRP1, PRP2, or both) you run over a JOINED logical (F4102JO) of F4102/F41021, and also does not have the Vendor # in the key.

Programmer/Developer/Anything-They-Give-Me
World A7.3 c11

 

[jlucey]

Not sure if this is related but we had a similar exeperience with delivery
tickets - going from a few seconds to 20 minutes! This had to do with the
OPNQRYF process not picking a good logical to use. We spoke to IBM about
this and eventually they agreed there was a problem with their optimizer -
so they had to work on a PTF - which we now have and does seem to have fixed
the problem.



We're running R4.5

John Lucey
Consultant @ Superior Propane in Calgary
7.3 cume 6

 

[ssolberg]

Yep, JDE confirmed it is the same thing as the SAR mentioned above. They are reluctant to call it a "bug" though... but promised to look into more because they have had calls recently about it. If anyone is experiencing this problem (really only occurs if you choose to use the Cat Codes for searching and we HAVE to), I encourage you to get your company on the list for this SAR. The more folks on it give it weight to getting it corrected. Thanks in advance.

Programmer/Developer/Anything-They-Give-Me
World A7.3 c11

 

[jhasl1]

We added 3 logicals on the F4102 to speed processing. Had to change P43011.
It helped a lot.

 

[cgraves]

Did this performance hit happen to take place when you went to OS/400
v4r5...and if so, what was the PTF, cuz I haven't been able to correct
it even with the Data Base Group PTF's

 

[cbower]

John

What was the ptf#

Chuck

----- Original Message -----
From: "jlucey" <[email protected]>
To: <[email protected]>
Sent: Thursday, November 29, 2001 5:18 PM
Subject: RE: PO Generator Performance


fixed
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=W&Number=2
4521

 

[ssolberg]

Re: RE: PO Generator Performance

If you are responding to my inquiry about PO Performance, no it was not related to a OS upgrade. It is absolutely a flaw in the logic of the program when it comes to using a decent logical for the lookup when using the Cat Codes in your search. Jhasl (previous post to me) hit it on the head because they had run into the same thing. They had to mod P43011 to add a logical for better performance, which is the very direction I've got to head since I doubt JDE will attend to this in any quick sort of fashion.

I believe what John had run into with delivery tickets is unrelated to my problem. (note: we are V4R5)

Programmer/Developer/Anything-They-Give-Me
World A7.3 c11

 

[fshort]

Re: RE: PO Generator Performance

Haslauer -- that you?

Short

Fred Short
Senior Systems Engineer
Alltel Information Services
World Vison 7.3
One World B7.33

 

[jlucey]

yes the hit did happen with the move to V4R5! The PTF IBM has created is
SF66724 - but I believe we are the test site for it which may mean no one
else can download. Anyway you can speak to your IBM contact and they may be
able to get you to test as well

John Lucey
Consultant/Programmer/general dogsbody @ Superior Propane
7.3 Cume 6

 

[jlucey]

SF66724 - but see my previous note as this may not be available to you

 

[ssolberg]

Thanks to Jim Haslauer for a solution. 20 lines of code in the program and a new logical and the PO Generator screams! I would have spent twice (or more) the time doing the same mod but he was kind enough to share his source to speed my changes into production.

Programmer/Developer/Anything-They-Give-Me
World A7.3 c11

 

[cbower]

Jim Haslauer, could you share the source that speeds up PO Generator with
me?

Chuck Bower
[email protected]

----- Original Message -----
From: "ssolberg" <[email protected]>
To: <[email protected]>
Sent: Friday, November 30, 2001 6:27 PM
Subject: Re: PO Generator Performance


and a new logical and the PO Generator screams! I would have spent twice
(or more) the time doing the same mod but he was kind enough to share his
source to speed my changes into production.
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=&Board=W&Number=24
611

 

[ssolberg]

Bringing this thread back up because as you will read in previous posts, we were able to solve the problem. But... JDE now apparently has a code change that they want someone to test for them that will also resolve the problem (hopefully). They don't have the data to test it so they are looking for someone to test it for them. I just don't have the time to do the paper fix they sent me and just thought someone else out there may want to check it out.

The SAR # they gave me is 5610326 and I do have a copy of the paper fix document if you can't find it on the KG. I would be happy to forward it if interested.

Programmer/Developer/Anything-They-Give-Me
World A7.3 c11