GDG base as inout to sort step

cvishu · Active User Joined: 31 Jul 2007 Posts: 136 Location: india

Hi ,
I have a GDG base as input to a sort step.
The generations of the GDG will be created and deleted in an upstream process. So i will not know if for a given day , there were any generations left or not.

The problem is when there are no generations for a given day , my job abend with "No input dataset". Is there any way to put some check, so that my job runs only if generations are available.

Bill Woodger · Posted: Mon Apr 30, 2012 4:56 pm

Can you show the exact messages you get please. "No input dataset" is not an abend and, as it contains lower case, is not an actual message from anything.

Anuj Dhawan · Posted: Mon Apr 30, 2012 5:55 pm

LISTCAT on GDG Base, SORT-NULLOUT RC=0/4?

cvishu · Active User Joined: 31 Jul 2007 Posts: 136 Location: india

@ bill, i just typed a info

This is the mesage "S010 IN - DATA SET NOT FOUND"

@ Anuj
i came acrross some sites suggesting the LISTCAT option , but i am not very clear as to how to use it . Could you please throw some light.

I tried the LISTCAT , and the sysprint was something like below, how should i prceed after this.

LISTCAT ENT('xxx.xxx.xxxxx') GDG ALL
GDG BASE ------ xxx.xxx.xxxxx
IN-CAT --- CATALOG.DWR.TSO2
HISTORY
DATASET-OWNER-----P13BNC CREATION--------2012.121
RELEASE----------------2 LAST ALTER------2012.121
ATTRIBUTES
LIMIT------------------5 SCRATCH NOEMPTY
ASSOCIATIONS
NONVSAM--xxx.xxx.xxxxx.G0001V00
NONVSAM--xxx.xxx.xxxxx..G0002V00
IDCAMS SYSTEM SERVICES TIME: 06:38:06
THE NUMBER OF ENTRIES PROCESSED WAS:
AIX -------------------0
ALIAS -----------------0
CLUSTER ---------------0
DATA ------------------0
GDG -------------------1
INDEX -----------------0
NONVSAM ---------------0
PAGESPACE -------------0
PATH ------------------0
SPACE -----------------0
USERCATALOG -----------0
TAPELIBRARY -----------0

Thanks for the response

Phrzby Phil · Posted: Mon Apr 30, 2012 6:35 pm

You see

dbzTHEdinosauer · Posted: Mon Apr 30, 2012 6:44 pm

if there are files 'associated' with the GDG,
then:

dick scherrer · Posted: Mon Apr 30, 2012 9:00 pm

Hello,

One easy way to resolve this is to create a new +1 generation that is empty, with an end-of-file mark. Then, if this the only generation cataloged, there will be no problem.

Many places i've supported have situations where there may be dozens/hundreds of "input" files and rarely there are none. This approach works well for them.

expat · Posted: Tue May 01, 2012 12:07 pm

Dick,

exactly what I have used in the past and was about to suggest

cvishu · Active User Joined: 31 Jul 2007 Posts: 136 Location: india

Thnks for all the suggestions, as the gdg is owned by an upstream application , i dont think i would be able to drive them to implement the last two suggestions(which obviously seems to be the simplest) . I will go ahead with the liscat option for now . Thnks again all

Nic Clouston · Posted: Wed May 02, 2012 2:44 pm

There is nothing to stop you creating a +1 before your sort step - or even concatenating a non-gdg empty file to SORTIN. You can delete the +1 afterwards if you want. If 'they' object then raise a priority 1 change request on them as they are breaking your application.

cvishu · Active User Joined: 31 Jul 2007 Posts: 136 Location: india

lol @ Nic , it may sound a lot easier than it actually is . There is always a politics among different applications and you can always raise a Pri 1 to resolve it ;)

Bill Woodger · Posted: Wed May 02, 2012 3:18 pm

Even if there is politics, there has to be coordination.

What the heck are you doing, anyway, if sometimes there is data, but sometimes they've removed everything? Don't you have to look at all the data that they produce? Or is it OK for you to go with random bits of data, like crumbs fallen from the table?

Do you have a "window" to run your stuff while there is no activity "upstream". If not, then you will be heading for confusion.

Your job checks. Finds associations. Kicks off the Sort job. Sort job fails as microseconds after you checked they deleted. Or vice versa. No associations found by your checker, but microseconds later...

Don't, please, go cataloging/deleting a +1 just to make your job "work". Even if you can do that, it will potentially leave a "hole" that will bug the upstream guys at some point.

There sounds like there is more to this than the "technicality" of getting your sort job to work.

Nic Clouston · Posted: Wed May 02, 2012 7:13 pm

It is NOT lol, it is ds - deadly serious. You and your clients are being mucked about - and possibly charged - because something is not right.

dick scherrer · Posted: Wed May 02, 2012 7:13 pm

Hello,

My suggestion is NOT to catalog a new +1 and then uncatalog it later. . .

When we've used the method i suggested, all of the cataloged generations (including the generation that is only an end-of-file) are deleted by the job that reads them. . . This way everything in the gdg is processed only one time.

What purpose does processing all generations serve if they are not then deleted? Sounds like a whole boatload of problems. Things run at different times will have different content. This could make your accounting people, inventory control people, shipping department people and the auditors (to name a few) somewhat crazy. . .

As we don't have much of an understanding about the entire process, it is difficult make suggestions.

cvishu · Active User Joined: 31 Jul 2007 Posts: 136 Location: india

ok , i am not exactly sure for the reason of deletion , but to my guess its like this.

For a give day the job runs thrice creating 3 generations ( runs at different times ). Some days there could be chances of some aben or for any other reason the jobs could be restarted resulting in addidtional generations.

Now there are different down streams, which need to process data from generations for that day alone , since they cant go by the generations ( -1, -2 ... because there could be addtional genrations) . They are clearing up the generations every day , so the downstreams could directly use the GDG base instead of giving the generation numbers.

Note : I never said this is the best practice, all i am saying is i cant go for a upstream change withough roughing up with other teams , which we are trying to avoid .

Thanks for all your valuable comments

David Robinson · Active User Joined: 21 Dec 2011 Posts: 199 Location: UK

You don't even need to read the output from the LISTCAT. If you append .* on to the end of your base GDG name, it will return a code 4 if there are no generations present, or 0 if there are.

cvishu · Active User Joined: 31 Jul 2007 Posts: 136 Location: india

David sorry , am not clear, you mean to say if i code a jcl with gdg name xxxx.* , and if no generation are avalable it will return a cc 4 ?

I am away from my mainframe desktop will give it a try and let you know.

Bill Woodger · Posted: Fri May 11, 2012 5:51 pm

If you run the listcat as David suggest, you'll get a 0 RC if generations are present, a 4 rc if they are not.

Pete Wilson · Posted: Tue May 15, 2012 8:41 pm

Be aware that if you have multiple 'down-streams' trying to access the same GDG's they can get locked out by any process that is reading the whole base, or anything creating a new (+1) because the GDG base gets exclusively ENQ'd. So you have to be careful with your scheduling as well.