How to check whether dataset exists or not

ibmmainframesyntel · Posted: Mon Feb 26, 2007 7:18 pm

How to check wether dataset exists or not .

i have to execute my job.

It may contains 3 or two input file.

so, before executing the above step, I have to check wether the dataset exists or not, so that I can override my PROC according to that
either 2 or 3 input file.

superk · Posted: Mon Feb 26, 2007 8:07 pm

Ask yourself, how would you do this in TSO (TSO, not ISPF) if you needed to? The method would be the same.

ofer71 · Posted: Mon Feb 26, 2007 9:35 pm

LISTDSI & SYSDSN are the methods for you.

O.

deepak.vl · New User Joined: 17 Feb 2007 Posts: 38 Location: Hyderabad

Have a step which tries to read the dataset, for which you want to check whether it exists.
Have 2 steps after this, one to execute if the step 1 fails and other when step 1 is successful.
Control the execution of Steps 2 & 3 by using either a COND or an IF-ELSE.
Something like this...

STEP1 EXEC ..... --> (will try to read the dataset)
STEP2 EXEC ......COND=(0,EQ) --> (this will assume there are 3 datasets)
STEP3 EXEC ......COND=(0,NE) --> (this will assume there are 2 datasets)

Thanks,
Deepak

dick scherrer · Posted: Tue Feb 27, 2007 12:00 am

Hello,

If step1 "tries to read" and the dataset does not exist, there will not be a recurn-code issued - the job will fail with a jcl error with the information that the dataset does not exist.

Please refer back to the post from superk. You need to verify existence before a step is executed with a dd statement that references the dataset which may or may not exist..

deepak.vl · New User Joined: 17 Feb 2007 Posts: 38 Location: Hyderabad

dick scherrer · Posted: Tue Feb 27, 2007 1:11 am

Hi Deepak,

It is not elegant, but the way i used to check for a dataset (in batch) was to run an IDCAMS/LISTCAT for the dataset in question. If the LISTCAT is successful, IDCAMS sets a zero RC. If the LISTCAT fails, it is non-zero. Testing the return code would be like your suggestion above. The only real difference is that i specify the dsn to a utility and not to a DD statement.

After years of this, i decided that i would make sure the file(s) were always on the system. If there was no data, the file would only have an end-of-file and raise the AT END condition on the first read.

I liked this approach better than doing some kind of dance to use or not use a file that may or may not be cataloged.

deepak.vl · New User Joined: 17 Feb 2007 Posts: 38 Location: Hyderabad

Thanks Dick

dick scherrer · Posted: Tue Feb 27, 2007 9:18 am

You're welcome

superk · Posted: Tue Feb 27, 2007 7:13 pm

Sigh.

So much for trying to get someone to think through the problem and arrive at a solution for themself.

We can't keep spoon-feeding everyone, folks.

dick scherrer · Posted: Tue Feb 27, 2007 9:11 pm

Actually, my mission with that last bit of info was to "air" my preference that "checking for existence" need not be an issue if the design ensures that is an e-o-f when there is no data for the file. It is at least as easy to ensure there is an empty file as it is to check for existence and code around that (IMHO).

I agree that it would be excellent if there was a way to get folks to do some research/attempt, explain what they've researched/tried, and then post where they are stuck, Then we could help.

Answers are sometimes nice, but becoming self-sufficient is far more valuable.

We can only hope.

ibmmainframesyntel · Posted: Thu Mar 01, 2007 8:33 pm

ibmmainframesyntel · Posted: Thu Mar 01, 2007 8:34 pm

ibmmainframesyntel · Posted: Thu Mar 01, 2007 8:43 pm

hi people.........
give a good solution........
it will be more helpful for me....

can we use symbolic parameter in the control statement

superk · Posted: Thu Mar 01, 2007 8:47 pm

William Thompson · Posted: Thu Mar 01, 2007 8:49 pm

superk · Posted: Thu Mar 01, 2007 8:50 pm

The best thing for you to do is, of course, use a PARM:

William Thompson · Posted: Thu Mar 01, 2007 8:51 pm

ibmmainframesyntel · Posted: Fri Mar 02, 2007 6:39 pm

superk · Posted: Fri Mar 02, 2007 7:33 pm

Let me try to illustrate:

Let's say your job looks like this:

ibmmainframesyntel · Posted: Thu Mar 08, 2007 8:34 pm

In the PROC

//MYPROC PROC MIDAS=,NUD=
//*
//SCHECK1 EXEC PGM=IEFBR14,PARM='NUMGL.&MIDAS..&NUD..MAININD'
//*

using IEFBR14,what u are doing?

After this if the dataset is not found what it will be?

weather it will be JCLERROR or RC=08

let me knw

thank you

dick scherrer · Posted: Thu Mar 08, 2007 9:26 pm

Hello,

IEFBR14 is just an example in this case to demonstrate the symbolic parameter substitution. Instead of IEFBR14 you would use your own program (TTT to continue the example above).

From the dataset name constructed from the constants and the variables, create the LISTCAT control statement shown above.

When you execute the IDCAMS step, you will get a return code not a JCL error.

ibmmainframesyntel · Posted: Fri Mar 09, 2007 10:12 am

Ok..

hw to mention the the listcat entries...

LISTCAT ENTRIES('resolved entry name passed from PARM')

I don't knw...

pls tell me....

the output of the program will have the expanded dataset name
by substituting the symbolic parameters..

that i got itt.

but hw i mention the output in LISTCAT(' ')

let me knw...

dick scherrer · Posted: Fri Mar 09, 2007 10:26 am

Hello,

In your program that recieves the parm info, define an output file with a length of 80. In the jcl for running the program inclulde a DD statement for this file and it will have DISP=(,CATLG,DELETE). Place this output file on dasd and use it as input to a following IDCAMS step.

In your working-storage create an 01 LISTCAT-CTL that has fillers set up with the constant LISTCAT control info. In that same 01, declare a field where you will move the complete dsn.

Once you have the control statement in working-storage, write the 80-byte output record from that LISTCAT control statement.

Good luck and let us know if you would like anyh other info.