MVSFORUMS.com

maverick05 · Beginner Joined: 15 May 2005 Posts: 72 Topics: 24

Hi,

How to move forward if a prior step fails with JCL errors in a JCL with many steps. Ex: If previous step fails because a dataset doesn't exist,my job fails. I can't use condition code here since it is not applicable for JCL errors.

Is their any alternative or shortcut to bypass jcl errors caused by prior steps?

Tx

papadi · Supermod Joined: 20 Oct 2009 Posts: 594 Topics: 1

No.

What you might do is ensure the datasets exist in an early step and then the run would not terminate in the middle due to a missing dataset.
_________________
All the best,

di

maverick05 · Beginner Joined: 15 May 2005 Posts: 72 Topics: 24

As specified in the post,there are many steps and verifying the existence of each step takes time.

papadi · Supermod Joined: 20 Oct 2009 Posts: 594 Topics: 1

The suggestion is not to verify each step - one step in the beginning will suffice.

And verifying should not take any significant time.
_________________
All the best,

di

Anuj Dhawan · Posted: Tue Nov 10, 2009 5:04 am Post subject:

jhandu · Posted: Tue Nov 10, 2009 8:06 am Post subject:

1) if there is a utility to check jcl syntax, use it before submitting
2) If there is no such utility, use TYPRUN=JSCAN in the jobcard

dbzTHEdinosauer · Posted: Tue Nov 10, 2009 8:28 am Post subject:

kolusu · Posted: Tue Nov 10, 2009 12:31 pm Post subject:

dbzTHEdinosauer · Posted: Wed Nov 11, 2009 7:07 am Post subject:

I stand corrected.
_________________
Dick Brenholtz
American living in Varel, Germany

Anuj Dhawan · Posted: Wed Nov 11, 2009 7:33 am Post subject:

I believe response from Dick is appropriate keeping the origianl request from TS in mind...
_________________
Regards,
Anuj

RonB · Posted: Wed Nov 11, 2009 7:47 am Post subject:

JCLCheck et. al., can only test for the existence of a dataset ( or GDG Base ) at the time that THEY, themselves, are run. Obviously, they cannot accurately determine whether OLD/SHR datasets or GDG generations will exist at the time the JOB actually RUNS.
Some job schedulers ( e.g. CA-7 ) "simulate" job step executions when staging a JOB to determine whether such datasets will exist by the time a given step is executed ( assuming that prior steps execute successfully ). E.G. if STEP7 references a dataset that doesn't exist at JOB initiation, but will be created in STEP3.
A good solution in any case would be to run an IDCAMS VERIFY step where appropriate to set return codes that can be tested.
_________________
A computer once beat me at chess, but it was no match for me at kick boxing.

dbzTHEdinosauer · Posted: Wed Nov 11, 2009 9:56 am Post subject:

Anuj Dhawan · Posted: Fri Nov 13, 2009 6:07 am Post subject:

dbzTHEdinosauer · Posted: Fri Nov 13, 2009 9:02 am Post subject:

well, since I did not elaborate as did RobB, my comment about JCLcheck - at face value - was false.

I appreciate your 'protection', but my big mouth often gets me into trouble,
no need for others to drown, also.

My biggest grip about these boards is lack of 'proper terminology'.
Computers are precision instruments and require precise verbiage.
I crossed that line with my post, and what Kolusu's posted was valid.
RonB, as is his habit, made a precise comment;
I felt no need to add or attempt to defend myself.

thx anyway, did not mean to off you.
_________________
Dick Brenholtz
American living in Varel, Germany

kanitha-mvs · Beginner Joined: 17 May 2006 Posts: 26 Topics: 17

COND=EVEN could be used to execute a step, even if a prior step abends.