Handle ROLLBACK in EXCI interface

Pandora-Box · Posted: Fri Sep 21, 2012 2:38 pm

I have 3 programs

Program A(Batch program) calling program B (which has synconreturn).Program B links CICS program C which does table updates.

1 St call C Updates table
before 2nd call program abends in A
No roll back happens

Program A(Batch program) calling program B (which has no synconreturn).Program B links CICS program C which does table updates.

1st call C updates table
2nd call happens program abends in B for the below error

dbzTHEdinosauer · Posted: Fri Sep 21, 2012 6:12 pm

I believe when PGM C returns to PgmB
a commit is made by virtue of the CICs Task Terminating.

Why do you want a RollBack, anyway?

A and B are batch,
B has the execi link

you are talking about rollback if A abends, but here B has abended.

also, terminology. A program ABENDs
because the OPS-SYS determines that the offending Program
can be no-longer allowed to continue.

what you have showed us is the execi response code to a LINK.
nothing abended.

Pandora-Box · Posted: Fri Sep 21, 2012 6:27 pm

Thanks Dick

In my first case

Program A(Batch program) calling program B (which has synconreturn).Program B links CICS program C which does table updates.

1st call to program C through B Updates table on gets commited due to SYNCONRETURN if my program abends due to some issue say SOC7, or due to any other table processing failure within program A I actually wanted the program to roll off updates happend through program C.

I wanted to rollback because after fixing the issues on restart I tend to process the same data again and encounter a duplicate entry and my program fails again

dbzTHEdinosauer · Posted: Fri Sep 21, 2012 6:34 pm

well,
i think you first design error was encountered
when you did not realize that a COMMIT or ROLLBACK
is a task oriented activity
and you essentially have two tasks:
the batch task
the cics task.

since you can not issue a COMMIT or ROLLBACK in one task
that will affect the other,
sounds as if you need a different approach in your RESTART process.

Pandora-Box · Posted: Fri Sep 21, 2012 7:15 pm

I understand Dick.

I guessed so just thought if there was an approach to ROLLBACK things

Thanks Dick for your inputs.

Pandora-Box · Posted: Fri Sep 21, 2012 7:35 pm

Hi Dick,

Btw removing SYNCONRETURN from the EXCI will that make any help??

Because I tried that

What I did is the following actions

A links C through B returns successful
A links C through B abends with

dbzTHEdinosauer · Posted: Fri Sep 21, 2012 8:47 pm

I would separate the db2 logic in the cics program
and create a new/non cics program that can be called from batch or cics.

why does the db2 logic require cics?

reusable code does not mean figure out how to jump into a convoluted mess of old code to take advantage of one function.

it means separate the function - make a separate module -.

yeah I know, fixing it so that it conforms to modern program techniques
is harder
than figuring out how to set 13 different switches
and do this in the batch program to offset that
which is done in the quagmire of existing code.

at some point, you just have to change something.

the modular programming technique was so named
because people could not grasp the idea of each module only performing one function.

the days of 15000 lines of spaghetti code should come to an end.

CALLs do not take that much time.
and even if they do, computers run at such speeds
that writing function oriented modules is much easier to maintain
and implement.

so you have 100 200-line modules instead of 1 20000-line modue. HFC?

Pandora-Box · Posted: Fri Sep 21, 2012 9:29 pm

Initially we had this program to be accessed for different functions over a series of tables for CICS

Later point it was realised we need them for Batch so we decided to reuse this rather developing one similar for batches

so that is the history

Thanks again :-)

What is HFC?

dbzTHEdinosauer · Posted: Fri Sep 21, 2012 9:44 pm

dbzTHEdinosauer · Posted: Fri Sep 21, 2012 10:17 pm

sorry Pandora_Box,

i should not subject you to my sermons
when you are not in a position to make any major changes.

you are stuck trying to debug you exci stuff.

without further info about your unique implementation,
it is tough to give you suggestions,
considering the many varied ways exci can be implemented.

good luck.
possibly others with better first hand knowledge of the interface
can help you.

Pandora-Box · Posted: Fri Sep 21, 2012 10:33 pm

No please never be sorry

I just treat them as invaluable inputs

Its not that I cant change them Considering the timeline we have its tough

Also I am sorry I thought I gave you all I want

Plz let me know what details you need I will share them asap

Yes as said I will wait

dbzTHEdinosauer · Posted: Fri Sep 21, 2012 10:43 pm

here is a writeup on SYNCONRETURN:
do a right click and open in new tab:
publib.boulder.ibm.com/infocenter/cicsts/v3r1/index.jsp?topic=%2Fcom.ibm.cics.ts31.doc%2Fdfhtm%2Fdfhtm49.htm

as you can see, depending on your site, things can get confusing.

at any rate, you now have a link to the manual.

Pandora-Box · Posted: Sat Sep 22, 2012 11:07 am

Thanks for the link Dick

Will Call to SRRBACK in program A when there is use of SYNCONRETURN resolve?