Help with SOC4 on DSNHLI/DSNELI

Kiran K Menon · New User Joined: 25 May 2011 Posts: 7 Location: India

Hello All,
Do you guys happen to encounter SOC4 exception on Cobol DB2 program on the DSNHLI/DSNELI call ?

In my case, we had an implementation to a higher test environment which involved changes to few db2 tables and identified program changes. As a customory practise with our client, they tend to rebind all DBRMs (irrespective of whether a change has happened on the modules involved). Unforunately, once the rebind had happened, some of the existing modules started failing with a SOC4 on DSNHLI call.
Batch programs failing with SOC4 and some onlines showing an abend code of AD2R

The message from the batch job is as below (with the offset address pointing to an INSERT query)
eg :
RUN PROG (abcdefg) PLAN (abcdefg)
IKJ56641I DSN ENDED DUE TO ERROR+
IKJ56641I SYSTEM ABEND CODE 0C4 REASON CODE 00000038

We have tried to recompile the abended module in a lower environment and promote it back to the higher environment followed by a rebind again but it wasn't helping either. Please note that the same module works fine in a lower environment and we are not aware of any DB2 version difference between the 2 regions. Also, note that there are other DBRM that got rebinded (without changes) holding good. And everything was working fine until this changes has gone in.

The abendaid was showing the below message
COMMAND ===> SCROLL ===> CSR

A storage reference exception occurred during execution of program DSNELI.
The expected completion code is S0C4.

Analysis of Error

The next sequential instruction to be executed in program DSNELI was at
displacement 000001DE.

The execution was a SELECT ID INTO :ID
FROM FINAL TABLE (INSERT INTO DB.TABLENAME(FIELD1,FIELD2, ID) VALUES('FIELD1','FIELD2',Current timestamp) kind of SQL

Any help in the right direction is highly appreciated

Thanks

Kiran K Menon · New User Joined: 25 May 2011 Posts: 7 Location: India

Just wanted to add that the table on which the INSERT / SELECT is happening is not modified as part of the changes

Rohit Umarjikar · Posted: Mon Apr 25, 2016 3:08 am

It looks to me a table space issue for the batch programs and for online AD2R yoU will have see the dump , start here.
at first what did you site support advised you on these abends? Did you even approach them ? Such issues only them can resolved as they know your system but we don't.

Kiran K Menon · New User Joined: 25 May 2011 Posts: 7 Location: India

Hello Rohit,
The support group (including the DBAs) were unavailable during the time to comment on it. I have tried to look through the SYSTABLESPACESTATS and SYSTABLEPART with my limited knowledge but didn't notice anything suspicious.

The AD2R abend is happening for a similiar SELECT / INSERT combination explained above.

I guess the DBAs would be back in office today to take a look at it.

Thanks.

Rohit Umarjikar · Posted: Mon Apr 25, 2016 11:13 am

Thats good. please share us back with your fix so that others can also be benefited.

Nic Clouston · Posted: Mon Apr 25, 2016 7:51 pm

"The abendaid was showing the below message
COMMAND ===> SCROLL ===> CSR"

Please review your posts before submitting (or immediately after) and fix stuff like this. You should know by now when, and how, to use the code tags.

Kiran K Menon · New User Joined: 25 May 2011 Posts: 7 Location: India

Update : No solution on this so far from DBAs either

On a separate note, we are unable to run the above query even using IKJEFT01 (as a stand alone query) in the higher enviornment but is working perfectly fine in the lower regions. When the INSERT was run seperately there were no issues but the moment we use SELECT FROM FINAL TABLE with an INSERT it goes for a toss.

Rohit Umarjikar · Posted: Tue Apr 26, 2016 10:18 pm

You will have to share us the error message and the sqlcodes and also the DB2 version on which you made your tries.
what do you mean by higher and lower environments? DBA's should be able to help you.

Kiran K Menon · New User Joined: 25 May 2011 Posts: 7 Location: India

fyi,
Finally got a resolution to the issue. As part of testing the IBM Accelerator product the DBAs had recently changed a subsysem parameter called ACCELMODEL from NO to YES. Once this parameter was changed back to NO, we rebinded all the packages again at it worked fine.

Rohit Umarjikar · Posted: Wed Apr 27, 2016 7:19 pm

Thanks for letting us know.

Robert Sample · Posted: Wed Apr 27, 2016 7:50 pm

This is a perfect example of why your FIRST point of contact should be your site support group, not a forum (this or any). Nobody on this forum works for your site and hence we would have no way of knowing that your site support group had changed site parameters; if you had contacted them first, perhaps they would have realized that their change was impacting users sooner.

Kiran K Menon · New User Joined: 25 May 2011 Posts: 7 Location: India

Robert, i wasn't waiting for an update from the forum before reaching out to the support group. I have mentioned that in my third post above.