Jobs getting abended with S0C1

Nivas_mumbai · New User Joined: 19 Jul 2008 Posts: 11 Location: Mumbai

Hi,

We have 8 Mainframe Batch jobs where the same program(Ex:Prog1) is being used in all the Jobs. This program(Prog1) is calling another program(prog2).Similarly Program2 is calling another program(prog3).

In the above scenario program3(Prog3) is a cobol program which can be called from both batch and online jobs.

All the 8 jobs ran at a time and only one job is running successfully and the remaining jobs getting abended with S0C1 error.

Below is the Sysout details:E0712MLP is my third program(Prog3)

[color=red]The system detected an operation exception (System Completion Code=0C1).
From compile unit E8721MLP at entry point E8721MLP at compile unit offset
-21210160 at entry offset -21210160 at address 00000000.

. Possible Bad Branch: Statement: Offset: +00003D66
.<> LEAID ENTERED (LEVEL 10/15/2007 AT 09.21)
.<> LEAID PROCESSING COMPLETE. RC=0 [/color]

Can any one please help me out why we are getting this abend.

dbzTHEdinosauer · Posted: Wed Jul 23, 2008 11:35 am

I would start here:

. Possible Bad Branch: Statement: Offset: +00003D66

Nivas_mumbai · New User Joined: 19 Jul 2008 Posts: 11 Location: Mumbai

Hi Dick,

Could you give me more details about bad branch.How we can we resolve it.

dbzTHEdinosauer · Posted: Wed Jul 23, 2008 12:42 pm

?? I have no idea what your code does.

Guessing................

since the same run unit is invoked in more that one job, I would start by analyzing the input.

Soc1 means you are trying to modify data at an address outside your pgms address space.
1. Branches to nowhere,
2. indexes (or subscripts) going beyond the max value,
3. poorly linked modules,
4. modules linked for dynamic yet the submodule is linkedited to be statically CALLed.
5. incorrect USING statements

since it works, i would not spend too much time on items 3 or 4. 1 can happen when 4 exists, or if you are using ALTER statements or assembler modules are involved.

5 is possibly not involved, but would not bet on it.

2 would be my best guess.

the above list is not complete, but is a starting point. IBM has very thick manuals that you can read that discuss abend handling.

Nivas_mumbai · New User Joined: 19 Jul 2008 Posts: 11 Location: Mumbai

Hi dick,

Thank you very much for detail explanation.

expat · Posted: Wed Jul 23, 2008 1:33 pm

dbzTHEdinosauer · Posted: Wed Jul 23, 2008 1:45 pm

expat,

probably not.

dick scherrer · Posted: Wed Jul 23, 2008 7:58 pm

Hello,

The 0c1 is an operation execption (rather than an addressing execption) - very often caused by trying to execute an op-code of x'00'.

Are these new programs? Did these program work at one time? If so, what was recently changed?

If program3 is to work both in batch and online, it needs special attention. . . There is recent discussion in the forum about code that will work in both environments.

Bill Dennis · Posted: Wed Jul 23, 2008 11:42 pm

A clue may be all pgms trying to run at once. Perhaps only one job gets a common file allocated and the others get a bad RC from OPEN but try to WRITE anyway?

Nivas_mumbai · New User Joined: 19 Jul 2008 Posts: 11 Location: Mumbai

Hi Dick,

Prog3 works both in Batch and online.Lot of other applications are using these module.First time we included this module in our batch process.

dick scherrer · Posted: Thu Jul 24, 2008 8:30 pm

Hello,

Nivas_mumbai · New User Joined: 19 Jul 2008 Posts: 11 Location: Mumbai

Hi Dick,

In our application only we included this program first time.Lot of other applications are using this module for long time.Infact this module is compliant to both online and batch process.

If it is not compliant to batch process,then how few of our jobs running with out any problem.

In our appliation alone we included this module in 19 jobs.All the jobs run at the same time.First 11 jobs running successfully.Remaining 8 jobs abended with S0C1.

One more interesting fact is after the abend we are simply check out the Calling program into a changeman package and did recompile.Then we are running job by giving this Load module,then these jobs running successfully.

Anuj Dhawan · Posted: Fri Jul 25, 2008 10:24 am

Nivas_mumbai · New User Joined: 19 Jul 2008 Posts: 11 Location: Mumbai

Hi Anuj,

We are running it in production only.All the jobs are using the same files.

Anuj Dhawan · Posted: Fri Jul 25, 2008 1:00 pm

If so, then with the information so far in hand, I would go with Bill's susggesion..

dick scherrer · Posted: Fri Jul 25, 2008 7:37 pm

Hello,

It sounds like the original promote to production did not link all of the modules correctly. Your checkout/re-compile has probably corrected the "corrupt" load module.