Error Handling with OPEN MACRO

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

Hi,

I am trying to open a file with "INVALID BLOCK SIZE". But the program abends as soon as it finishes OPEN MACRO instruction. Is there a way we can capture the return code and do error handling so that we can display some messages and exit gracefully.

Thanks
Ravi

Bill O'Boyle · Posted: Tue May 29, 2012 6:13 pm

First, you can test R15 (which will contain the Return-Code) for zero or non-zero. The popular method is via a Load And Test Register (LTR R15,R15) with BZ or a BNZ CC.

Another method uses the DCBD Macro -

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

Thanks for responding Bill,

For this kind of error "INVALID BLOCK SIZE FILE" the OPEN Macro does not return so the next instruction "LA R1,"FILENAME"" onwards would never get executed.

I am thinking if there is a way to modify the OPEN macro itself to hanle this kind of error

Bill O'Boyle · Posted: Tue May 29, 2012 6:49 pm

Try inserting the LTR R15,R15 right after the OPEN Macro and before the LA of R1. R15 should contain a value to interrogate. I'm assuming this is a variable-length QSAM file and not VSAM?

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

Hi Bill,

I have tried both the above but were not successful.

Thanks
Ravi Kuchi

Bill O'Boyle · Posted: Tue May 29, 2012 7:10 pm

The maximum LRECL = (LRECL + 4) and the BLKSIZE should not be greater than 32768 and be a multiple of ((32768 / (LRECL + 4) + 4) without a remainder as an optimum value. I have seen 32768 subsituted with the maximum value of one-half track associated with the DASD device/model, which lends itself to a more optimum BLKSIZE.

Are these similar to the calculations you're using?

This is a QSAM file, correct?

Review the SYNAD optional keyword of the DCB Macro. If there's an error, MVS issues an internal BAL/BAS to the label/routine specified for the SYNAD, performs your error processing and returns control to the NSI after the OPEN.

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

Hi Bil,

Not much of luck with SYNAD, not sure if I used it in the right way, here is the code in the attachment. Please see if u can find the issue:

By the way the file is just a flat file

enrico-sorichetti · Posted: Tue May 29, 2012 9:25 pm

DO NOT POST ATTACHMENTS, NOT EVERYBODY CAN SEE THEM
REDUCING THGE NUMBER OF PEOPLE WHO CAN HELP

USE A PLAIN TEXT CUT AND PASTE WITH THE CODE TAGS.

the attachment has been deleted!

Bill O'Boyle · Posted: Tue May 29, 2012 10:38 pm

My error. The SYNAD label/exit is used to capture errors in actual I-O (GET, PUT, DELETE, etc).

I don't know where else to point you to.

Are you Assembling/Linking with RMODE 24?

In other words, the program is not using an extended DCB?

dick scherrer · Posted: Tue May 29, 2012 11:59 pm

Hello,

Typically, what is done to resolve the error is to determine the cause of the open error and correct the code, the jcl, or re-create the input file.

Once the open error is resolved, there should be no need to "handle" an open error.

Possibly, there is something i am missing.

Bill O'Boyle · Posted: Wed May 30, 2012 12:20 am

In your JCL/DCB, does the LRECL and BLKSIZE match what's coded in the program's DCB LRECL and BLKSIZE or vice versa?

Just a SWAG....

dbzTHEdinosauer · Posted: Wed May 30, 2012 12:31 am

just a thought.

in cobol, when you use the file-status clause,
any i/o error can be trapped by the code,
and you can 'gracefully' exit the step,
instead of what is probably happening to the TS
which is he is getting about 400,000 lines of dump
because he can't get his jcl and code in sync,
and someone said 'hey, WTF is wrong with your program?'

dick scherrer · Posted: Wed May 30, 2012 2:49 am

Hello,

Maybe it is time to ask which specific abend are you dealing with?

Please post the error message and id along with the jcl being submitted and the file attributes for the problem file. Posting the DCB from the code may help also.

Bill Woodger · Posted: Wed May 30, 2012 3:09 am

If you know that the file you are trying to open has an invalid blocksize, why are you trying to open it? And why do you want to report and exit gracefully?

The combinations and possibilities for errors you get on an open are already dealt with by system messages. There are quite a number and variety of them.

Why do you want to do what you are trying (or trying to get us) to do?

PeterHolland · Posted: Wed May 30, 2012 11:55 am

Define an abend exit in the DCB via the EXLST parameter.
Info can be found here : DFSMS Using Data Sets

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

I think ACB and EXLST work for VSAM dataset only, but the file I would be testing is a flat file.

The problem or issue is as follows:
Before opening the file we do not know if the file is valid or invalid, so incase the file is invalid(block size) I would like to trap the error and perform some record keeping stuff and end the job

Below is the code, I am trying to work on:

PeterHolland · Posted: Wed May 30, 2012 5:08 pm

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

Hi Peter,

Sorry for the misunderstanding, I am a beginner in assembler so please excuse.

Here is the piece of code I am trying to execute:

Bill O'Boyle · Posted: Wed May 30, 2012 6:23 pm

I've been under the impression this is a variable-length file. But, your DCB RECFM specifies FB?

Is it fixed or variable?

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

it is fixed..

PeterHolland · Posted: Wed May 30, 2012 9:54 pm

All the time i see something like a "MEMBER" variable. Are you reading a PDS?

Grant Goodale · Posted: Wed May 30, 2012 10:54 pm

The EXLST= parameter specifies the address of an exit list, not an exit routine.

PeterHolland · Posted: Thu May 31, 2012 12:01 pm

Your EXLIST should look like this :

PeterHolland · Posted: Thu May 31, 2012 1:38 pm

Ravi Kuchi · New User Joined: 29 May 2012 Posts: 14 Location: INDIA

Hi Peter,

Thanks a lot !

The EXLST seems to be working, now I am into the FERROR para when I encounter the corrupt file. Just was wondering how I can set the return code to zero and exit gracefully.

No, I am not reading a PDS here, sorry the variable would have got carried as it is a work in progress module.

Thanks again!