Re: CPF7CFD Error occurred running REXX procedure

On 1/3/11 3:42 AM, David FOXWELL wrote:

I have a job that has crashed with this message. Looks like a bug
to me but maybe someone will spot something. Thanks.

A CLP is called that issues a STRREXPRC with an SQL statement
'DELETE FROM MYFILE'

CHGVAR VAR(&SQL) VALUE('DELETE FROM ' *CAT &myfile)
EXECSQL CMD(&SQL)
MONMSG MSGID(SQL0000)

This command calls the rexx procedure below :

PARSE ARG iSql
RC=0

ADDRESS EXECSQL,
EXECSQL 'SET OPTION COMMIT = *NONE'
ADDRESS EXECSQL,
EXECSQL iSql

IF RC<> 0 THEN DO
MSGID = "SQL"RIGHT(STRIP(ABS(SQLCODE)),4,'0')
'SNDPGMMSG MSGID(&MSGID) MSGF(QSQLMSG) MSGDTA(&SQLERRMC)',
'MSGTYPE(*ESCAPE)'
END

The same CLP has been called over 5000 times for 5 different
multimember files. An override on the member is issued by the CLP
calling program. Mostly, the rexx procedure is returning SQL100
that is ignored on returning to the CLP.

In the log, I've got this :

Erreur. MCH1210 non intercepté par QSQREXX à la spécif *N, inst
X'0617'
6 *-* Address EXECSQL EXECSQL iSql;
+++ RC(CPF9999)
10 +++ MSGID = 'SQL'RIGHT(STRIP(ABS(SQLCODE)), 4, '0');
Error in line 10: Bad arithmetic conversion.
Error occurred running REXX procedure EXECSQL.

And in the dump :
<<SNIP & replace with symptom kwd string:
msgMCH1210 f/cblabranch x/0040 t/QSQREXX x/0617 >>

Instead of failing with an SQL condition, the REXX call-out to the SQL [i.e. per address EXECSQL] failed with an unmonitored exception [prior to the SQL being able to establish the SQLCODE variable\value. Thus the MsgId assignment is attempting the "Bad arithmetic conversion" of the string "SQLCODE" in the ABS() function; that function requires a numeric argument.

Searching the more generic symptom keywords msgMCH1210 and T/QSQREXX finds an old\similar [but improbable match, per failing instruction number] SE20177 at v5r4 and another [but without any noted failing instruction number to infer similarity] SE37756 at v6r1. Note: The symptoms appear as add-on versus actually associated with the APAR SE37756, and are found in a PTF cover letter versus the APAR text, so it may be that the PTF is corrective for the symptom, but as a side-effect of a change included since [although possibly before] that PTF SI35319:
http://www-01.ibm.com/support/docview.wss?uid=nas39de8d6a33b8e81c0862575fa0052b494

Regards, Chuck