Dan Rasch wrote:
I was monitoring a job that was not producing the results I
expected, and found out there was an OPNQRYF that redefined my
world.
Is there a simple way to check for the existence of an OPNQRYF
in a job? I did not see it in the overrides.
Both TRCJOB and STRDBG enable a simple way to determine if an
OPNQRYF was used [in a serviced process; with STRSRVJOB for another
process]. The output from TRCJOB to either spool or database file
can be reactively searched for the CPP [command processing program]
for the OPNQRYF command; or command if *DATA is included in the
trace. For an immediate notification of its use in a job being
serviced and debugged [with STRDBG], the ADDBKP command [Add
Breakpoint] can specify to break on statement '/1' of the CPP; that
quoted slash-one indicates the debugger should stop on the first
*instruction* of an OPM program object [syntax for instruction
versus statement on the STMT() parameter].
FWiW the breakpoint could perform a notification by some means
other than by the defaulted interactive "Display Breakpoint" panel.
That is accomplished using the BKPPGM parameter on ADDBKP to
specify that a user-defined breakpoint program should be invoked
instead; see the help text for the parameter requirements of a
breakpoint program.
When the debugged program is ILE versus OPM, or just as another
option is to CHGOBJAUD to track *ALL activity on the CPP or the
OPNQRYF *CMD command object. Then like with TRCJOB [reactively vs
actively notified], after the job runs, review the output [but for
example by DSPJRN of the QAUDJRN but with even more limited access
than for debug activity] to see if an OPNQRYF was used in a
particular job.
Unlike with only either query engine debug messages or auditing,
trace and breakpoint can enable easily detecting from the flow or a
stack, that the feature is invoked by\at the point pertinent for the
application debug. That is, the breakpoint and trace enable context
of the OPNQRYF invocation to be accessible whereas the relevance of
any detected\logged OPNQRYF activity may not be intuitive from
either of the joblog or auditing methods.
Regards, Chuck
