On 19 Dec 2012 16:42, Darryl Freinkel wrote:
I have a need to list all objects on a system daily.
DSPOBJD *ALL/*ALL *ALL
to an *outfile takes too long.
I will need the creation and change dates and size.
Is there another way to achieve this?
If "all objects" is sufficiently defined by what is available in that
DSPOBJD request, then the speed can be improved using parallel requests.
Ages ago when only /QSYS.LIB objects were of interest to me, I had
used a CLP which submitted jobs for each library to a somewhat throttled
job queue. The CLP gets a list of libraries using a DSPOBJD QSYS/*ALL
*LIB *FULL OUTPUT(*OUTFILE) and then did a RCVF against the output file
doing something like SBMJOB CMD(DSPOBJD &ODOBNM/*ALL ...) or SBMJOB
CMD(CALL PGM(ObjDbyLib) PARM(&ODOBNM)) for each record [while the
&ODOBTP *EQ '*LIB' because the same OUTFILE() is being used in all of
the concurrent requests]. The biggest issue is ensuring all of the jobs
complete without error and if desirable, also are awaited batch job
completion to allow for a synchronous operation of the submitter.
Since the database can effect parallel processing and allows an
external program to be called, another option is to have a SELECT of the
data from the DSPOBJD of the OBJTYPE(*LIB) whereby a scalar UDF lists
the objects. Something like the following for example:
setup:
dspobjd qsys/*all *lib *full output(*outfile) outfile(qtemp/od)
crtdupobj od qtemp *file *curlib data(*yes)
chgpf *curlib/od size(*nomax)
create function dspobjd
( libname varchar(10)
)
returns char
language sql
not deterministic modifies sql data
external action returns null on null input
allow parallel
set option dbgview=*SOURCE
begin /* note: no error handling */
declare cmdstr varchar(300);
set cmdstr = 'dspobjd ' concat rtrim(libname)
concat '/*all *all output(*outfile) '
concat 'outfile(*curlib/od) outmbr(*n *add)';
call execcmd(cmdstr); /* Proc invokes QCMDEXC */
return '1';
end
Activate output; e.g. the following SELECT:
create table qtemp/genObjDesc as
( select dspobjd(odobnm), odobnm from qtemp/od )
; -- as a report, but use in insert or create stmt:
....+....1....+....2....+....3
DSPOBJD ( ODOBNM ) Object
1 LIB1
1 LIB2
******** End of data ********
Output would be the list of libraries from the DATA(*YES)
duplicated from the original DSPOBJD into the QTEMP/OD, then a record
for each object added to the OUTFILE(*CURLIB/OD) as generated by the
DSPOBJD LibName/*ALL *ALL *FULL request... which if the database SQL is
allowed to, can run parallel. Of course each DSPOBJD requests would
effect\log the diagnostic CPD000D rc2 "Command *LIBL/DSPOBJD not safe
for a multithreaded job."
midrange.com
