On Thu, 1 Jun 2017 06:37:04 -0500, John Gateley wrote:
>I had a similar problem. In my case I was reading JESYSMSG to check the return
>codes of preceding steps and usually JESYSMSG was there but occasionally it
>was not, and in these cases the job elapsed time was quite low.
>I added a call to an assembler program at the start of the REXX to issue a
>STIMER for 0.7 seconds and then JESYSMSG was always there.
>The 0.7 works even when the only previous step is an IEFBR14.
>Since then I have written an Assembler program to do the browse using DYNALLOC
>of USERID.JOBNAME.JOB01234.JESYSMSG and that does not need to sleep at all. So
>there does seem to be a delay in REXX/SDSF but I have no idea why.
>
I looked at a job a few months old where I had coded:
address 'SYSCALL'
...
RC = bpxwdyn( 'alloc dd(TESTO1) sysout lrecl('Len') msg(WTP)' )
/* Write a long record to SYSOUT. */
address 'MVS' 'EXECIO 1 DISKW TESTO1 (finis stem L.'
'sleep 03' /* Race condition? */
address 'TSO'
'TIME'
address ISPEXEC 'vget ( ZSEQ ) shared'
'%ISFUNLD -z JOB'ZSEQ
...
so there seems to be a lot of that going around. 3 seconds worked;
I doubt that I experimented with smaller values.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
