Re: Controlled versus immediate.

rob@xxxxxxxxx wrote:

Been suggested that I replace using ENDSBS *ALL *IMMED
with ENDSBS *ALL *CNTRLD DELAY(120)
That I can understand because of the .0005% of programmers who actually use the shutdown op code or check for that particular status when doing an i/o or something.

It was also suggested that I change
PWRDWNSYS OPTION(*IMMED) RESTART(*YES *FULL) IPLSRC(B)
to
PWRDWNSYS OPTION(*CNTRLD) DELAY(120) RESTART(*YES *FULL) IPLSRC(B)
If I've already done the ENDSBS, and waited for the appropriate message in QSYSOPR, that what's the benefit?

Rob:

I regularly use DELAY() to give system code time to do its clean-n-neat shutdown work rather than to let 'applications' shutdown nicely. I.e., I rarely expect that some everyday application code path is going to poll for shutdown status; but I _do_ expect things like the Mail Server Framework or whatever to be very well behaved in that respect.

If it so happens that a Payroll app or some app based on stored procs or whatever handles system shutdown nicely, so much the better.

PWRDWNSYS doesn't matter quite so much to me because the "system" is generally already ended when PWRDWNSYS is executed. But adding DELAY() rarely makes a lot of difference there since I also signoff right after the command is entered. The signoff ends the last job, so the delay period actually ends when the console job ends, not two minutes later.

Tom Liotta