On Fri, Jan 31, 2020 at 6:34 PM James H. H. Lampert
<jamesl@xxxxxxxxxxxxxxxxx> wrote:
Is there a way, short of setting up a separate job schedule entry for
every time the job should run during a typical day?
Um, sort of. But not really in the job scheduler itself.
Basically, you have to control it yourself in the program being
launched by the job scheduler. You could either loop with delay, or
resubmit for the next iteration before ending. This is not necessarily
*completely* self-perpetuating, since it relies on the initial launch
by the scheduler, and then self-perpetuates for the rest of the day.
Why would you do this (having the called program only handle multiple
runs for a day) rather than fully self-perpetuating? Well, I can think
of two plausible reasons: (1) The logic may be simpler, so it's easier
to write and maintain. (2) If you have a policy of always having a job
schedule entry for automated jobs, you may be able to find them more
easily. Fully self-perpetuating jobs are harder to discover if you
aren't the one who wrote/initiated them. They are less
"self-documenting" so to speak.
Note that this also applies to jobs which are run too *infrequently*
for the (basic) job scheduler to handle directly, such as quarterly,
or every third Monday. In those cases, you also could either have
fully self-perpetuating programs, or do the hybrid "ride the scheduler
but then have your own refinements within the launched program".
I will say that I really, really dislike having tons of scheduled jobs
to run the same program multiple times per day. It's something we do
at my shop a lot, I think because most of my coworkers and especially
my predecessors cannot handle programming any time or date arithmetic
whatsoever.
John Y.
midrange.com
