Tom, is no way to directly limit the numbr of tries. First, given the
variety of TCPs and the platforms they operate on, and that TCPs don't
keep historical data on failed connects, it would be impractical to have
any sort of limiter at the TCP layer. That leaves us with the ADSM
Schedule Prompter. The prompters sole job is to try to connect with clients
for which there is scheduled work to do. The prompter will attempt every 60
seconds to contact such a client until the window for the schedule elapses.
Once the window elaoses, the related node is removed from the prompters list
of nodes to contact. So given this, you could reduce the schedule duration
as way of limit how long the attempts are made. Long durations are often
only useful if your clients run in polling mode with randomization. I would
however recommend a duration of at least 15 minutes. If the ability was
added to have the prompter remove a node after n nbr of tries, then depending
on what the client is using for "retry period" and "maxcmdretries" you
may wind up with clients that don't get backed for long periods of time.
Unless you are monitoring client success/failures frequently with the
server  Q Event, this may go un-noticed. In some situations you may have to
manually restart the DSMC SCHED clients to get them back on track. All in all,
most customers prefer to know immediately when clients are failing.
 All that being said, one of the major benefits for prompting mode is if you
frequently change schedule starttimes or durations where you would want the
clients to be aware of the changes in a timely manner. If your schedules
are fairly stable, or if you have problematic cients (TCP not loaded, or
DSMC SCHED not loaded), then you may want consider polling mode. Polling
mode leaves it up to the client to contact the server for schedule information
and then run at the appropraite time. This would eliminate the resources usage
from the prompter trying to contact where contact is failing.
 I hope this isn't too confusing. In summary, try adjusting durations,
retryperiods, and maxcmdretires, to find a happy medium in terms of resources
used. And consider polling mode where appropriate.

Tom Brooks
ADSM Development