>The only way I have found so far that would be certain that I do not have duplicate record is to create a task with the minimum interval, thus it would run every 15 seconds. That task will simply have a responsibility to create the daily record. If other robots are in need of the daily record and it would not be yet available, that would be from midnight + 15 seconds, I could establish a 15-second wait time with a retry to grab the daily record after, as it would then be created from the new task.
>
>If someone has an alternate design, I would be happy to see if I can use it.

candidate index on date of dayly rec and robots not thrown by error from dupliacte creation tries come to mind

update: If Upsert/Merge can be made to work, it might be less code and be cleaner as it should delegate handling to the server. Upsert not in all SQL dialects might be a show stopper.