OK... for grins let's recap what we know so far from the thread and the searches I'm sure you've done here:
a. 'Reset if required then Run' in a Job Activity stage does exactly what it sounds like it does, allowing an Aborted job to be rerun without a manual reset first.
b. Enabling 'checkpoints' in a Job Sequence allows it to (if desired) restart a sequence of jobs from the point of failure.
c. Restart is a synonym for rerun. And the way you rerun
anything (regardless of tool) multiple times in an unattended fashion is with a
loop of some kind. Multiple choices here with yet-another-Sequence-job being the prime candidate mentioned. Use that in conjunction with the Loop Start/End stages, otherwise you'd still need something
else to force iteration. You could even use a script for this with 'dsjob' inside a loop.
d. Do
not loop forever. Personally, I would suggest no more than two times, exiting after a successful run of the job or failure on the second iteration, meaning the reset/restart did absolutely no good. Of course, you could configure any kind of loop limit in that you felt was appropriate, including notification after it starts to loop 'too long'. How robust and bullet-proof you make it is totally up to you.
Hopefully, all that should get you going, even if it isn't the entire procedure, detailed out step-by-step in glorious minutia.
