We're working on "condor_off -drain", so that you can use the interuptible method of draining for this task, instead.
You could also have used "condor_drain -exit" instead of "condor_off -peaceful -startd" (if you wanted the startds to stay off until you turned them back on). You'd have to cancel the drain for the machines that were still draining as well as "condor_on -startd" the others.
- ToddM