Pick a few subnets that are failing and break down the job to focus on one subnet at a time. Check 100% that you assigned subnet rep is actually on and that the details for it in LANDesk are correct. If you then schedule a job to that subnet do you get a failure? If you do, try choosing another machine as a rep and see if that works.
When you migrated, did you upgrade the agenets on all of the subnet rep machines? If not, make sure failed reps are fully up to date.
If you have followed these steps and you can reproduce the failure then we can look into other possibilities.