Depending on your usecase you can integrate using your backups occasionally into your normal data processing. Again, depends on usecase but then it becomes int

Depending on your usecase you can integrate using your backups occasionally into your normal data processing. Again, depends on usecase but then it becomes integral to your existing workflow instead of an addendum that you end up forgetting to do The whole purpose is to make the failing of one be effectively extremely noisy and irritating It's like what I do with raid. I have a script that will shut the machine down on drive failure and then will use dialog(1) to say something like "hey bozo replace the fucking drive first" when you boot it up and then it will shutdown again and be unusable. Make the complaining show stopping, loud, rude, and disruptive. Because if the next one fails you're screwed