This would work if you have sufficient time between alarm and failure. If some issue or process uses up all of your available disk space in a short time span, you won't have that luxury. Hopefully, the author is using alerts on top of having this failsafe