WD Caviar Green disks (WD15EADS-00S2B0 in my case) like to unload the drive heads to save power. The ReadyNAS likes to load them again. This causes two problems: Frequent eight-second delays as the drive heads move away from the platters and then back towards them, and in my case, the drives would reach their stated lifetime (300,000 load cycles) in less than a year.
Several solutions exist. Upgrading the ReadyNAS to 4.2.8 should solve it, but that version isn't available for the NV+. Removing the drives, one by one, to run WD's wdidle3 using a different computer should also help.
What I did instead: Install EnableRootSSH, log in as root, and tune down the greenness using hdparm.
hdparm -B 254 /dev/hdc
hdparm -B 254 /dev/hde
hdparm -B 254 /dev/hdg
hdparm -B 254 /dev/hdi
This setting survives reboot.
The argument to -B ranges from 1 to 254, plus the magic value 255 (
disable power saving entirely). The original value was 128 and made the drive much too eager to load/unload. At 254, I see just a few load/unload cycles per day (much less than 100), so the disks are unlikely to reach 300,000 load-unload cycles at all. I think I'm going to try 230 or so later, to see what that does to load cycle count and drive temperature.
hdparm can also report the -B setting:
hdparm -I /dev/hdc
Advanced power management level: 254
Each drive's current load cycle count is displayed by the ReadyNAS' SMART+ button (in System → Health).
Update: Even at 250 the drives suffer too many load cycles. I'm running with -B 255 now, and the drives spin down after 20 minutes of inactivity.