I haven't had that happen in many years. Stiction was a problem back in the days when manufacturers landed their heads on the platter instead of using a head lifter. The spindle motor didn't have the torque to break loose the stiction between two parallel surfaces. See item #44 for life with stiction:
Also gone are the days when the drives loaded all their firmware from below track zero off the platter. Even though there were 3 or more copies of the "firmware" (more correctly bootware), dimensional variations in the aluminum case was sometimes sufficient to cause a boot failure. That's because the servo tracks were NOT being used until after the bootware was loaded and therefore could not compensate for these changes.
More common are calls like "I turned off the machine, cleaned out the dust, and now it won't boot or turn on". What that really means is that the electrolytics in the power supply have cooled down and now have a higher ESR than when hot. Replacing the power supply and leaving the HD alone usually fixes that.
That's exactly the scenario I described to DecadentLinuxUserNumeroUno and the reason I don't like RAID.
Try blade or nano-ITX servers and SSD drives for lower power and less noise. At this time, they're probably too expensive for consumers but as prices drop, I think these will be the norm.
Nice. My plans are to do something like that. All storage goes on NAS boxes. All services on a dedicated ITX box running FreeBSD or Debian. Add FreeRadius to your list of services so that I don't have to deal with shared wireless pass phrases (WPA2-PSK). I have it mostly working at home.