pontcysyllte writes: "Some years ago I was local tech support for cc:Mail. This product had the uncanny ability to predict a hard drive head crash or similar failure. I had direct experience of this capability at two different customer sites. That is, when I was contacted to explain error messages being sent to the email administrator, I could look them up in the manual(*) and advise customers to: (1) back up your email database immediately; and (2) swap the hard drive out and replace with a new unit as a matter of urgency (the old drive could be used in a non-critical area). In both cases the old hard drive failed a few weeks later. Does anyone know how this was achieved? I can make guesses of course, but it would be interesting to know the facts. cc:Mail was highly dependent on its mail database, which was a single file containing all users' messages at that post office (excluding attachments), so it was crucial. I have a vague memory of a white paper... (*) Perhaps some readers remember the days when decent software produced meaningful error codes and came with technical manuals which in many cases not only listed each code but explained which program module(s) might flag the error, what conditions might cause them to do so and what this might mean to the user."