Failing harddrive?
 

I've got a server with two IDE harddrives. I monitor the temperature of the harddrives and have noticed that one of the disks shows intermittent drops in temperature (see http://www.welzien.se/misc/hddtemp-week.png). Does this indicate that the disk is about to fail and should be replaced? The two disks are configured as a software RAID-1 array.

No, it's more likely a firmware error. I've seen that issue as well, and work around it by bezier smoothing the values (gnuplot). A failure would look like a gradual ramp-up of drive temperature, that cannot be explained (loss of AC, fan, clogged air intake, etc.). The smoothing removes the minor variations, while still letting you see the trend.