We just found out earlier this week that one of the CPU fans on our main web server which hosts cfdev.com, this site, and a number of other sites had burnt out. We think that the fan has been out for a while, and had been causing some strange crashes on the weekends when the AC is not as cold in our building. Hopefully the CPU is not damaged - it appears that it still works but once the new fan arrives we will have to do some testing.

So we now run lm_sensors, and if your servers running linux I would recommend you do the same. Most modern motherboards provide CPU temperatures and fan speeds often found in the BIOS, but you can also get this info from OS.

Here's a good article on lm_sensors from the linux journal.

Our next step is to find or write some software to automate this monitoring process, and send out alerts. Anyone know of any?