As a bit of background, over the weekend the main air conditioning system for our server room went down. As I was out of town I was unable to stop in and check to make sure the auxiliary system was functioning properly. When I got to work this morning all the servers seem happy.
The question I have, is there any sort of temperature logging functionality that runs on unix based servers? I would love to know how the auxiliary aircon units went and keeping things cool! We have servers running Solaris, FreeBSD, Ubuntu (and if it helps Windows server 2003), however currently have no monitoring applications running. Thanks.
A general option to read temperature data from Linux servers might be with lmsensors. If your servers are IPMI capable, that could be another topic to evaluate. When you found out how to get the temp data out of your system, you can log it with a myriad of options, from a simple script logging into plain text to graph creation via rrdtool or a full fledged monitoring solution like Cacti, MRTG or Nagios.
I do what SvenW recommends. If you have lmsensors set up, then munin and munin-node are used together to measure the mainboard, CPU, and drive temperatures. This provides an indirect measure of how warm the computer room is.
I also have a couple of systems with USB-attached temperature sensors, which are read by a munin script and graphed that way.
Finally, I have a hardware environmental monitor system that can generate alerts via email when thresholds are crossed (The AVTECH RoomAlert 11E is the one we are using). This device is also queryable via SNMP and those values are graphed by cacti.
Depending of your servers, you should take a look at "DELL OpenManage IT Assistant", "HP System Insight Manager" or any equivalent to monitor all your servers sensors.