Kimball Larsen wrote:
Possibly not - but if the temp ever gets to 150 in the first place,
the new A/C unit I installed is no longer working properly, and I'll
be alerted to the fact I gotta get some cold air in that server room
asap.
That's the only reason I needed this. :)
I am working on a project at work that does exactly this. Unfortunately,
it's not exactly available to the public. :) We allow users to customize
alarms and thresholds and can take preventative action like shutting
down equipment that is over-heating or alerting the staff in an
operations center. We've actually built quite a framework of it, all
using Qt with a pretty nifty ORM backed by MySQL on Fedora 10, but I
digress.
We manage multi-rack systems of hundreds of pieces of equipment
(including servers and signal processing gear). In our case, we usually
use a separate box for monitoring temperature as they are more reliable
and don't fail if a server fails (like from overheating!). If you're
interested, I can give you more info, but the boxes we use tend to make
their temperature data available via HTTP or SNMP, both of which are
readily fetchable with command line scripts.
Separate temperature boxes are also nice because they usually have
probes that you can place where you want them, like right near the A/C
unit so you can tell quickly if it has failed and fix the problem before
it damages your gear. They are pretty cheap too. I can't imagine
operating any kind of data center, small or large, without one.
--Dave
/*
PLUG: http://plug.org, #utah on irc.freenode.net
Unsubscribe: http://plug.org/mailman/options/plug
Don't fear the penguin.
*/
