UNSOLVED
Controlled response to server overheating
In our rack we have 3 Poweredge 2950, one 2850 and one 2600 servers with one dual Amari graphics render server, together with an Powervault autoloader, UPS and 4 switches.
In our 4cu mtr server room we also have a telephone switch and two standalone UPS.
Air conditioning in our server room is compromised by the small volume, high thermal load and obstructed airflow. Experience has shown that the A/C unit will freeze and the temperature rise uncontrollably if the A/C thermostat is set too low. We have established a stable environment at 27deg C - rather warmer than we would like.
We will address some of the server room issues in the medium term but want to take belt and braces precautions now to avoid damage from overheating - especially out of hours.
We installed have a 3rd party (ServersCheck) application on one of the servers with a USB temperature probe in the cupboard and a GPRS modem that will alert the IT team if the room temperature exceeds 30deg and escalate to a futhter team if the message is not responded to or if the temperature rises above 40deg.
We can configure ServersCheck to issue a psshutdown to the least critical servers on the 40deg alert.
However, we want to automate safe shutdown
of the remainder by setting the Dell Openmanage warning and failure limits to a lower level than default.
Thanks for assistance on any of these questions.
Q) Do the servers automatically shut down on reaching the fail limit or do you have to configure psshutdown for this purpose?
Q) If not, do they simply struggle on until they break down
Q)
Is there a How To on any of this?
Stephen Solt
IT Manager
CZWG Architects
OMNavigator
291 Posts
555
0
Posted December 5th, 2007 02:00