HP Integrity rx7620 Server - User Service Guide, Sixth Edition

Server Management Behavior
This section describes how the system responds to over-temperature situations, how the firmware
controls and monitors fans, and how it controls power to the server.
Thermal Monitoring
The manageability firmware is responsible for monitoring the ambient temperature in the server
and taking appropriate action if this temperature becomes too high. To this end, the ambient
temperature of the server is broken into four ranges: normal, overtemp low (OTL), overtemp
medium (OTM), and overtemp high (OTH). Figure 5-11 shows the actions taken at each range
transition. Actions for increasing temperatures are shown on the left; actions for decreasing temps
are shown on the right.
Figure 5-11 Temperature States
On large temperature swings, the server transitions through all the states in order. It might go
to the following state immediately, but each of the preceding actions will occur. If the temperature
reaches the highest range, the server shuts down immediately by the manageability firmware.
Fan Control
There are three sets of fans in the system: those on the I/O bay, the front and rear fans that are
connected to the main backplane, and those on the cell boards. The front fans are run off of
standby power, and run any time AC input power is supplied to the server. All of the fans turn
on when 48 V power is supplied to the system.
As shown in Figure 5-11, the fan behavior is related to the temperature state. The fans are set to
high-speed when the ambient temperature is anywhere above the normal operating range. The
front and rear fans are set to high-speed any time a chassis intrusion switch is triggered when
removing a side cover.
Altimeter Circuit
The PCI-X backplane contains an altimeter circuit. This circuit is used to adjust the chassis fan
speeds for the operating altitude at power on and during MP initialization. The chassis fans
consist of the two front fans, two rear fans, and the six PCI-X I/O assembly fans. If an altimeter
failure is detected, the information is logged as an Event ID then propagated to the OS level to
be picked up by monitoring diagnostics.
The altimeter circuit is checked at power on by the MP. If an expected value is returned from
the altimeter circuit, the altimeter is determined good. The altimeter reading is then set in
non-volatile random access memory (NVRAM) on board the core I/O card. If the value is ever
lost like for a core I/O replacement, the NVRAM is updated at the next boot provided the altimeter
Server Management Behavior 87