Penguin Computing Relion 1900e, Technical Manual

Page: 79 / 174

Relion 1900e/2900e Manual
Revision 1.0 64
All auto-rearm sensors that show an asserted event status generate a de-assertion SEL event at the time the
BMC detects that the condition causing the original assertion is no longer present; and the associated SDR is
configured to enable a de-assertion event for that condition.
7.3.3 BIOS Event-Only Sensors
BIOS-owned discrete sensors are used for event generation only and are not accessible through IPMI sensor
commands like the Get Sensor Reading command. Note that in this case the sensor owner designated in the
SDR is not the BMC.
An example of this usage would be the SELs logged by the BIOS for uncorrectable memory errors. Such SEL
entries would identify a BIOS-owned sensor ID.
7.3.4 Margin Sensors
There is sometimes a need for an IPMI sensor to report the difference (margin) from a non-zero reference
offset. For the purposes of this document, these type sensors are referred to as margin sensors. For instance,
for the case of a temperature margin sensor, if the reference value is 90 degrees and the actual temperature
of the device being monitored is 85 degrees, the margin value would be -5.
7.3.5 IPMI Watchdog Sensor
The BMC supports a Watchdog Sensor as a means to log SEL events due to expirations of the IPMI 2.0
compliant Watchdog Timer.
7.3.6 BMC Watchdog Sensor
The BMC supports an IPMI sensor to report that a BMC reset has occurred due to action taken by the BMC
Watchdog feature. A SEL event will be logged whenever either the BMC FW stack is reset or the BMC CPU
itself is reset.
7.3.7 BMC System Management Health Monitoring
The BMC tracks the health of each of its IPMI sensors and report failures by providing a “BMC FW Health”
sensor of the IPMI 2.0 sensor type Management Subsystem Health with support for the Sensor Failure offset.
Only assertions should be logged into the SEL for the Sensor Failure offset. The BMC Firmware Health sensor
asserts for any sensor when 10 consecutive sensor errors are read. These are not standard sensor events
(that is, threshold crossings or discrete assertions), these are BMC Hardware Access Layer (HAL) errors. This
means the BMC is unable to get a reading from the sensor. If a successful sensor read is completed, the
counter resets to zero.
7.3.8 VR Watchdog Timer
The BMC FW monitors that the power sequence for the board VR controllers is completed when a DC power-
on is initiated. Incompletion of the sequence indicates a board problem, in which case the FW powers down
the system.
The BMC FW supports a discrete IPMI sensor for reporting and logging this fault condition.
7.3.9 System Airflow Monitoring
This sensor is only available on systems at Intel
®
chassis. BMC provides an IPMI sensor to report the
volumetric system airflow in CFM (cubic feet per minute). The air flow in CFM is calculated based on the
system fan Pulse Width Modulation (PWM) values. The specific PWM or PWMs, used to determine the CFM is
SDR configurable. The relationship between PWM and CFM is based on a lookup table in an OEM SDR.
The airflow data is used in the calculation for exit air temperature monitoring. It is exposed as an IPMI sensor
to allow a datacenter management application to access this data for use in rack-level thermal management.