SUMMARY
Red LED on the front of my Unitrends appliance is lit - diagnosing it with IPMI
ISSUE
Purpose
Why is the red system LED on the front lit? (different than the disk LEDs)
Description
Newer Unitrends platforms use IPMI firmware to set the red LED if a hardware event occurs. For example:
|
Recovery-712 Recovery-713 Recovery-813 (and several other models) |
Recovery-833-100 Recovery-833-200 Recovery-943
|
Use IPMI commands to discover the cause of the fault causing the red LED.
Cause
On prior platforms that did not support IPMI, we used lm_sensors (the ‘sensors’ command) to check sensor status, but with IPMI, lm_sensors is not valid and conflicts with the IPMI firmware functions. For IPMI functions, use the ipmiutil package.
Resolution
*** Use if you are comfortable with shell command-line ***
1) Stop any lm_sensors activity: pkill sensors
2) If using a unitrends release prior to 7.2.0, then install ipmiutil rpm on the DPU:
a) Install it from the repo by doing 'yum install ipmiutil'
b) Or see one of these links to download ipmiutil:
Unitrends ftp repo: http://updateftp.unitrends.com/pub/RecoveryOS/CentOS/6/x86_64/RPMS/
upstream external site http://ipmiutil.sourceforge.net
and then run 'rpm -i ipmiutil*.rpm'
3) Run ‘ipmiutil sensor -c’ on the DPU and check for any sensors not ‘OK’ .
(CPU Temp will usually say ‘Low’). Sample output is below on a Recovery-822 where the sensor values are normal (OK).
-- BMC version 2.60, IPMI version 2.0
ID | SDRType | Type |SNum| Name |Status| Reading
0004 | Full | Temperature | 11 | System Temp | OK | 21.00 C
0047 | Full | OEM(c0) | 12 | CPU Temp | 00 Low
008a | Full | Fan | 41 | FAN 1 | OK | 4470.00 RPM
00cd | Full | Fan | 42 | FAN 2 | OK | 2250.00 RPM
0110 | Full | Fan | 43 | FAN 3 | OK | 4285.00 RPM
0153 | Full | Fan | 44 | FAN 4 | OK | 4285.00 RPM
0196 | Full | Fan | 45 | FAN 5 | Absent | 0.00 na
01d9 | Full | Voltage | 21 | CPU Vcore | OK | 0.89 V
021c | Full | Voltage | 23 | +3.3VCC | OK | 3.31 V
025f | Full | Voltage | 22 | +12 V | OK | 12.30 V
02a2 | Full | Voltage | 24 | CPU DIMM | OK | 1.53 V
02e5 | Full | Voltage | 25 | +5 V | OK | 5.12 V
0328 | Full | Voltage | 26 | -12 V | OK | -12.10 V
036b | Full | Voltage | 50 | VBAT | OK | 3.15 V
03ae | Full | Voltage | 4f | +3.3VSB | OK | 3.31 V
03f1 | Full | Voltage | 27 | AVCC | OK | 3.31 V
0434 | Full | Platform Securi | 51 | Chassis Intru | OK | 0.00 na
0477 | Full | Power Supply | 55 | PS Status | 01 Present
Even if all the sensors are OK now, an event may have occurred in the past which will be shown in the IPMI System Event Log (SEL).
4) Run ‘ipmiutil sel -e’ to look for recent events with severity other than ‘INF’. This is the key information to know why the LED came on. If there is no recent event in the IPMI SEL, please save the output of this command for later analysis.
Below is a sample Non-critical System Temperature event on a Recovery-813:
0005 12/31/69 22:37:00 MIN BMC Temperature #11 System Temp Hi Noncrit thresh actual=75.00 C, threshold=75.00 C
0006 01/01/70 08:54:56 INF BMC Temperature #11 System Temp HiN thresh OK now actual=72.00 C, threshold=75.00 C
4) If there is a Chassis Intrusion event, you can use this command to clear it:
5) If nothing is abnormal in these outputs but the red LED is still lit, you can do this to restart the firmware and turn off the LED:
# ipmiutil reset -k
6) After the firmware restarts, and ‘ipmiutil sensor –c’ shows the sensors again, the red LED should be off.
7) In many cases, although the condition has been cleared by the firmware, the BIOS also needs to be invoked for the Red LED to turn off, so doing a warm reboot will invoke the BIOS and clear the event.
8) If for some reason these steps did not clear it, shut down the system, pull the input power plug for 10 seconds, then power it up again.
Third-Party Sources
See http://ipmiutil.sourceforge.net for a UserGuide and other files.
Related Articles
Using IPMI LAN for remote access