freeipmi-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Freeipmi-users] Ipmi-sensors threshold values


From: Al Chu
Subject: Re: [Freeipmi-users] Ipmi-sensors threshold values
Date: Tue, 03 Nov 2009 09:43:59 -0800

Hey Maxim,

This definitely looks like a bug on the motherboard:

=====================================================
Get Sensor Reading Response
=====================================================
[              2Dh] = cmd[ 8b]
[               0h] = comp_code[ 8b]
[              B1h] = sensor_reading[ 8b]
[               0h] = reserved1[ 5b]
[               0h] = reading_state[ 1b]
[               1h] = sensor_scanning[ 1b]
[               1h] = all_event_messages[ 1b]
[               3h] = sensor_event_bitmask1[ 8b]

the sensor event bitmask is indicating "at or below lower critical
threshold" (and also "at or below lower non-critical", but FreeIPMI only
outputs the "worst" one).

Using your debug output for sensor #15, I'm calculating:

sensor reading = 10266
lower non critical threshold = 3944
lower critical threshold = 3422

The 3422 matches the output, so I'm pretty sure I'm calculating
correctly :-)  So clearly, the sensor reading is no where near the
thresholds.

At first, I thought it could be a bug w/ endian, perhaps the bits were
flipped and the upper critical threshold had been triggered instead.
However:

[               1h] =
readable_thresholds.lower_non_critical_threshold[ 1b]
[               1h] = readable_thresholds.lower_critical_threshold[ 1b]
[               0h] = readable_thresholds.lower_non_recoverable_threshold[ 1b]
[               0h] = readable_thresholds.upper_non_critical_threshold[ 1b]
[               0h] = readable_thresholds.upper_critical_threshold[ 1b]
[               0h] = readable_thresholds.upper_non_recoverable_threshold[ 1b]

indicates there are no upper thresholds.  Only lower ones.

I was hoping I could add a workaround to deal with this, but I'm not
sure if I can.  It just seems to be an outright bug in the firmware.
The "lower critical threshold" flag shouldn't be set.

I suppose I could try and add some type of "check the thresholds
manually" type of option.  In other words, manually check the thresholds
instead of trusting the motherboard.  Would that be something you guys
would be interested in having to help deal with this problem?

Al

On Tue, 2009-11-03 at 04:00 -0800, Maxim Kuleshov wrote:
> 
> Ashish Ray wrote:
> > 
> > I have inte'l SR2500 server running with freeipmi-0.3.1. I see ipmi-sensor
> > output as "At or Below
> > (<=) Lower Critical Thre shold" even the actual RPM is greater  than
> > threshold. Can someone please
> > 
> The same strange thing happens to me. I have intel SR1500SAS chassis and
> S5000PAL board. One of three fans have correct status, and others - invalid.
> Is it MB bug or something else?
> 
> ipmi-sensors -s 13,14,15
> 
> 13: Fan 1A (Fan): 10730.00 RPM (3422.00/NA): [At or Below (<=) Lower
> Non-Critical Threshold]
> 14: Fan 2A (Fan): 13224.00 RPM (3422.00/NA): [OK]
> 15: Fan 3A (Fan): 10382.00 RPM (3422.00/NA): [At or Below (<=) Lower
> Critical Threshold]
> 
> ipmi-sensors --version
> 
> ipmi-sensors - 0.7.14
> 
> Debug output for impi-sensors --debug -s 13,14,15 command attached.
> http://*old.nabble.com/file/p26160080/debug.log debug.log 
> 
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory





reply via email to

[Prev in Thread] Current Thread [Next in Thread]