Re: [Freeipmi-devel] HP, iLO 2.21 and thresholds

[Albert Chu]

Hey Diego,

On Wed, 2012-08-22 at 14:09 -0700, Diego Elio Pettenò wrote:
> Hi all,
> 
> Today I've updated the iLO firmware of the HP servers, in hope to fix
> the reporting of one CPU's temperature as 0.00 instead of N/A. This
> fixed the problem because... now it reports no temperature at all.
> 
> But even on this latest version, it seems like they have issues with the
> thresholds... but then I'd like to know if it might not be a problem
> with FreeIPMI (I don't count on that, but who knows).

Definitely could be some strange corner case interpretation in the IPMI
spec I missed.

> 
> 1104 | System Fan 2     | Fan                      | 14106.36   | RPM
> | N/A        | N/A        | N/A        | 3996.80    | 3475.48    | N/A
>       | 'OK'
> 
> Here's the problem: the fans are reported as being above threshold when
> running at 14K RPM which is their nominal normal speed; it seems to me
> like the Upper NC and Upper C should be instead Lower NC and Lower C
> respectively — it feels extremely strange that the Upper NC is higher
> than the Upper C.

Yeah, you would think it's be amongst the "lower" thresholds than the
"upper" ones.

> This happens on the DL140 G3 firmware 2.21, but it's similar to what
> happens on a DL160 G6 firmware 4.20:
> 
> 3440  | FAN1_OUTLET    | Fan                      | 12621.48   | RPM   |
> N/A        | 0.00       | N/A        | N/A        | 999.20     | N/A
>     | 'OK'
> 
> In this case the Lower C is correct, but the Upper C is wrong (and
> should probably be Lower NC).
> 
> 2128  | FAN_3           | Fan                      | 2049.64    | RPM
> | N/A        | 0.00       | N/A        | N/A        | 978.81     | N/A
>       | 'OK'
> 
> And on the ML160 G6 firmware 4.23 (which is the only one I could update
> at this point) there's the same issue (in this case it's not a rackmount
> so the fans are slower).
> 
> Albert do you think there might be something to be done to fix this by
> chance?

Could you send me the --debug output for one of these sensors.  I'm
wondering if HP put the thresholds into the wrong fields in the SDR.
That would definitely mess things up.

I assume the output was good/normal before the firmware update?

Al

-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory