Re: [Freeipmi-devel] sensors ok vs. alert

[robing]

I agree with AB that if these sensor values are actually "ok" then
Intel ought to change the nominal min/max values to reflect this.
When we first set up our host monitoring software on thunder, I
noticed a large number of sensor values out of range and tried to 
get Bryan Rice from Intel to investigate but I never succeed in 
getting him onto our system.

AB, would you please go through CDC's channels with Intel to
open a problem report on this and get a determination from Intel
as to whether these sensor readings are really ok or if they 
indicate a real hardware problem?  If the former, then I wold like them
to change their nominal limits.  If the latter, then we will
want CDC or Intel to assist us with identifying and replacing the
affected hardware components, since this affects nearly half of 
thunder.

Regards,
Robin G.

> 
> ,----[ Albert Chu <address@hidden> ]
> | Hey AB,
> | 
> | We've noticed on a thunder that as many as 50% of nodes are showings
> | alerts for voltage sensors.  I don't believe this is bad, b/c the
> | voltages are within the lower and upper critical range.  However,
> | since fish/sensors uses the normal min/max range, many sensors are
> | showing alerts (IMHO) more often than is appropriate.
> | 
> | I'm in discussion with the admins about their preference for output of
> | ranges and alerts, but I think something should be changed in
> | fish/sensors.  If I understand this correctly, we can use the readable
> | threshold mask (field 20 of the full sensor data record) to determine
> | which ranges are available.
> | 
> | I'm going to add a feature request on this to FreeIPMI so we don't
> | forget.
> | 
> | Al
> `----
> Hi Al,
> How about, Intel increasing the nominal min/max if they believe that
> the condition is normal. 
> 
> Otherwise, I will work with Bala and come up with a good alerting
> policy for sensors tool. 
> 
> I think,
>   crossing nominal  min/max limits should still be ALERT and 
>   crossing critical min/max limits as CRITICAL.
> 
> I will take a look at threshold mask too.
> 
> -- 
> Anand Babu
> Free as in Freedom <www.gnu.org>
> 
> 
> _______________________________________________
> Freeipmi-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/freeipmi-devel
>