Hardware monitor needed

[Alex Zbyslaw]

Laszlo Nagy wrote:

>
>  Hi,
>
> My FreeBSD 6.2 server restarts suddenly once or twice a day. I believe 
> it is because the processor is overheated, but I'm not sure. Is there 
> a way to check this from software? I would like to install a hardware 
> monitor program that can log out processor temperature in every 
> minute. The mainboard is ASUS P5LD2, if that matters. Is there a 
> software out there that can do this for me?

Check out healthd or mbmon.  One or other has worked OK for me on other 
Asus boards, and both are in ports (sysutils/ I think).

If you have ACPI and your board supports thermal zones, then you can 
check those. 

    sysctl -a | egrep 'acpi.*therm'
or
    sysctl -a | egrep 'acpi.*tz'

one or other should be a good enough incantation.  None of my ASUS mobos 
do have thermal zones so I can't be sure -- it's much more commonly 
supported in laptops.

Or just

    sysctl -a | egrep acpi

and eyeball for anything that looks like temp information.


Also beware of constant monitoring.  Every now and again (once a day on 
average), I find mbmon sits chewing CPU and pushing the temperature up 
itself.  So I run in with

  (ulimit -t 1; /usr/local/bin/mbmon -p winbond -c 1)

which means that the parent shell kills it if it uses more than 1 second 
of CPU, which is far more than it needs.

--Alex

PS Many disks which support SMART can display their apparent temp as one 
of the SMART parameters (see sysutils/smartmontools).  Not 100% 
trustworthy, but better than nowt.  I'd rather fry the processor than a 
disk :-)