Topic: Monitoring HD temperatures

[cpuffalt]

I just discovered by accident today that a lot of hard drives these days contain temperature sensors.  This information is accessible via a utility called hddtemp.  I found an rpm for it over at http://rpm.pbone.net/index.php3/stat/4/idpl/960440/com/hddtemp-0.3-0.fdr.0.10.beta10.rh80.i386.rpm.html and it seems to work great with SME v6.0.1.  What I was wondering though is if someone could integrate this with rrdtool and sysmon?  Or maybe someone could tell me how to go about integrating it?  

The website for hddtemp itself is http://coredump.free.fr/linux/hddtemp.php.  Someone has also put together a bit of a howto on using it with rrdtool here http://martybugs.net/linux/hddtemp.cgi but I'm not sure how to go about integrating this with the existing sysmon contrib.

Thanks,
Corey

[Reinhold]

cpuffalt,

I agree S.M.A.R.T. has been around some time, it is supported by 'our' SME kernel and YES I always activate it (and get 'some' useful info into my SME message log).

hddtemp however is a different story...
The underlying information is NOT standardized and these days it seems proprietary information again
With 40+60 Gb HDs on the list you may be successful, however with current 120-200Gb I still have to find one that REALLY works - correctly that is - even if it's in the list I had temp readings of 76 degC on a HD that was "not even warm" on touch .-/   (i.e. l.t.40 deg)

Samsung and Hitachi (IBM) do not answer to emails requesting info I sent ...
I assume the temp-'scale' has changed to "non linear" ...

If you find out more let me know integrating it for display into sysmon shouldn't be such a task...
 
Regards
Reinhold

[cpuffalt]

I agree S.M.A.R.T. has been around some time, it is supported by 'our' SME kernel and YES I always activate it (and get 'some' useful info into my SME message log).

What do you mean when you say you activate it?  Is there a setting I can change so that SMART information is logged automatically?

hddtemp however is a different story...
The underlying information is NOT standardized and these days it seems proprietary information again
With 40+60 Gb HDs on the list you may be successful, however with current 120-200Gb I still have to find one that REALLY works - correctly that is - even if it's in the list I had temp readings of 76 degC on a HD that was "not even warm" on touch .-/   (i.e. l.t.40 deg)

I guess I got lucky.  The HDs I tested were all in that 40-60GB range and the temperature readings looked reasonable.    I have had a lot less success with lm_sensors which I'm even more interested in getting working.

If you find out more let me know integrating it for display into sysmon shouldn't be such a task...

I definitely want to integrate the information especially ... can you explain how to go about doing that?

Thanks,
Corey

[Reinhold]

Have a look at those two:

ucsc SMART Suite
http://csl.cse.ucsc.edu/smart.shtml
...which (check!) I believe is included in 6.x of SME due to the RH Kernel

smartmontools
http://smartmontools.sourceforge.net/

...and I'm off for the weekend  

Enjoy
Reinhold

[Jesper Knudsen]

Its actually pretty simple to do MRTG statistics on this.

http://www.swerts-knudsen.dk/mrtg/

First determine where you want MRTG to create the stats. I have a mrtg ibay and a sub-folder called hw. Create mrtg .cfg file with the following for 2 disks.

workdir: /home/e-smith/files/ibays/mrtg/html/hw

Target[temp1]: hddtemp -n /dev/hda&hddtemp -n /dev/hda
MaxBytes[temp1]: 200
Options[temp1]: gauge,nopercent
Title[temp1]: Disk hda Temperature
PageTop[temp1]: <H1> Disk hda Temperature</H1>
WithPeak[temp1]: dwmy
YLegend[temp1]: Temperature
ShortLegend[temp1]: Celcius
LegendI[temp1]: &nbsp;Temp:
LegendO[temp1]:


Target[temp2]: hddtemp -n /dev/hda&hddtemp -n /dev/hda
MaxBytes[temp2]: 200
Options[temp2]: gauge,nopercent
Title[temp2]: Disk hdb Temperature
PageTop[temp2]: <H1> Disk hdb Temperature</H1>
WithPeak[temp2]: dwmy
YLegend[temp2]: Temperature
ShortLegend[temp2]: Celcius
LegendI[temp2]: &nbsp;Temp:
LegendO[temp2]: