Re: [smartmontools-support] raw values – help required [ATTACHMENTS]

[Franc Z. <fz...@in...>]

At 04:44 AM 23/04/10, you wrote:

>I've never seen the interiour of a HDD before!)

Here is an excellent article.

HDD from inside - Main parts:
http://hddscan.com/doc/HDD_from_inside.html

>I wander if the SAMSUNG drive will brake down due to it's incredibly high LCC
>value…

The Samsung drive's SMART numbers don't make 
sense to me. ISTM that, if the Samsung's load 
cycle frequency was indeed once every 3.9 sec, 
then it would be spending its entire life waking 
up and going back to sleep. Surely 3.9s is not 
enough time for a drive to spin down and then up 
again ??? Or perhaps the Samsung just unloads its 
heads without spinning down ???

Furthermore, the SAMSUNG HM500JI hasn't lost a 
point from its normalised POH count after 3712 
hours, so it appears that there is a bug in that 
attribute. Or perhaps it is logarithmic, in which 
case the drive will lose points at a much faster rate as it ages ???

> > normalised SER = -10 log(total lifetime errors / total lifetime seeks)
>
>-10 log (value a over value b) ­ got it.
> > If errors = 0, then let errors = 1.
>
>Sorry. Don't get this one. Why "let" errors=1, if errors=0?

Because log(0) is undefined. :-)

>Okay. I try to keep up. > 60 is good then, right? and < 60 would be more than
>1 error in 1 million seeks i.e. bad, yes?

Actually, the SMART threshold is 30, which 
equates to 1 error in 1000. Seagate's product 
manuals rate the typical seek reliability at less than 1 error in 1 million.

>You know, it would be good if there was a program that could utilize smartctl
>and any other necessary tool, like hdparm, and just adjust everything for a
>specific *known* disk. It's really hard work otherwise to make all these
>calculations by hand, and different for each drive (not even vendor specific,
>but within one vendor different!).

The log relationship for the SER is something I 
determined on my own, after investigating many 
SMART reports. I wouldn't accept it as gospel. 
However, it *seems* to fit the data ...

> > To prevent a drive
> > from being dropped from the array, some models
> > incorporate ERC (Error Recovery Control, an ATA
> > standard, used by Seagate), or TLER (Time Limited
> > Error Recovery, Western Digital), or CCTL
> > (Command Completion Time Limit, Samsung). These
> > commands define the time limits for the drive to
> > complete a read or write command.
>
>I'll have to find a way to set the timeouts.

WD has a utility called wdtler.exe. It makes 
non-volatile mods to certain older drives, but 
may not be supported on modern drives. Setting 
the timeouts using ERC commands may work on 
supported drives, but I suspect that the changes 
may not survive a power cycle, in which case I 
don't know how you will be able to apply them to your DiskStation. :-(

Here is an article that discusses ERC support in smartmontools:
http://www.csc.liv.ac.uk/~greg/projects/erc/

There is also some discussion at this forum thread:
http://hdat2.getphpbb.com/hdat2-common-f3/wd-tler-samsung-cctl-seagate-erc-t186.html

Regards,
Franc