[geeks] Murphy, instantiated

[Shannon Hendrix]

On Jun 4, 2009, at 17:42 , nate at portents.com wrote:

>> SMART: I find it more useful than Google evidently did, but it's
>> definitely not that great.  It has false positives, and there are  
>> some
>> drive errors that it never seems to report, and some that it cannot
>> report, like firmware errors.
>
> A little known fact is that SMART is not actually a standard (well  
> it was
> briefly but was withdrawn)

...and it was only a "standard" on paper.  No one ever got around to  
really making their drives respond to all the commands.

That's why smartmontools is such a PITA to maintain.  It's full of  
exceptions and custom code.

> - which frankly is just pathetic.  I think it's
> an indication of just how lazy and purely profit-driven the whole  
> storage
> industry is:

Actually, their abuse of and failure regarding SMART is just the tip  
of the iceberg.

Maxtor, for years, falsified drive responses to various ATA commands,  
including things like synchronous writes (and thus ordered write), as  
well as quite a few times of drive errors, and their firmware put off  
drive calibration.

All of that, and some other idiocy, was done to win benchmark tests.

Quite a few Maxtor and some WD drives also lie about write caching.   
They will respond it is turned off even when it isn't, just like they  
respond a write is finished when it isn't.

That said, I don't find SMART totally useless, and on some drives it  
has been pretty accurate.

I don't really give a damn what name they use for their error  
handling, so the issue of SMART no longer being real isn't important.   
What's important is getting everyone to support this in the command  
and data standards.

D

-- 
"Where some they sell their dreams for small desires."