[ale] Seagate Hdd Not Lining Up!? -- Diverging from the problem...

[aaron]

Just for fun I'll take a stab at the sideline questions here.
Y'all feel free to laugh and deride me if I get some of it wrong :-)

On 2009, Jul, 11, , at 7:48 PM, Tim Watts wrote:
> Just out of curiosity, why doesn't the reported total bytes on the  
> drive
> divide evenly by the bytes/cylinder? Where would those extra bytes  
> reside if
> not on a cylinder? And why isn't the total == 320GB? (I suppose the  
> answer to
> the last question is tied up in the bad sector map. But how do they  
> keep those
> out of the usable cylinder space? Or do they?)

As I understand it, the total space quoted for a drive, eg "320 GB",
is never what's actually available to partition and format into usable
space for data storage. File system housekeeping elements also cut
into the available space for actual data on the drive.

It also seems to have become the common practice of manufacturers to
misrepresent and overstate the size of any storage device by calculating
their size numbers from "decimal" byte values (1,000 bytes = 1 mB)
instead of the traditional and, of course more accurate binary values
(1,024 Bytes = 1 MiB // 1024 MiB = 1GiB).  And then, as you note,
there is the issue of bad sectors which are mapped in the firmware
of device and updated automatically on modern disc and flash devices.

The "bytes per cylinder" issues are part of the answer to your other
questions.

> And OK, I'm not afraid to look totally stupid (but I'll qualify my  
> question by
> saying it's been /years/ since I've stuck my head inside a drive):  
> Am I to
> believe that these drives, which can't be more than 3/4" high, / 
> really/ have
> 255 platters spinning inside them with an arm between each platter?  
> Or has the
> geometry all been virtualized today? If you tightly stacked 255  
> crisp $1
> bills, do you know how high it would be? I don't, but I'm pretty  
> sure it'd be
> more than 3/4"! (BTW, if you do know, you're spending way too much  
> time at the
> strip clubs.)

You actually guessed it in suggesting that the physical drive geometry
has been "virtualized" and moved into the drive firmware.  As storage
devices expanded in capacity and incorporated more of the controlling
software internally, the legacy mechanical references to "cylinders" and
"surfaces" used to communicate data transfer with the bus and bios and
OS were outgrown both in scope and in need.  Rather than rewrite the
interfaces and drivers of the bios and OS components, it was simpler
and easier to just virtualize the reference values inside the drive
firmware and let the smarts on the drive translate the values.

> Thanks,
>   Tim
>   (who is awed by all this and just choc full o'questions)


HTH!
peace
aaron

>
> On Saturday 11 July 2009 5:28:22 pm Michael B. Trausch wrote:
>> On Sat, 11 Jul 2009, Brian Pitts wrote:
>>> Marc Ferguson wrote:
>>>> [root at fergatron ~]# fdisk -l /dev/sdb
>>>>
>>>> Disk /dev/sdb: 320.0 GB, 320072933376 bytes
>>>> 255 heads, 63 sectors/track, 38913 cylinders
>>>> Units = cylinders of 16065 * 512 = 8225280 bytes
>>>> Disk identifier: 0x0b99f72f
>>>>
>>>>    Device Boot      Start         End      Blocks   Id  System
>>>> /dev/sdb1               1       38913   312568641   83  Linux
>>>> [root at fergatron ~]#
>>>>
>>>> I'm not fully comprehending these outputs. Do they indicate  
>>>> anything of
>>>> significance?
>>>
>>> That looks like one (roughly) 320 GB partition to me. You could  
>>> try to
>>> grow the filesystem by running the following as root
>>>
>>> umount /dev/sdb1
>>> e2fcsk -f /dev/sdb1
>>> resize2fs -p /dev/sdb1
>>> mount /dev/sdb1 /media/backup
>>
>> If the kernel is recent enough, and the filesystem is extXfs, it  
>> should be
>> growable online.
>>
>>  	--- Mike