On Wed, 3 Aug 2005, Shachar Shemesh wrote:
Peter wrote:
Afaik modern drives do the mapping silently using spare sectors in each
track. You get bad blocks when the drive runs out of spare sectors.
It seems that this is not exactly correct. Modern drives do automatic
remapping if and only if they managed to correctly read the data once after
finding out it's a bad sector. Otherwise, they will not create silent data
corruption, and will report said sector as bad despite having free sectors to
map to (http://smartmontools.sourceforge.net/, search for "Where can I find
manufacturer-specific disk-testing utilities?").
In my experience the first few failures are fixed by the drive. Maybe it
finds the data as you said. Anyway to the user it looks like a 'hiccup'
often followed by recalibration (often sharp click when heads travel to
detent). Without SMART you don't know you have a problem. Morale: do not
ignore noises from the drive and use SMART. Afaik the sector remapping
also happens during 'low level' formatting. This makes a not-so-good
drive appear perfect. A drive I have here shows 19k hours with zero
reallocated sectors. Hmm not so bad. (IBM drive). The last faults were
at 156hours. If the bathtub curve theorem is correct I should have ample
warning for any coming disaster a long time before it happens.
A low level format is, indeed, the right thing to do in such a case.
The above link points to a place
(http://www.benchmarkhq.ru/english.html?/be_hdd2.html) that list a
whole set of low level bad sectors remapping tools.
Peter
=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
