[ale] possible to use hard drives that have bad blocks ?

[Greg Freemyer]

Not really.

Blocks (sectors) slowly fail on disks over time.  Not much you can do
about that, and it is pretty much true of all drives.

I have read that you should write to every sector at least once every
5 years because the magnatism slowly fails, and after 5 years the
sectors need a re-fresh.

If your worried about cost and want to save a disk, you can do a dd
if=/dev/zero of=/dev/hdX to refresh the drive.  Who knows it may last
another 5 years.

For true block failures as opposed to magnetic degradation, the blocks
are only tested by the drive logic on write, so if you have a sector
that is never written and it goes truly bad, you will have a single
bad sector for the remaining life of the drive.

Performing a write to the sector not only refreshes the magnetic
field, it allows the drive to recognize the failed sector and "do its
job" by re-assigning the bad sector to a good sector from an internal
reserved bad sector list.

I suspect RAID logic is used as much to protect from these isolated
block failures as much as it is for a true disk failure.

OTOH, if you write to the sector and it remains bad, then you more
than likely have a drive on its way out the door.

FYI: High-end RAID systems can monitor the disk sector failure rate
and once the failure rate exceeds a given rate, they can pro-actively
declare the drive failing.  HP for one accepts this designation as a
valid reason to swap a drive if you are a supported customer.

Greg

On 6/26/05, Courtney Thomas <ccthomas at joimail.com> wrote:
> Makes sense :-)
> 
> Thanks,
> Courtney
> 
> 
> On Sun, 2005-06-26 at 11:40, Jim Lynch wrote:
> > As cheap as disks are these days, if this is an internal 3.5" drive, I'd
> > replace it.  If one block is bad, how long before another one goes?
> >
> > Jim.
> > Courtney Thomas wrote:
> >
> > >I've got a couple of HDs that when I:
> > >
> > >dd if=/dev/hdX of=/dev/null bs=1m
> > >
> > >I get a single instance of something like:
> > >
> > >FAILURE - READ_DMA status=51 <READY, DSC, ERROR> error=40
> > ><UNCORRECTABLE> LBA=19194112 Input/output error
> > >
> > >My question is: does this signify a single bad block and is there a way
> > >to identify this error to some program that can block it's use, so the
> > >drive can be used ?
> > >
> > >Thank you,
> > >Courtney
> > >
> > >_______________________________________________
> > >Ale mailing list
> > >Ale at ale.org
> > >http://www.ale.org/mailman/listinfo/ale
> > >
> > >
> > >
> > >
> > >
> >
> > _______________________________________________
> > Ale mailing list
> > Ale at ale.org
> > http://www.ale.org/mailman/listinfo/ale
> 
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale
> 


-- 
Greg Freemyer
The Norcross Group
Forensics for the 21st Century