Mailinglist Archive: opensuse-factory (279 mails)

[Greg Freemyer <greg.freemyer@xxxxxxxxx>]

On Tue, Mar 10, 2009 at 9:52 AM, Carlos E. R.
<robin.listas@xxxxxxxxxxxxxx> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Content-ID: <alpine.LSU.2.00.0903101448121.9084@xxxxxxxxxxxxxxxx>
>
>
> On Tuesday, 2009-03-10 at 06:45 -0400, Greg Freemyer wrote:
>
> > On Tue, Mar 10, 2009 at 12:05 AM, Carlos E. R.
> > > On Monday, 2009-03-09 at 23:26 -0400, Greg Freemyer wrote:
> > >
> > > > > dd if=/dev/sda of=/dev/sdc bs=1G conv=noerror
> > > >
> > > > Please note, never use noerror without also using sync.
> > >
> > > Why? I'm curious. I can make a guess, but I'm not sure, it is not
> > > mentioned
> > > in the manual:
>
> ...
>
> > Look at the output of a dd run.
> >
> > dd if=/dev/zero of=/dev/null count=100
> > 100+0 records in
> > 100+0 records out
> > 51200 bytes (51 kB) copied, 0.000172439 s, 297 MB/s
> >
> > The first 100 says 100 blocks read without error.  the first zero is 0
> > failed blocks failed to read.
> >
> > The second 100 is the blocks written without error, and the second 0
> > is the failed block writes.
>
> Aha, ok...
>
>
> > If you don't use sync and there are 2 errors I believe you get:
> >
> > 98+2 records in
> > 98+0 records out.
> >
> > i.e The failed blocks are not replicated in the destination at all
>
> I see... I don't know how to confirm this, though. It could be:
>
> 98+2 records out
>
> instead. :-?
>
> I wonder why they don't document this in the manual.

If you're really curious, hdparm can create and recover bad blocks on
your media.  (I think it needs a IDE/SATA hdd to work).

see hdparm --make-bad-sector  and   --repair-sector

When it creates them it uses a ATA diagnostic write to cause the per
sector crc to be bad, then on read you get a media error.

(ie on disk sectors are bigger than 512 bytes.  Part of the overhead
is a crc used to verify the media has not failed.)

Not sure how it does the fix side, but you are not supposed to
permanently loose the sector, thus it is more or less safe to test
with.

I would use a drive that does not have critical data.  And I assume
hdparm will be counting sectors from the beginning of the drive, not
the partition.

Greg
-- 
Greg Freemyer
Litigation Triage Solutions Specialist
http://www.linkedin.com/in/gregfreemyer
First 99 Days Litigation White Paper -
http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf

The Norcross Group
The Intersection of Evidence & Technology
http://www.norcrossgroup.com
--
To unsubscribe, e-mail: opensuse-factory+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse-factory+help@xxxxxxxxxxxx