Partition with bad blocks cannot be resized

[Synapse Syndrome]

I was trying to extend a partition into free space that I got after deleting
another partition on a hard drive using Acronis Disk Management Suite.

It would not allow it, and there was a red cross icon on the partition. I
ran CHKDSK within the program, with the option to fix all errors, and it
took a while to do that.

The red cross went away, and I could use the wizard to extend the partition
and pressed the Commit button. Then I get an error saying that I cannot
resize a partition with bad blocks.

Is there anything I can do here that I am not thinking of? I suppose I can
copy the contents of the partition to another drive through network and then
try reformatting it, but it will be hassle, and I am not sure that it will
even work.

Any thoughts?

Cheers

ss.

 

[Jaimie Vandenbergh]

I was trying to extend a partition into free space that I got after deleting
another partition on a hard drive using Acronis Disk Management Suite.

It would not allow it, and there was a red cross icon on the partition. I
ran CHKDSK within the program, with the option to fix all errors, and it
took a while to do that.

The red cross went away, and I could use the wizard to extend the partition
and pressed the Commit button. Then I get an error saying that I cannot
resize a partition with bad blocks.

Does running chkdsk again show anything more than zero bytes of bad
blocks?

Is there anything I can do here that I am not thinking of? I suppose I can
copy the contents of the partition to another drive through network and then
try reformatting it, but it will be hassle, and I am not sure that it will
even work.

It's the only thing to do - but consider that bad blocks only become
visible when the hard drive has already reallocated all its spare
blocks to hide the bad blocks. You have a failing drive, get it
replaced under warranty.

Cheers - Jaimie

 

[Alex Fraser]

[snip]

It's the only thing to do - but consider that bad blocks only become
visible when the hard drive has already reallocated all its spare
blocks to hide the bad blocks.

As far as I can see, a drive can only relocate a sector to a spare when it
is written to or if it detects that the sector is failing on a read. So
seeing bad sectors doesn't necessarily mean all possible (ie physically
nearby) spare sectors have been used.

You have a failing drive, get it replaced under warranty.

Agreed. And if the warranty has expired (or even if it hasn't), dispose of
it - a replacement will be cheap and/or bigger & faster.

Alex

 

[Jaimie Vandenbergh]

[snip]

It's the only thing to do - but consider that bad blocks only become
visible when the hard drive has already reallocated all its spare
blocks to hide the bad blocks.

As far as I can see, a drive can only relocate a sector to a spare when it
is written to or if it detects that the sector is failing on a read.

Such as the chkdsk /r full scan which SS ran.

So
seeing bad sectors doesn't necessarily mean all possible (ie physically
nearby) spare sectors have been used.

There's some difference between when Windows identifies and marks a
bad block, and when the drive firmware remaps an unusable sector. I
don't know the details, but a reformat with full check *can* clear a
bad block list.

Temporarily. Long enough to make you incorrectly trust the drive
again, sometimes...

Agreed. And if the warranty has expired (or even if it hasn't), dispose of
it - a replacement will be cheap and/or bigger & faster.

Which was my main point, though I'm sure SS knows that well.

Cheers - Jaimie

 

[me]

I was trying to extend a partition into free space that I got after deleting
another partition on a hard drive using Acronis Disk Management Suite.

It would not allow it, and there was a red cross icon on the partition. I
ran CHKDSK within the program, with the option to fix all errors, and it
took a while to do that.

The red cross went away, and I could use the wizard to extend the partition
and pressed the Commit button. Then I get an error saying that I cannot
resize a partition with bad blocks.

Is there anything I can do here that I am not thinking of? I suppose I can
copy the contents of the partition to another drive through network and then
try reformatting it, but it will be hassle, and I am not sure that it will
even work.

Any thoughts?

Cheers

ss.

http://www.grc.com/spinrite.htm

It's a good program, especially if imminent death (of the drive) is
suspected.

me (not connected with grc)

 

[Mike]

http://www.grc.com/spinrite.htm

I'm not taking sides here, but you might want to read this:
http://www.radsoft.net/news/roundups/grc/20060123,00.shtml

Mike

 

[DaveW]

The bad blocks are areas of the harddrive disks that are no longer useable.
No way around it.

 

[Synapse Syndrome]

Such as the chkdsk /r full scan which SS ran.


There's some difference between when Windows identifies and marks a
bad block, and when the drive firmware remaps an unusable sector. I
don't know the details, but a reformat with full check *can* clear a
bad block list.

Temporarily. Long enough to make you incorrectly trust the drive
again, sometimes...


Which was my main point, though I'm sure SS knows that well.

Running chkdsk didn't help.

The drive is actually one of two identical Maxtor D740X drives that I have,
from 2001. So it's well out of warranty. Sad to see it failing as they
were top-notch - quietest and fastest IDE drives of their time.

I just use them for p2p and tv recording, so there is no valuable data to
lose.

ss.

 

[Synapse Syndrome]

I'm not taking sides here, but you might want to read this:
http://www.radsoft.net/news/roundups/grc/20060123,00.shtml

I actually have Spinrite on this really dodgy CD I have (Hiren's BootCD), so
I will try using it, just to see what happens.

The Spinrite website is so amateurish and full of tacky marketing that it
does not look like a serious program, but I have nothing to lose, and it
will be interesting if it allows me to resize this partition afterwards.
Even if what it does leads to later problems, I have nothing to lose in this
instance. I'll tell you know how it goes.

ss.

 

[Jonathan Buzzard]

[snip]

It's the only thing to do - but consider that bad blocks only become
visible when the hard drive has already reallocated all its spare
blocks to hide the bad blocks.

As far as I can see, a drive can only relocate a sector to a spare when it
is written to or if it detects that the sector is failing on a read. So
seeing bad sectors doesn't necessarily mean all possible (ie physically
nearby) spare sectors have been used.

Not quite true. Typically while on a specific cylinder the drive reads
everything passing it's head, and should it find a suspect sector
immediately relocates it. So while the computer has not actually requested
data from the sector it has in fact been read.

In addition a lot of drives do patrol reads when otherwise idle
looking for suspect sectors to relocate before they become a problem.

That said I have seen one or two drives report bad sectors that
have mysteriously disappeared on giving the drive a full format. That
would suggest the sector rapidly deteriorated and the drive did not get a
chance to relocated it but there where spare sectors available for
relocation and the full format forced this. Regardless any drive I come
across that reports bad sectors is retired from use ASAP.


JAB.

 

[Alex Fraser]

Not quite true. Typically while on a specific cylinder the drive reads
everything passing it's head, and should it find a suspect sector
immediately relocates it. So while the computer has not actually
requested data from the sector it has in fact been read.

That makes sense (did you mean track rather than cylinder, ie sectors
passing one head?). Also there is whatever read-ahead caching the drive
does. But there is still a good chance that none of the sectors over several
cylinders are read (aside from patrol reads) for a long period, eg a large,
unfragmented file that isn't accessed.

In addition a lot of drives do patrol reads when otherwise idle
looking for suspect sectors to relocate before they become a problem.

Ah, yes, I had forgotten about that. Any idea how effective it is in hiding
problems?

Alex

 

[Daniel James]

Typically while on a specific cylinder the drive reads
everything passing it's head, and should it find a suspect sector
immediately relocates it.

How would a drive detect that a sector was "suspect"? If we're talking
about read errors then surely there's nothing that can safely be relocated
as the data cannot reliably be read?

That sounds dodgy ... wouldn't it be better to leave the sector as it is
until someone comes to read it (which may never happen) and to report a
read error when they do (so at least they'll get an error and know that the
data are unreliable). I'd have thought that the only safe time to relocate
the sector was on a write ...

I'm not saying you're wrong ... just trying to understand what you said.

... any drive I come across that reports bad sectors is retired from
use ASAP.

That's the safest course, certainly. If a drive suddenly started reporting
a small number of bad sectors after (say) a power outage I might be
prepared to put that down to the outage (especially if the bad sectors were
all on the same track) and assume that the rest of the drive was OK.

Cheers,
Daniel.

 

[CBFalconer]

How would a drive detect that a sector was "suspect"? If we're
talking about read errors then surely there's nothing that can
safely be relocated as the data cannot reliably be read?

Hard disks protect their contents with a checksum per sector. That
doesn't change, once written, but it, or the data that creates it,
may be misread. After multiple reads a faulty sector may finally
be read correctly, and the correct checksum identifies this event.
Now steps can be taken to retire the sector from active use.
Remember that disk read/writing is an analog process.

 

[Alex Fraser]

How would a drive detect that a sector was "suspect"? If we're talking
about read errors then surely there's nothing that can safely be
relocated as the data cannot reliably be read?

An ECC is used, so the drive can suspect a sector is failing before it
cannot be reliably read.

Alex

 

[CBFalconer]

An ECC is used, so the drive can suspect a sector is failing
before it cannot be reliably read.

It's not normally an ECC, but it is a reliable CRC checksum, such
that all faulty reads can be detected. ECC is an error correcting
code.

 

[Alex Fraser]

An ECC is used, so the drive can suspect a sector is failing
before it cannot be reliably read.

It's not normally an ECC, [...]

As far as I can tell, for modern drives, an ECC is used. ECC seems likely to
be a good trade-off with today's high densities.

Alex

 

[Daniel James]

Hard disks protect their contents with a checksum per sector.

Yes, but only a CRC32.

After multiple reads a faulty sector may finally
be read correctly ...

IME (mostly with duff floppies) if it takes more than a couple of reads
you're just as likely to have bad data and a bad checksum that just happen
to match as to have the correct data. I tried, not so long ago, to recover
some 'vital' information from a dodgy floppy for a family member, and I
found that I could read the bad sectors without error about one time in
10-15 tries, but that I did NOT get consistent results.

Such a read error NEEDS to be flagged as an error, and the user given the
chance to discard the data ... and the error message should preferably be
shown in the context of a read of the relevant data, or the user may be a
little confused as to whey the message is cropping up there and then ...

Error nnnn reading some data you didn't actually ask me to read.
Is it OK to move this file to a good part of the disk and fill it
with some data that may turn out to be random noise?

I don't think I'd like that ....

OTOH if the drive actually uses an ECC, as Alex Fraser suggests, then there
is scope for the controller to detect an error and, if it's not too
catastrophic (depending on the number of ECC bits) to correct it. The data
can then be moved to a good sector fairly safely. There's still some
potential for data corruption if the data contain more bit errors than the
ECC can check/correct, but I think it's the best we can hope for.

Cheers,
Daniel.