Erasing a hard disk

[Grimble]

I have a Buffalo Linkstation with 2 x 2TB disks in a Raid 1
configuration. I found out that one disk was defective, so I bought an
identical one on ebay, secondhand because my model is no longer in
production. I need to erase it (it was formatted as XFS) so that it can
be used to recreate the raid array. I connected it via a USB adaptor and
gave the command
dd if=/dev/zero of=/dev/sdc bs=1M

but got the message
dd: error writing '/dev/sdc': No space left on device
32117+0 records in
32116+0 records out
33676349440 bytes (34 GB, 31 GiB) copied, 9.70154 s, 3.5 GB/s

which surprised me, because I thought that dd command was highly
destructive. If that means it only wrote to about 15% of the disk, does
that mean that 85% of the disk cannot be used?
Can someone provide more insight please?

[William Unruh]

Yes, dd is "highly destructive". I would worry about writing 0 to the
whole disk that it would also destroy the low level formatting of the
disk.

On https://askubuntu.com/questions/17640/how-can-i-securely-erase-a-hard-drive
it says
"dd halts at the first bad block, and fails to clobber the rest (unless
I painfully use skip=... to jump ahead each time it stops)."

so it seems like that disk has bad blocks. Do you really want to be
using a hard drive which has bad blocks ? It sounds like it is on its
last legs which is probably why it was sold on ebay in the first place.

Why not just buy two new hard disks to replace both of your disks. One
of them has failed. What are the chances the other will fail in the next
while?


Also why wipe the disk? Why not just reformat it for use as a raid?

Post by Grimble
I have a Buffalo Linkstation with 2 x 2TB disks in a Raid 1
configuration. I found out that one disk was defective, so I bought an
identical one on ebay, secondhand because my model is no longer in
production. I need to erase it (it was formatted as XFS) so that it can
be used to recreate the raid array. I connected it via a USB adaptor and
gave the command
dd if=/dev/zero of=/dev/sdc bs=1M
but got the message
dd: error writing '/dev/sdc': No space left on device
32117+0 records in
32116+0 records out
33676349440 bytes (34 GB, 31 GiB) copied, 9.70154 s, 3.5 GB/s
which surprised me, because I thought that dd command was highly
destructive. If that means it only wrote to about 15% of the disk, does
that mean that 85% of the disk cannot be used?
Can someone provide more insight please?

[Grimble]

Post by William Unruh
Yes, dd is "highly destructive". I would worry about writing 0 to the
whole disk that it would also destroy the low level formatting of the
disk.
On https://askubuntu.com/questions/17640/how-can-i-securely-erase-a-hard-drive
it says
"dd halts at the first bad block, and fails to clobber the rest (unless
I painfully use skip=... to jump ahead each time it stops)."
so it seems like that disk has bad blocks. Do you really want to be
using a hard drive which has bad blocks ? It sounds like it is on its
last legs which is probably why it was sold on ebay in the first place.
Why not just buy two new hard disks to replace both of your disks. One
of them has failed. What are the chances the other will fail in the next
while?

I wanted to protect my 2 years' backups by repairing the current
degraded raid array. Your point about potential failure had not escaped
me, but I thought I could get some safety first.

Post by William Unruh
Also why wipe the disk? Why not just reformat it for use as a raid?

Buffalo technical help say the drive should not be pre-formatted before
recovering the RAID array, which is why I am trying to erase current xfs
format.

Post by William Unruh

Post by Grimble
I have a Buffalo Linkstation with 2 x 2TB disks in a Raid 1
configuration. I found out that one disk was defective, so I bought an
identical one on ebay, secondhand because my model is no longer in
production. I need to erase it (it was formatted as XFS) so that it can
be used to recreate the raid array. I connected it via a USB adaptor and
gave the command
dd if=/dev/zero of=/dev/sdc bs=1M
but got the message
dd: error writing '/dev/sdc': No space left on device
32117+0 records in
32116+0 records out
33676349440 bytes (34 GB, 31 GiB) copied, 9.70154 s, 3.5 GB/s
which surprised me, because I thought that dd command was highly
destructive. If that means it only wrote to about 15% of the disk, does
that mean that 85% of the disk cannot be used?
Can someone provide more insight please?

[William Unruh]

Post by Grimble

Post by William Unruh
Yes, dd is "highly destructive". I would worry about writing 0 to the
whole disk that it would also destroy the low level formatting of the
disk.
On https://askubuntu.com/questions/17640/how-can-i-securely-erase-a-hard-drive
it says
"dd halts at the first bad block, and fails to clobber the rest (unless
I painfully use skip=... to jump ahead each time it stops)."
so it seems like that disk has bad blocks. Do you really want to be
using a hard drive which has bad blocks ? It sounds like it is on its
last legs which is probably why it was sold on ebay in the first place.
Why not just buy two new hard disks to replace both of your disks. One
of them has failed. What are the chances the other will fail in the next
while?

I wanted to protect my 2 years' backups by repairing the current
degraded raid array. Your point about potential failure had not escaped
me, but I thought I could get some safety first.

Post by William Unruh
Also why wipe the disk? Why not just reformat it for use as a raid?

Buffalo technical help say the drive should not be pre-formatted before
recovering the RAID array, which is why I am trying to erase current xfs
format.

One suggestion I read somewhere is to do a destructive bad-blocks test
on the drive. This writes to every location on the drive and reports all
bad-blocks found. If it is true that dd stops on bad blocks, (and I have
no reason to doubt it) then that would tell you exactly how bad your
drive is, and erase all the data at the same time. All current informed
opinion seems to be be that anything that is written over, the previous
data cannot be discovered in any way whatsoever, but then you are not
interested in removing old data, just in removing the formatting on the
disk so you can let the raid software format it. It might be sufficient
simply to remove all formatting from the disk-- ie, just erase the
formatting. You probably do not care about what data is on the disk now,
just that the raid software can use the disk, overwriting the current
data.

But from the evidence I suspect you have a disk with bad blocks. A few
might not be so bad, but I would suggest that the reason the disk was
sold on ebay was that it is going-- ie it was developing lots of bad
blocks. From what I know these things tend to avalanch. Eg a tiny piece
of dirt got in there ( eg the surface of the drive flaked off a tiny
piece) at that point you get an avalanch as it gets trapped between the
head and the disk and scratches another part of
the disk, releasing more particles, which then release more and more,
until the inside of the disk is a ruststorm.

So, I would strongly urge you to do a badblocks test on the disk.
Someone decided to sell it. Now they could have just replaced a
perfectly good full disk with a larger one. But more likely in my mind
is that they started having some trouble with the disk and decided to replace it.
And then irresponsibly decided to sell the old one (buyer beware).

I presume that you can still backup all of the data on your one good
raid disk (depends on the type of raid you had on them). Just buy two
new disks.

[David W. Hodgins]

Post by Grimble
I have a Buffalo Linkstation with 2 x 2TB disks in a Raid 1
configuration. I found out that one disk was defective, so I bought an
identical one on ebay, secondhand because my model is no longer in
production. I need to erase it (it was formatted as XFS) so that it can
be used to recreate the raid array. I connected it via a USB adaptor and
gave the command
dd if=/dev/zero of=/dev/sdc bs=1M
but got the message
dd: error writing '/dev/sdc': No space left on device
32117+0 records in
32116+0 records out
33676349440 bytes (34 GB, 31 GiB) copied, 9.70154 s, 3.5 GB/s
which surprised me, because I thought that dd command was highly
destructive. If that means it only wrote to about 15% of the disk, does
that mean that 85% of the disk cannot be used?
Can someone provide more insight please?

Are you sure sdc isn't a usb stick and not the hard disk?

Regards, Dave Hodgins

[Grimble]

Post by David W. Hodgins

Post by Grimble
I have a Buffalo Linkstation with 2 x 2TB disks in a Raid 1
configuration. I found out that one disk was defective, so I bought an
identical one on ebay, secondhand because my model is no longer in
production. I need to erase it (it was formatted as XFS) so that it can
be used to recreate the raid array. I connected it via a USB adaptor and
gave the command
dd if=/dev/zero of=/dev/sdc bs=1M
but got the message
dd: error writing '/dev/sdc': No space left on device
32117+0 records in
32116+0 records out
33676349440 bytes (34 GB, 31 GiB) copied, 9.70154 s, 3.5 GB/s
which surprised me, because I thought that dd command was highly
destructive. If that means it only wrote to about 15% of the disk, does
that mean that 85% of the disk cannot be used?
Can someone provide more insight please?

Are you sure sdc isn't a usb stick and not the hard disk?
Regards, Dave Hodgins

Hello Dave. Yes, lsblk recognises /dev/sdc as 2TB/1.8TiB SCSI disk.

[David W. Hodgins]

Post by Grimble
Hello Dave. Yes, lsblk recognises /dev/sdc as 2TB/1.8TiB SCSI disk.

What I do with an old dead drive is drill a few holes through it.

Regards, Dave Hodgins