Groups | Search | Server Info | Login | Register

Groups > comp.os.os2.setup.storage > #75

bad sectors versus bad blocks

Show all headers | View raw

I'm trying to get a better understanding of how disk drives and
filesystems work.

I have a Seagate 2 TB external USB drive that I use strictly for
backup purposes.  I would often perform an incremental backup
when archival data were written to one of my main disks.  I did
that twice on a single day a few days ago.  The first one seemed
to work fine.  The second one failed.  I used CHKDSK to check the
disk, and it reported an unrecoverable error reading M from the
disk.  JFS was used for the file system on that disk.  I know from
past experience that "M" refers to metadata (why the error message
just doesn't come right out and say that, I don't know), and I know
it's bad news.

I then attached the drive to a Windows machine that has Seagate's
SeaTools installed on it, and the Long Generic test on the drive
found hundreds of bad sectors.  The drive did not pass the test.
SeaTools recommended replacing the drive.  I did that, and thanks
to OS/2's rather slow USB 2.0 driver, it took 2 days to recopy the
500 GB of data on my main disks to the new external USB backup
drive.

With a new backup in place, I was free to experiment with the
failed drive.  First, I tried SeaTools' repair option, in which
it tries to copy data from bad sectors to spare sectors on the
disk.  I gather there were too many bad sectors.  (Amazing how
the drive could fail so suddenly.  There were no power glitches
between the successful and failed backups.)  Then I attached the
disk to an eCS laptop and tried reformatting the disk with JFS
and the /L (long) option, which does media testing.  That took
something like six days.  Unclear whether it's because of the
disk being bad or the slowness of eCS 2.0's USB 2.0 drivers that
caused the long option to run so slowly.  Anyway, between the
99 percent completion and 100 percent completion stages, a bunch
of general failure error messages occurred, so many that they
scrolled off the top of the window, but the final message indicated
that the partition had been "successfully" formatted.  Then I ran
CHKDSK on it, and this time there was no recurrence of the
unrecoverable error reading M error.  Rather interestingly,
CHKDSK reported 0 bad blocks.  That got me wondering about the
difference between sectors and blocks, and how much self-repair
happens within the disk drive itself, with bad sectors being marked
off, quite independently of the operating system and file system,
and within the JFS filesystem, with bad blocks being marked off.
What can others tell me about this?

The disk drive appears to be useable again.  I'm certainly not
going to trust it with important data, and I can't imagine having
enough scratch data to utilize 2 TB worth of storage, so it may
get relegated to the scrap heap sooner rather than later, but I
may continue to experiment with it, to get a better idea of just
how reliable or unreliable a drive like this can be after a
complete reformatting.

Back to comp.os.os2.setup.storage | Previous | Next — Next in thread | Find similar

Thread

bad sectors versus bad blocks tholen@antispam.ham - 2013-01-12 01:42 +0000
  Re: bad sectors versus bad blocks Marcel Müller <news.5.maazl@spamgourmet.org> - 2013-01-13 11:29 +0100

csiph-web