Re: hard drive repair problem
On Mon, Jun 05, 2006 at 03:26:36PM -0400, Marty Landman wrote:
> At 12:54 PM 6/5/2006, Digby Tarvin wrote:
> >On Mon, Jun 05, 2006 at 12:35:30PM -0400, Marty Landman wrote:
> >>
> >> Ok does this result indicate a hardware error then?
> >>
> >> penskefile:/home/marty# dd if=/dev/hdb1 of=/dev/null
> >> dd: reading `/dev/hdb1': Input/output error
> >> 40+0 records in
> >> 40+0 records out
> >> 20480 bytes transferred in 4.123283 seconds (4967 bytes/sec)
> >> penskefile:/home/marty#
> >
> >I am afraid so. At a minimum, you have a bad sector at a critical
> >place in the filesystem which fsck is unable to deal with. But it
> >may be a failed head or worse, meaning that you will have lost
> >much if not all of the data beyond the first 20K of your partition.
> >
> >If you have valuable data that is not backed up, try using the
> >'skip=blocks' option on dd to jump further into the partition
> >to see if the disk is good once you are beyond the initial
> >fault region.
>
> Things are looking a bit up over here. For one thing Maxtor (now
> Seagate) told me there's a 30 day grace period on my 6 mo warrantee
> which is kind of cool because I bought this HD about 6 mo plus 2 1/2
> weeks ago. So once the dd finishes up I can get the serial # and call
> back for an RMA.
>
> Thing is, I seem able to dd everything past relatively close to the
> beginning of the drive. I mean
>
> penskefile:/home/marty# dd if=/dev/hdb1 of=/dev/null skip=130
> dd: reading `/dev/hdb1': Input/output error
> 0+0 records in
> 0+0 records out
> 0 bytes transferred in 2.666983 seconds (0 bytes/sec)
> penskefile:/home/marty# dd if=/dev/hdb1 of=/dev/null skip=150
>
> the skip 150 is still chugging along after half an hour.
>
> Anyone out there know of a way that I can keep some or all of this data?
>
> I have the further complication that I bought this drive specifically
> because I don't have this kind of space available otherwise on my
> LAN. It's probably got < 50 GB of data, mostly captured home videos
> I'd rather not re-capture though I certainly could. Available other
> space is in blocks or maybe 30 GB here (1), 1-5 GB there (5 or 10). I
> do have a DVD burner... on my Windows workstation. Wonder if I could
> burn an ISO image or three. Have yet to burn a DVD - less than 5 GB
> per that's a lot of burning.
>
> Well it's Monday, certainly wasn't expecting it all to be good news.
IDE drives are pretty cheap these days, so if you don't have a backup
device of sufficient capacity, what I would do in your place is..
1. Buy a replacement drive of same or greater capacity as the damaged one
2. Creat a partition of identical size as the one of the old drive
3. Use dd to copy every good sector from the old drive to the new
drive (some trial and error should determine the first good sector
after the bad region)
4. return the old drive for warranty replacement
5. use some recovery tools, like perhaps the 'testdisk' suggested
by Florian to attempt to recover as much as possible from your disk.
I'm not an ext3 expert, but a lot will probably depend on weather or
not your inode data can be recovered. If it can, then all of the
files not stored in the bad part of the disk should be recoverable.
6. Use the replacement drive to do your nightly backups.
I have been using rsync to do nightly backups this way ever since hard
drives got larger than capacities that my tape drive could comfortably
handle.
It might be interesting to check what 'smartctl' reports for your bad
disk before you send it back. It should give you some concrete indication
of what to say when requesting your rma if nothing else...
Regards,
DigbyT
--
Digby R. S. Tarvin digbyt(at)digbyt.com
http://www.digbyt.com
Reply to: