Re: [gentoo-user] [OT] Badblocks on my harddisk

Dale <rdalek1967@×××××.com> [14-07-26 09:54]:
> meino.cramer@×××.de wrote:
> > Hi,
> >
> > After running smartctl for an extended offline test I got
> > a badblock (information extracted from the report):
> >
> > SMART Self-test log structure revision number 1
> > Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
> > # 1  Extended offline    Completed: read failure       90%     14460         4288352511
> > 197 Current_Pending_Sector  0x0032   200   200   000    Old_age   Always       -       1
> >
> > I found a explanation to map the LBA to a partition here:
> > http://smartmontools.sourceforge.net/badblockhowto.html
> >
> > My partition layout is:
> > #> sudo fdisk -lu /dev/sda
> >
> > Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
> > Units: sectors of 1 * 512 = 512 bytes
> > Sector size (logical/physical): 512 bytes / 512 bytes
> > I/O size (minimum/optimal): 512 bytes / 512 bytes
> > Disklabel type: dos
> > Disk identifier: 0x07ec16a2
> >
> > Device     Boot      Start        End    Blocks  Id System
> > /dev/sda1  *          2048     104447     51200  83 Linux
> > /dev/sda2           104448   12687359   6291456  82 Linux swap / Solaris
> > /dev/sda3         12687360  222402559 104857600  83 Linux
> > /dev/sda4        222402560 1953525167 865561304   5 Extended
> > /dev/sda5        222404608  232890367   5242880  83 Linux
> > /dev/sda6        232892416  442607615 104857600  83 Linux
> > /dev/sda7        442609664  652324863 104857600  83 Linux
> > /dev/sda8        652326912  862042111 104857600  83 Linux
> > /dev/sda9        862044160 1071759359 104857600  83 Linux
> > /dev/sda10      1071761408 1281476607 104857600  83 Linux
> > /dev/sda11      1281478656 1491193855 104857600  83 Linux
> > /dev/sda12      1491195904 1953525167 231164632  83 Linux
> >                 4288352511  <<< The number reported by smartctl
> >
> >
> > Following the linked document...
> > It seems the bad LBA is not on the checked harddisk.
> >
> > Or (more obvious) I did something wrong...
> >
> > How can I correctly identify the partition, which contains the bad
> > block?
> > How can I get a full list of all bad blocks (if any) from a mounted
> > file systems?
> > How severe is the problem?
> >
> > Thank you very much for any help in advance!
> > Best regards,
> > mcc
> >
>
> I ran into this recently on the drive that has my home partition on it.
> Someone posted that it *may* be fixable without moving data etc etc.  I
> didn't have a backup at the time and nothing large enough to make one so
> I just ordered a new drive.  When I got the new drive in and moved my
> data over, then I played with the drive a bit.  I used dd to erase the
> drive, then stuck a file system back on it and filled it up.  After
> doing that, the drive seems to have marked that part as bad and doesn't
> use it anymore.  It has passed every test since then.
>
> My point is this, backups for sure just in case but you may be able to
> get the drive to mark that area as bad by moving that data off there.
> In my case, the files were corrupted and gone.  Yea, I might could have
> sent it somewhere but I ain't into that.  To much money for files I can
> replace if needed.  I think it was like 3 or 4 video files.  I'd find
> out what files are there, see what damage has occurred so that you can
> correct later, then find one really good howto and follow it.   From my
> understanding, if you can move that data in the bad spot off there, the
> drive sort of fixes itself.  If yours works like mine did, you should be
> OK but I'd use it for stuff that ain't so important.  I use mine as a
> backup drive and test it a lot.  ;-)  I may trust it again, one day.
>
> So, most likely you will have some files corrupted at least.  The drive
> *may* be fixable if you can figure out what files to move so that the
> drive can do its magic.  Key thing is, finding out what to move so that
> the drive can do its work.  Two options, try to move files so the drive
> can do its thing or move all the data to another drive, do like I did
> mine with dd and give it a fresh start that way.   I didn't feel I had
> the experience to try and move the files so I took the 2nd option.  Now
> I wish I had done option #1 and took notes that I could pass on.  That
> would likely help you more.
>
> BTW, my drive gave that error for weeks and never got worse.  I could be
> lucky on that one so do what needs doing as soon as you can, just in
> case.  The last drive that really failed on me years ago, I got a
> serious warning from SMART.  It even said I had like 24 hours to get my
> data off.  It needs attention in your case but hopefully you will have
> the results I did in the end and you have time to deal with it.
>
> Dale
>
> :-)  :-)
>


Hi Dale,

thank you very much for the explanations you gave...and for the hope
in it ;) :)

In the meanwhile I found ddrescue... :)

It took me five hours to copy the disk (1T) binaryly (this word looks
wrong...) to another identical one with ddrescue. This beast is
smart...it first copies all what it is able to read instantly and
writes out a logfile, which contains the informations, what is wrong
with the disk and where:


# Rescue Logfile. Created by GNU ddrescue version 1.16
# Command line: ddrescue -f -n /dev/sda /dev/sdb ddrescue.log
# current_pos  current_status
0x36220000     +
#      pos        size  status
0x00000000  0x3621F000  +
0x3621F000  0x00000E00  /
0x3621FE00  0x00000200  -
0x36220000  0xE8AAB96000  +

In my case it report one errornous read and a defective size of 4096
bytes.

After that it is called a second time with different parameters and
the name of the logfile.

It then tries to read the sector again and retries it several times.
after that the logfile looks like this:


# Rescue Logfile. Created by GNU ddrescue version 1.16
# Command line: ddrescue -d -f -r3 /dev/sda /dev/sdb ddrescue.log
# current_pos  current_status
0x3621FE00     +
#      pos        size  status
0x00000000  0x3621F000  +
0x3621F000  0x00001000  -
0x36220000  0xE8AAB96000  +

What has been fixed has gone from the logfile.

So there is something left...

I will start a complete smartctl scan again and will see, whether
the bad block has been mapped and replaced.

I *hope* that this is a single accident, because only one spot (and a
small one) one a 1T disk is affected...we will see (fingers crossed).

By the way: There are other tools similiar to ddrescue called
dd_rescue and similiar. I found dddrescue recommended over the others
on the net.

Now....will start a smartctl complete check...this will take hours...
I will report later, what happened...

Best regards and have a nice weekend!
mcc