Mailinglist Archive: opensuse (2912 mails)

< Previous Next >
Re: [SLE] fsck failed
  • From: "Carlos E. R." <robin1.listas@xxxxxxxxxx>
  • Date: Thu, 10 Feb 2005 02:36:48 +0100 (CET)
  • Message-id: <Pine.LNX.4.58.0502100228100.12684@xxxxxxxxxxxxxxxx>

The Tuesday 2005-02-08 at 22:57 -0500, Carl E. Hartung wrote:

> | I think your partition probably developped some badblocks. I'm not
> | sure of the correct procedure nowdays with reiser partitions.
> |
>
> Hi Amanda & Carlos,
>
> I found a really fascinating post on this topic in another list archive.
> ~ The author is a BSEE, CISSP & RHCE and it looks like he's following a
> well designed "boilerplate" protocol for analyzing and responding to
> this problem. What do you think?

Let's see.

> Quote
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> By the numbers...
...

> # dd if=/dev/hda5 of=/dev/null
> dd: reading `/dev/hda5': Input/output error
> 4825684+0 records in
> 4825684+0 records out

Right, I used that method myself time ago. It just triggers read errors.

>
> # cat /var/log/messages | grep LBA
> Nov 15 18:15:01 lioness kernel: hda: dma_intr: error=0x40
> { UncorrectableError }, LBAsect=42088723, sector=42088559
> ...
> Nov 15 18:24:30 lioness kernel: hda: dma_intr: error=0x40
> { UncorrectableError }, LBAsect=42088723, sector=42088723

Right, he gets the LBA sector number of the errors.

>
> # smartctl -t /dev/hda
> ...

And he confirms the error in smartctl log.

> -- -- -- -- -- -- --
> 40 51 5c 13 39 82 e2 Error: UNC 92 sectors at LBA = 0x02823913 = 42088723

Same place.

> /Quote

But I was referring to "repair procedure". Reiserfs has some badblock
handling capability now (since SuSE 9.1, I think), but I'm unsure how it
works or how to use it.

Another method is to trigger the HD remapping feature. Simply writing to
the bad block will cause the HD firmware to remap the sector - but loosing
the affected files, I think.

--
Cheers,
Carlos Robinson


< Previous Next >