[Linux-disciples] Lots of errors

[Stephen R Laniel]

On Tue, Dec 07, 2004 at 09:41:34AM -0500, Adam Rosi-Kessel wrote:
> Ah, that might explain the nonstandardness in the SMART report (weird 
> percentages). Older drives tend to have SMART oddities. I would 
> suggest replacing, or at least supplementing with an identical copy, 
> the drive right away. I have a whole bunch of spare drives probably in 
> that size range, but they might be in the "bad age" category as well.
> 
> The smallest you can find on pricewatch these days is 20.0GB for $20.

I'll probably get a new hard drive soon, but did you notice
the weirdness in the SMART output? E.g.,

Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Vendor offline      Fatal or unknown error        130%         1         218104125
# 2  Vendor offline      Fatal or unknown error        130%         1         2130706749

That's a very large LBA: 2,130,706,749. That's a
(sequential) block number, right? So that would be the
two-billionth block on the disk, at 4K per block. Hence the
drive would have to be something like 8 TB.

The whole thing is screwy, obviously, to the point that I
can't even really decide whether anything is wrong. I assume
it is, and it's not a good idea to take chances.

I ran e2fsck on my unmounted root volume the other day. It
took much less time than my 200G drive did to scan, and was
done within an hour or so. Even though the scan was
read-only, it said something at the end like
"*****FILESYSTEM WAS CHANGED*****". I assume that's good,
since the idea would be to mark bad blocks, which would
sometimes involve zeroing out files from bad blocks. But it
looks like the SMART report is still showing lots of bad
blocks.

I realize we don't have much of a knowledge base about this
subject on l-d, but I thought I'd toss this in as a datum.

-- 
``I'm not up on my knitting terminology, but I assert that
  a 'uterus doll' should never, ever be made from anything
  called 'Cascade 128 Chunky'''
 -Brian Cooke, 9 December 2004

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.bostoncoop.net/pipermail/linux-disciples/attachments/20041209/8068605e/attachment.pgp