"One or more disks are failing" ?
Bill Davidsen
davidsen at tmr.com
Mon Jul 6 00:17:08 UTC 2009
Alan Cox wrote:
> On Sat, 04 Jul 2009 16:07:18 -0700
> Scott Beamer <geekboy at angrykeyboarder.com> wrote:
>
>> On 07/04/2009 10:25 AM, Bruno Wolff III wrote:
>>> Drives typcially won't reallocate bad sectors if they can't get a good read
>>> or the operation is a write. This is to give you a chance to recover the data
>>> if you want to try. And if you want to spend some effort, you can figure out
>>> what files, if any, were using these blocks.
>> Wouldn't checking for bad sectors (finding none) followed by formatting
>> the drive eliminate this problem?
>
> Most drives will reallocate a bad sector providing you write over it.
> fsck will do this for problematic metadata (block counts, bitmaps, inodes
> etc) if it has to.
>
> For data hdparm --repair-sector offers a very low level interface. As
> there is no easy way of finding out which file owns the problematic block
> or how many there are and which files they are in without accessing that
> bit of data a backup and restore is normally wise.
>
> When ever possible I use raid 1 (mirroring). Drives are fairly cheap,
> sizes are so big that capacity isn't a problem. Reliability without raid
> isn't good enough IMHO.
>
Why raid-1 rather than raid-10,f2? The performance seems significantly better,
although write performance is still a place where a good hardware raid
controller can beat software raid, by only sending one copy of the data over the
system bus to the controller.
--
Bill Davidsen <davidsen at tmr.com>
"We have more to fear from the bungling of the incompetent than from
the machinations of the wicked." - from Slashdot
More information about the fedora-list
mailing list