Siegbert Marschall <siggi@xxxxxxxxxxxxx> wrote:
> > Aug 31 11:40:01 curly /bsd: SENSE KEY: Media Error
> > Aug 31 11:40:01 curly /bsd: INFO: 0x8ced8b (VALID flag on)
> > Aug 31 11:40:01 curly /bsd: COMMAND INFO: 0xeb00175
> > Aug 31 11:40:02 curly /bsd: ASC/ASCQ: Uncorrected Read Error -
> > Recommend Rewrite The Data
>
> 1. Rewrite the Data.
Or just rewrite the block with any data. Zeroes are fine. The
data in the block is already lost, but you may be able to make the
file system fully operational again. If the block contained meta
data you want to force a file system check afterwards. You can use
dd(1) on the raw disk to locate the block and rewrite it, that's
where dd's skip, seek, and count options come in handy.
> 4. Get a fresh disk since this one is going doowwwnnnn.....
Not necessarily. Modern ATA and SCSI disks are supposed to relocate
defective blocks on their own--for SCSI drives there's a bit in a
mode page to control this--but sometimes they fail to do so in time
and if the block just can't be read any longer, there's nothing the
drive can do but signal an error. Rewriting the block will give
the drive the opportunity to transparently swap it out.
I've successfully rewritten single defective blocks on both ATA and
SCSI disks that have continued to work fine afterwards.
--
Christian "naddy" Weisgerber naddy@xxxxxxxxxxxx
|
