I/O error in filesystem console message
Calum Wilkie
such messages. The first time, I booted off of my old disk and
ran fx to search for bad blocks and then ran xfs_repair to repair the
file system. This then allowed the disk to boot. I then wiped the
disk completely, ran fx again and reformated and re-installed the
disk. It has been working fine for a week... now I get the same
message, but different blocks. Is the disk bad, or do I need to
patch IRIX for large disks (this is a 36GB disk)?
[/] uname -a
IRIX64 despair 6.5 05190004 IP28
and the console message is:
ALERT: I/O error in filesystem ("/") meta-data dev 0x35 block 0x2730c60
("xfs_trans_read_buf")
dksc0d1: [Alert] Media error: Unrecovered data block read error
(asc=0x11, asq=0x0), (data byte 160), Block #41094240 (41360480)
dksc0d1: retrying request
dksc0d1: [Alert] Media error: Unrecovered data block read error
(asc=0x11, asq=0x0), (data byte 160), Block #41094240 (41360480)
dksc0d1: retrying request
dksc0d1: [Alert] Media error: Unrecovered data block read error
(asc=0x11, asq=0x0), (data byte 160), Block #41094240 (41360480)
dksc0d1: retrying request
dksc0d1: [Alert] Media error: Unrecovered data block read error
(asc=0x11, asq=0x0), (data byte 160), Block #41094240 (41360480)
dksc0d1: retries exhausted
Any help would be appreciated,
Calum
Ralf Beyer
>
> I have just installed a new disk... this is the second time I have seen
> such messages. The first time, I booted off of my old disk and
> ran fx to search for bad blocks and then ran xfs_repair to repair the
> file system. This then allowed the disk to boot. I then wiped the
> disk completely, ran fx again and reformated
Obviously you had very good reasons to reformat the disk ...
> and re-installed the
> disk. It has been working fine for a week... now I get the same
> message, but different blocks. Is the disk bad, or do I need to
> patch IRIX for large disks (this is a 36GB disk)?
Looks like the disk is going bad.
Using 'fx -x' and the 'exercise' option you may exercise the
disk _READ-ONLY_ if you have data on it or (better) READ-WRITE
if you don't care about its contents.
Exercise the disk in sequential mode several times until you
get one of these results:
- all bad blocks allocated, no bad blocks
detected anymore: disk may be in good working
condition again. Write-out the new label.
- bad block(s) can't be allocated (format error
or bad block list exhausted): try to reformat
and to exercise the disk and to start all over
again. Reformat may succeed or damage the disk
beyond repair for the ordinary user.
- new bad blocks show-up all the time and get
allocated to the bad block list: try to reformat
the disk (see above) and to exercise the disk
thereafter, but better get a new one.
Regards
Ralf Beyer
--
beyer.bra...@freenet.de
Walter Roberson
Ralf Beyer <beyer.bra...@freenet.de> wrote:
:Looks like the disk is going bad.
:Using 'fx -x' and the 'exercise' option you may exercise the
:disk _READ-ONLY_ if you have data on it or (better) READ-WRITE
:if you don't care about its contents.
wr-c (write compare) is the name for the read/write option.
:Exercise the disk in sequential mode several times until you
:get one of these results:
One small amendment:
In a situation where a disk starts reporting errors soon after
it was formatted, one possibility is that there is something wrong
with the seek mechanisms. To check that, as well as doing a
sequential test, I would also do a butterfly or random sequence test.
I have seen problems detected by sequential mode that random did not
find, and I have seen problems detected in random mode that sequential
did not find.
Tony Mantler
Roberson) wrote:
: In article <GpItw...@arl.army.mil>,
: Ralf Beyer <beyer.bra...@freenet.de> wrote:
[...]
: :Exercise the disk in sequential mode several times until you
: :get one of these results:
:
: One small amendment:
:
: In a situation where a disk starts reporting errors soon after
: it was formatted, one possibility is that there is something wrong
: with the seek mechanisms. To check that, as well as doing a
: sequential test, I would also do a butterfly or random sequence test.
:
: I have seen problems detected by sequential mode that random did not
: find, and I have seen problems detected in random mode that sequential
: did not find
I should second that. I just recently had a drive that would run the sequential
test perfectly, then fail miserably in the butterfly (and in real usage) about
half way through the disk.
Luckily, the operation of rescuing the data from the drive almost exactly
matched the seek profile of the 'sequential' test, so I didn't lose anything. ;)
Cheers - Tony 'Nicoya' Mantler :)
--
Tony "Nicoya" Mantler - Renaissance Nerd Extraordinaire - nic...@apia.dhs.org
Winnipeg, Manitoba, Canada -- http://nicoya.feline.pp.se/
Calum Wilkie
it keeps turning out more and more bad blocks...
Calum