xfs
[Top] [All Lists]

Re: XFS corruption

To: tima@xxxxxxxxxxxxxx, linux-xfs@xxxxxxxxxxx
Subject: Re: XFS corruption
From: Seth Mos <knuffie@xxxxxxxxx>
Date: Mon, 29 Oct 2001 23:17:57 +0100
In-reply-to: <20011029133012.A6980@bud>
Sender: owner-linux-xfs@xxxxxxxxxxx
At 13:30 29-10-2001 -0800, Artem Veremey wrote:
Hi all,
I have a case of XFS corruption to report.

After poweroff, /dev/hda8 (/home) did not mount - "unknown fs...".
xfs_repair produces the following results:

xfs_repair /dev/hda8
Phase 1 - find and verify superblock...
superblock read failed, offset 370130944, size 2048, ag 4294967295, rval 4

fatal error -- Input/output error

Kernel logged a few dozen errors like this during xfs_repair (that happens with DMA disabled):
Oct 29 09:56:21 bud kernel: hda: dma_intr: status=0x51 { DriveReady SeekComplete Err
or }
Oct 29 09:56:21 bud kernel: hda: dma_intr: error=0x40 { UncorrectableError }, LBAsec
t=19197740, sector=722913
Oct 29 09:56:21 bud kernel: end_request: I/O error, dev 03:08 (hda), sector 722913


xfs_check produced many errors too of the following two types:
block 7/305 expected type unknown got free2
and
block 6/22564 expected type free1 got unknown

Your disk is b0rken. (Like so many other IBM drives). Better see your retailer since it will probably go _very_ soon.


Disk (IBM 60GXP 40Gb) seems to be OK (only one partition failed and filling new fs on /dev/hda8 with dd doesn't produce errors)

There is a thorough explanation for this in the FAQ. Sometimes drives only start remapping clusters after a reboot or after a too long time. Sometimes you run out of spare clusters and you actually get to see this. Get your disk replaced imediately. It think it will hold out for a day if you are lucky.


Machine runs a kernel build from cvs update on Oct 25.

Linux version 2.4.13-xfs-fast (root@bud) (gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)) #1 Thu Oct 25 22:23:37 PDT 2001
(-fast is just a configuration for specific hardware on that box)


/sbin/xfs_repair -V
xfs_repair version 1.3.9

MB chipset - ALi Magic1.

Your disk is having problems and not the IDE controller.

The question is who's fault is that? :)

The disk and mine ofcourse.

Cheers
--
Seth
Every program has two purposes one for which
it was written and another for which it wasn't
I use the last kind.


<Prev in Thread] Current Thread [Next in Thread>