Re: How to fix bad superblock or xfs_repair: error - read only 0 of 512 bytes

[Eric Sandeen <sandeen@sandeen.net>]

On 1/23/12 3:23 AM, Christian Kildau wrote:
> On Jan 23, 2012, at 5:31 AM, Dave Chinner wrote:
> 
>> On Sat, Jan 21, 2012 at 11:29:15AM +0100, Christian Kildau wrote:
>>> Sorry if this message appears twice!
>>>

Argh.  ;)

>>>
>>> Hello,
>>>
>>> I'm having some very serious issues with XFS after upgrading from a
>>> Linux Distro running Ubuntu 2.6.32 to 3.2.
>>>
>>> It seems like my filesystems are damaged after attaching them to a
>>> Linux 3.2 server. I am also no longer able to mount the hdd on the old
>>> server that is still running 2.6.32!
>>
>> I take it that you are using external storage of some kind? Can you
>> describe it?
> 
> This hdd is connected via e-sata, but it doesn't make any difference if I directly connect it internally via sata.
> Also it doesn't make any difference if I connect it back to the 'old' server
> 
>>
>>> (I created the xfs filesystem on the entire hdd, not on a partition,
>>> so /dev/sdd is not a typo)

I wonder if your installer helpfully scribbled something on it since it
had no partitions (which should be safe, but there are dumb apps out there).

>>> $ sudo mount -t xfs /dev/sdd /media/
>>> mount: /dev/sdd: can't read superblock
>>> (dmesg)
>>> [236659.912663] attempt to access beyond end of device
>>> [236659.912667] sdd: rw=32, want=2930277168, limit=2930275055
>>> [236659.912670] XFS (sdd): last sector read failed
>>
>> So XFS has asked to read 2113 sectors beyond the size of the device
>> that the kernel is reporting. What is the output of
>> /proc/partitions?
> 
> $ grep sdd /proc/partitions 
>    8       64 1465137527 sdd

so 1465137527*1024 = 1500300827648 bytes

>From the strace repair is trying to read at:

pread(4, "", 512, 1500301909504)        = 0

which is about 1 meg past the end of the device.

>>
>>> $ sudo xfs_check /dev/sdd
>>> xfs_check: error - read only 0 of 512 bytes
>>>
>>> $ sudo xfs_repair /dev/sdd
>>> Phase 1 - find and verify superblock...
>>> xfs_repair: error - read only 0 of 512 bytes
>>
>> So both buffered and direct IO to the first block in the block
>> device are failing. I'd say your problems have nothing to do with
>> XFS. However, can you strace them and find out what the error that
>> is occuring actually is?
> 
> Strace is giving me:
> wait4(-1, xfs_check: /dev/sdd is not a valid XFS filesystem (unexpected SB magic number 0x00000000)

now that is something else...

> xfs_check: WARNING - filesystem uses v1 dirs,limited functionality provided.
> xfs_check: read failed: Invalid argument
> cache_node_purge: refcount was 1, not zero (node=0x21ecef0)
> xfs_check: cannot read root inode (22)
> bad superblock magic number 0, giving up

those are different failures than first reported....

xfs_db -c "sb 0" -c "p" /dev/sdd still might be interesting.

-Eric

> I attached the entire strace logs to this email.
> 
> 
> 
> 
> 
> Do you have any idea what has caused this or how to fix it?
> 
> Thanks in advance!
> Chris
> 
> 
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs