Hi Emmanuel,
thanks for your reply, sorry for the delayed answer...
Am 23.01.2014 um 15:19 schrieb Emmanuel Florac <eflorac@xxxxxxxxxxxxxx>:
> Le Wed, 22 Jan 2014 17:09:10 +0100
> Sascha Askani <saskani@xxxxxxxxx> écrivait:
>
>> Internal error xfs_bmbt_read_verify at line 789
> ...
>> metadata I/O error: block 0x1f0 ("xfs_trans_read_buf_map") error 117
>> numblks 8
>
>
> A quick glance at the source code seems to indicate that there was an
> actual CRC error (on disk) at this point, so that could be a RAID
> problem;
(Un)fortunately, I don’t see any kind of controller errors, but I definitely
second your opinion
>
> OTOH it looks really similar to an older bug :
>
> http://oss.sgi.com/archives/xfs/2013-02/msg00482.html
>
> AFAIK this one could be related to the current Ubuntu LTS lvm stack...
> We need Dave's advice on this one.
>
> Was there nothing in the xfs_repair output? nothing in lost+found
> either?
xfs_repair output (which I unfortunately do not have) was sparse (ignoring the
many dots it printed after having failed to find the primary superblock),
basically it removed invalid references. lost+found contains 1 directory (which
is empty)
As for the bug you referenced above: Yes, it looks similar, but we have no
pvmove going on anywhere at any time, but I’ll see if I can try to reproduce
the error with the many files (kernel source) as described in the bug.
For reference, the LVM versions currently in use:
root@hsoi-gts3-de02:/# lvdisplay --version
LVM version: 2.02.66(2) (2010-05-20)
Library version: 1.02.48 (2010-05-20)
Driver version: 4.25.0
Thanks so far :)
Best regards,
Sascha
signature.asc
Description: Message signed with OpenPGP using GPGMail
|