bad version from xfs_repair after raid crash

bad version from xfs_repair after raid crash

[Andreas Vogler]

Hi all,

we had a crash of a raid 5 system. After rebuilding we could access one of  two partitions. Upon mounting the second parttion we got an error "mount: Structure needs cleaning". So we tried using xfs_repair.

Xfs_repair returns: "bad primary superblock - bad version number !!!"

and tries to find a secondary superblock but gives up after scanning the volume.

Looking at the superblock with xfs_db looks good as far as I can tell. Can anybody give me a hint what "bad version number" means?

Thank you very much in advance.

  Andreas Vogler

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

Re: bad version from xfs_repair after raid crash

[Eric Sandeen]

On 8/11/12 9:59 AM, Andreas Vogler wrote:
> Hi all,
> 
> we had a crash of a raid 5 system. After rebuilding we could access one of  two partitions. Upon mounting the second parttion we got an error "mount: Structure needs cleaning". So we tried using xfs_repair.
> 
> Xfs_repair returns: "bad primary superblock - bad version number !!!"
> 
> and tries to find a secondary superblock but gives up after scanning the volume.
> 
> Looking at the superblock with xfs_db looks good as far as I can tell. Can anybody give me a hint what "bad version number" means?

There is a version number in the superblock itself, which must have features recognized by repair.

Is your xfsprogs version up to date?

You could also include another email with the output from xfs_db -c "sb 0" -c "p" /dev/blah

-Eric

> Thank you very much in advance.
> 
>   Andreas Vogler
> 
> _______________________________________________
> 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

AW: bad version from xfs_repair after raid crash

[Andreas Vogler]

> Is your xfsprogs version up to date?


I guess the xfsprogs are rather old but they should be on the same level as the filesystem as they are part oft he same distro (OpenFiler 2.3, package is called xfsprogs=2.6.25-4-0.1).

> You could also include another email with the output from xfs_db -c "sb 0" -c "p" /dev/blah

Here ist he output of xfs_db:

magicnum = 0x58465342
blocksize = 4096
dblocks = 332800000
rblocks = 0
rextents = 0
uuid = 567f4c2f-ce9a-42a3-bbc2-e791cf43f8ae
logstart = 268435460
rootino = 128
rbmino = 129
rsumino = 130
rextsize = 1
agblocks = 83200000
agcount = 4
rbmblocks = 0
logblocks = 32768
versionnum = 0xb4a4
sectsize = 512
inodesize = 256
inopblock = 16
fname = "\000\000\000\000\000\000\000\000\000\000\000\000"
blocklog = 12
sectlog = 9
inodelog = 8
inopblog = 4
agblklog = 27
rextslog = 0
inprogress = 0
imax_pct = 5
icount = 1079360
ifree = 80
fdblocks = 178779506
frextents = 0
uquotino = 131
gquotino = 132
qflags = 0x77
flags = 0
shared_vn = 0
inoalignmt = 2
unit = 0
width = 0
dirblklog = 0
logsectlog = 0
logsectsize = 0
logsunit = 1
features2 = 0x8

But I have already experimented a little bit, originally versionnum was 0xb4f4 but the next superblock (sb 1) contained 0xb4a4, so I tried to set it with xfs_db to this value, but it didn't make a difference.

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

Re: bad version from xfs_repair after raid crash

[Dave Chinner]

On Sun, Aug 12, 2012 at 09:49:48AM +0200, Andreas Vogler wrote:
> 
> > Is your xfsprogs version up to date?
> 
> 
> I guess the xfsprogs are rather old but they should be on the same level as the filesystem as they are part oft he same distro (OpenFiler 2.3, package is called xfsprogs=2.6.25-4-0.1).

That's -very- old:

-- Nathan Scott <nathans@debian.org>  Tue, 29 Mar 2005 11:57:57 +1000

xfsprogs (2.6.26-1) unstable; urgency=low

  * New upstream release.
  * Man page updates (closes: #295397)
  * Fix compilation with gcc version 4 (closes: #297876)
  * Switch build dependency from readline4 to readline5.


> > You could also include another email with the output from xfs_db -c "sb 0" -c "p" /dev/blah
> 
> Here ist he output of xfs_db:
> 
> magicnum = 0x58465342
> blocksize = 4096
> dblocks = 332800000
> rblocks = 0
> rextents = 0
> uuid = 567f4c2f-ce9a-42a3-bbc2-e791cf43f8ae
> logstart = 268435460
> rootino = 128
> rbmino = 129
> rsumino = 130
> rextsize = 1
> agblocks = 83200000
> agcount = 4
> rbmblocks = 0
> logblocks = 32768
> versionnum = 0xb4a4
.....
> features2 = 0x8

So it's got the ATTR2 feature bit set.

 -- Nathan Scott <nathans@debian.org>  Tue, 31 Jan 2006 13:35:39 +1100

xfsprogs (2.7.7-1) unstable; urgency=low

  * New upstream release.
  * Add support for (optional) ATTR2 format extension (closes: #336350)
  * Allow gcc -pedantic option for C++ <xfs.h> users (closes: #249429)
  * Fix segv in xfs_db frag command (closes: #338207)

So, yeah, upgrade your userspace to something recent, and it will
just work.

> But I have already experimented a little bit, originally
> versionnum was 0xb4f4 but the next superblock (sb 1) contained
> 0xb4a4, so I tried to set it with xfs_db to this value, but it
> didn't make a difference.

Don't stuff around with feature bits if you don't know what they
mean - you'll render parts of your filesystem
unusable/corrupt/unreadable by doing so.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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

AW: bad version from xfs_repair after raid crash

[Andreas Vogler]

> So, yeah, upgrade your userspace to something recent, and it will just work.

Ok, the most recent xfs_repair doesn't complain about the version and at least starts repairing even though the system seems extremely corrupt. Thanks.

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