[linux-lvm] Re: [reiserfs-list] lvm-list pointed to reiserfs: fs/lvm-problems?

[linux-lvm] Re: [reiserfs-list] lvm-list pointed to reiserfs: fs/lvm-problems?

[Alexander Zarochentcev]

 PB> Heinz Mauelshagen pointed me to send my error-report to this list:
 PB> ----- Forwarded message from Peter Bartosch <peter@bartosch.net> -----

 PB> Date: Sun, 27 May 2001 23:41:17 +0200
 PB> From: Peter Bartosch <peter@bartosch.net>
 PB> To: linux-lvm@sistina.com
 PB> Subject: lvm-problems?

 PB> hello,

 PB> i recently made my three LV new (caused by an accident as i tried to
 PB> extend an existing (0.8i) LV)

 PB> so i created a new kernel (2.4.2) with lvm 0.9.1_beta7

Linux-2.4.4 has serious reiserfs bugs fixed.

 PB> the first problems came when i tried to create more than one VG - the
 PB> second VG couldn't be activated (always inactive after reboot and unable
 PB> to activate)

 PB> so i'm using only one VG

 PB> the first two LV's of this VG could i create without problems, but the
 PB> third makes problems:

 PB> i created the reiserfs on it, but when i tried to mount it, the mount
 PB> failed
 PB> an reiserfsck --rebuild-sb and after that reiserfsck --rebuild-tree
 PB> clears it and i could mount my LV -- UNTIL the next reboot

What kernel messages were ?

Sorry, but it looks for me like a lvm problem.  Can you test you LV by
coping some random data (~ 64 MB) to a file in reliable location and
to LV, and compare them (or md5 hashes) after reboot ?

 PB> i had to reboot because something oopsed:

 PB> ----

Reiserfs error handling not an ideal one.  Kernel oops below is a
normal :( reaction on damaged data coming from disk.

Same reason (damaged data), I think, prevented reiserfs volume from
mounting. 


 PB> is_leaf: item location seems wrong (second one): *NEW* [9 10 0x0 SD],
 PB> item_len 44, item_location 2308, free_space(entry_count) 65535
 PB> vs-5150: search_by_key: invalid format found in block 8271. Fsck?
 PB> kernel BUG at namei.c:343!
 PB> invalid operand: 0000
 PB> CPU:    0
 PB> EIP:    0010:[<c018ff08>]
 PB> EFLAGS: 00010286
 PB> eax: 0000001b   ebx: c6655e58   ecx: 00000001   edx: 00000001
 PB> esi: c02d4fb4   edi: c6655edc   ebp: c6655e70   esp: c6655e40
 PB> ds: 0018   es: 0018   ss: 0018
 PB> Process cp (pid: 1799, stackpage=c6655000)
 PB> Stack: c02d4c7e c02d4d7e 00000157 00000000 c02d4fb4 c6655e9c 00000002 0000000e 
 PB> 	4a0507ff 000001f4 00000000 00000003 c3e55a20 c0190030 c7742c20 c7dd3b20 
 PB>         0000001b c6655e9c c6655edc fffffff4 c6654000 c7742c20
 PB> 	c3e55a20 00000001 
 PB> Call Trace: [<c0190030>] [<c01468e9>] [<c013de9a>]
 PB>         [<c013e6e9>] [<c013db5a>] [<c013ee4c>] [<c013b8c6>] 
 PB>         [<c0133047>] [<c0108f5b>] 
 PB> Code: 0f 0b 83 c4 0c 8d 76 00 8b 4d 10 51 8b 45 0c 50 57 8d 55 e8 

 PB> -----

 PB> this oops let the process which works on the third LV freeze - so i
 PB> couldn't kill it

 PB> the first and the second LV seem to work well

 PB> when i reboot the third LV won't be mountet automagically and the
 PB> filesystem seems to be corrupt


 PB> any help will be appreciated

 PB> ----- End forwarded message -----

 PB> as this happened just again, i got some more informations:


 PB> vs-13048: reiserfs_iget: bad_inode. Stat data of (9 10) not found
 PB> is_leaf: item location seems wrong (second one): *NEW* [9 10 0x0 SD],
 PB> item_len 4
 PB> 4, item_location 2308, free_space(entry_count) 65535
 PB> vs-5150: search_by_key: invalid format found in block 8271. Fsck?
 PB> vs-13070: reiserfs_read_inode2: i/o failure occurred trying to find stat
 PB> data of
 PB>  [9 13 0x0 SD]
 PB>  vs-13048: reiserfs_iget: bad_inode. Stat data of (9 13) not found
 PB>  is_leaf: item location seems wrong (second one): *NEW* [9 10 0x0 SD],
 PB>  item_len 4
 PB>  4, item_location 2308, free_space(entry_count) 65535
 PB>  vs-5150: search_by_key: invalid format found in block 8271. Fsck?
 PB>  kernel BUG at namei.c:343!
 
-- 
Thanks,
Alex.

[linux-lvm] Re: [reiserfs-list] lvm-list pointed to reiserfs: fs/lvm-problems?

[Peter Bartosch]

Hi!


>  PB> extend an existing (0.8i) LV)
> 
>  PB> so i created a new kernel (2.4.2) with lvm 0.9.1_beta7
> 
> Linux-2.4.4 has serious reiserfs bugs fixed.

hmm, i tried to make a LVM patch for it but this fails - yesterday i saw a
message, that there's somewhere a patch for the 2.4.4 that LVM-sources
will make a patch

>  PB> i created the reiserfs on it, but when i tried to mount it, the mount
>  PB> failed
>  PB> an reiserfsck --rebuild-sb and after that reiserfsck --rebuild-tree
>  PB> clears it and i could mount my LV -- UNTIL the next reboot
> 
> What kernel messages were ?
> 
> Sorry, but it looks for me like a lvm problem.  Can you test you LV by
> coping some random data (~ 64 MB) to a file in reliable location and
> to LV, and compare them (or md5 hashes) after reboot ?

hmm, i haven't really tried another filesystem, maybe i should try ext2?

> Reiserfs error handling not an ideal one.  Kernel oops below is a
> normal :( reaction on damaged data coming from disk.
> 
> Same reason (damaged data), I think, prevented reiserfs volume from
> mounting. 

yes you're right - but why does it oops when i _only_ want to do an "ls
-l"?

:wq - until next mail B-), l8r

Peter
-- 
   :~~~~~~~~~~~~~~~~~~  peter@bartosch.net  ~~~~~~~~~~~~~~~~~~:
   :          student of technical computer science           :
   :     university of applied sciences krefeld (germany)     :
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   FD314F21   C7 AE 2F 28 C1 33 71 77  0D 77 CD 6E 58 E9 06 6B

Re: [linux-lvm] Re: [reiserfs-list] lvm-list pointed to reiserfs: fs/lvm-problems?

[Heinz J. Mauelshagen]

On Tue, May 29, 2001 at 10:23:04PM +0200, Peter Bartosch wrote:
> Hi!
> 
> 
> >  PB> extend an existing (0.8i) LV)
> > 
> >  PB> so i created a new kernel (2.4.2) with lvm 0.9.1_beta7
> > 
> > Linux-2.4.4 has serious reiserfs bugs fixed.
> 
> hmm, i tried to make a LVM patch for it but this fails - yesterday i saw a
> message, that there's somewhere a patch for the 2.4.4 that LVM-sources
> will make a patch

You didn't tell what the failure cause was.

It could likely be the renaming of get_hardblocksize to get_hardsect_size
in the 2.4.4 kernel which hits lvm.c.

The recent CVS software at www.sistina.com has that fix already or
you change it in lvm.c end recompile/reinstall.

Hope this helps.

Regards,
Heinz    -- The LVM Guy --

> 
> >  PB> i created the reiserfs on it, but when i tried to mount it, the mount
> >  PB> failed
> >  PB> an reiserfsck --rebuild-sb and after that reiserfsck --rebuild-tree
> >  PB> clears it and i could mount my LV -- UNTIL the next reboot
> > 
> > What kernel messages were ?
> > 
> > Sorry, but it looks for me like a lvm problem.  Can you test you LV by
> > coping some random data (~ 64 MB) to a file in reliable location and
> > to LV, and compare them (or md5 hashes) after reboot ?
> 
> hmm, i haven't really tried another filesystem, maybe i should try ext2?
> 
> > Reiserfs error handling not an ideal one.  Kernel oops below is a
> > normal :( reaction on damaged data coming from disk.
> > 
> > Same reason (damaged data), I think, prevented reiserfs volume from
> > mounting. 
> 
> yes you're right - but why does it oops when i _only_ want to do an "ls
> -l"?
> 
> :wq - until next mail B-), l8r
> 
> Peter
> -- 
>    :~~~~~~~~~~~~~~~~~~  peter@bartosch.net  ~~~~~~~~~~~~~~~~~~:
>    :          student of technical computer science           :
>    :     university of applied sciences krefeld (germany)     :
>     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>    FD314F21   C7 AE 2F 28 C1 33 71 77  0D 77 CD 6E 58 E9 06 6B
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] Re: [reiserfs-list] lvm-list pointed to reiserfs: fs/lvm-problems?

[Alexander Zarochentcev]

 PB> Hi!
 PB> extend an existing (0.8i) LV)
 >> 
 PB> so i created a new kernel (2.4.2) with lvm 0.9.1_beta7
 >> 
 >> Linux-2.4.4 has serious reiserfs bugs fixed.

 PB> hmm, i tried to make a LVM patch for it but this fails - yesterday i saw a
 PB> message, that there's somewhere a patch for the 2.4.4 that LVM-sources
 PB> will make a patch

 PB> i created the reiserfs on it, but when i tried to mount it, the mount
 PB> failed
 PB> an reiserfsck --rebuild-sb and after that reiserfsck --rebuild-tree
 PB> clears it and i could mount my LV -- UNTIL the next reboot
 >> 
 >> What kernel messages were ?
 >> 
 >> Sorry, but it looks for me like a lvm problem.  Can you test you LV by
 >> coping some random data (~ 64 MB) to a file in reliable location and
 >> to LV, and compare them (or md5 hashes) after reboot ?

 PB> hmm, i haven't really tried another filesystem, maybe i should try ext2?

Yes, try to run this script over ext2:
http://namesys.com/stress.sh

 >> Reiserfs error handling not an ideal one.  Kernel oops below is a
 >> normal :( reaction on damaged data coming from disk.
 >> 
 >> Same reason (damaged data), I think, prevented reiserfs volume from
 >> mounting. 

 PB> yes you're right - but why does it oops when i _only_ want to do an "ls
 PB> -l"?

I agree with you, such behaviour is not good, but I do not think
that you want to run reiserfs on an unstable block device.

-- 
Thanks,
Alex.

Re: [linux-lvm] Re: [reiserfs-list] lvm-list pointed to reiserfs: fs/lvm-problems?

[Peter Bartosch]

Hi!


>  >> 
>  >> Sorry, but it looks for me like a lvm problem.  Can you test you LV by
>  >> coping some random data (~ 64 MB) to a file in reliable location and
>  >> to LV, and compare them (or md5 hashes) after reboot ?
> 
>  PB> hmm, i haven't really tried another filesystem, maybe i should try ext2?
> 
> Yes, try to run this script over ext2:
> http://namesys.com/stress.sh

i've done that, but it gave me some cruel errors - like an ext2 panic or
so (can't find anything about it in the logs)

anyway - the PC was frozen


after the reboot e2fsck showed me that:

Inode 1313467 has imagic flag set.  Clear<y>? yes

Inode 1313468 is in use, but has dtime set.  Fix<y>? yes

and - of course not only once :-( but hundred times

>  >> Same reason (damaged data), I think, prevented reiserfs volume from
>  >> mounting. 
> 
>  PB> yes you're right - but why does it oops when i _only_ want to do an "ls
>  PB> -l"?
> 
> I agree with you, such behaviour is not good, but I do not think
> that you want to run reiserfs on an unstable block device.

it's (mostly) only because of the fschk-times -- this is (or should be)
an 125GB directory tree for music


the next step i'll do ist to give the new kernel (2.4.5) a try (hmm,
eventually with ac4-patch - i've read that there are some
reiserfs-issues are solved ...)


thanks for your attention


:wq - until next mail B-), l8r

Peter
-- 
   :~~~~~~~~~~~~~~~~~~  peter@bartosch.net  ~~~~~~~~~~~~~~~~~~:
   :          student of technical computer science           :
   :     university of applied sciences krefeld (germany)     :
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   FD314F21   C7 AE 2F 28 C1 33 71 77  0D 77 CD 6E 58 E9 06 6B