snapshot file zombies

Hi dear list,

I created a snapshot from my /home to do my usual backup with tar. Then I got 15 messages of the kind

tar: lvm-snapshot/vollmer/Mail/inbox/,120: can not stat, no permission
(or so it's translated from german to englich by me)

doing a
find /lvm-snapshot/vollmer/Mail/inbox/ -name ",120"
rsults in
/lvm-snapshot/vollmer/Mail/inbox/,120

so the file exists, but:
ls -l /lvm-snapshot/vollmer/Mail/inbox/,120
results in
ls: /lvm-snapshot/vollmer/Mail/inbox/,120: Permission denied

get some infos via find:
find /lvm-snapshot/vollmer/Mail/inbox/ -name ",120" -print "-A@"
results in:
find: /lvm-snapshot/vollmer/Mail/inbox/,120: Permission denied

The "original file" ~vollmer/Mail/inbox/,120 is ok and readable.

Two weeks ago everything worked fine.

So what's wrong with it? Is my corrupted? (tar --compare seems to complain only at those files).

I use SuSe 8.1, Logical Volume Manager 1.0.5(mp-v6)

With best regards

Jürgen

Dr.rer.nat. Juergen Vollmer, Viktoriastrasse 15, D-76133 Karlsruhe
Tel: +49(721) 9204871 Fax: +49(721) 24874
[email protected],[email protected],[email protected]
www.informatik-vollmer.de

Can you find any nasty messages in your syslog related to the access fault or _before_ you weren't able to access it?

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

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

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

Dear heinz,

"Heinz J . Mauelshagen" <[email protected]>

the only obscure message I found was:

Dec  7 01:22:07 joergli kernel: vs-7000: search_by_entry_key: search_by_key returned item position == 0<4>vs-7000: search_by_entry_key: search_by_key returned item position == 0<4>vs-7000: search_by_entry_key: search_by_key returned item position == 0<4>zam-7001: io error in reiserfs_find_entry

Dec  7 01:22:07 joergli kernel: vs-7000: search_by_entry_key: search_by_key returned item position == 0<4>zam-7001: io error in reiserfs_find_entry

but the failure was around 15:00 on that day

Jürgen

Dr.rer.nat. Juergen Vollmer, Viktoriastrasse 15, D-76133 Karlsruhe
Tel: +49(721) 9204871 Fax: +49(721) 24874
[email protected],[email protected],[email protected]
www.informatik-vollmer.de

Jürgen,

those are Reiserfs errors telling that the filesystem isn't able to search an entry by key. Could well be an io error on the snapshot LV but then we should have a related entry in the syslog.

If you are able to read the snapshot fine (for eg. with
(dd if=/dev/SNAPSHOT of=/dev/null bs=1024k), I assume a reiserfs error.

If reading the snapshot LV fails, let's go forward analyzing LVM, if not there might be a reiserfs fix for that.

Regards,
Heinz -- The LVM Guy --

*** Software bugs are stupid.
Nevertheless it needs not so stupid people to solve them ***

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

Heinz Mauelshagen Sistina Software Inc.
Senior Consultant/Developer Am Sonnenhang 11
56242 Marienrachdorf Germany
[email protected] +49 2626 141200

                                                       FAX 924446

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

"Heinz J . Mauelshagen" <[email protected]>:

Dear Heinz,
A good new year to you!

I've just some spare time :-)

I created the snapshot as usual and let tar access the files:
for i seq 1 20
do
tar -cvf /dev/null /lvm-snapshot
done

after some time tar produces the following:
tar: /lvm-snapshot/vollmer/.exmh/exmh_addrs: can not stat, permission denied
(message translated by me from german)

During that I did some work on the original (reading email and news, running some compile jobs, running some seti@home jobs).

It seems to me, that when I modify or delete a file in the original, those files trigger the problem! (I just modified a Makefile and deleted a file and the next time tar visits that Makefile and the deleted file it was flagged with the above error message);

The problem is reproducible.

The files exists and are accessible in the original.

ls -l /lvm-snapshot/vollmer/.exmh/
ls: /lvm-snapshot/vollmer/.exmh/exmh_addrs.bak: no permission
ls: /lvm-snapshot/vollmer/.exmh/exmh_addrs: no permission insgesamt 34

-rw-r--r--    1 vollmer  users       11282 2003-01-01 20:03 exmh-defaults
-rw-r--r--    1 vollmer  users       11115 2002-09-25 20:25 exmh-defaults~
drwxr-x---    2 vollmer  users        1872 2002-07-06 10:18 exmh-images/
-rw-r--r--    1 vollmer  users          65 2001-01-19 11:09 exmh-shell-history
-rw-r-----    1 vollmer  users        1564 1999-10-27 23:45 exmhsedit

The files seem to be in the directory catalog, but not in the filesystem anymore.

Here the original:
ls -l /home/vollmer/.exmh/
insgesamt 466

-rw-r--r--    1 vollmer  users       11282 2003-01-01 20:03 exmh-defaults
-rw-r--r--    1 vollmer  users       11115 2002-09-25 20:25 exmh-defaults~
drwxr-x---    2 vollmer  users        1872 2002-07-06 10:18 exmh-images/
-rw-r--r--    1 vollmer  users          65 2001-01-19 11:09 exmh-shell-history
-rw-r--r--    1 vollmer  users      220115 2003-01-04 17:20 exmh_addrs
-rw-r--r--    1 vollmer  users      220115 2003-01-04 17:14 exmh_addrs.bak
-rw-r-----    1 vollmer  users        1564 1999-10-27 23:45 exmhsedit

Currently I use a SuSe 8.1 system with the actual (i.e. patch-40) kernel from ftp.suse.com/pub/people/mantel/next.

I use an SMP System (2 x XENON 2GHz, 1GB RAM) with 4 SCSI devices. Three of them under control of LVM. As filesystem I use reiserfs

ReiserFS version:

$ mkreiserfs -h
mkreiserfs 3.6.2 (2002)

LVM version:
$lvcreate --version
lvcreate: Logical Volume Manager 1.0.5(mp-v6) Heinz Mauelshagen, Sistina Software 15/07/2002 (IOP 10)

Do you need more information, how can I help to find the problem.

With best regards

Jürgen

Dr.rer.nat. Juergen Vollmer, Viktoriastrasse 15, D-76133 Karlsruhe
Tel: +49(721) 9204871 Fax: +49(721) 24874
[email protected],[email protected],[email protected]
www.informatik-vollmer.de