Strange error from dump/restore since upgrading to squeeze
Tim Woodall
backup /var
I'm using LVM on a md raid1. I take an LVM snapshot, backup, mount the
snapshot read only and then verify and the news spool overview IDX files
are failing.
I've rebooted and forced an fsck and that hasn't helped.
I've deleted all the overview and rebuilt it and that hasn't helped
either.
Here's snippets from a test run I've just done (I did this with inn
stopped just in case that was the problem)
einstein:/tmp# /sbin/lvcreate -A n -L500M -s -nvar-backup /dev/vg0/var
WARNING: This metadata update is NOT backed up
WARNING: This metadata update is NOT backed up
Logical volume "var-backup" created
einstein:/var/log/news# fsck -f /dev/vg0/var-backup
fsck from util-linux-ng 2.17.2
e2fsck 1.41.12 (17-May-2010)
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
/dev/mapper/vg0-var--backup: 66520/256000 files (2.0% non-contiguous),
212183/512000 blocks
einstein:/tmp# /sbin/dump -z9 -9 -f test.dmp /dev/vg0/var-backup
DUMP: Date of this level 9 dump: Tue Feb 15 21:59:46 2011
DUMP: Date of last level 1 dump: Tue Feb 15 02:32:34 2011
...
DUMP: Wrote 280060kB uncompressed, 71722kB compressed, 3.905:1
DUMP: DUMP IS DONE
einstein:/tmp# mkdir x
einstein:/tmp# mount -o ro /dev/vg0/var-backup /tmp/x
einstein:/tmp# restore -C -D /tmp/x -f test.dmp
Dump tape is compressed.
Dump date: Tue Feb 15 21:59:46 2011
Dumped from: Tue Feb 15 02:32:34 2011
Level 9 dump of an unlisted file system on
einstein.home.woodall.me.uk:/dev/vg0/var-backup
Label: none
filesys = /tmp/x
./spool/news/overview/c/l/c/comp.lang.c.IDX: tape and disk copies are
different
Missing blocks at the end of
./spool/news/overview/c/l/c/comp.lang.c.IDX, assuming hole
restore: ./spool/news/overview/c/l/c/comp.lang.c.IDX: ftruncate: Invalid
argument
...
restore: ./spool/news/overview/c/t/t/comp.text.tex.IDX: ftruncate:
Invalid argument
Some files were modified! 30 compare errors
einstein:/tmp# umount x
einstein:/tmp# lvremove -A n -f /dev/vg0/var-backup
WARNING: This metadata update is NOT backed up
Logical volume "var-backup" successfully removed
einstein:/tmp#
Any ideas what could be wrong? Note that there does seem to be something
wrong with the backup:
einstein:/tmp# restore -i -f test.dmp
Dump tape is compressed.
restore > cd spool/news/overview/u/f/
restore > add uk.finance.IDX
restore > extract
You have not read any volumes yet.
Unless you know which volume your file(s) are on you should start
with the last volume and work towards the first.
Specify next volume # (none if no more volumes): 1
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of <name unknown>, assuming hole
restore: <name unknown>: ftruncate: Invalid argument
Missing blocks at the end of ./spool/news/overview/u/f/uk.finance.IDX,
assuming hole
set owner/mode for '.'? [yn] n
restore > quit
einstein:/tmp# ls -l spool/news/overview/u/f/uk.finance.IDX
-rw-rw-r-- 1 news news 14292 Feb 15 19:41
spool/news/overview/u/f/uk.finance.IDX
einstein:/tmp# ls -l /var/spool/news/overview/u/f/uk.finance.IDX
-rw-rw-r-- 1 news news 14292 Feb 15 19:41
/var/spool/news/overview/u/f/uk.finance.IDX
einstein:/tmp# diff spool/news/overview/u/f/uk.finance.IDX
/var/spool/news/overview/u/f/uk.finance.IDX
Binary files spool/news/overview/u/f/uk.finance.IDX and
/var/spool/news/overview/u/f/uk.finance.IDX differ
einstein:/tmp#
Tim.
--
God said, "div D = rho, div B = 0, curl E = - @B/@t, curl H = J + @D/@t,"
and there was light.
Tim Woodall
Tim Woodall <dev...@woodall.me.uk> wrote:
> Since I upgraded to squeeze, I've been getting a strange error when I
> backup /var
>
> I'm using LVM on a md raid1. I take an LVM snapshot, backup, mount the
> snapshot read only and then verify and the news spool overview IDX files
> are failing.
>
> I've rebooted and forced an fsck and that hasn't helped.
>
> I've deleted all the overview and rebuilt it and that hasn't helped
> either.
>
> Here's snippets from a test run I've just done (I did this with inn
> stopped just in case that was the problem)
>
> ./spool/news/overview/c/l/c/comp.lang.c.IDX, assuming hole
>
This is a bug in dump where it fails to correctly dump some files with
leading zeros.
http://sourceforge.net/tracker/?func=detail&aid=3133762&group_id=1306&atid=101306
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=614068
It was reported upstream at the end of December. I've tracked it down to
one particular change. I've also now logged a bug on the debian
bugtracker.