Data loss (corruption with rsync)

[kordex -]

I forgot to paste the line I used for the real testing but it's here:

navi:/mnt# rsync --verbose --sparse --append-verify --fuzzy --progress
--stats --recursive --times --delete /mnt/miracle/ /mnt/storage/

-Mikko

2009/12/5 kordex - <kordex at gmail.com>:
> Hello,
>
> Yes I ran exclusive testing:
>
> Checking source (/mnt/miracle):
>
> navi:/mnt# fsck.jfs -f /dev/sda1
> fsck.jfs version 1.1.12, 24-Aug-2007
> processing started: 12/5/2009 16.11.38
> The current device is:  /dev/sda1
> Block size in bytes:  4096
> Filesystem size in blocks:  244190000
> **Phase 0 - Replay Journal Log
> **Phase 1 - Check Blocks, Files/Directories, and  Directory Entries
> **Phase 2 - Count links
> **Phase 3 - Duplicate Block Rescan and Directory Connectedness
> **Phase 4 - Report Problems
> **Phase 5 - Check Connectivity
> **Phase 6 - Perform Approved Corrections
> **Phase 7 - Rebuild File/Directory Allocation Maps
> **Phase 8 - Rebuild Disk Allocation Maps
> 976760000 kilobytes total disk space.
>    13986 kilobytes in 2316 directories.
> 766065198 kilobytes in 29268 user files.
>     1532 kilobytes in extended attributes
>   212356 kilobytes reserved for system use.
> 210494900 kilobytes are available for use.
> Filesystem is clean.
> navi:/mnt#
>
> Wiping the target (/mnt/storage) filesystem:
>
> navi:/mnt# mkfs.jfs /dev/mapper/void-helix
> mkfs.jfs version 1.1.12, 24-Aug-2007
> Warning!  All data on device /dev/mapper/void-helix will be lost!
>
> Continue? (Y/N) y
>   \
>
> Format completed successfully.
>
> 976748544 kilobytes total disk space.
>
> navi:/mnt# fsck.jfs -f /dev/mapper/void-helix
> fsck.jfs version 1.1.12, 24-Aug-2007
> processing started: 12/5/2009 16.25.53
> The current device is:  /dev/mapper/void-helix
> Block size in bytes:  4096
> Filesystem size in blocks:  244187136
> **Phase 0 - Replay Journal Log
> **Phase 1 - Check Blocks, Files/Directories, and  Directory Entries
> **Phase 2 - Count links
> **Phase 3 - Duplicate Block Rescan and Directory Connectedness
> **Phase 4 - Report Problems
> **Phase 5 - Check Connectivity
> **Phase 6 - Perform Approved Corrections
> **Phase 7 - Rebuild File/Directory Allocation Maps
> **Phase 8 - Rebuild Disk Allocation Maps
> 976748544 kilobytes total disk space.
>        0 kilobytes in 1 directories.
>        0 kilobytes in 0 user files.
>        0 kilobytes in extended attributes
>   182256 kilobytes reserved for system use.
> 976566288 kilobytes are available for use.
> Filesystem is clean.
>
> navi:/mnt# mount /dev/sda1 /mnt/miracle
> navi:/mnt# mount /dev/mapper/void-helix /mnt/storage/
> navi:/mnt# mount
> /dev/sda1 on /mnt/miracle type jfs (rw)
> /dev/mapper/void-helix on /mnt/storage type jfs (rw)
>
> navi:/mnt# cp "/mnt/miracle/./waffcd/music/Moving Past The
> Boundaries/Negative Format - 07 Momentum.flac" /mnt/storage/
>
> navi:/mnt# cksum "/mnt/miracle/./waffcd/music/Moving Past The
> Boundaries/Negative Format - 07 Momentum.flac"
> 2863733908 34132620 /mnt/miracle/./waffcd/music/Moving Past The
> Boundaries/Negative Format - 07 Momentum.flac
> navi:/mnt# cksum /mnt/storage/Negative\ Format\ -\ 07\ Momentum.flac
> 2863733908 34132620 /mnt/storage/Negative Format - 07 Momentum.flac
>
> navi:/mnt# mkdir /mnt/storage/bar
>
> navi:/mnt# rsync --verbose --sparse --append-verify --fuzzy --progress
> --stats --recursive --times --delete
> "/mnt/miracle/./waffcd/music/Moving Past The Boundaries/Negative
> Format - 07 Momentum.flac" /mnt/storage/bar
> sending incremental file list
> Negative Format - 07 Momentum.flac
>    34132620 100%   23.79MB/s    0:00:01 (xfer#1, to-check=0/1)
>
> Number of files: 1
> Number of files transferred: 1
> Total file size: 34132620 bytes
> Total transferred file size: 34132620 bytes
> Literal data: 34132620 bytes
> Matched data: 0 bytes
> File list size: 49
> File list generation time: 0.001 seconds
> File list transfer time: 0.000 seconds
> Total bytes sent: 34136885
> Total bytes received: 31
>
> sent 34136885 bytes  received 31 bytes  13654766.40 bytes/sec
> total size is 34132620  speedup is 1.00
>
> navi:/mnt# cksum /mnt/storage/bar/Negative\ Format\ -\ 07\ Momentum.flac
> 2863733908 34132620 /mnt/storage/bar/Negative Format - 07 Momentum.flac
>
> Works fine, eh? Well let's see how the real thing goes:
>
> Well I don't want to cksum the whole thing again as I have done it two
> times already, and my slow controller (SiL 3114) and G4 powerpc cpu
> (7447) are bit slowing doing that, so I assume partitial check is ok:
>
> Hitting ^C. Stops here:
>
> music/1998 - Cathedron/Sephiroth - 01 - Wolftribes.flac
>       32768   0%   78.82kB/s    0:06:10  ^C
> rsync error: received SIGINT, SIGTERM, or SIGHUP (code 20) at
> rsync.c(541) [sender=3.0.3]
> rsync: writefd_unbuffered failed to write 97 bytes [generator]: Broken pipe (32)
>
> Checksuming the files I copied:
>
> navi:/mnt/storage# find ./ -type f > ../storage.txt
>
> navi:/mnt/storage# foreach i ( "`cat ../storage.txt`" )
> foreach? cksum "$i" >> ../storage.cksum.txt
> foreach? end
>
> Comparing checksums (the one I canceled might fail ofcourse):
>
> navi:/mnt/storage# cd ..
>
> navi:/mnt# wc -l storage.txt
> 223 storage.txt
>
> navi:/mnt# head -n 223  miracle.cksums.txt > bar
>
> navi:/mnt# diff storage.cksum.txt bar | wc -l
> 182
>
> ...
>
> 213,215c213,215
> < 2249859435 41520924 ./music/1992 - How to Destroy Angels/04 -
> Tectonic Plates.flac
> < 1562355406 26647298 ./music/1992 - How to Destroy Angels/05 - Dismal Orb.flac
> < 4252661771 56535930 ./music/1992 - How to Destroy Angels/06 - How to
> Destroy Angels II.flac
> ---
>> 1965594369 41520924 ./music/1992 - How to Destroy Angels/04 - Tectonic Plates.flac
>> 3734256630 26647298 ./music/1992 - How to Destroy Angels/05 - Dismal Orb.flac
>> 667032915 56535930 ./music/1992 - How to Destroy Angels/06 - How to Destroy Angels II.flac
>
> ...
>
> Few last lines from diff.
>
> Let's try the cp:
>
> navi:/mnt# cp "/mnt/miracle/./music/1992 - How to Destroy Angels/06 -
> How to Destroy Angels II.flac" /mnt/storage/
>
> navi:/mnt# cksum "/mnt/miracle/./music/1992 - How to Destroy Angels/06
> - How to Destroy Angels II.flac"
> 667032915 56535930 /mnt/miracle/./music/1992 - How to Destroy
> Angels/06 - How to Destroy Angels II.flac
>
> navi:/mnt# cksum /mnt/storage/06\ -\ How\ to\ Destroy\ Angels\ II.flac
> 667032915 56535930 /mnt/storage/06 - How to Destroy Angels II.flac
>
> -Mikko
>
>
> 2009/12/5 Matt McCutchen <matt at mattmccutchen.net>:
>> On Fri, 2009-12-04 at 18:21 +0200, kordex - wrote:
>>> I just rsync'd my 1TB partitions formated with jfs filesystem
>>> containing ~720GB data. I saw -3G difference between source and
>>> destination which made me suspect possible dataloss. Also I had
>>> experienced loss of integrity on my previous copies but I did not find
>>> the cause until now.
>>>
>>> As command line I used following:
>>> sh-3.2# rsync --verbose --sparse --append-verify --fuzzy --progress
>>> --stats --recursive --times --delete /mnt/miracle /mnt/storage
>>
>>> As we all know the second field of output from cksum is the size: so
>>> the size matches but the content does not.
>>
>> OK... can you reproduce the problem by copying some of the source files
>> to a different place on the destination filesystem, or on a different
>> filesystem?  What if you use cp instead of rsync?  Have you run fsck on
>> the source and destination filesystems?
>>
>>> This maybe due fuzzy option but append-verify should (?) confirm that
>>> the already existing data matches the appended one (?).
>>
>> Yes, when a file transfer is performed.  If the destination already has
>> a file that passes the quick check with the source file, no file
>> transfer is performed.  In that case, consider --ignore-times to disable
>> the quick check.
>>
>> --
>> Matt
>>
>>
>