On Sat, Dec 17, 2005 at 12:39:44PM -0600, Michael S. Zick wrote:
> Group,
>
> I find differences between lsattr and showattr outputs
> which the script marks as errors.
>
> Background:
> Linux 2.6.14 on x86 with vserver 2.0.1 applied.
> util-vserver 0.30.209
> ReiserFS 3.6.14
> Distribution layout is basicly Debian Sarge
>
> Note 1: script expects a mkfs.reiserfs so I faked it with:
> ln -s /usr/sbin/mkreiserfs /sbin/mkfs.reiserfs
the mkfs.<fs> linkx/tools are somewhat standartized
so there _should_ be an mkfs.<fs> for every supported
filesystem (doesn't mean debian has it :)
> Note 2: script calls lsattr (part of e2fsprogs) which
> means that versions prior to 1.37 will silently fail.
ahem, why?
rpm -qf /usr/bin/lsattr
e2fsprogs-1.32-2mdk
and that one works pretty fine too :)
> I built e2fsprogs 1.38 to get past that one.
>
> Here is what I am seeing, all that I can tell is that
> the attributes reported are different:
>
> Without the (-o attrs) option to mount command
(for reiserfs, which absolutely _requires_ attrs to
work with attributes :)
> [mount /dev/loop0 /mnt]
>
> # showattr /mnt
> ---bui- /mnt
> ----ui- /mnt/file_9079
> ---bui- /mnt/dir_9079
yep, means the attributes were ignored, that's what
reiserfs _wiithout_ attrs does ...
> # ./lsattr /mnt/
> ./lsattr: Inappropriate ioctl for device While reading flags on /mnt/file_9079
> ./lsattr: Inappropriate ioctl for device While reading flags on /mnt/dir_9079
> (Which seems reasonable)
yeah, no attributes no ioctl :)
> Note 3: Some of the on-line documentation mentions an "xattrs" option to
> reiserfs - my 3.6.14 only understands "attrs" (correct in the script).
>
> Next, umount and mount the test file with the "attrs" option
> [mount -o attrs /dev/loop0 /mnt]
>
> Now we have (without any changes to the files)
>
> # showattr /mnt
> ---bui- /mnt
> ----UI- /mnt/file_9079
> ---Bui- /mnt/dir_9079
yes, attributes are now available ...
(iunlink and barrier is set)
> (Notice the changes of case with the addition of "attrs"
> that is going to bite somebody running Reiserfs. )
>
> # ./lsattr /mnt/
> ----i-------- /mnt/file_9079
> ------------- /mnt/dir_9079
again ioctl and attributes are there ...
> Which translates as:
> # ./lsattr -l /mnt/
> /mnt/file_9079 Immutable
> /mnt/dir_9079 ---
>
> The only conclusion I can reach is either
> something is broke or incomplete in lsattr
humm, why?
> or Perhaps the script should not be expecting
> that the output of lsattr on a reiser3 filesystem
> to make any sense.
again, why? for me it looks pretty fine?
please elaborate what am I missing ...
TIA,
Herbert
> PS: anyone tried this on reiser4?
PS: did it get into to the kernel yet?
> Mike
> _______________________________________________
> Vserver mailing list
> Vserver@list.linux-vserver.org
> http://list.linux-vserver.org/mailman/listinfo/vserver
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Mon Dec 19 03:39:47 2005