[Fwd: Re: xfsdump/restore confusion about ACLs]

[Stephen Lord <lord@sgi.com>]

Looks like we are having some internal email difficulties,
here is another explanation for xfsrestore losing acls.

Steve

-----Forwarded Message-----

From: Tim Shimmin <tes@sgi.com>
To: mkulima@sgi.com
Cc: slinx-xfs@cthulhu.engr.sgi.com
Subject: Re: xfsdump/restore confusion about ACLs
Date: 14 May 2002 11:43:45 +1000

Hi,

Nathan has responded to the dump/acl question to Greg and the linux-xfs
list but our emails to oss.sgi.com are currently bouncing (Arghhh).

John, perhaps you (or someone else) can send the info on to the 
linux-xfs list.

Thanks,
Tim.


On Mon, May 13, 2002 at 06:52:34PM -0400, Greg Freemyer wrote:
> 
> I'm trying to verify that xfsdump/xfsrestore handle ACLs.
> 
> When I do:
> 
>   su    # ie. set effective ID to root.
> 
>   getfacl gaf/samba/Junk/Test1.txt
>   xfsdump -J -s gaf -L dummy-label -f /tmp/xfs.out /home
>   xfsrestore -J -f /tmp/xfs.out restore_temp
>   getfacl restore_temp/gaf/samba/Junk/Test1.txt
> 
> I loose the test acl I put on the test file.  (Output below my signature)
> 
> Am I missing an argument, or is something wrong.  
> I did RTM, but I must have missed something.
> 
> xfsdump and xfsrestore claim to be version 3, 
> so I don't know what version they really are.
This should really be fixed, so that the VERSION info is
output by xfsdump and xfsrestore.
Will be on the todo list.

> 
> I'm using a pure SuSE 8.0 release of XFS and the utilities.
>
The problem reported above is likely to be a bug in the xfs-kernel
that was fixed around 16/April: 
http://marc.theaimsgroup.com/?l=linux-xfs&m=101893920409629&w=2

Nathan replied thus:
    Date: Tue, 14 May 2002 09:37:09 +1000
    To: Greg Freemyer <freemyer@NorcrossGroup.com>
    Cc: linux-xfs@oss.sgi.com
    User-Agent: Mutt/1.2.5i
    From: Nathan Scott <nathans@sgi.com>
    Subject: Re: xfsdump/restore confusion about ACLs

    Try a CVS kernel, there has been fixes in this area since 
    the code the Suse folks had at the time they released 8.0.

    Not sure what the answer to the second question is, there
    must surely be a cleaner solution though. ;)

    cheers.


> 
> Also, in my experiments, I had several occasions where I had an 
> interrupted restore I wanted to delete.
> 
> I just deleted /var/lib/xfsdump/inventory/*.   
> Is there a better way to resolve this complaint.
For an interrupted restore and cumulative restore, state about
the restore session is stored within the xfsrestorehousekeepingdir
within the destination directory.
So I'd normally rm the destdir/xfsrestorehousekeepingdir if
one wanted to cleanup the interrupted state.

--Tim