samba - Jan 2007 - Charset problems, for the first time

Hi list!

I've been using Samba (as packaged in Debian) for years without any
charset conversion issus, but now one bit me:

Server:
Samba 3.0.23d-2 (Debian testing),
system locale en_US.UTF-8, no charset options in smb.conf specified

Client-DE: MS Windows XP Pro SP2 German, locale for non-Unicode apps
set to German / Austria

Client-JP: MS Windows XP Pro SP2 German, locale for non-Unicode apps
set to Japanese

There's a few clients of the former type, the single client of the
latter type is a recent addition, needed for a few Japanese legacy
apps. Now the Samba server and this last machine obviously do not like
each other:

- Client-JP can't open anything on server that is extended latin-1:
trying to open "?bung 1" yields "Ubung 1 - file not found.

- file names display correctly but somehow are coded differently:
creating a folder titled "???" via Windows Explorer works, and the
folder displays correctly in a shell on the server, but can't be
deleted via shell because it isn't found. If I create the ??? folder
via command line it doesn't show up on the samba share at all, but can
be removed again from command line.

- interestingly, Japanese files work fine on Client-JP

- Clients-DE have none of the above problems (except the fact that my
legacy apps won't run)

All this leads me to believe that samba and Windows do not actually
use (the right?) Unicode variant to exchange file names. Quite
possibly this is Windows' fault but I'd appreciate some help
nonetheless :)

Regards,

Christian

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 01/09/2007 09:05 AM, Christian Pernegger escreveu:
> Hi list!
	Hi Christian!

> I've been using Samba (as packaged in Debian) for years without any
> charset conversion issus, but now one bit me:
> 
> Server:
> Samba 3.0.23d-2 (Debian testing),
> system locale en_US.UTF-8, no charset options in smb.conf specified
	I'm using Debian "etch", Samba 3.0.23d with a few locales
installed, including pt_BR and en_US, both with latin1 and utf8
versions.

> Client-DE: MS Windows XP Pro SP2 German, locale for non-Unicode apps
> set to German / Austria
> 
> Client-JP: MS Windows XP Pro SP2 German, locale for non-Unicode apps
> set to Japanese
> 
> There's a few clients of the former type, the single client of the
> latter type is a recent addition, needed for a few Japanese legacy
> apps. Now the Samba server and this last machine obviously do not like
> each other:
> 
> - Client-JP can't open anything on server that is extended latin-1:
> trying to open "?bung 1" yields "Ubung 1 - file not found.
> 
> - file names display correctly but somehow are coded differently:
> creating a folder titled "???" via Windows Explorer works, and
the
> folder displays correctly in a shell on the server, but can't be
> deleted via shell because it isn't found. If I create the ??? folder
> via command line it doesn't show up on the samba share at all, but can
> be removed again from command line.
> 
> - interestingly, Japanese files work fine on Client-JP
> 
> - Clients-DE have none of the above problems (except the fact that my
> legacy apps won't run)
> 
> All this leads me to believe that samba and Windows do not actually
> use (the right?) Unicode variant to exchange file names. Quite
> possibly this is Windows' fault but I'd appreciate some help
> nonetheless :)
	I migrate from Debian "sarge" (Samba 3.0.14a) to Debian
"etch" (Samba 3.0.23d) without big problems, except that after
the rsync from the old server to the new one, we got strange
characters.

	I solve that using convmv, Samba3 uses UTF-8 and tries
to figure out by itself the best way to handle the charset
stuff between server and clients, after converting a couple of
files just to test it, we did it in something like 0.5TB and
everything works without any problems, either by console,
MS Windows or ssh.

> Regards,
> Christian
	I hope this helps, kind regards.

- --
Felipe Augusto van de Wiel <felipe@paranacidade.org.br>
Coordenadoria de Tecnologia da Informa??o (CTI) - SEDU/PARANACIDADE
http://www.paranacidade.org.br/           Phone: (+55 41 3350 3300)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Debian - http://enigmail.mozdev.org

iD8DBQFFo7XUCj65ZxU4gPQRAtWWAJ0af+kLjt0u7u32G6SZ2wxi55WWzgCfao7i
VckqpfbdPmOYx1pMnBIhGg0=xsDv
-----END PGP SIGNATURE-----