[Samba] client code page and valid chars

[Stefan G. Weichinger]

Hello, Christoph,

Freitag, 05. Dezember 2003, 13:58 you wrote:

CL> this would be the hard way I think ... I am afraid of this, because I
CL> don't know where our users got the old "corrupted" file names stored in.
CL> Maybe in CVS-Repositories, network drives with synchronisation, etc. So
CL> renaming the files without interaction of the files owner is not an
CL> option for us.

CL> I read the manual page of convmv and discovered: "As a result of that
CL> the files which contain non-ASCII characters are screwed up if you
CL> ``ls'' them on the Unix server. If you change the ``character set''
CL> variable afterwards to iso8859-1, newly created files are okay, but the
CL> old files are still screwed up in the Windows encoding."

CL> It would help me very much if I could see the old file names as "screwed
CL> up". But I cannot see the at all if I change the charset configuration.

Where? On the Samba-host (in the shell) or on the win-clients?
You should definitely see them in the shell as they should still BE
there.

---

Something else:

Depends on the size of data, weŽre talking about.

A pretty simple but somewhat brute-force approach would be:

- BACKUP YOUR DATA.
- sync the data to another host (samba or win)
- shut down samba
- DELETE stuff (I know, that letŽs your heart beat faster)
- adjust smb.conf

OR EVEN BETTER

- upgrade to Samba-3, then adjust smb.conf (remove that charset stuff,
  hello Unicode !)
- restart Samba
- sync data back to your host.
- enjoy.


I admit that this works well for letŽs say <= 100 Gb when you have
appropriate temp space for the sync. This might be unusable for big
servers.

At least it worked out for me that way in a pretty small installation
of about 20Gb.

Just to let you know.

-- 
best regards,
Stefan G. Weichinger
mailto:monitor at oops.co.at